Highly proton conducting MoS₂/graphene oxide nanocomposite based chemoresistive humidity sensor

Abstract

This paper reports the development of MoS₂/GO nanocomposite based sensing layers for resistive humidity sensors. The MoS₂ nanoflakes were synthesized through liquid exfoliation and GO was synthesized using modified Hummers method. The nanocomposite was drop-cast on a Si/SiO₂ substrate containing aluminium electrodes to fabricate the sensor device. The best performance was shown by the 1 : 4 (MoS₂/GO) composite. Various characterization techniques like Scanning Electron Microscopy (SEM), Transmission Electron Microscopy (TEM), X-Ray Diffraction (XRD), X-ray Photoelectron Spectroscopy (XPS), and Fourier Transform Infrared Spectroscopy (FTIR) were used to verify the composite formation. The sensing response was found to lie between 55 times at 35% RH and 1600 times at 85% RH. Such a high response is believed to be because of proton conductivity in the water layer for both MoS₂ and GO. The sensor performance was found to be repeatable even after three months of the first measurement with quick response and recovery. Thus the authors believe that the excellent sensitivity coupled with low cost synthesis and resistive sensing will make their work useful to develop new generation humidity sensors.

---

Introduction

The necessity of continuous monitoring of humidity in various fields starting from industrial processing, environmental control, manufacturing of highly sophisticated integrated circuits in the semiconductor industry, the automobile industry, and agriculture¹ has led to an increased demand for highly sensitive, low cost, wide range, fast humidity sensors. In a recent survey it has been estimated that the humidity sensor market will reach ∼$ 2.3 billion by 2020.²

Based on transduction techniques, humidity sensors can be classified as resistive,³ capacitive,⁴ optical fibre based,⁵ field effect transistor (FET),⁶ Surface Acoustic Wave (SAW),⁷ Quartz Crystal Microbalance (QCM)⁸ etc. Among these, capacitive humidity sensors are most widely used. In fact, most of the commercially available humidity sensors are based on a capacitive technique where the dielectric constant of a polymer layer changes in the presence of humidity. However, capacitive sensors have their own limitations like hysteresis, stability at high temperature and humidity levels as well as durability against some organic vapours.⁹ Also, such sensors are expensive. On the contrary, resistive humidity sensors have simpler structure and they are easier to integrate with Complementary Metal Oxide Semiconductor (CMOS) technology than capacitive sensors. Hence a lot of research is currently being focussed on development of resistive humidity sensors which is believed to reduce the cost of per unit sensor and also their long time stability will be more.

The most commercially used sensing materials for humidity detection are porous ceramics and polymer films¹⁰ (e.g. Greystone,¹¹ Johnson Controls¹²). The polymer films are unable to operate at high humidity levels, show hysteresis, slow response time and long term drift. Although humidity sensors using metal oxides as sensing layer are also commercially available (e.g. Honeywell humidity sensors¹³), the disadvantage is such sensor requires periodic heat cleaning to recover its humidity sensing properties. The sensors also exhibit a progressive shift of resistance due to gradual formation of chemisorbed hydroxyl ions on the surface of the sensing layer.¹⁴ Again, several groups are involved to develop high performance humidity sensing nanomaterials like silicon nanostructures,¹⁵ ceramic nanomaterials,¹ semiconductor nanoparticles¹⁶ and metal oxide nanowires.^3,14,17–21 There have also been recent reports of graphene and its derivatives (e.g. reduced graphene oxide (RGO) and graphene oxide (GO)) as humidity sensing layers.^3,21–23 GO can act as a good humidity sensor because of the presence of large functional groups, inherent defect sites and formation of large proton conducting paths in presence of humidity. However pure graphene lacks in defect sites and thus its sensitivity towards humidity is low.²⁴ Recently, two dimensional transition metal dichalcogenides like molybdenum disulphide (MoS₂) have received increasing attention for ultrasensitive sensor applications.^25,26 MoS₂ is a layered metal dichalcogenide which consists of hexagonal rings of S–Mo–S atoms held together by weak van der Waals force of interaction. The presence of layered structure and inherent defects make it an ideal candidate as the sensing material for humidity sensors. However, there have been very few reports on MoS₂ based humidity sensors as per the knowledge of the authors.^6,27–29 The MoS₂ based FET in Rao et al.'s paper did not show significant response till a relative humidity (RH) level of 60%.⁶ Zhang et al. showed a response value of only 3 times using their MoS₂ based thin film sensor device.²⁷ Using their capacitive sensor, Zhu et al. observed a sensitivity of around 32.19 nF/%RH at 89.5% RH.²⁸

Thus humidity sensors made of pure MoS₂ suffer from low sensitivity at room temperature as has been reported already. Hence to increase the sensitivity of MoS₂ based humidity sensor, it can be functionalized with other nanomaterials or metal nanoparticles. However, doping with noble metals would increase the per capita cost of the sensor.

