Superior Room-Temperature Ammonia Sensing Using a Hydrothermally Synthesized MoS2/SnO2 Composite.

Layered two-dimensional transition metal dichalcogenides, due to their semiconducting nature and large surface-to-volume ratio, have created their own niche in the field of gas sensing. Their large recovery time and accompanied incomplete recovery result in inferior sensing properties. Here, we report a composite-based strategy to overcome these issues. In this study, we report a facile double-step synthesis of a MoS2/SnO2 composite and its successful use as a superior room-temperature ammonia sensor. Contrary to the pristine nanosheet-based sensors, the devices made using the composite display superior gas sensing characteristics with faster response. Specifically, at room temperature (30° C), the composite-based sensor exhibited excellent sensitivity (10%) at an ammonia concentration down to 0.4 ppm along with the response and recovery times of 2 and 10 s, respectively. Moreover, the device also exhibited long-term durability, reproducibility, and selectivity toward ammonia against hydrogen sulfide, methanol, ethanol, benzene, acetone, and formaldehyde. Sensor devices made on quartz and alumina substrates with different roughnesses have yielded almost an identical response, except for slight variations in response and recovery transients. Further, to shed light on the underlying adsorption energetics and selectivity, density functional theory simulations were employed. The improved response and enhanced selectivity of the composite were explicitly discussed in terms of adsorption energy. Lowdin charge analysis was performed to understand the charge transfer mechanism between NH3, H2S, CH3OH, HCHO, and the underlying MoS2/SnO2 composite surface. The long-term durability of the sensor was evident from the stable response curves even after 2 months. These results indicate that hydrothermally synthesized MoS2/SnO2 composite-based gas sensors can be used as a promising sensing material for monitoring ammonia gas in real fields.

Introduction

Our environment consists of various gases that are naturally available, and some of these are generated by human activities on earth. Among these, ammonia (NH3) is the most prominent one, which is used in the agricultural sector and other research-related activities. This has resulted in significant growth in the production of NH3 and concerns related to human health have also grown.[1−3] An increase in the ammonia level beyond the naturally available ammonia level (sub-ppb range) may result in various health-related issues among living beings. As a result, the Occupational Safety and Health Administration (OSHA) has fixed the exposure time limit (8 h) and the maximum ammonia concentration (25 ppm) at workplaces.[4−6] Irrespective of the workplace or industrial plants, accurate monitoring of the NH3 level is essential to mitigate accidental threats. Consequently, real-time monitoring of ammonia demands sensitive, faster, and selective ammonia sensors that can operate under ambient conditions.[7−11]

In the past decade, various gas sensing strategies utilizing metal oxides, conducting polymers, and composite based on carbon nanotubes and graphene have been used extensively for monitoring various hazardous gases and volatile organic compounds.[12−16] Among these, metal-oxides (SnO2, WO3, ZnO, etc.), due to their larger sensing response, ease of fabrication, and faster response-recovery features, have gained significant attention. Despite all of these advantages, their high-temperature operation and inferior selectivity behavior have limited their use for commercial applications due to the risk of ignition.[15,17−19] Usually, a lower operating temperature ensures reduced power consumption and enables expansion of the application range of gas sensing devices. In an attempt to lower the operating temperature, the composites of tin-oxide (SnO2) with carbon nanotubes (CNTs) and graphene have been used. Gas sensing operation in these cases depends upon electron transfer between the metal oxide and carbon nanostructures. But, electron-transfer-induced perturbations due to charge transfer have limited their gas sensing response.[20−24]

Recently, transition metal dichalcogenides (TMDCs), such as MoS2, MoSe2, WS2, etc., have attracted significant attention due to their two-dimensional (2D) layered structures.[25−27] They possess layer-dependent properties and are semiconducting in nature. Their semiconducting nature together with the large surface-to-volume ratio and most importantly room-temperature working conditions make them attractive for gas sensing applications. Indeed, 2D MoS2 and MoSe2 nanosheets have been used for the detection of various gases, for instance, H2S, NO, NO2, and NH3.[16,28−34] Yuan et al. have shown a highly responsive MoS2 nanosheet-based formic acid sensor.[35] Previous results indicate that the sensing features of MoS2 nanosheets are strongly influenced by ambient oxygen.[28] Cui et al. have used a composite-based strategy to overcome this problem and successfully shown that stable NO2 gas sensing can be achieved at room temperature.[28] In one of our recent reports, significant enhancement in response and improved selective behavior using a MoS2/multiwalled carbon nanotube (MWCNT) composite were shown.[16] In a rather different approach, the sensitivity and selectivity of oxide-based sensors were found to be strongly dependent on the choice of dopants (Pt, Pd, and Rh) used in synthesis.[36−38] While platinum and rhenium are an expensive choice as dopants, palladium-based precursors on the other hand are highly toxic and carcinogenic in nature. Therefore, a composite-based strategy is useful to address these challenges by opening various avenues to optimize material parameters. A composite maintains the unique features of the individual material and gives rise to novel properties arising from the interaction between the two materials.[28,39−41]

