Aerosol-Printed MoS2 Ink as a High Sensitivity Humidity Sensor.

Molybdenum disulfide (MoS2) is attractive for use in next-generation nanoelectronic devices and exhibits great potential for humidity sensing applications. Herein, MoS2 ink was successfully prepared via a simple exfoliation method by sonication. The structural and surface morphology of a deposited ink film was analyzed by scanning electron microscopy (SEM), Raman spectroscopy, and atomic force microscopy (AFM). The aerosol-printed MoS2 ink sensor has high sensitivity, with a conductivity increase by 6 orders of magnitude upon relative humidity increase from 10 to 95% at room temperature. The sensor also has fast response/recovery times and excellent repeatability. Possible mechanisms for the water-induced conductivity increase are discussed. An analytical model that encompasses two ionic conduction regimes, with a percolation transition to an insulating state below a low humidity threshold, describes the sensor response successfully. In conclusion, our work provides a low-cost and straightforward strategy for fabricating a high-performance humidity sensor and fundamental insights into the sensing mechanism.

Introduction

In recent years, printing technologies have increasingly been used to replace conventional semiconductor device technologies for the fabrication of ultra-low-cost electronic components.[1] Inks containing different classes of materials can be used to print functional components such as semiconducting layers, resistors, and dielectrics[2] over different types of (flexible) substrates.[3] In this context, a blooming field is the development of sensors that cover a wide area of applications such as real-time biomonitoring, wearables, environmental monitoring, industrial process control, and personal safety.[4−6] In particular, an area of interest is humidity sensors, which have had an increasing demand for use in environmental control and industrial processing.[7] For instance, humidity is closely monitored in the semiconductor industry, as devices and integrated circuits, printed circuit boards, electronic components, and data are highly sensitive to humidity.[8] Similarly, several industrial processes such as chemical purification of gases, using dryers and ovens, paper production and textiles, and food processing, require high control of humidity.[7] Thus, it is highly desirable to develop low-cost humidity sensors integrated into different surfaces and portable devices.[9,10] Several materials have been studied for their humidity sensing capabilities, such as polymers,[11,12] metal oxides,[13,14] carbon nanomaterials,[9,15,16] cellulose,[17,18] and transition metal dichalcogenides (TMDs).[19−23] Two-dimensional (2D) transition metal dichalcogenides, such as molybdenum disulfide (MoS2), are becoming significant materials in various research fields and, specially, for sensing applications. Recently, MoS2 has received increasing attention in ultrasensitive sensor applications to detect different gases, such as NO2, NH3, H2, and water molecules.[21,24−26]

Conductive ink formulations are still far from ideal, as they require substrate functionalization[27] or lengthy and expensive formulation processing.[28,29] They are also usually based on toxic and expensive solvents and additives or need relatively high temperatures to dry,[30] restricting their application to specific substrates. Alternatively, the use of aqueous dispersions as inks for printing on multiple devices can be an approach to replace nonsustainable and expensive methods.[31,32] However, the development of new inks composed of functional materials, with different readily available properties, that are both sustainable and inexpensive still is in its infancy.[33] Recently, several works in the literature have shown the development of water-based inks of 2D materials.[23,34] For instance, Casiraghi et al.[23] developed water-based and biocompatible graphene and hBN inks to fabricate all-2D material and inkjet-printed capacitors. In another work, McManus et al.[35] developed photodetectors inkjet-printed on paper using graphene as electrodes and TMDs such as MoS2, WS2, MoSe2, and MoTe2 as photoactive components.[34,36] Among various types of TMD preparation, liquid-phase exfoliation (LPE) by sonication is considered a low-cost, simple, and versatile method with high potential for scale-up.[37] This work presents a low-cost, stable, and printable MoS2 ink for high-performance humidity sensors. No harsh conditions, solvent exchange, or chemical treatments were used, which allow the application to several substrates, including flexible substrates. The aerosol-printed MoS2 sensor exhibit high sensitivity (105–106%) with response/recovery times comparable to commercial humidity sensors and excellent repeatability. Possible mechanisms for the water-induced conductivity increase are discussed, and an analytical model that encompasses two conduction regimes, depending on relative humidity (RH), is proposed and describes the sensor response to RH successfully. We believe that our device can significantly influence the application of MoS2 ink for humidity sensors.