In our paper, we are reporting for the first time the development of an excellent humidity sensing layer based on MoS₂/GO nano-composite. The MoS₂ nanoflakes were obtained using liquid exfoliation technique and GO was prepared by the well-known modified Hummers method. The proportion of GO was varied to get optimized performance. The composites were extensively characterized using different material characterization techniques. The response of the sensors was carried out at six different humidity levels between 35% and 85% RH. The optimized sample showed response as high as 1600 times at 85% RH. The optimized sample was further tested for stability, repeatability and hysteresis. The detailed results and the probable sensing mechanism have been explained in this manuscript.

Experimental section

Chemicals required

MoS₂ powder (Sigma Aldrich, <2 μm, 99%) and absolute ethanol were required for the exfoliation of MoS₂ nanoflakes. Graphite powder, NaNO₃, H₂SO₄ (98%), H₂O₂ (30%), KMnO₄ were used to exfoliate graphene oxide from bulk graphite powder. All the chemical reagents were of analytical grade.

MoS₂ synthesis

In our laboratory, MoS₂ nanoflakes were exfoliated from bulk MoS₂ powder using a sonication assisted exfoliation technique. 75 mg of MoS₂ powder (<2 μm, 99%, Alfa Aesar) was added to 45% ethanol water. The solution was then sonicated for 8 hours in a Microclean 102 ultrasonic bath. In order to study the effect of particle size on the sensing response, three other samples with different sonication times of 2 hours, 4 hours and 12 hours were also prepared. After sonication, the dispersion was centrifuged at 4500 rpm to remove the residue/precipitate and the yellowish green supernatant containing the nanoflakes was collected.

Graphene oxide synthesis

GO was synthesized using modified Hummers method.³⁰ For our experiment, 0.5 g of graphite and 0.5 g of NaNO₃ were taken in a round bottomed flask and 25 mL of 98% H₂SO₄ was added to it. Next, 3 g of KMnO₄ was slowly added to the mixture kept in an ice bath with constant stirring. The solution was then transferred to a 40 °C water bath and constantly stirred for about 1 hour to form a thick paste. 50 mL of water was added to the solution which was again stirred for 30 minutes at 90 °C. The reaction mixture was diluted with 100 mL of water. Finally, 3 mL of H₂O₂ (30%) was slowly added to the dark brown solution which turned yellow. The graphene oxide cake was then filtered (vacuum filtration) and washed with 800 mL of water. The visible particles and other water soluble impurities were removed by repeated washing cycles of centrifugation and sonication.

MoS₂/GO composites

The MoS₂ dispersion (8 hours sonicated sample) was mixed with different volumes of GO to get a number of samples with different volume ratios as MoS₂/GO 1 : 0.33, 1 : 0.5, 1 : 1, 1 : 2, 1 : 3, 1 : 4, 1 : 6, 1 : 8. MoS₂/GO sample 1 : 8 means 1 mL of MoS₂ dispersion was mixed with 8 mL of GO and so on. The samples were named respectively as S0–S8 where S0 stands for the sample without any GO content and S1–S8 for the samples with progressive increase of GO proportion. Each mixture was stirred for 20 minutes to ensure proper mixing and uniform dispersion of the flakes. To perform the humidity tests, 0.4 μL of each sample was drop-casted on aluminum electrodes designed and fabricated on Si/SiO₂ substrate using conventional lithography techniques. Fig. 1 shows our chemoresistive humidity sensing device.

Fig. 1 Schematic of our humidity sensing device.

Material characterizations

In order to ensure successful exfoliation of few-layer MoS₂ ​nanoflakes from the bulk MoS₂ powder and graphene oxide from bulk graphite powder, various material characterizations were done using a number of different techniques like SEM, TEM, XRD and FT-IR. The samples were also characterized to ensure proper mixing of MoS₂ and GO (i.e. presence of both the flakes). The Field Emission SEM (FESEM) was performed using SUPRA 40 field emission scanning electron microscope for the surface morphological studies. The samples were prepared by spin coating the dispersions on Si/SiO₂ substrate. The High Resolution TEM (HRTEM) measurements were carried out using JEOL JEM-2100 HRTEM operated at an acceleration voltage of 200 kV. The samples for HRTEM characterization were prepared with a few drops of diluted dispersion of MoS₂/GO on a standard holey carbon covered copper grid. The X-ray diffraction patterns were collected using a Panalytical X'Pert Pro Diffractometer with a conventional X-ray tube (Cu Kα radiation). The XPS characterization was carried out using PHI 5000 Versa Probe II (ULVAC-PHI Inc., Japan). FTIR spectrometry of a few samples were carried out using Fourier transform infrared spectrometer (FT-IR spectrometer), model no: NEXUS-870. The samples were prepared using KBr pellets and FT-IR was carried out in transmittance mode.