Relying on these significant results, here, we propose to explore the properties of a composite made from MoS2 and SnO2. As mentioned earlier, MoS2 is a semiconductor and tin-oxide (SnO2) is an n-type semiconductor metal oxide with a wide band gap of 3.6 eV at room temperature. The latter has been used in various fields, for instance, in lithium-ion batteries, supercapacitors, and gas sensors.[42−44] Its encouraging characteristics as a gas sensing material have enabled its use in monitoring NH3 but at elevated temperatures.[45] Very recently, two other reports employing MoS2/SnO2 composites for room-temperature ammonia sensing have appeared.[38,39] In one of our reports, we have also reported ammonia detection down to 50 ppm with response and recovery times of 35 and 15 s, respectively.[38] In another work, Wang et al. have shown that the same composite displays an improved relative response with faster response (23 s) and recovery (1.6 s) times.[39] In a rather different study, Zhang et al. have shown the use of MoS2-modified SnO2 hybrid for humidity sensing.[46] However, it should be noted that in these three cases different precursors were used for obtaining a composite and different morphology of the obtained sample may give rise to different sensing behaviors. Among these, the last report focuses on humidity sensing, whereas the former two deal with ammonia sensing and lack in the understanding of the adsorption kinetics. Therefore, enough room exists for further exploration of gas sensing properties of these composites.

Herein, we report significant advancements over previously reported results. The first part of this article deals with the synthesis and structural aspects of the MoS2/SnO2 composite. Here, we present a successful low-thermal-budget synthesis of the composite via a two-step hydrothermal method. The second major part of the article is devoted to the electrical and gas sensing properties of two-terminal devices made from MoS2, SnO2, and their composite on alumina and quartz substrates. These results show significant improvement as compared to previously reported data. In short, we report ammonia gas sensing down to 0.4 ppm with a relative response of 10% accompanied by faster response and recovery times (2 and 10 s, respectively). Further, to support the experimental data, we also present density functional theory (DFT) simulations to understand the adsorption kinetics of various gas molecules and subsequent changes in the electronic structure of the composite. The DFT simulations were found to be helpful in understanding the observed sensing and selectivity behavior of the MoS2/SnO2 composite.

Experimental Section

Materials, Methods, and Chemicals

A two-step hydrothermal synthesis method was used for preparing the MoS2/SnO2 (1:1 by weight) composite. In a typical synthesis procedure, ammonium tetrathiomolybdate ((NH4)2MoS4), tin-chloride pentahydrate (SnCl4·5H2O), and sodium hydroxide (NaOH) were used as precursors. All precursors including Nafion solution were acquired from Sigma-Aldrich, and sulfuric acid (H2SO4) was purchased from Loba Chemicals India. Initially, 1 g of ammonium tetrathiomolybdate was dissolved in 50 mL of distilled water. This was followed by the addition of 10 mL of hydrazine hydrate to the mixture obtained in the previous step. The obtained mixture was bath-sonicated for an hour and transferred to a 100 mL autoclave. The autoclave was kept in a furnace at 180° C for 10 h and allowed to cool naturally. After cooling, the black precipitates were filtered and washed several times with distilled water. Finally, the obtained MoS2 powder was dried in a vacuum oven at 60° C for 6 h. In the second step, 0.5 g of the synthesized MoS2 powder was mixed with 50 mL of distilled water. To this solution, we added 0.2 g of SnCl4·5H2O and performed bath sonication for another 2 h. The pH value of the obtained solution was adjusted to 7 via addition of 0.15 g of NaOH. The obtained solution was again transferred into an autoclave, which was kept at 180° C for 16 h. The obtained black precipitates were repeatedly washed with distilled water and dried at 60° C for 2 h in a vacuum oven. The powder so obtained was used for structural characterization and fabrication of two-terminal devices for gas sensing measurements. A schematic illustration of various synthesis steps is shown in Figure . Further, a nebulizer was used for creating variable relative humidity (RH) levels for performing gas sensing measurements. The RH level was monitored using a digital hygrometer placed inside the measurement chamber.

Figure 1

Schematic representation of the two-step synthesis process used for obtaining the MoS2/SnO2 composite. In the first step, MoS2 is obtained via a hydrothermal method. The second step involves another hydrothermal synthesis routine, where MoS2 obtained in the previous step is mixed with precursors to obtain the MoS2/SnO2 composite.

