FlatChem 5 (2017) 1–8

Contents lists available at ScienceDirect

FlatChem

journal homepage: www.elsevier .com/locate /flatc

Conductometric gas sensing behavior of WS2 aerogel

http://dx.doi.org/10.1016/j.flatc.2017.08.0032452-2627/� 2017 Elsevier B.V. All rights reserved.

⇑ Corresponding author at: Department of Chemical and Biomolecular Engineer-ing, University of California, Berkeley, CA 94720, United States.

E-mail address: maboudia@berkeley.edu (R. Maboudian).

Wenjun Yan a,b,c, Anna Harley-Trochimczyk a,b, Hu Long a,b, Leslie Chan a,b, Thang Phamd,e,f, Ming Hu c,Yuxiang Qin c, Alex Zettl d,e,f, Carlo Carraro a,b, Marcus A. Worsley g, Roya Maboudian a,b,⇑aBerkeley Sensor & Actuator Center, University of California, Berkeley, CA 94720, United StatesbDepartment of Chemical and Biomolecular Engineering, University of California, Berkeley, CA 94720, United Statesc School of Electronic Information Engineering, Tianjin University, Tianjin 300072, ChinadDepartment of Physics, University of California, Berkeley, CA 94720, United StateseMaterials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, United StatesfKavli Energy NanoSciences Institute at the University of California, Berkeley and the Lawrence Berkeley National Laboratory Berkeley, CA 94720, United Statesg Physical and Life Science Directorate, Lawrence Livermore National Laboratory, Livermore, CA 94550, United States

a r t i c l e i n f o

Article history:Received 19 April 2017Revised 2 August 2017Accepted 8 August 2017Available online 9 August 2017

Keywords:WS2 aerogel2D materialsGas sensorLow-power consumption

a b s t r a c t

The gas sensing characteristics of porous tungsten disulfide (WS2) aerogel are investigated. The sensor isfabricated by integrating WS2 aerogel onto a low power polycrystalline silicon microheater platform toprovide control over the operating temperature. The WS2 aerogel is characterized by scanning electronmicroscopy, transmission electron microscopy, Raman spectroscopy, X-ray diffraction, X-ray photoelec-tron spectroscopy, and thermal gravimetric analysis. The sensing performances of the WS2 aerogel-based sensor to NO2, O2, NH3, H2, and humidity are investigated, indicating a p-type behavior. The opti-mum sensing temperature is found to be about 250 �C, when considering sensitivity, power consumptionand response time. The role of O2 in the sensor performance is probed and is found to be helpful forenhancing the sensitivity and recovery of the sensor in H2, humidity and NH3.

� 2017 Elsevier B.V. All rights reserved.

Introduction

There remains a strong need for miniaturized, low-power gassensors that can be deployed in wireless applications for improvedenvironmental protection and safety. The detection of gases suchas NO2 which is toxic at ppb-level concentrations [1], H2 which isexplosive and flammable at 4–75% concentrations in air [2], andNH3 which can result in lung damage and death at high concentra-tions [3], requires advances in sensing materials to provide thenecessary performance characteristics.

Two-dimensional (2D) transition metal dichalcogenides(TMDCs) with lamellar structures similar to that of graphite havereceived significant attention in recent years. As with graphene,TMDCs can be isolated into two-dimensional sheets, but unlikegraphene, they are often semiconducting with sizable band gaps.For example, bulk tungsten disulfide (WS2) is an indirect-bandgap (1.4 eV) semiconductor, but becomes a direct-bandgap(�2.1 eV) material when exfoliated into the monolayer state [4].More recently, 2D TMDCs are being explored as promising alterna-tives to graphene for many device applications including gas sen-

sors, due to their high surface-to-volume ratio and adjustablebandgap [5–8].

Single- or few-layer MoS2 has been found to have high sensitiv-ity for NO2, H2, NH3, and triethylamine gas [9–12]. However, lim-ited reports on the gas sensing characteristics of WS2 [4,13–17]are available. Furthermore, the recovery of WS2-based gas sensorsis typically incomplete at room temperature because of strong gasadsorption, which limits their practical applications [13,18].