Humidity sensing setup

The humidity test set up which was made in-house consisted of a stainless steel chamber to probe the samples, a test fixture AGILENT 16442A, Semiconductor Parametric Analyzer (SPA) Agilent 4155C, gas bubbler and compressed air supply. The nano-composite was initially drop-casted on patterned aluminum electrodes on Si/SiO₂ substrate to fabricate the sensor device. The sensor was then placed on a vero board using a double sided tape. Copper wires were then connected to the two electrodes using silver paste. The sensor samples were probed inside the airtight stainless steel chamber which was connected to the test fixture. Fig. 2 shows the schematic of the gas sensing setup. The chamber where the sample was probed had two gas lines-one for the compressed air to enter the chamber whiles the other for humid air which reached the chamber through a gas bubbler. The relative humidity was set with the help of a commercial humidity sensor (Dolphin Automation, model no. TH-382) capable of measuring both humidity and temperature. The entire test set-up was interfaced to the SPA using the test fixture. The SPA was connected to a PC having Agilent's Desktop Easy EXPERT software, a GUI based characterization software, for efficient and repeatable device characterization in the entire characterization process.

Fig. 2 Schematic of gas sensing setup.

Results and discussion

Material characterizations results

During the humidity sensing process, it was found that MoS₂/GO 1 : 4 sample showed the maximum response. So we have tried to highlight the morphology as well as structural analysis of that sample through our characterizations.

Fig. 3 shows the FESEM images of bulk MoS₂, 8 hours sonicated and exfoliated MoS₂, exfoliated GO and MoS₂/GO 1 : 4 samples respectively in Fig. 3(a), (b), (c) and (d). It is clear from the SEM images (Fig. 3(a)–(b)) that the number of stacked layers of MoS₂ as well as the lateral dimension of the flakes got reduced with sonication. The MoS₂ flake sizes varied from 168 nm to 1.1 μm. Fig. 3(d) shows the co-existence of both MoS₂ flakes and GO sheets, thus showing a composite structure. The MoS₂ layers are not uniform throughout the sample and some parts contain more number of layers than the other. This might have occurred because of restacking of nanosheets upon removal of solvent from the dispersion. The SEM images of various composites of MoS₂/GO are given in the ESI (Fig. S1†). With the increase in GO proportion, it can be seen that the GO films partially or fully covering the MoS₂ flakes or the latter lying on GO film. The formation of composite was further revealed from the EDAX images provided in the ESI (Fig. S2–S4†) which shows successful formation of the composite. In Tables S1 and S2,† the atomic% of Mo and S show that they exist in the atomic ratio of 1 : 2. Again the weight% of Mo and S reduced in MoS₂/GO composite compared to that of pristine MoS₂ confirming the formation of a composite with GO. Also, the presence of high proportion of carbon and oxygen confirms the presence of GO in the composite. The silicon peaks were seen as the sample was spin coated on Si substrate and the peak of gold arose from the gold coating that was required to characterize the sample. The lateral size of the GO flakes was found to be around 1.37 μm. The increase in proportion of GO was evident from the SEM images in ESI† showing a gradual increase of GO layers and decrease of MoS₂ flakes.

Fig. 3 FESEM image of (a) bulk MoS₂ powder (b) exfoliated MoS₂ (c) exfoliated GO (d) MoS₂/GO 1 : 4.

In order to obtain the microstructural information of the samples, the HRTEM measurements were carried out. The TEM images and selected area diffraction (SAED) pattern of MoS₂/GO 1 : 4 are shown in Fig. 4. The characterization reveals that the MoS₂/GO composites deliver a composite structure of layered MoS₂ on GO. From the TEM images (Fig. 4(a) and (b)), it was found that few layered MoS₂ flakes co-exist with partially or fully overlapped layers of GO.

Fig. 4 (a and b) TEM images of MoS2/GO 1 : 4 (c) HRTEM image showing fringes of GO (d) SAED patterns of MoS₂ and GO in the sample.

The SAED pattern (Fig. 4(d)) confirmed the presence of various planes of MoS₂ ((002) and (004)) and GO (001). It also indicated a multiple spot pattern, one of which was due to polycrystallinity of stacked GO sheets and the other one due to MoS₂. The hexagonal spot pattern also confirms the crystallinity as well as hexagonal structure of the exfoliated MoS₂. Moire patterns (Fig. 4(c)) were also observed that arose because of overlapping of different planes of GO. The TEM images clearly reveal a successful chemical doping of MoS₂ flakes by GO.

To study the crystallographic feature of nanoflakes, the diffraction pattern was collected. The HRTEM data was verified with XRD measurement. Fig. 5 shows the comparative XRD pattern of MoS₂/GO 1 : 4 and exfoliated MoS₂ samples with the lines indexed according to ICDD-JCPDS card no. 0037-1492 for MoS₂ and JCPDS card no. 75-1621 for hexagonal carbon. The interlayer spacing d₀₀₂ for MoS₂ has been calculated as 6.15 angstrom which matches with the reported data and it is believed that exfoliation of MoS₂ occurs along the (002) plane.³¹ The diffraction pattern of the samples showed MoS₂ peaks at 2θ = 14.4°, 29.02°, 35.99°, 39.925°, 43.045°, 49.786° and 55.01° corresponding to (002), (004), (102), (103), (006), (105) and (106) planes respectively. Sample S6 (MoS₂/GO 1 : 4) also showed a broad peak at 2θ = 10.4° which corresponds to the (001) plane of GO. The presence of (002) plane and (004) plane of MoS₂ and (001) plane of GO were also confirmed from the SAED pattern of HRTEM images.