Material Characterization and Sensor Fabrication

The structural and morphological characteristics of the MoS2/SnO2 composite were investigated using an X-ray diffractometer (XRD), a field emission scanning electron microscope (FESEM, Supra 55, Carl Zeiss, Germany), and a transmission electron microscope (JEOL JEM-2100, Japan). X-ray diffraction was performed using a Bruker D8 Focus diffractometer with angle 2θ ranging from 10 to 80° using Cu Kα radiation (λ = 0.154 nm). For FESEM analysis, SnO2 and MoS2–SnO2 powder were transferred on a carbon tape fixed on an aluminum stub and used for further analysis inside the FESEM chamber. High-resolution transmission electron microscopy (HRTEM) imaging was performed to explore the structural properties of the MoS2/SnO2 composite. For TEM analysis, a small amount of the composite material was probe-sonicated in distilled water to obtain a suspension. A drop of the suspension was placed on a carbon-coated copper grid and dried under a lamp in air. This grid was used for structural analysis using HRTEM imaging. For evaluating the surface roughness of the substrates, atomic force microscopy (AFM) was performed using a Tosca 400, Anton Paar system. X-ray photoelectron spectroscopy (XPS) was employed to confirm the chemical states of the as-prepared MoS2/SnO2 sample using a Thermo Scientific K-Alpha XPS spectrometer. Brunauer–Emmett–Teller (BET) technique was employed to obtain the specific surface area based on the nitrogen adsorption–desorption isotherms using a Micromeritics (ASAP 2020) surface area and porosity analyzer. An electrochemical workstation (M204 Autolab, the Netherlands) was used to perform Mott–Schottky (MS) measurements.

For electrical measurements, two-terminal sensor devices were fabricated on alumina and quartz substrates with predeposited silver contacts having a separation of ∼3 mm. For fabricating two-terminal devices, a small amount of the dried powder of the MoS2/SnO2 composite was mixed with distilled water to make a paste. The paste was applied on alumina and quartz substrates with the help of a paintbrush. The devices so obtained were dried in a vacuum oven at 60° C for 2 h. These devices were used for performing various electrical and gas sensing measurements. The schematic representation of various steps involved in device fabrication and data collection is given in Figure S1.

Electrical and Gas Sensing Measurements

Two-terminal current–voltage (I–V) characteristics were obtained using a Keithley 2612A SourceMeter. Gas sensing measurements were performed in a homebuilt measurement setup described elsewhere.[16,29] In short, it consists of a 40 L volume test chamber with a sample holder, a temperature oven, and a small gas-mixing fan. To obtain the desired concentration of a particular gas, the corresponding liquid was injected into the furnace inside the test chamber with a syringe. For this purpose, pure liquid methanol, ethanol, benzene, chloroform, hydrogen sulfide, and acetone were vaporized to obtain the corresponding vapors in the test chamber. Aqueous solutions of concentrated formaldehyde (40 wt %) and ammonia (25 wt %) were used[47−50] for preparing the corresponding gas vapors. Based on the content of the liquid injected (say NH3) into the chamber, the concentration of the vapors was calculated using the equation C = (22.4ρTVs/273MV) × 100, where C is the concentration of NH3 (or other gas) in ppm, ρ is the density of liquid ammonia (g/mL), T is the temperature (K), Vs is the volume of liquid ammonia (μL), M is the molecular weight of ammonia (g/mol), and V is the volume of the testing chamber (L).[51] At a fixed concentration of a gas, the influence of relative humidity on device performance was assessed. Over the investigated range of concentration, the variations in relative humidity were within ±0.5%.

Computational Methodology

We employ the state-of-the-art density functional theory (DFT)-based first-principles method, which is implemented in the QUANTUM ESPRESSO (QE) code.[52] Norm-conserving scalar relativistic pseudopotentials with the exchange–correlation functional of Perdew–Burke–Ernzerhof (PBE) type under the generalized gradient approximation (GGA) have been utilized, which replace the ionic potential and account for the electronic effect arising from the valence electrons based on the Martins–Troullier treatment.[53] The ground state of the system was obtained by a proper convergence method using the bisection technique to find the optimized values of the kinetic energy cutoff of the plane wave along with the Monkhorst–Pack momentum grid and the lattice constant. The optimized values of the kinetic energy cutoff were found to be 80 Ry with a uniform Monkhorst–Pack momentum (k) grid of 9 × 9 × 9.[54] Following this, self-consistent calculations were performed by setting a convergence threshold of <10–6 Ry until the forces on each atom were <10–4 Ry/Bohr with cell pressure <0.5 kbar.

The 3 × 3 × 1 supercell of MoS2 and SnO2 pristine monolayers was optimized, following which the heterostructure (HS) was constructed. Upon optimization of the HS, we performed investigations for its interaction with the toxic gases to confirm the experimental trend of interactions. The size of the supercell is sufficient enough since it considers the adatom repulsions and concentration effects while avoiding the adsorbate–adsorbate interaction, which might occur due to the periodic lattice repetition in our calculations. To confirm the experimental trend of interactions, we considered four species of toxic gas molecules: NH3, H2S, CH3OH, and HCHO. The molecular structures of these molecules were also optimized prior to the calculation of their adsorption energies and Lowdin charges. Since the interaction of the toxic gases with the HS is governed by the long-range van der Waals forces (i.e., ∼3 Å), we employ Grimme’s dispersion corrections[55] in our calculations.