In order to keep the low-power consumption of sensor whileheating the sensing material to enhance its response and recovery,a microfabricated heater is needed. However, integration of single-layer materials onto a microfabricated sensor device is difficult,and the surface area is limited by the device footprint. Assembling2D sheets into 3D aerogel structure provides a new way forenhanced sensing performances by maintaining a high surface areain an accessible porous network [19–22].

Given the potential of aerogels for gas sensing and the relativelack of knowledge of the gas sensing properties of WS2, an explo-ration of the gas sensing capability of WS2 aerogel is in order. Here,we report the first fabrication and characterization of a low-powergas sensor based on WS2 aerogel. The WS2 aerogel is synthesizedusing thermal decomposition of ammonium thio-metal salts [23].The aerogel is integrated onto a low-power siliconmicroheater plat-form, in order to control the operating temperature and enhance the

2 W. Yan et al. / FlatChem 5 (2017) 1–8

sensing response and recovery of the aerogel. The sensing behaviorsof the WS2 aerogel to NO2, O2, H2, NH3, and humidity in dry air andN2 are reported, including the sensor’s dependence on temperature.The role of background oxygen in sensing performances is alsoreported and a possible gas sensing mechanism is proposed.

Experimental

Synthesis and characterization

WS2aerogel synthesis follows theprocedure reported inRef. [23].Briefly, 36 mg of ammonium thio-tungstate (ATT) is dissolved in1 mL of deionized water in a vial. The vial is then subjected to afreeze-drying cycle to produce an ATT aerogel. The ATT aerogel isannealed in 2% H2/98% Ar mixture at 750 �C for 1 h to yield theWS2 aerogel.

WS2 aerogel is characterized using scanning electron micro-scopy (SEM, FEI Sirion XL30) and transmission electron microscopy(TEM, JEOL 2010). Raman spectroscopy (HORIBA Jobin Yvon systemequipped with an Olympus microscope, and a 632.8 nm laser as theexcitation source), X-ray diffraction (XRD, AXS D8 Discover GeneralArea Detector Diffraction System form Bruker with radiation from aCu target, Ka, k = 0.15406 nm), and X-ray photoelectron spec-troscopy (XPS, Omicron Dar400 system with an achromatic Al KaX-ray source) are used for elemental and structural analyses.

Sensor fabrication and sensing testing

The details of themicroheater fabricationprocess can be found ina previous report [24]. A silicon substrate is coated with silicon-richlow-stress silicon nitride (100 nm) followed by the deposition ofphosphorus-doped polysilicon (100 nm), which is patterned usingphotolithography to form the microheater. Another silicon nitridelayer (100 nm) is deposited and patterned for the microheater con-tacts, which are electron-beam evaporated Ti/Pt (10/90 nm). Theback side of the wafer is patterned and etched using KOH to releasethe silicon nitride membrane. The Ti/Pt electrodes overlaid on thesilicon nitride membrane are used as electrodes for conductometricsensing. The chip is wire bonded into a 14-pin ceramic dual inlinepackage for electrical characterization and sensor testing.

Conductometric gas sensors are prepared by drop casting a sus-pension of WS2 aerogel material onto a low-power microfabricatedheater platform. The WS2 aerogel is dispersed in isopropyl alcohol(0.5 mg/mL) by sonication and drop cast (five drops of 0.2 mL each)onto the microheater chip while the microheater is heated to200 �C and maintained there for 3 h to promote solvent evapora-tion and material deposition at the center of the microheater.

For sensor testing, the sensor is placed in a gas flow chamber(�1 cm3 volume). Gas exposure and signal collection are controlledusing Labview and an open-source Java-based instrument and con-trol and measurement software suite, Zephyr. Gas tanks of O2

(Praxair, 99.995%), H2 (Praxair, 5% in N2), NO2 (Praxair, 20 ppm inN2), and NH3 (Praxair, 1000 ppm in N2) are controlled with massflow controllers (Bronkhorst) and mixed with N2 or dry air to givethe desired gas concentration with a total flow rate of 300 sccm.Relative humidity (RH) is adjusted by evaporating deionized waterand mixing it with gases. For sensor signal collection, a bias voltageis applied to the microheater and the microheater resistance ismeasured with a Keithley 2602A source-meter.