Fig. 5 XRD patterns of MoS₂ nanoflakes and MoS₂/GO 1 : 4.

FT-IR spectrometry of exfoliated MoS₂ flakes, GO nanosheets as well as MoS₂/GO 1 : 4 composite are shown in Fig. 6. Since MoS₂ is featureless in the wavenumber range of 4000 cm⁻¹ to 700 cm⁻¹, the characteristic peaks in that range are due to the presence of GO. The presence of oxygen containing groups reveals that graphite has been oxidized. The peaks at 1052 cm⁻¹, 1225 cm⁻¹ and 1415.1 cm⁻¹ correspond to C–O stretching vibration (alkoxy), C–O stretching vibrations (epoxide/ether) and C–OH (carboxyl) respectively. The peak at 1641 cm⁻¹ corresponds to the stretching and bending vibration of OH groups of water molecules absorbed on graphene oxide. The surface hydroxyl groups result in the formation of hydrogen bonds between graphite and water explaining clearly the hydrophilic nature of graphene oxide. The absorbed water in GO is shown by a band between 3226 cm⁻¹ and 3600 cm⁻¹ contributed by OH stretch of water molecules.³² A MoS₂ peak can be observed (red curve) at 473 cm⁻¹ and it can be accounted for the bridging Mo–S bond (Mo–S vibration perpendicular to the basal plane).³³

Fig. 6 FT-IR spectrometry of MoS₂, GO flakes and MoS₂/GO 1 : 4 composite.

The peak at 658 cm⁻¹ can be assigned to S–H vibrations confirming H bonding at S edges (blue curve) and 588 cm⁻¹ to S–S bonding strongly. The wavenumber shift strongly suggests chemical interactions between MoS₂ and GO thus showing a chemical functionalization of MoS₂ by GO.

The XPS characterization of the 2 hours and 8 hours sonicated MoS₂ samples were also done to investigate the presence of S-vacancies in the samples. The details of the characterization have been given in the ESI (Fig. S5†).

Humidity sensing results

The chemoresistive sensor device was first exposed to compressed air for around 30 minutes to achieve a stable baseline resistance. The conductivity of the sensing layer changes in the presence of humidity.

The changes in current through the sensor were measured using a SPA. The response of the sensors was calculated as

I_humidity/I_baseline

where I_baseline is the baseline current when compressed air was passed through the chamber (the commercial sensor used for calibration in our sensor chamber showed a relative humidity of around 25% RH of this compressed air and this was the lowest humidity level possible in our setup) and I_humidity was the current value in presence of higher humidity levels. The entire experiment was carried out at room temperature. The general response of the samples was an increase of current (i.e. decrease in resistance) with increase in humidity.

A number of humidity tests on various samples were carried out which are briefly described as follows.

The response of the MoS₂ based sensor devices (i.e. the sensing layer containing only MoS₂ and no GO) against different humidity levels was studied as shown in Fig. 7. All the four samples with different sonication times were exposed to three different levels of humidity viz. 45% RH, 55% RH and 65% RH, each for 10 minutes followed by compressed air purging for the next 10 minutes. The bar plot shown in Fig. 7(a) clearly shows the response of 8 hours sonicated sample was better than the remaining samples. Hence the response of the 8 hours sonicated sample was plotted separately in Fig. 7(b). The response of the sensor device was found to increase with increase in humidity levels.

Fig. 7 (a) Comparative response values of 2 hours, 4 hours, 8 hours and 12 hours sonicated MoS₂ samples. (b) Response of 8 hours sonicated MoS₂ sample for three different humidity levels.

Although the 8 hours sonicated sample showed the optimised response, however, from Fig. 7(b), it is clear that pristine MoS₂ nanoflakes show poor response (∼10 times for a RH level of 65%) towards humidity. The response of pristine MoS₂ based humidity sensor was found to be much lower than other nanomaterial based humidity sensors.

Thus to enhance the performance of MoS₂ based humidity sensor, we have tried to functionalize MoS₂ nanoflakes with GO.

The GO was mixed with MoS₂ in various proportions (the details have been given in the Experimental section). In order to carry out the humidity tests, each sample was tested against three different humidity levels viz. 45%, 55% and 65% RH as before. This test was repeated for all the nine samples from S0–S8. The response for one particular cycle viz. 65% RH level has been shown in Fig. 8(a). Since the optimized response was observed for MoS₂/GO 1 : 4 composite, the sample was tested against six different humidity levels (35% RH to 85% RH) and the responses have been plotted in Fig. 8(b).

Fig. 8 (a) Comparative plot of response of different samples at 65% RH. (b) Response of MoS₂/GO 1 : 4 sample for 6 different humidity levels.