Results and Discussion

Material Characterization

Figure shows a schematic illustration of the sensor preparation, described in detail in the Experimental Section (Figure a,b), with a representative picture of the device of the MoS2 film sensor on a flexible PET substrate (Figure c) and typical scanning electron microscopy (SEM) micrograph of the surface morphology of the film (Figure d). Details of the MoS2 ink sensor deposited on flexible (PET) substrate response to humidity are shown in the Support Information.

Figure 1

(a) Schematic representation of the humidity sensor preparation: a spray pen was employed to apply a uniformly MoS2 ink layer on a PET substrate with previously Au-deposited interdigitated electrodes and (b) humidity sensor with the MoS2 sensing material deposited. (c) Picture of the MoS2 film sensor on a flexible PET substrate. (d) SEM micrograph of the surface morphology of the MoS2 film.

The SEM image presented in the inset of Figure d reveals that the deposited MoS2 sensing film is continuous and composed of an interconnected array of MoS2 flakes that densely cover the entire device area forming a 3D-like nanostructure film. A large number (∼50) of spray scans were performed to ensure film continuity in the active region. Raman spectroscopy was employed to investigate the structural properties of the as-deposited MoS2 films (Figure a). Two active Raman active modes, E2g1 at 381.1 cm–1 and A1g at 406.8 cm–1, associated with the hexagonal MoS2 structure, are observed as expected. The A1g mode is associated with the out-of-plane vibration of S atoms only in opposite directions, whereas E2g1 results from the opposite vibration of two S atoms with respect to the Mo atom.[39−41] The frequency difference between the A1g and E2g1 Raman modes can be correlated to the number of MoS2 layers in the crystal. The obtained value of 25.7 cm–1 indicates that MoS2 flakes have multiple layers (>7 layers).[42]

Figure 2

(a) Raman spectrum of the MoS2 film. The separation between the position of the E2g1 and A1g peaks (25.7 cm–1) indicates that the flakes have multiple layers (>7 layers). (b) AFM image of MoS2 flakes. (c) Histogram of the average height distribution of the MoS2 flakes. (d) Distribution of Feret length of MoS2 flakes. A lognormal distribution (solid line) was fitted to the data to determine the mode of the MoS2 height and length distributions.

To perform the AFM measurements, the MoS2 ink was deposited on a Si wafer by drop-casting. To provide a representative statistical analysis of the material, nine images over distinct regions of the sample were acquired with a lateral size of 5 μm with 500 × 500 pixels. The lateral size of the MoS2 flakes was defined as the maximum Feret length and the height as the mean height value of the flakes.[43] This analysis followed the methodology developed by Fernandes et al.,[43] which provided a semiautomated statistical analysis of the thickness and size of graphene systems.

A representative AFM image of the MoS2 nanoflakes is shown in Figure b. The corresponding height and Feret length distributions are shown in Figure c,d, respectively. The modal height of the MoS2 flakes was 6.2 ± 5.7 nm (Figure c), which corresponds to around 10 layers, which is consistent with the number of layers estimated by Raman analysis (separation between the position of the E2g1 and A1g peaks (25.7 cm–1)), which indicates that the number of layers would be higher than 7 nm.[42] The modal Feret length obtained was 28.2 ± 104.1 nm (Figure d). The characteristic length/thickness aspect ratio, obtained from the modal values discussed above, is 4.5. Such value of ratio can be considered intermediate between the near-unity values of quantum-dot-type nanoparticles[44] and the values typically larger than 10 of nanoflake-type nanoparticles.[45−47] Previous investigations on films of quantum-dot-type MoS2 nanoparticles have also reported humidity-dependent conductivities.[44] Therefore, we can consider the possibility that small aspect ratio values can be relevant to humidity sensing characteristics.

Humidity-Sensing Results of the MoS2 film

To probe the sensor’s performance, we have carried out dynamic measurements, where the relative humidity is varied continuously. The sensor response was recorded as a function of time with a relative humidity variation of 10–95%. Figure a depicts the sensor conductance variation for a fixed applied voltage (+10 V) as a function of RH. Such high voltage is needed because of the high resistance of MoS2 films. One can note that the conductance changes by 6 orders of magnitude between 10–95% of RH. Results for different devices can be found in the Supporting Information. The sensitivity (S) of the MoS2 humidity sensor can be defined as[48]where IRH and I10 represent the device’s current values at the humidity atmosphere and at 10% RH, respectively. The inset of Figure a shows the device sensitivity S in log scale versus RH, for RH between 10 and 95%. A detailed view of S in the range of 10–20% is shown in Figure S5 of the Supporting Information file. The sensitivity reaches a maximum value of 106%. The sensitivity results for different devices can be found in Figure S6 of the Supporting Information file.