Results and discussion

Characterization methods

Each single layer of WS2 comprises of a tungsten layer sand-wiched between two sulfur layers in a trigonal prismatic coordina-

tion as shown schematically in Fig. 1a. Fig. S1 (in SupportingInformation) shows an optical image of as-synthesized WS2 aero-gel. Scanning electron microscopy images in Fig. 1b-c show theconnected nanosheet structure of the WS2 aerogel. The aerogeldensity is 35 mg/cm3, which is 0.5% of the bulk density of WS2,with a surface area of 9 m2/g calculated from nitrogen adsorptionanalysis [23] (Supporting Information, Fig. S2). After depositionon the microheater device, the sensing material maintains theporosity and the connected nanosheets of the aerogel (as shownin Fig. S1, Supporting Information), as judged by comparing thematerial on SEM stub and microheater device after deposition.Fig. 1d shows a representative low-magnification TEM image ofWS2 stacked nanosheets. The measured spacing between thefringes in the lattice is approximately 6.1 Å (Fig. 1e), whichmatches well the (0 0 2) d-spacing for the layered WS2 structure.The high-resolution TEM image in Fig. 1f shows Moiré fringes indi-cating that the WS2 sheets contain many layers, with thicknessesbetween 2 and 8 nm (Fig. S3 in Supporting Information, andFig. 1f).

To confirm the structure and composition of the aerogel, Ramanspectroscopy is performed. The WS2 Raman spectrum is character-ized by the first order modes at the Brillouin zone center, known asE2g and A1g. Previous studies on the Raman spectroscopy of WS2report that the A1g mode blueshifts with an increasing number oflayers, while the E2g mode exhibits a redshift [25–27], with theA1g and E2g peaks of monolayer WS2 reported at 417 and355 cm�1, respectively [25]. Fig. 2a shows that the major peaksfound for the WS2 aerogel are at 420 cm�1 (A1g) and 350 cm�1

(E2g). Hence, the WS2 aerogel used in the present study is com-posed of multiple stacked layers, consistent with the TEM analysisof the material.

X-ray photoelectron spectroscopy is performed to determinethe oxidation states of the W in the aerogel. In Fig. 2b, the peakslocated at 33.1, 35.2, and 38.6 eV correspond to W 4f7/2, 4f5/2,and W 5p3/2 binding energies (BE), respectively. The energy posi-tions of these peaks indicate a W valence state of +4, which is inaccordance with the previous reports, indicating the formation ofpure WS2 phase [28]. At the same time, nearly all the S 2p signalis from S2� (2p3/2, BE = 162.6 eV; 2p1/2, BE = 163.7 eV), also typicalof WS2. The sulfur to tungsten ratio is calculated to be 2.05. Theanalysis also indicates the absence of any WO3 in the aerogel.

The X-ray diffraction pattern (Fig. 2c) shows characteristicpeaks for WS2 at 2h = 14.2�, 28.5�, 33.3�, 39.5�, 43.3� and 58.5�,which respectively correspond to the (0 0 2), (0 0 4), (1 0 0),(1 0 3) and (1 1 0) planes of hexagonal 2H-WS2 (JCPDS-ICDD 71-4832). The (0 0 2) peak suggests that the turbostratic stacking isdominant in the aerogel [29]. The broadness of the peaks is indica-tive of the randomly connected nanostructures that make up theaerogel. This result is consistent with the SEM and TEM images(Fig. 1b–f).

Thermogravimetric analysis (TGA) performed under 20% O2 andAr mixed gas is presented in Fig. 2. The aerogel shows some weightloss at temperatures below 100 �C, which is attributed to theremoval of adsorbed water. It begins to degrade to form WO3 ataround 400 �C, with a final weight loss of 6.5%, which is consistentwith the full conversion of WS2 to WO3. Therefore, sensing testsmust be performed at temperatures below 400 �C to ensure thatthe WS2 is not oxidizing during testing.

Gas sensing performances

The microheater used in this work can reach 300 �C with only5 mW peak power. The schematic diagram of the sensor is shownin Fig. 3a, indicating the suspended microheater and the WS2 aero-gel deposited above the heater. An optical image of the micro-heater chip (3.5 � 3.5 mm2) containing four microheater sensors

Fig. 1. (a) Primitive cell and 3D schematic representation of a WS2 structure with sulphur atoms in yellow and the tungsten atoms in blue. (b and c) SEM images of WS2aerogel. (d) Low magnification TEM image, (e) and (f) high resolution micrograph of a flake with layered crystal structure of WS2 aerogel. The inset in image (e) is themagnification of the green box.