From the above two figures, it can be seen that the response of the samples increases with increase in humidity levels. The sensor response varied from 3 times (with 35% RH) to 1600 times (85% RH). This is much higher than previously reported results of MoS₂ based humidity sensor devices (∼3 times at 60% RH,²⁷ 32.19 nF/% RH at 89.5% RH²⁸).

The response time (ζ_response) is defined as the time taken by the sensor to reach 90% of the response after the humid air was switched on and recovery time (ζ_recovery) is defined as the time taken to reach 10% of the baseline resistance after the humidity was switched off. The response time and recovery time of sample S6 (MoS₂/GO 1 : 4) for a single cycle response of 65% RH were found to be 43 seconds and 37 seconds respectively which are better than (ζ_response = 90 s, ζ_recovery = 110 s for 71.8% RH cycle²⁸) or comparable to (ζ_response = 9 s and ζ_recovery = 17 s (although response was poorer than our device²⁷))] previously reported MoS₂ based humidity sensors. The results show that we can make a fast highly sensitive humidity sensor using MoS₂/GO composite.

The MoS₂/GO 1 : 4 sample was also tested for reproducibility. It was exposed to three cycles of 65% RH, each for 10 minutes followed by 10 minutes of compressed air purging. The response of the sensor was also found to be repeatable over the three iterative cycles (Fig. 9(a)). So, it can be concluded that the MoS₂/GO based sensor is reusable over a number of times.

Fig. 9 (a) Repeatability of MoS₂/GO 1 : 4 sample at 65% RH. (b) Stability test of MoS₂/GO 1 : 4 sample.

In order to ensure the stability of the sensor, the responses of the sensor device against different concentrations of humidity was measured after 45 days and 90 days. The device shows quite a stable response with a very small degradation in performance (maximum error is 0.143 at 55% RH after 45 days and 0.203 after 90 days) as shown in Fig. 9(b). Thus it can be concluded that the MoS₂/GO based humidity sensor device shows good stability and reliability.

In order to show the hysteresis of the sensor device, the current through the device was measured at various humidity levels (with increasing humidity from 25% to till 85% and then decreasing it back to 25% again) are shown in Fig. 10.

Fig. 10 Hysteresis of MoS₂/GO 1 : 4 sample.

The hysteresis measurements were done by calculating the difference between the logarithmic impedance values during the adsorption and desorption processes for various levels of humidity.³⁴ The results are shown in the table below (Table 1):

Table 1 Hysteresis values at different levels of %RH

%RH	Hysteresis
25	0.13
35	0.29
40	0.21
45	0.28
50	0.24
55	0.12
60	0.26
65	0.34
70	0.27
75	0.3
80	0.05
85	0.03

---

From the table, it can be seen that the hysteresis values are quite small for most of the humidity levels. The hysteresis performance can be improved even further by using heat treatment.³⁵

In order to ensure proper connectivity between the electrodes and the sensing layer, the I–V measurements were done at each humidity level, the plot of which is shown in Fig. 11(a). The response shows a perfect ohmic contact. The response variation with increase in RH level was linear as plotted in Fig. 11(b).

Fig. 11 (a) I–V plots of MoS₂/GO 1 : 4 sample at various humidity levels. (b) Response versus RH plot (Adj. R-square = 0.909).

To get an idea about the variation of response with change in thickness of the sensing layer, four different samples were prepared. The first one was drop-casted with only one drop (0.2 μL) of the MoS₂/GO dispersion and the thickness was measured using Dektak D150 Surface Profilometer. Similarly the next three samples were prepared with two, three and four drops of the dispersion. The thicknesses of the four samples respectively from the first to the fourth sample were 559.8 nm, 656.6 nm, 729.2 nm and 919.8 nm. The response of the samples for three different humidity levels was measured and the results are shown in Fig. 12. It can be seen that the responses increased slightly with increase in thickness of the films.

Fig. 12 Comparative plot of response of MoS₂/GO 1 : 4 samples with different thickness at different humidity levels.

The possible sensing mechanisms of the MoS₂ based humidity sensor as well as the improved response of the MoS₂/GO based sensor devices have been explained in the following section:

Humidity sensing mechanism

i. Sensing mechanism of MoS₂ based humidity sensor. It has been observed in Fig. 7 that the response of the MoS₂ based humidity sensor increased with increase in humidity levels. Two conduction mechanisms take place simultaneously – one is the electron conduction through the MoS₂ sensing channel (water acts as electron donor) and the second is the proton hopping through continuous water layers by the well-known Grotthuss mechanism.³⁶

It has been reported elsewhere that charge transfer between MoS₂ molecules and water molecules enhances the conductivity of MoS₂ sensing layer in presence of humidity.²⁷ The water vapour molecules get adsorbed on the sensing material and serve as electron donors. Thus the conductivity of the inherently n-type³⁷ semiconducting MoS₂ sensing layer increases in presence of humidity.