Figure 3

(a) Conductance values of a MoS2 sensor in a copper-clad phenolic sheet, determined from the fixed applied voltage of 10 V, as a function of the relative humidity (RH). Inset: Device sensitivity S (see eq ) versus RH. (b) Repeatability performance of the MoS2 sensor exposed to cyclic variations of RH between 20 and 50%. (c) Stability of the MoS2 humidity sensor at 50 and 95% RH during a period of 40 days. (d) Response and recovery dynamics of the MoS2 (black) and commercial (blue) sensors to sudden changes in relative humidity.

Table displays a comparison between the type of electrical response monitored (impedance, capacitive, and resistive) for different RH sensor materials reported in the literature, along with their sensitivity, recovery, and response time. Several materials have been used for humidity sensors in composite or in a single phase, having different properties like high thermal conductivity, high electrical conductivity, and good chemical and mechanical stabilities. Our sensor sensitivity is one of the highest achieved for resistive-type sensors and comparable with those obtained by other measurement techniques.

Table 1

Properties Comparison of the MoS2 Humidity Sensor and Reported Humidity Sensors

sensing material	sensor type	detection range (RH) (%)	sensitivity (%)	response time (s)	recovery time (s)	reference
MoS2	FET	0–35	104	10	60	(49)
MWCNT/Nafion nanofibers film	surface acoustic wave resonator	10–80	427.6	3	63	(50)
MoS2 QDs synthesized in NMP	impedance	10–95	2.27 × 106	14	280	(44)
graphene/ZnO	impedance	0–85	NA	1	2	(51)
glycidyl trimethyl ammonium chloride/cellulose	impedance	11–95	>67.3	25	188	(52)
2D hBN-poly(ethylene oxide)	impedance	0–90	2160	2.6	2.8	(53)
dendritic MoS2	impedance	11–95	3031	11	17	(21)
2D MoS2-PEDOT:PSS	impedance	0–80	4000	0.5	0.8	(54)
PEDOT:PSS/GO	impedance	0–100	2.6 × 104	1	3.5	(55)
halloysite nanotubes	impedance	0–91.5	105	0.7	57.5	(56)
TiO2/(K,Na)NbO3	impedance	12–94	1.6 × 105	25	38	(57)
BEHP-co-MEH: PPV-PAA.PSS	capacitive	0–80	NA	3.5	5	(58)
poly(dimethylsiloxane)/CaCl2	capacitive	30–95	10.2	120		(59)
GO	capacitive	30–90	209	∼200	∼100	(9)
ITO/alumina	capacitive	5–95	737.2	47.2	49.5	(60)
MoS2/nanodiamond	capacitive	11–97	∼3500	<1	0.9	(61)
carbon dots	capacitive	20–90	6300			(62)
GO/MWCNT	capacitive	11–97	7980	5	2.5	(63)
poly(ethylene oxide)/CuO/MWCNT	capacitive	30–90	53837.6	20	11	(64)
MoS2/SnO2	capacitive	0–97	3.3 × 106	5	13	(65)
MWCNT/HEC	resistive	20–80	3.84	∼20	∼35	(66)
SnO2/rGO	resistive	11–97	45.02	∼90	∼100	(67)
PEDOT:rGO-PEI/Au NPs	resistive	11–98	51.6	20	35	(68)
printed MWCNTs	resistive	30–60	57.6			(69)
MoS2/PVP	resistive	11–94	80	5	2	(70)
MoS2/GO	resistive	35–85	∼1700	43	37	(71)
MoS2/SiNWA	resistive	11–95	2967	22.2	11.5	(26)
Pt/MoS2	resistive	35–85	4000	91.2	153.6	(72)
2D MoS2	resistive	0–80	6800	0.6	0.3	(73)
TiO2 nanoflowers	resistive	20–95	4.61 × 104	4	<1	(74)
N-doped TiO2	resistive	0–90	3.28 × 105	18	299	(75)
MoS2-flakes	resistive	10–95	5.3 × 106	8	22	this work

The repeatability of the MoS2 ink sensor was assessed via consecutive tests of adsorption and desorption processes, switching the RH level for 10 cycles between ∼20% RH (for 4 min) and 50% RH (for 1.5 min); see Figure b. In the adsorption (desorption) process, the sensor’s current increases (decreases) with the increase (decrease) of the relative humidity. As shown in Figure b, the current variation was almost identical for all cycles performed for more than 50 min indicating excellent repeatability. Next, the sensor stability was studied for several consecutive days, evaluating the sensitivity at 50 and 90% relative humidity, as shown in Figure c. The current variation was less than 10% for each humidity region probed. The device also shows a very stable response with minimal performance fluctuation after 35 days, demonstrating critical long-term stability.