The adsorption energies (Eads)[56] of gases with the HS were evaluated using eq , where EMoS is the total energy of the gas and the HS system and EMoS and Egas are the total energies of the HS and the gas systems, respectively, which are obtained from the ground states of the optimized structuresFor the analysis of charge transfer[57] (using the definition as presented in eq ) between the gas molecule and the HS, we calculate the corresponding Lowdin charges. In eq , Qbefore and Qafter are the total charges carried by the gas molecules before and after interaction with the HS, respectively, such that, for Qtransfer > 0, electrons transfer from the gas to HS, and for Qtransfer < 0, electrons transfer from the HS to the gas.

Results and Discussion

Structural Characterizations of the MoS2/SnO2 Composite

Prior to assessing the gas sensing performance of the two-terminal devices, the structural analysis, morphology, and microstructure of the composite were investigated using XRD, FESEM, and HRTEM. X-ray diffractograms for SnO2, MoS2, and their composite are shown in Figure S2a,c, along with standard data in Figure S2b,d. The diffraction pattern in Figure S2a indicates the polycrystalline nature of SnO2, and all peaks have been indexed according to standard data (card: 211250) given in the lower panel. Figure S2c displays the corresponding X-ray diffractograms for MoS2 and its composite with SnO2. In Figure S2c, broad diffraction peaks corresponding to the (002), (101), (102), and (110) planes of hydrothermally synthesized MoS2 (JCPDS: 037-1492) are clearly visible. The upper diffractogram corresponds to the composite and shows additional peaks corresponding to (110) and (211) planes of SnO2. Synthesized SnO2 and its composite with MoS2 were observed under a scanning electron microscope, and the images are displayed in Figure S3. Figure S3a shows the SEM image of as-synthesized SnO2 particles. As is evident from Figure S3a, SnO2 appears as having a spherical shape with a size ranging from submicrons to a few microns. Figure S3b shows the SEM image of the MoS2–SnO2 composite at the same magnification. As evident from Figure S3, one can see the spherical-shaped SnO2 particles lying on MoS2 sheets. Therefore, this figure gives a visual confirmation of the formation of the composite. Further, the elemental composition was revealed by an energy-dispersive X-ray spectrum, shown in Figure S3c. Here, Mo and S elements have a stoichiometry ratio of ≈1:2. It is to be noted that the stoichiometry ratio between Sn and O does not follow the expected ratio of 1:2. The excess oxygen might originate from the adsorbed oxygen on the sample.[28]

To obtain further structural information about the MoS2/SnO2 composite, HRTEM was used to investigate the nanostructure. Figure a displays a low-resolution image of the composite. The highlighted region corresponds to MoS2. Figure b represents the HRTEM image of the same sample and displays clear fringes corresponding to SnO2. Further, HRTEM images in Figure c show the hexagonal symmetry of MoS2. Both insets in Figure c are the digital filtered images of the highlighted area in the image. One can see the hexagonal arrangement of Mo and S atoms and the corresponding fringes with a separation of 3.5 Å belonging to the (300) plane of MoS2. Figure d gives the selected area electron diffraction (SAED) pattern of the composite sample. Figure e shows another HRTEM image of the sample. The left inset obtained after noise filtering indicates two important points: (i) the fringes with a spacing of 3.3 Å correspond to the (110) plane of SnO2 and (ii) a discontinuity in parallel running lines can be seen in the upper part of the left inset. As a result, the separation between parallel running planes changes from 3.3 to 3.5 Å. This change arises from line defects (clearly visible in the inset) in the synthesized sample. (iii) Further, we also note that the typical size of SnO2 nanoparticles in this processed sample ranges from 2 to 5 nm. Figure f is the HRTEM image of the same sample illustrating the simultaneous presence of SnO2 and MoS2. The fringe spacing of 3.3 Å corresponds to the (110) plane in SnO2, whereas a spacing (right inset) of 3.5 Å belongs to the (300) plane in MoS2. Figure f is an enlarged view of the left inset in Figure e. One can see how the presence of line defects causes an increase in the interplanar spacing. Such types of defects might be quite useful for gas sensing applications, as these defect sites act as highly active locations for the adsorption of gas molecules.

Figure 2

(a) Low-resolution image of the MoS2/SnO2 composite. (b) HRTEM image of the composite showing fringes corresponding to SnO2. (c) HRTEM image of hydrothermally synthesized MoS2. (d) Corresponding SAED image of MoS2 showing the polycrystalline nature of the MoS2 sample. (e) HRTEM image of SnO2 nanoparticles. Insets correspond to the highlighted region in the image. (f) Another image showing MoS2 and SnO2 side by side. For detailed information, see the text.