Besides, the surface area of MoS₂ nanoflakes increased after exfoliation from its bulk powder enhancing the sensing response. During sonication, the ultrasound generates high energy in the solution and increases the local temperature and pressure leading to the exfoliation of MoS₂. This along with the micro-jet effect leads to a reduction of flake size in MoS₂.²⁷

For each S–Mo–S layer, the atoms are divided into basal plane sites terminated by sulphur atoms and edge sites which are terminated by Mo or S atoms with missing coordination bonds.²⁷ Studies have shown that S-vacancies are the most stable defect sites in MoS₂.³⁸ The presence of S-vacancy sites (such defects increases with increase in sonication time, predominantly presence at the flakes edges) in the sample was confirmed by the XPS results (details given in the ESI†) and they were found to play a significant role in the sensing process.

For lower concentration of humidity, charge transfer from water to the MoS₂ sensing layer dominates and this helped in increasing the conductivity at lower levels of humidity. The water molecules get chemisorbed on the surface of the sensing layer and dissociate to form hydroxyl ions (OH⁻) which are preferably adsorbed at the S-vacancy sites. However, the chemisorbed layer and the first physisorbed water layer do not help in proton conduction³⁶ because the water layer formation is not continuous and the protons do not get a continuous path for hopping. Hence for lower humidity levels, the response is quite low.

With increase in RH level, a large number of H₂O molecules get physisorbed on the MoS₂ surfaces preferably on the top of MoS₂ hexagons.³⁹ The water vapour condenses on the surface of the sensing layer and proton conduction takes place through the aquatic layers.³⁶ At higher humidity levels, the water molecules also dissociate to form hydronium (H₃O⁺) ions. So both H⁺ and H₃O⁺ contribute to the conduction mechanism and this increases the response at higher humidity levels. The recovery of the MoS₂ based humidity sensor was also found to be quite fast. We believe, the intrinsic hydrophobic property of MoS₂ ²⁷ accelerates the desorption of water molecules and leads to a fast recovery of the sample.

The sensing mechanism has been explained through a schematic representation in Fig. 13. The OH– ions are chemisorbed at the defect sites and water molecules are physisorbed on top of them. More water layers are adsorbed on top of the first physisorbed and chemisorbed layers and proton conduction takes place through them. Thus the combination of charge transfer between water molecules and sensing layer and also proton conduction between water molecules increases the conductivity of the sensing layer in presence of humidity.

Fig. 13 Schematic representation of humidity sensing mechanism of MoS₂ sensing layer.

ii. Response variation with sonication time. In order to get insights into the role of S-vacancies in humidity sensing, the response of 2 hours, 4 hours, 8 hours and 12 hours sonicated samples were tested against three different humidity levels. The improved response with increase in sonication time (as shown in Fig. 7(a)) further confirms the role of vacancy sites in humidity detection. The number of defect sites increased with increase in sonication time and this is evident from the XPS results shown in the ESI (Fig. S5†). More the number of S-vacancies, more number of OH– gets attached to these sites and facilitates in proton conduction for the subsequent water layers. However, the response of the 12 hours sonicated sample was found poorer than that of the 8 hours sonicated sample. This might have occurred because of agglomeration of the 12 hours sonicated nanoflakes due to the high energy of sonication. The agglomeration was observed from the AFM image of the sample (Fig. S6 in ESI†).

It was also observed that the response of the MoS₂ based humidity sensors was quite low. This is because undoped MoS₂ layers have an extremely low density of defect sites and exhibit a very low affinity to external molecules. The water molecules face difficulty getting attached to the MoS₂ sensing surface to create a high density of surface adsorption sites. Hence to increase the number of adsorbent sites, functionalization was done by mixing different proportions of GO with MoS₂.

iii. Composite response. The nano-composite made of MoS₂ and GO was used as humidity sensing layer to take the advantage of the both the materials. Thus the humidity response of the composite was found to be much better than that of MoS₂ or GO based sensor response. The MoS₂/GO 1 : 4 sample exhibited a record high response of 1600 times at 85% RH sensor. The superior performance from our sensing layer might be because of the following reasons:

(I) GO usually contains many functional groups (e.g. carboxyl, carbonyl, epoxy and hydroxyl) and inherent defect sites. These can act as adsorption sites for water molecules. Apart from this, large proton conductivity among water molecules plays a crucial role in GO based humidity sensor. Thus, GO gives reasonably good response towards humidity, as shown in Fig. 8(a).

(II) MoS₂ is inherently n-type³⁷ and GO is p-type.²⁴ The work function of MoS₂ is around 4.55 eV and that of graphene is around 4.66 eV.⁴⁰ Hence a depletion layer is formed at the interface of MoS₂ nano flakes and GO nano sheets and this might increase the number of adsorption sites.

(III) In the nano-composite, water molecules are going to attach at defect sites present in both MoS₂ and GO layers. Apart from this, functional groups present in the GO layer help in water molecules' adsorption. Owing to the large surface area of MoS₂ nanoflakes as well as GO nanosheets, the content of adsorbed water is very high thus increasing the density of charge carriers.

Apart from such charge transfer between the sensing layer and water molecules, proton conduction occurs through the GO layer as well as MoS₂.