Figure d compares the current response of the MoS2 sensor with a commercial capacitive humidity sensor model AM2303 DHT22 response, both normalized. The MoS2 sensor variation between the two states is fast, stable, and reversible. Figure d shows that the MoS2 sensor response time (to a suddenly increased humidity) is similar to that of the commercial sensor, but it also shows that the MoS2 sensor recovery time (to a suddenly reduced humidity) is faster than that of the commercial sensor investigated. The response time relative to an RH increase from 85 to 87% is 8 s. Conversely, the recovery time to an RH decrease from 87 to 85% is 22 s. The difference between the response and recovery times can be ascribed to the higher humidity sensitivity response and higher bonding energy between the adsorbed water molecules and the surface of the sensor material.[76,77] Also, Table shows that recovery and response times of our sensor are comparable to most of the humidity sensors reported. Response and recovery times for a wider range variation of humidity can be found in the Supporting Information.

Humidity-Sensing Mechanisms

We will now address possible mechanisms behind the change of conductivity of the MoS2 film upon interaction with water molecules. Previous studies have investigated this interaction for MoS2-based composites and exfoliated MoS2 flakes. The issue is controversial since some works report that H2O adsorption on MoS2 results in a decrease of conductivity.[15,49,78] An electron charge transfer process from MoS2 to water molecules was predicted via density functional theory calculations[79] and also observed in few-layer MoS2 transistors.[80] On the other hand, other works report an increase of MoS2 conductivity due to water interaction.[21,26,61,65,71] For instance, conductivity increase with humidity was observed for MoS2/GO,[71] MoS2/nanodiamond composites,[61] MoS2/Si nanowires,[26] and dendritic MoS2.[21]

In the following, we will consider two analytical models[81,82] that can describe the dependence of the conductivity σ (and, consequently, of the conductance G = 1/R) on the relative humidity, in limits of low and high relative humidity (<55 and >55%, respectively). Figure a shows √G versus RH for the sensor at the low RH regime. Up to a given value of RH, there is no measurable current through the device. Above a threshold value RH0, a linear behavior of √G versus RH–RH0 is shown. By extrapolating such linear behavior down to G = 0 (a value that is not experimentally accessible due to maximum resistivity limitations), we obtain an estimation of RH0 = 6.07%.

Figure 4

(a) Square root of the conductance, √G, versus RH for humidity values below 55%. (b) Fitting of the conductance G as a function of relative humidity with eq (in blue) and (5) (in red).

The observed linear behavior of √G implies a power-law behavior of σ above the conduction threshold,with an exponent t = 2. As the amount of adsorbed water n has been predicted and observed[83] to increase continuously with RH, we conclude that a power law with the same exponent will occur as a function of n,

Now, let us consider that a contiguous adsorbed water layer is the only medium, where conduction can occur, without the possibility of either tunneling or thermal-induced crossing of charge carriers through “dry” regions. Then, the physical situation can be described by standard percolation conductivity,[84] where the conductivity, as a function of n, is predicted to be null up to a critical density n0, and, at the threshold for conduction, to behave exactly as in eq , with universal exponents that only depend on the dimensionality of the network of conducting channels. These exponents can be calculated numerically, leading to values such as t = 1.310 ± 0.001 in two dimensions[85] and t = 1.998 ± 0.004 in three dimensions.[86,87] Therefore, if standard percolation describes the conduction in our devices, the observed exponent t = 2 would indicate that the MoS2 film behaves as a “sponge” of finite thickness, where the adsorbed water can infiltrate and form a 3D network of water channels: a simple 2D water adsorption at the surface would lead to a smaller exponent. Interestingly, Figure a also shows that the approximately linear behavior of √σ versus RH extends for a wide range of relative humidity, well beyond the humidity RH0 at the conduction threshold up to about RH ≅ 55%. This provides us a very simple fitting formula for RH < 55%,