The surface compositions and the elemental chemical states of the MoS2/SnO2 composite were investigated using X-ray photoelectron spectroscopy. Figure S4 displays the complete survey spectra, illustrating the presence of Mo, S, Sn, and O in the MoS2/SnO2 composite. Figure a–d displays the high-resolution spectrum of Mo 3d, S 2p, Sn 3d, and O 1s, respectively. As we can see, most of the Mo signal (Figure a) comes from the peaks positioned at around 229.2 and 232.2 eV matching with Mo4+ 3d5/2 and Mo4+ 3d3/2 of 2H-MoS2, respectively.[58] The peaks at ∼233.2 and ∼235.8 eV are attributed to Mo6+, which might arise due to partial oxidation of MoS2 in air to form MoO3 or MoO4.[58,59] Therefore, the shoulders at the higher energy side are attributed to the formation of oxide due to exposure to air. This is further evident from the presence of a shoulder in the S 2p region at higher binding energies. The deconvolution of S 2p spectra (Figure b) has resulted in four main peaks located at 162.1, 163.4, 163.8, and 169.2 eV. The first two of these have been assigned to S2– 2p3/2 and S2– 2p1/2 of 2H-MoS2, respectively.[58] The latter two, at higher binding energies, further confirm the +6 oxidation state of Mo. Figure c displays the Sn 3d region of the spectra. It shows two symmetrical peaks of SnO2 with binding energies of 487.2 eV (3d5/2) and 495.6 eV (3d3/2). The O 1s spectra in Figure d show two peaks at 530.8 and 532.6 eV. The former corresponds to the O 1s region in SnO2. The latter, i.e., at 532.6 eV, might arise from adsorbed oxygen on the sample surface.[60]

Figure 3

High-resolution XPS spectra and peak positions of MoS2/SnO2 hybrid: (a) Mo 3d, (b) S 2p, (c) Sn and (d) O 1s spectrum of MoS2/SnO2.

A typical gas sensing material with a sufficiently large surface-to-volume ratio provides enough adsorption sites for the gas molecules. Not only the surface area but also the porosity have a great influence on its gas sensing capabilities. To gain insight into the influence of these parameters, the specific surface area and porosity of MoS2 and its composite with SnO2 were evaluated by N2 adsorption–desorption isotherms. Figure S5 presents nitrogen adsorption–desorption curves with a small hysteresis loop for both samples. The latter together with type-IV isotherms indicates that mass transport is dominated by mesopores. Table S1 summarizes the physical parameters obtained from the N2 adsorption–desorption data of these two samples. Approximately two times enhancement in the specific surface area can be seen in the case of the MoS2/SnO2 composite.

Since the roughness of the top surface of a sensing device may influence its sensing properties, we performed roughness characterization of the top surface of two different substrates chosen for fabricating two-terminal devices. Figure S6 (top panel) shows the atomic force microscopy (AFM) image of the surface of the quartz substrate. Further, we also investigated the influence of the type of substrate and hence its roughness on sensing characteristics of the sensor made from the same material. The AFM images of top surfaces of quartz and alumina are shown in Figure S6. In the case of the former, the cross-sectional height profile in Figure S6a reveals a peak-to-peak roughness of about 1 nm, while the lateral variation ranges between 0.1 and 0.15 μm. For an alumina substrate, a similar height profile in Figure S6b reveals a very large (15–20 nm) variation in the peak-to-peak height roughness, and the substrate has a lateral roughness variation of 0.4 μm. This roughness is copied to the bottom surface of the active device layer made from the composite material. To assess the influence of the type of substrate on the sensing performance, two sets of devices were made. The gas sensing characteristics of the prepared devices were tested inside a homebuilt sealed acrylic chamber.

Electrical and Gas Sensing Characteristics of the MoS2/SnO2 Composite

In the beginning, two-terminal current–voltage (I–V) measurements were performed to assess the electrical characteristics of the composites. The detailed measurements are shown in Figure S7. Figure S7a depicts an increase in the current level with an increase in the temperature, and the corresponding negative temperature coefficient α = −9.3 × 10–3 K–1 (Figure S7b) assures us of the semiconducting nature of the MoS2/SnO2 composite. Further, in Figure S7c, we notice that with the increase in the ammonia concentration, the current level increases, implying a decrease in the resistance of the device. Before we discuss the gas sensing behavior of various devices, we define the response and recovery times of a sensor, as shown in Figure S8. The sensor’s response time (tresponse) is defined as the time taken by the sensor to attain a 90% change in resistance w.r.t. base resistance in air. On the other hand, the recovery time (trecovery) is defined as the time taken by the device to recover between the minimum resistance value and the resistance value, which is 10% below the base resistance. Further, gas sensing measurements were performed on devices made from MoS2, SnO2, and their composite.