W. Yan et al. / FlatChem 5 (2017) 1–8 3

is shown in Fig. 3b. Each microheater sensor has four electricalcontacts: two for the microheater leads and two for electrical prob-ing of the sensing layer. Fig. 3c shows a zoomed-in view of a singlemicroheater sensor showing the polysilicon heater, the siliconnitride membrane and the Ti/Pt sensor leads. Fig. 3d shows thepower input to the microheater versus temperature of the micro-heater, which is calibrated by measuring the radiation spectrumwith the heater powered to the glow point and fitting it to Planck’sdistribution, as described in Ref. [30].

The current vs. voltage characteristics across the WS2 aerogel(Fig. 4a) display a nearly linear behavior in the temperatures rangefrom 20 to 300 �C, suggesting the formation of an Ohmic contactwith the Pt/Ti sensing electrodes. Fig. 4b shows a decreasing resis-tance with temperature increasing, consistent with the typicalbehavior for semiconductors. Namely, as the temperatureincreases, carriers are thermally activated and the resistancedecreases following an exponential relationship, R(T) = R0e

(E/kT),where E is the activation energy and k is the Boltzmann constant.The thermal-activation model yields a good fit to the dataset asshown in Fig. 4b, with an activation energy of 0.30 eV. This behav-ior is in agreement with previous report on WS2 nanowires [31].For gas sensing measurements, the sensor voltage is maintainedat 1 V.

The gas sensitivity is calculated as DR/R0 = (R0 � Rgas)/R0, whereR0 and Rgas represent the resistance of the device to the backgroundgas and the analyte gas, respectively. The response time (tresponse) isdefined as the time taken to reach 90% of the full response after theintroduction of the target gas. The recovery time (trecovery) isdefined as the time taken to return to 90% of the baseline resis-tance after the flow of target gas is stopped.

Fig. 5a shows the sensor response to 2 ppm NO2 in pure N2 at20 �C for 15 min duration pulses. The Fig. 5a indicates that theWS2 aerogel can detect NO2 gas at room temperature, althoughthe response does not stabilize and the rate of signal recovery isvery low. Fig. 5b shows the temperature dependence of theresponse to 2 ppm NO2 in pure N2, with the temperature rangingfrom 20 to 300 �C. There is a clear trend from slower response timeand incomplete recovery at room temperature to faster responseand recovery at higher temperatures, with a similar recoverybehavior at 250 and 300 �C. Additionally, the response rate andsensitivity appear to increase with increasing temperature,although the sensitivity at low temperatures is difficult to quantifywhen the sensor does not reach a stable value. The sensing behav-ior versus temperature is similar to the response and recoveryresults for other TMDCs-based gas sensors [32,33]. Consideringthe improved response and recovery to NO2, the thermal stabilityof the WS2 aerogel, and the power consumption of the sensor,the heater temperature of 250 �C is considered the optimum oper-ating temperature. Further sensing tests are taken at 250 �C tofacilitate comparison of the response to other gases.

The NO2 exposure tests reported in Fig. 5 were performed innitrogen, but in practical applications, the sensor is exposed toair. In order to determine whether this has an impact on the sensorcharacteristics, the WS2 aerogel sensor is exposed to oxygen at var-ied concentrations. The resistance of the WS2 aerogel sensor versustime to different concentrations of O2 in N2 ranging from 500 to5000 ppm with the heater at 250 �C was tested. Fig. 6 shows theresponse increasing monotonically with O2 concentration in thisrange. The resistance of a WS2 aerogel sample decreases uponexposure to O2, and increases when the O2 flow is turned off. For

Fig. 2. WS2 aerogel (a) Raman spectrum, (b) X-ray photoelectron spectra fromW 4f and S 2s regions, (c) X-ray diffraction spectrum, and (d) thermogravimetric analysis in 20%O2/Ar.