For low humidity levels, the charge carriers (protons) migrate through hydrogen hopping between adjacent hydroxyl ion sites. But the transfer through the discontinuous water layer is quite difficult and hence the sensor exhibits low response for lower humidity levels as can be seen from Fig. 8(b). At higher humidity values, presence of continuous water layers facilitates the dissociation of water to produce hydronium groups. The composite was found to be slightly acidic (pH value was ∼5) further helping in proton generation in acidic medium. Due to the existence of a continuous water layer and existence of an electrostatic field between the adsorbed layer and surface, the physisorbed water molecule dissociates into H₃O⁺ and OH⁻ ions. The reactions are as follows:

H₂O ⇔ H⁺ + OH⁻ (for lower humidity levels)

2H₂O ⇔ H₃O⁺ + OH⁻ (for higher humidity levels)

For higher humidity levels, the diffusion of H₃O⁺ on hydroxide ions is dominant, although protons (H⁺) transfer by OH⁻ between neighboring water molecules still exist. Charge transfer occurs when a proton is released from one H₃O⁺ and transferred to the adjacent water molecule. This proton tunneling between adjacent water molecules lead to a significant increase in sensitivity of our sensor device.

From Fig. 8(a), we see that increasing the proportion of GO enhances the humidity transport properties of the sensing layer but till a certain level (MoS₂/GO 1 : 4) beyond which the response degrades. It has been seen from the SEM images that with increase in the amount of GO in the solution, the GO sheets, being larger in size, tend to cover the MoS₂ flakes. Hence the water molecules do not get a chance to get adsorbed on the exposed edges of MoS₂ as well the depletion layer created at the junction of GO and MoS₂. Hence the response decreases with larger proportion of GO beyond 1 : 4.

Conclusions

To conclude, MoS₂/GO composites based resistive humidity sensors were fabricated on Si/SiO₂ substrate. The humidity sensing mechanism of MoS₂ as well as MoS₂/GO was also explained in details. The optimised MoS₂/GO sample showed a record high response (1600 times at 85% RH level), excellent stability, repeatability with a very small amount of hysteresis. Thus we believe that the MoS₂/GO nano composite can act as a promising sensing material in future to develop highly sensitive, low power & ultrafast resistive humidity sensors.

Acknowledgements

D. Burman acknowledges Milan Sasmal for providing the mask of electrodes. S. Santra acknowledges Department of Science and Technology (DST), India for the partial support of the work (project no SR/S2/RJN-104/2011). P. K. Guha acknowledges SRIC, IIT Kharagpur (ISIRD) for supporting this work partially. All the authors extend their acknowledgements to Central Research Facility, IIT Kharagpur to let us use the instruments to carry out different material characterizations (AFM, FESEM, HRTEM, and XRD). We would also like to acknowledge DST FIST Lab (in the Department of Physics, IIT Kharagpur) for allowing us to use its XPS facility and Microelectronics Lab in Department of Electronics and Electrical Communication Engineering for humidity sensing facility.