Three different devices made on alumina substrate were tested, and the results are shown in Figure . To check the influence of a substrate and hence the roughness, one more device from the composite was made on a quartz substrate. As we can see, all four devices display a decrease in resistance when they are exposed to ammonia and recover to the initial resistance value (in air) when ammonia is removed from the measurement chamber. One can clearly see (Figure c,d) that the choice of the substrate (and roughness of the top surface) influences the response and the recovery feature of the sensor device. The top surface roughness of the substrate may play an important role in determining the roughness of the top surface of the active layer of the device, provided the thickness of the active layer is smaller than the roughness level of the substrate. Since, in our case, we have used thick-film active layers, they do not seem to influence the overall response of the device (Figure c). However, response and recovery transients have slight differences, as shown in Figure c,d. These initial measurements indicate that all devices display n-type behavior. To clarify its n-type semiconducting characteristics, Mott–Schottky (MS) plots were measured in a 0.5 M H2SO4 acid as an electrolyte, as shown in Figure S9. As we can see, MoS2, SnO2, and its composite MoS2/SnO2 display an n-type character (positive slope).[32] Therefore, MS plots independently support the observed n-type semiconducting nature in gas sensing measurements. In Figure d, ΔR = |Rb – Rg| gives the absolute change in resistance. Here, Rb is the base resistance of the sensor in air and Rg is its resistance in the presence of gas. It is to be noted that the MoS2-based sensor displays a slow response as compared to the device made from the composite. More detailed measurements are shown in Figure . Figure a,b compares the relative response for various ammonia concentrations ranging from 0.4 to 200 ppm. In Figure c, one can see that no significant change in the relative response is observed for devices made on two different substrates.

Figure 4

Room-temperature sensing response to 10 ppm of ammonia shown by devices (a) Alumina/MoS2/Ag, (b) Alumina/SnO2/Ag, (c) Alumina/MoS2-SnO2/Ag, and (d) Quartz/MoS2-SnO2/Ag.

Figure 5

Relative response (in %) for various ammonia concentrations given in ppm for (a) Alumina/MoS2-SnO2/Ag and (b) Quartz/MoS2-SnO2/Ag sensors. (c) Absolute relative response comparison between composite-based sensors made on different substrates and (d) the selectivity behavior of the Quartz/MoS2-SnO2/Ag device toward various analytes. All measurements are performed at room temperature.

Figure d compares the response/recovery characteristics of a sensor toward 50 ppm of various analytes. The data reveals the highest response of 53% for NH3 along with the second highest of 8% for H2S, which is only 15% of the response for NH3. The response for other analytes was negligible. Therefore, the sensing response for NH3 is much higher than those for other reducing gases, indicating higher selectivity of the composite toward NH3 at ambient temperature. It is to be noted that the sensor displays a slightly positive response for ethanol and methanol, indicating p-type behavior. The oxidizing or reducing characteristics of adsorbed gas molecules depend upon the interaction between the base material and the target gas molecule. Certain exceptions have occurred where a particular analyte displays a reducing character on certain base materials and an oxidizing character on some other materials. For instance, H2O behaves as a reducing agent for n-type metal oxides, whereas it acts as an oxidizing agent in the case of graphene- and MoS2-based sensors.[61]

For a sensor, repeatability is another important concern. The device was tested for three different concentrations and showed stable results (Figure a). Figure b displays the long-term durability of the device. However, it should be noted that with the passage of time, a slight change in the base resistance of the device was observed. Slight changes in the response time are also evident from the data. These changes might arise from environment-assisted changes in the properties of the active device layer. Hence, all of these results collectively indicate that composite-based devices can be used for stable and rugged ammonia sensing down to 0.4 ppm.

Figure 6

(a) Repeatability test of the sensor at three different ammonia concentrations and (b) its long-term durability.

Further, a highly humid environment has a great impact on the response of the sensing devices at room temperature. In this respect, a humidity sensing experiment was performed at an ambient temperature of 30 °C. For this purpose, the relative humidity level inside the measurement chamber was maintained and monitored using a digital hygrometer. The relative response of the sensor for various RH levels (40–90%) is given in Figure S10. As we can see, the relative response decreases with increasing levels of RH, whereas the response/recovery times on the other hand have stable and smaller values. This might be attributed to water-poisoning effects, which greatly reduce the response of the sensor. When the adsorption between H2O and NH3 molecules is competitive and the relative humidity is high (>50%), excess water molecules prevent adsorption between NH3 molecules and the sensing channel, which causes the sensitivity value to decrease.[62,63]