Fig. 3. (a) Schematic diagram of the sensor cross-section. (b) Optical image of a 3.5 � 3.5 mm2 chip containing four microheaters. (c) Zoom-in of one microheater with sensingelectrodes on top. (d) Typical microheater temperature vs. applied power curve.

4 W. Yan et al. / FlatChem 5 (2017) 1–8

sensors operating in air, this indicates that the sensor baseline willbe lower than expected. It also suggests the importance of carefullycontrolling the oxygen concentration during exposure tests bal-anced in air such that any sensor response is attributed to the

change in the target gas concentration, and not oxygenconcentration.

Theoretical calculations and experimental results on the gassensing characteristics of graphene and MoS2-based devices have

Fig. 4. (a) I/V characteristics of the WS2 aerogel sensor at different temperatures from 20 to 300 �C. (b) Typical sensor resistance vs. temperatures with a sensor voltage of 1 V.

Fig. 5. (a) Response of the WS2 aerogel sensor to three cycles of 2 ppm NO2 in pure N2 at 20 �C. (b) Response of the sensor to 2 ppm NO2 in pure N2 at different temperatures(20, 100, 200, 250, 300 �C) showing improved response and recovery times with increasing temperature.

Fig. 6. Sensitivity of the WS2 aerogel sample to 500–5000 ppm O2 in pure N2 at250 �C. The error bars represent the standard deviation of response of the 5exposures to a given O2 concentration.

W. Yan et al. / FlatChem 5 (2017) 1–8 5

demonstrated high sensitivity to various gases due to the chargetransfer between target gases and sensing materials, where oxidiz-ing gases act as electron acceptors, and reducing gases are electrondonors [12,34,35]. In our case, when the WS2 aerogel sensor deviceis exposed to oxidizing gases (NO2 and O2), the observed decreasein the sensor resistance suggests p-type behavior, which is inagreement with some previous reports [36–38]. McDonnell et al.[39] demonstrated systematically by experiments that sulfur-deficient and sulfur-rich MoS2 would be expected to result in n-type and p-type behavior respectively. Probably, a somewhat richS from the precursor (ammonium thiotungstate) results in the p-type behavior of WS2 aerogel. This is consistent with the XPS spec-tra (Fig. 2f), indicating S-rich WS2.

The sensor response to varied concentrations of NO2 from 0.2 to3 ppm in dry air is shown in Fig. 7a with the heater at 250 �C. Theresistance of the sensor decreases upon exposure to NO2, consis-tent with p-type behavior. Fig. 7b shows the response of the sensorincreasing with increasing NO2 concentration, saturating above2 ppm. Using the baseline noise level and the conventionalsignal-to-noise ratio of 3 implies a limit of detection of 8 ppb forNO2, which is well below the 100 ppb threshold of interest [1].The response time decreases from 6 to 2 min with increasingNO2 concentration while the recovery time ranges from 10 to

Fig. 7. (a) Resistance versus time and (b) the sensitivity of the sensor to NO2 in dry air with the microheater at 250 �C. (c) Response versus time and (d) the related sensitivityof the sensor to 500–5000 ppm H2 in dry air with the microheater at 250 �C. The error bars represent the standard deviation of response of the 5 exposures to a given NO2 (orH2) concentration.

6 W. Yan et al. / FlatChem 5 (2017) 1–8

14 min over the examined concentration range (Supporting Infor-mation, Fig. S4).

For the typical reducing gas H2, the resistance of the sensorincreases upon exposure to H2 with the heater at 250 �C demon-strating the p-type of WS2 aerogel, as shown in Fig. 7c. Fig. 7dshows the response increasing monotonically with H2 concentra-tion in the range of 50–5000 ppm. The signal-to-noise ratio of 3yields a limit of detection of 60 ppm for H2, which is well belowthe lower explosive limit of H2 (4000 ppm) [2]. The response andrecovery times range from 1 to 2 min for all hydrogenconcentrations.