References

1. H. Farahani, R. Wagiran and M. Hamidon, Sensors, 2014, 14, 7881–7939.
2. http://www.marketsandmarkets.com/PressReleases.
3. R. Ghosh, A. Midya, S. Santra, S. Ray and P. Guha, in Physics of Semiconductor Devices, ed. V. K. Jain and A. Verma, Springer International Publishing, 2014, ch. 180, pp. 699–701.
4. M. Matsuguchi, Y. Sadaoka, Y. Sakai, T. Kuroiwa and A. Ito, J. Electrochem. Soc., 1991, 138, 1862–1865.
5. C. Bariáin, I. R. Matías, F. J. Arregui and M. López-Amo, Sens. Actuators, B, 2000, 69, 127–131.
6. D. J. Late, Y.-K. Huang, B. Liu, J. Acharya, S. N. Shirodkar, J. Luo, A. Yan, D. Charles, U. V. Waghmare, V. P. Dravid and C. N. R. Rao, ACS Nano, 2013, 7, 4879–4891.
7. S. Lei, D. Chen and Y. Chen, Nanotechnol., 2011, 22, 265504.
8. Y. Yao, X. Chen, H. Guo and Z. Wu, Appl. Surf. Sci., 2011, 257, 7778–7782.
9. W. Lim, Y. Yap, W. Chong and H. Ahmad, Sensors, 2014, 14, 24329.
10. E. Traversa, Sens. Actuators, B, 1995, 23, 135–156.
11. http://www.greystoneenergy.com.
12. http://www.johnsoncontrols.com/content/us/en/products/building_efficiency/products-and-systems/integrated_hvac_systems/hvac/sensor_product_family/humidity.html.
13. http://sensing.honeywell.com/products/humidity-sensors?Ne=2308%26N=3217.
14. C. Wang, L. Yin, L. Zhang, D. Xiang and R. Gao, Sensors, 2010, 10, 2088–2106.
15. O. Erdamar, B. Bilen, Y. Skarlatos, G. Aktas and M. N. Inci, Phys. Status Solidi C, 2007, 4, 601–603.
16. M. Parthibavarman, V. Hariharan and C. Sekar, Mater. Sci. Eng., C, 2011, 31, 840–844.
17. R. Ghosh, A. Singh, S. Santra, S. K. Ray, A. Chandra and P. K. Guha, Sens. Actuators, B, 2014, 205, 67–73.
18. T. Low and P. Avouris, ACS Nano, 2014, 8, 1086–1101.
19. S. Chen, P. Bao, L. Xiao and G. Wang, Carbon, 2013, 64, 158–169.
20. V. P. Verma, S. Das, I. Lahiri and W. Choi, Appl. Phys. Lett., 2010, 96, 203108.
21. S. Santra, G. Hu, R. C. T. Howe, A. De Luca, S. Z. Ali, F. Udrea, J. W. Gardner, S. K. Ray, P. K. Guha and T. Hasan, Sci. Rep., 2015, 5, 17374.
22. A. De Luca, S. Santra, R. Ghosh, S. Z. Ali, J. W. Gardner, P. K. Guha and F. Udrea, Nanoscale, 2016, 8, 4565–4572.
23. S. Ghosh, R. Ghosh, P. K. Guha and T. K. Bhattacharyya, IEEE Trans. Nanotechnol., 2015, 14, 931–937.
24. D. Zhang, J. Tong and B. Xia, Sens. Actuators, B, 2014, 197, 66–72.
25. F. K. Perkins, A. L. Friedman, E. Cobas, P. M. Campbell, G. G. Jernigan and B. T. Jonker, Nano Lett., 2013, 13, 668–673.
26. B. Cho, M. G. Hahm, M. Choi, J. Yoon, A. R. Kim, Y.-J. Lee, S.-G. Park, J.-D. Kwon, C. S. Kim, M. Song, Y. Jeong, K.-S. Nam, S. Lee, T. J. Yoo, C. G. Kang, B. H. Lee, H. C. Ko, P. M. Ajayan and D.-H. Kim, Sci. Rep., 2015, 5, 8052.
27. S.-L. Zhang, H.-H. Choi, H.-Y. Yue and W.-C. Yang, Curr. Appl. Phys., 2014, 14, 264–268.
28. Y. Tan, K. Yu, T. Yang, Q. Zhang, W. Cong, H. Yin, Z. Zhang, Y. Chen and Z. Zhu, J. Mater. Chem. C, 2014, 2, 5422–5430.
29. M. Chen, S. Wi, H. Nam, G. Priessnitz and X. Liang, J. Vac. Sci. Technol., B: Nanotechnol. Microelectron.: Mater., Process., Meas., Phenom., 2014, 32, 06FF02.
30. J. Song, X. Wang and C.-T. Chang, J. Nanomater., 2014, 2014, 6.
31. J. Xiao, D. Choi, L. Cosimbescu, P. Koech, J. Liu and J. P. Lemmon, Chem. Mater., 2010, 22, 4522–4524.
32. C. Liu, F. Hao, X. Zhao, Q. Zhao, S. Luo and H. Lin, Sci. Rep., 2014, 4, 3965.
33. J. Whelan, I. Banu, G. Luckachan, N. Banu, S. Stephen, A. Tharalekshmy, S. Al Hashimi, R. Vladea, M. Katsiotis and S. Alhassan, J. Anal. Sci. Technol., 2015, 6, 1–8.
34. http://www.onsetcomp.com/support/tech-notes/there-way-verify-accuracy-relative-humidity-sensors-hobo-loggers-i-tried-testing.
35. Z. Cheng-Long, Q. Ming and H. Qing-An, IEEE Sens. J., 2011, 11, 2986–2992.
36. Z. Chen and C. Lu, Sens. Lett., 2005, 3, 274–295.
37. B. Radisavljevic, A. Radenovic, J. Brivio, V. Giacometti and A. Kis, Nat Nanotechnol., 2011, 6, 147–150.
38. J. Hong, Z. Hu, M. Probert, K. Li, D. Lv, X. Yang, L. Gu, N. Mao, Q. Feng, L. Xie, J. Zhang, D. Wu, Z. Zhang, C. Jin, W. Ji, X. Zhang, J. Yuan and Z. Zhang, Nat. Nanotechnol., 2015, 6, 6293.
39. Q. Yue, Z. Shao, S. Chang and J. Li, Nanoscale Res. Lett., 2013, 8, 1–7.
40. S. Larentis, J. R. Tolsma, B. Fallahazad, D. C. Dillen, K. Kim, A. H. MacDonald and E. Tutuc, Nano Lett., 2014, 14, 2039–2045.

---

Footnote

† Electronic supplementary information (ESI) available: The SEM images of various MoS₂/GO composites, EDAX of 8 hours sonicated MoS₂ nanoflakes, GO sheets and MoS₂/GO 1 : 4 composite. The XPS analysis of exfoliated MoS₂ containing sulfur vacancy sites and AFM image of 12 hours sonicated sample showing agglomeration. See DOI: 10.1039/c6ra11961a

---