Density Functional Theory Studies

Structural Details

The optimized lattice constants “a” of pristine monolayers of 2H-MoS2 and 2H-SnO2 are 9.58 Å (with a for unit cell ∼3.193 Å, which matches with other theoretical and experimental studies[64]) and 9.25 Å (with a for unit cell ∼3.083 Å, which matches with other theoretical and experimental studies[65]), respectively, with 25.00 Å vacuum along the (001) crystal direction and the angles ∠O–Sn–O and ∠S–Mo–S being 80.92 and 80.50°, respectively. Also, the bond lengths of Sn–O and Mo–S are 2.10 and 2.42 Å, respectively. Following this, the HS that has an optimized lattice constant of 9.60 Å with a vacuum of 26.00 Å in the (001) crystal direction was created, which is slightly excess as compared to a monolayer, to avoid any possible interactions when the gas molecules are introduced on the HS. This HS exhibits a lattice mismatch of ∼3.51% with the distances between the HS being 5.74 Å (Mo–Sn) and 3.28 Å (S–O).

We place the toxic gas molecules, NH3, H2S, CH3OH, and HCHO, on the SnO2 surface of the HS to examine the affinity of the gas molecules toward the HS to complement our experimental observations (Figure ). The distance between gas molecules and the SnO2 surface is tabulated in Table along with the variation in ∠S–Mo–S and ∠O–Sn–O.

Figure 7

(a–d) Side views and (e–h) top views of the toxic gas molecules NH3, H2S, CH3OH, and HCHO, respectively, when placed on the SnO2 surface of the proposed HS.

Table 1

Structural Parameters When the Toxic Gas Molecules NH3, H2S, CH3OH, and HCHO Are Placed on the SnO2 Surface of the HS

gas species	distance gas–HS	distance between HS	angle
NH3	N–Sn = 2.42 Å	Mo–Sn = 6.14 Å	∠S–Mo–S = 80.50°
	N–O = 2.72 Å	S–O = 3.43 Å	∠O–Sn–O = 81.26°
H2S	H′–Sn = 3.56 Å	Mo–Sn = 5.63 Å	∠S–Mo–S = 80.50°
	H″–Sn = 3.59 Å	S–O = 3.32 Å	∠O–Sn–O = 80.89°
	H′–O = 3.74 Å
	H″–O = 2.42 Å
CH3OH	CH3–Sn = 4.04 Å	Mo–Sn = 5.72 Å	∠S–Mo–S = 80.64°
	OH–Sn = 2.85 Å	S–O = 3.52 Å	∠O–Sn–O = 80.97°
	C–O = 2.82 Å
	OH–O = 2.52 Å
H′CH″O	C–Sn = 4.33 Å	Mo–Sn = 5.64 Å	∠S–Mo–S = 80.50°
	O–Sn = 4.45 Å	S–O = 3.36 Å	∠O–Sn–O = 81.00°
	H′–Sn = 3.43 Å
	H″–Sn = 4.39 Å
	C–O = 2.82 Å
	O–O = 3.66 Å

Adsorption Energy and Charge Densities

We investigate the adsorption energy for different species of toxic gas molecules alongside their charge transfer in terms of Lowdin charges with the stannic oxide (SnO2) surface of the proposed HS. From the calculation of the adsorption energy, we observe that the SnO2 surface of the proposed HS shows an affinity toward the NH3 molecule as compared to other molecules. This matches and complements our experimental observations of relative responses. Overall, we observe a selectivity of the SnO2 surface of the HS toward NH3 as compared to other molecules, which exhibit weak interactions.

The charge transfer (tabulated in Table ) is studied quantitatively in terms of the Lowdin charges and qualitatively in terms of the charge density plots presented in Figure . The interaction of gas molecules is mainly governed by the transfer of electrons from the NH3 and CH3OH molecules to the SnO2 surface, whereas in the case of the H2S and HCHO molecules, we find a peculiar behavior. In the case of H2S, there is a partial transfer of electrons between the gas molecules and the SnO2 surface of the HS, i.e., the S atom gains electrons from the SnO2 surface, whereas the two H atoms donate electrons to the SnO2 surface. Similarly, in the case of HCHO, the O atom gains electrons from the SnO2 surface of the HS while the H and C atoms donate electrons.

Table 2

Charge Transfer (Qtransfer in Units of e) between Different Species of Toxic Gas Molecules and the SnO2 Surface of the Proposed HS and the Corresponding Adsorption Energies (Eads in Units of eV)

gas species	atom	Qtransfer (e)	Eads (eV)
NH3	N	0.0540	–0.4354
	H	0.2239
	H	0.2206
	H	0.2206
H2S	S	–0.0401	–0.0939
	H	0.2812
	H	0.2758
CH3OH	C	0.0016	–0.2680
	O	0.0731
	H	0.2172
	H	0.0859
	H	0.2782
	H	0.2949
HCHO	C	0.1014	–0.0789
	H	0.1953
	O	–0.0004
	H	0.2799

Figure 8

Charge density plots for (a, b) NH3 (along with the top scale panel of (c)), (d, e) H2S (along with the bottom scale panel of (c)), (f) CH3OH (along with the top scale panel of (g)), and (h, i) HCHO (along with the bottom scale panel of (g)).