For values of RH larger than 55%, we will consider that the MoS2 layer is fully wet and that a contiguous liquid water layer is formed atop it. In this regime, the humidity-induced variation in the electrical transport is associated with changes in the ionic conductivity of the water layer. Such ionic transport is usually ascribed to the Grotthuss mechanism,[81,82] where H3O+ ions act as charge carriers in proton-exchange reactions, H3O+ + H2O → H2O + H3O+. Skinner et al. have proposed an analytical description of the ionic conductivity of a humidity-induced water layer atop an otherwise insulating solid surface.[88] Their calculations are based on the thermodynamic equilibrium between the water layer and the water vapor, as well as on the thermodynamic equilibrium between unbound (free) ionic carriers in the water layer and corresponding bound ions at the insulating surface. One of their main results is that the 2D density of free carriers n, in the limit of n ≪ nb, where nb is the 2D density of ion binding sites at the surface, is given by , where nH is the 3D density of water molecules in the liquid phase, γ is the Euler constant, κ is the water permittivity, lB is the Bjerrum length, and d is the thickness of the water layer.[88] Based on that result, Skinner et al. obtained a simple fitting function for the ionic conductivity σ as a function of the relative humidity RH (as shown in Figure b),

Conclusions

In conclusion, we have investigated the humidity-sensing properties of the resistive-type humidity sensor based on MoS2 ink, which is produced via sonication-based exfoliation. The conductance readout of the sensor can increase up to 6 orders of magnitude upon relative humidity increase from 10 to 95% at room temperature. The MoS2 ink sensor showed very high sensitivity, excellent stability, repeatability, and good response and recovery times. The humidity-sensing mechanisms of MoS2 were also discussed in detail. The results showed ideal characteristics for the development of high-performance humidity sensors for real-life applications.

Experimental Section

Material Preparation and Fabrication of Humidity Sensor

Molybdenum disulfide ink was prepared by sonication-assisted exfoliation. Briefly, 400 mg of MoS2 powder (<2 mm, 99%, Aldrich) was dispersed in 250 mL of a 7:3 volume of deionized water and isopropyl alcohol. The exfoliation of MoS2 in suspension was performed using an ultrasonic probe processer for 3 hours (Sonics Vibra-Cell VCX 2500). After sonication, the dispersion was centrifuged at 2500 rpm for 20 minutes to remove the residue/precipitate, and the yellowish-green supernatant containing the nanoflakes was collected, as described by Coleman and collaborators.[38] This supernatant was stable for six months. The UV–visible spectra of exfoliated MoS2 are shown in the Support Information. The MoS2 sensor film was deposited using an aerosol deposition method employing a 0.2 mm airbrush pen (Importway). Briefly, 20 mL of MoS2 ink was uniformly sprayed by the airbrush in an area of 50 × 4 mm2 in a copper-clad phenolic sheet or using a PET flexible substrate containing Au/Cr interdigital electrodes (IDEs) previously deposited. During the spraying process, the substrates were kept at 80 °C for rapid evaporation of water from the ink.

Material Characterization and Testing System of Humidity Sensors

The surface morphology of MoS2 ink films was observed by scanning electron microscopy (SEM, FEI Quanta 200 FEG). The microstructure was characterized by Raman spectroscopy (Witec Alpha300 spectrometer). Atomic force microscopy characterization was carried out on a Bruker MultiMode 8 SPM using the intermittent contact mode and Si cantilevers (DPE/XSC11 hard) from Mikromasch, with spring constants of 7–42 Nm–1 and a tip radius of curvature of ∼10 nm. A home-built controlled environmental chamber was used to record the sensor’s electrical response upon changes in relative humidity. The humidity level inside the chamber was regulated using electronic mass flow controllers. Water vapor from a bubbler under heating was added to an argon stream to set the chamber’s humidity, whereas the second stream of argon was used for dehumidification and for purging. A commercial temperature and humidity sensor (AM2302 DHT22, precision of 2% RH), based on a capacitor with a hygroscopic polymer as dielectric, was used as a reference and feedback. The commercial sensor and the MoS2 film sensor responses were continuously monitored. The temperature for the whole experiment was maintained at 25 °C. In addition, the bubbler’s heating temperature was also controlled to keep the humidity approximately constant or to obtain different values of relative humidity inside the chamber. Figure S3 of the Supporting Information shows a detailed schematic of the characterization setup.