For NH3, another reducing gas, the resistance of the sensor alsoincreases upon exposure to NH3 with the heater at 250 �C (Fig. S5,Supporting Information). The sensitivity increases monotonicallywith NH3 concentration in the range of 50–800 ppm. The signal-to-noise ratio of 3 implies a limit of detection of 13 ppm for NH3

[3]. The response and recovery times range from 3 to 5 min forall NH3 concentrations. Compared to most TMDC-based sensors,which do not reach steady state upon exposure to NO2 or NH3,nor exhibit complete [13,34,40], the WS2 aerogel sensor reachessteady state with over 90% recovery in the examined concentrationrange at microheater temperature of 250 �C. Furthermore, the WS2aerogel is highly stable. After extended gas sensing measurementsat 250 �C, the WS2 remains unchanged, according to the Ramanspectra before and after one week of sensing tests at 250 �C(Fig. S7, Supporting Information).

The sensitivity values to H2 and NH3 are much lower than NO2

given that the responses are comparable but for very different con-centration ranges (0.2–3 ppm for NO2 versus 500–5000 ppm for H2

and 50–800 ppm NH3). The larger adsorption energy of NO2, com-

pared to that of H2 and NH3, may be responsible for the higher sen-sitivity and longer response/recovery time of the sensor in NO2

[41], as has been proposed earlier in the case of graphene-basedsensors [42,43].

Because the WS2 aerogel sensor is affected by the presence ofoxygen as shown in Fig. 6, the role of oxygen in the sensing perfor-mances of this sensor is further elucidated. In Fig. 8a, the sensor isexposed to 2000 ppm H2 with varied O2 content between 0 and21% balanced in dry N2. Fig. 8b-c show the sensitivity andresponse/recovery times as a function of O2 content in N2. The sen-sor resistance increases upon exposure to H2 in all measurementatmospheres (Fig. 8a), which is in agreement with Fig. 7c, whilethe baseline resistance of the sample decreases with increasingO2 content in N2, as expected from Fig. 6. However, the existenceof oxygen, in particular in 5–21% range, sharply enhances the sen-sitivity of the WS2 aerogel. Increasing the O2 content continuouslyimproves the recovery time, while it has no obvious effect on theresponse time.

Similarly, O2 enhances the sensing performances of WS2 aerogelto humidity and NH3, as shown in Fig. 8d-e. Clearly, recovery of thesensor is incomplete in pure N2, while the sensor recovers well indry air. Sensitivity values are increased and response time valuesare decreased to both humidity and NH3 in dry air (see SupportingInformation Table 1).

Although the majority of recent reports conclude that the pri-mary response mechanism for TMDC gas sensors is charge transferbetween the adsorbed gas and the sensing material, our resultsindicate that the effects of adsorbed oxygen cannot be ignored.The weak response to reducing gases (H2, humidity and NH3) inpure N2 is consistent with theoretical studies that predict a low

Fig. 8. (a) Response to 2000 ppm H2 with different O2 content (0, 5%, 10% and 21%) in N2 at 250 �C; (b) the sensitivity and (c) response/recovery times to 2000 ppm H2 withdifferent O2 content mixed in N2. Error bars are standard deviation of all data points taken at 2 ppm H2 during a 2 h test. The reproducible resistance vs. time with cyclicexposures to (d) 40% RH and (e) 800 ppm NH3 at 250 �C in dry air and N2 respectively. (f) Selectivity of the WS2 aerogel sensor in dry air and N2 respectively with themicroheater at 250 �C.

W. Yan et al. / FlatChem 5 (2017) 1–8 7

degree of charge transfer between these three molecules and WS2[16]. The presence of oxygen in dry air further enhances the sensi-tivity of WS2 to H2, humidity and NH3. Similar behavior is reportedon horizontally aligned carbon nanotubes for H2 sensing [44] andliquid-exfoliated MoS2 for ethanol sensing [45], and is attributedto the adsorbed oxygen promoting the interaction kineticsbetween the reducing gases and the sensing materials. On the con-trary, investigation of NO2 sensing in pure N2 compared to in dryair shows a small effect of O2 (Supporting Information, Fig. S6). Thisis most likely because the degree of charge transfer for NO2 is sostrong that WS2 resistance is unaffected by O2 presence, or NO2

has no interaction with adsorbed O2 molecules.Although a few conductometric gas sensors based on WS2 have

been reported previously [4,13–17], limited information is avail-able on the selectivity and NO2 sensing performance of these sen-sors. The comparisons for WS2-based gas sensors are listed inTable. 1. The WS2 aerogel-based sensor, in present work, shows p-type characteristics, allowing easy differentiation of oxidizing andreducing gases in air as shown in Fig. 8f, as well as excellent NO2

Table 1Comparison of WS2-based gas sensors reported in literatures.