Also, the charge density plots validate the results of the Lowdin charge qualitatively; from Figure a,b,f, we can observe the blue region between NH3, CH3OH gas, and the SnO2 surface of the HS, which indicates the charge accumulation-driven interaction between them. The blue region is more intense in the case of NH3 (Figure a,b) as compared to CH3OH (Figure f) gas; this makes it clear that NH3 exhibits a higher affinity toward the HS. In the case of H2S (Figure d,e) and HCHO (Figure h,i), we observe the green region between the molecule and the SnO2 surface of the HS. This can be attributed to the lower interaction between the absorbate (gas) and the absorbent (surface). Moreover, the green region is more intense in the case of the HCHO gas (Figure h,i) showing minimal interaction and correspondingly lower adsorption over the SnO2 surface of the HS as compared to other species of gas. With this, we conclude that NH3 interacts strongly with the SnO2 surface of the HS as compared to other gases, thus complementing our experimental results. This indicates that the MoS2-SnO2 HS with the SnO2 surface exposed to various toxic gas species is highly sensitive to NH3.

Sensing Mechanism

The experimental and theoretical results included in previous sections confirmed excellent sensing characteristics of MoS2/SnO2 toward NH3 at ambient temperature. DFT simulations have complemented the experimental data with preferential interaction and a larger amount of charge transfer between adsorbed ammonia and the SnO2 surface of the HS as compared with other gas molecules. Various factors may be assigned for such an improved behavior shown by the composite material. This includes enhanced specific surface area for the composite that facilitates adsorption of gas molecules by providing more adsorption sites. Second, as discussed earlier, both MoS2 and SnO2 have separately shown n-type conduction toward NH3. In the composite form, they lead to the formation of an n–n-type heterojunction. The work function of a material represents the energy required to remove an electron from the material surface to vacuum. These intrinsic material parameters have shown a change due to environmental effects. For instance, bare MoS2 has a work function of 4.04 eV in vacuum, whereas O2-adsorbed MoS2 has a value of 4.47 eV.[66] In air, SnO2 possesses a higher work function (4.7 eV) than MoS2.[67][67] When these different materials combine to form a heterostructure, the work function difference causes an electron flow from MoS2 to SnO2 until the Fermi levels in both materials are aligned. This process leads to the formation of a depletion layer at the interface between two different n-type materials. Any event that leads to the modulation of this depletion layer by the charge transfer process influences the overall electrical conduction across such a barrier. In air, this depletion layer gives rise to higher resistance. The air-exposed MoS2/SnO2 surface interacts with ambient oxygen and forms an oxygen ion after taking an electron from SnO2.[68] When this oxygen-adsorbed composite surface is exposed to a reducing gas such as ammonia, the NH3 molecules transfer electrons to adsorbed oxygen. Subsequently, the oxygen releases the electrons back to the metal oxide semiconductor, thereby reducing the depletion layer width, and hence the measured resistance across such junction reduces.

Conclusions

In summary, MoS2/SnO2 composite-based sensors were fabricated and tested for gas sensing at room temperature. Experimental and theoretical investigations have revealed selective and faster detection of ammonia as compared to other analytes. The composite was successfully prepared using a facile bottom-up approach. Structural analysis revealed the presence of various defects that might influence the adsorption energy and gas sensing properties of the composite. The composite-based sensor displayed a nearly 85% larger response to NH3 as compared to H2S and other tested analytes. This was further justified based on density functional theory simulations, which predicted the largest negative adsorption energy of −435 meV (≈41.95 kJ/mol) per ammonia molecule. Further, the Lowdin charge analysis implies a significant amount of charge transfer from the ammonia molecule to the composite. This, together with the Mott–Schottky plots and the experimentally observed reduction in the resistance of the device after exposure to ammonia, confirms the n-type behavior of the composite. Therefore, these results indicate that ammonia is physisorbed on the MoS2/SnO2 composite. A series of gas sensing tests were performed on sensors fabricated on quartz and alumina substrates. An almost identical response except for slight differences in transient times was observed. These slight variations might be due to different roughnesses of the top surface of the substrate. Various other tests such as repeatability and long-term durability confirmed their long-lasting properties. Most importantly, the device displayed detection down to 0.4 ppm with a superior signal-to-noise ratio, indicating that detection can be improved beyond this level by employing serpentine-type electrodes. The observed response and recovery times of 2 and 10 s, respectively, together with room-temperature operation further assure us that no additional recovery mechanism is required to bring the sensor back to the initial stage. Therefore, these results indicate that the MoS2/SnO2 composite can serve as an ideal sensing material for room-temperature ammonia detection.