Material Sensor type Detected gases Recovery time

WS2 thin films [13] FET NH3 IncompleteMetallic 1T-WS2 [14] Impedimetric Methanol and

water–

WS2 nanoparticle [17] Resistance Humidity 13 sWS2 nanosheets [46] Resistance NO2 500 ppm, >10 minWS2 aerogel (present) Resistance NO2 2 ppm, 10 min

sensing performance with great selectivity in N2. This behavior isdifferent from gas sensors based on metal oxides (such as WO3),working in rich oxygen environment [47]. Compared to previousWS2 nanosheets-based NO2 sensor [46], the aerogel-based sensorhere shows short recovery time to low concentrations of NO2.

Conclusions

We have experimentally demonstrated a low-power microsen-sor utilizing WS2 aerogel sensing material. The response of the sen-sor to NO2, O2, H2, humidity, and NH3 were investigated in dry airand N2 respectively. Increasing the operating temperature to250 �C (<5 mW) results in both increased sensitivity and shorterresponse and recovery time. Due to larger adsorption energy ofNO2, WS2 aerogel shows stronger response to NO2 than to othergases with longer response and recovery time. O2 enhances thesensing performances of the sensor to reducing gases (H2 andNH3) and humidity, while it has a very weak effect on NO2 sensing.The WS2 aerogel is promising for NO2 sensing in low or even with-out oxygen environment.

Acknowledgements

The authors acknowledge Hai Liu for help with the SEM charac-terizations. This work was supported by Berkeley Sensor & Actua-tor Center (BSAC) Industrial Members and National ScienceFoundation (NSF Grant No. IIP 1444950), which provided for thedesign of experiments, student support (W.Y. A.H.-T., H.L., L.C.)and sensor fabrication and performance characterization. T.P. and

8 W. Yan et al. / FlatChem 5 (2017) 1–8

A.Z. acknowledge the Air Force Office of Scientific Research undercontract FA9550-14-1-0323, which provided for student support(T.P.) and instrumentation; and the Director, Office of Science,Office of Basic Energy Sciences, Materials Science and EngineeringDivision, of the U.S. Department of Energy under contract no. DE-AC02- 05CH11231 within the sp2-bonded Materials Program(KC2207), which provided for TGA and TEM characterization. M.W. would like to acknowledge the support of Lawrence LivermoreNational Laboratory under the auspices of the U.S. Department ofEnergy under Contract DE-AC52-07NA27344, through LDRD award13-LW-099. W.Y., H.L. and A.H.-T. acknowledge additional supportthrough the China Scholarship Council and the NSF GraduateResearch Fellowship (DGE 1106400), respectively.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.flatc.2017.08.003.

References

[1] United States Environmental Protection Agency. https://wwwepagov/no2-pollution. 2016, Feb.

[2] W. Buttner, R. Burgess, M. Post, C. Rvkin. Summary and Fingdings from theNREL/DOE Hydrogen Sensor Workshop, 2011.

[3] N. Roney, F. Llados, S.S. Little, D.B. Knaebel. Toxibological Profile of Ammonia -Agency for Toxic Substances and Disease Registry. US Department of Healthand Human Services, Atlanta, GA, USA, 2004.

[4] N. Huo, S. Yang, Z. Wei, S.S. Li, J.B. Xia, J. Li, Photoresponsive and gas sensingfield-effect transistors based on multilayer WS2 nanoflakes, Sci. Rep. 4 (2014)5209.

[5] Q.H. Wang, K. Kalantar-Zadeh, A. Kis, J.N. Coleman, M.S. Strano, Electronics andoptoelectronics of two-dimensional transition metal dichalcogenides, Nat.Nanotechnol. 7 (2012) 699–712.

[6] M. Chhowalla, H.S. Shin, G. Eda, L.J. Li, K.P. Loh, H. Zhang, The chemistry of two-dimensional layered transition metal dichalcogenide nanosheets, Nat. Chem. 5(2013) 263–275.

[7] A.B. Kaul, Two-dimensional layered materials: structure, properties, andprospects for device applications, J. Mater. Res.. 29 (2014) 348–361.

[8] P.K. Kannan, D.J. Late, H. Morgan, C.S. Rout, Recent developments in 2D layeredinorganic nanomaterials for sensing, Nanoscale 7 (2015) 13293–13312.

[9] Q.Y. He, Z.Y. Zeng, Z.Y. Yin, H. Li, S.X. Wu, X. Huang, et al., Fabrication of flexibleMoS2 thin-film transistor arrays for practical gas-sensing applications, Small 8(2012) 2994–2999.

[10] D.J. Late, Y.K. Huang, B. Liu, J. Acharya, S.N. Shirodkar, J.J. Luo, et al., Sensingbehavior of atomically thin-layered MoS2 transistors, Acs Nano 7 (2013) 4879–4891.

[11] H. Li, Z.Y. Yin, Q.Y. He, X. Huang, G. Lu, D.W.H. Fam, et al., Fabrication of single-and multilayer MoS2 film-based field-effect transistors for sensing NO at roomtemperature, Small 8 (2012) 63–67.

[12] F.K. Perkins, A.L. Friedman, E. Cobas, P.M. Campbell, G.G. Jernigan, B.T. Jonker,Chemical vapor sensing with mono layer MoS2, Nano Lett. 13 (2013) 668–673.

[13] M. O’Brien, K. Lee, R. Morrish, N.C. Berner, N. McEvoy, C.A. Wolden, et al.,Plasma assisted synthesis of WS2 for gas sensing applications, Chem. Phys.Lett. 615 (2014) 6–10.

[14] C.C. Mayorga-Martinez, A. Ambrosi, A.Y.S. Eng, Z. Sofer, M. Pumera, Metallic1T-WS2 for selective impedimetric vapor sensing, Adv. Funct. Mater. 25 (2015)5611–5616.

[15] V.Q. Bui, T. Pham, D.A. Le, C.M. Thi, H.M. Le, A first-principles investigation ofvarious gas (CO, H2O, NO, and O2) absorptions on a WS2 monolayer: stabilityand electronic properties, J. Phys.-Condes. Matter 27 (2015) 11.

[16] C. Zhou, W. Yang, H. Zhu, Mechanism of charge transfer and its impacts onFermi-level pinning for gas molecules adsorbed on monolayer WS2, J. Chem.Phys. 142 (2015) 214704.

[17] A.S. Pawbake, R.G. Waykar, D.J. Late, S.R. Jadkar, Highly transparent wafer-scale synthesis of crystalline WS2 nanoparticle thin film for photodetector andhumidity-sensing applications, ACS Appl. Mater. Interfaces 8 (2016) 3359–3365.

[18] K.Y. Ko, J.G. Song, Y. Kim, T. Choi, S. Shin, C.W. Lee, et al., Improvement of gas-sensing performance of large-area tungsten disulfide nanosheets by surfacefunctionalization, ACS Nano 10 (2016) 9287–9296.

[19] M.A. Worsley, T.T. Pham, A.M. Yan, S.J. Shin, J.R.I. Lee, M. Bagge-Hansen, et al.,Synthesis and characterization of highly crystalline graphene aerogels, AcsNano 8 (2014) 11013–11022.

[20] M.A. Worsley, S.O. Kucheyev, H.E. Mason, M.D. Merrill, B.P. Mayer, J. Lewicki,et al., Mechanically robust 3D graphene macroassembly with high surfacearea, Chem. Commun. 48 (2012) 8428–8430.

[21] M.A. Worsley, P.J. Pauzauskie, T.Y. Olson, J. Biener, J.H. Satcher, T.F. Baumann,Synthesis of graphene aerogel with high electrical conductivity, J. Am. Chem.Soc. 132 (2010) 14067–14069.

[22] J.L. Mohanan, I.U. Arachchige, S.L. Brock, Porous semiconductor chalcogenideaerogels, Science 307 (2005) 397–400.

[23] M.A. Worsley, S.J. Shin, M.D. Merrill, J. Lenhardt, A.J. Nelson, L.Y. Woo, et al.,Ultralow density, monolithic WS2, MoS2, and MoS2/graphene aerogels, ACSNano 9 (2015) 4698–4705.

[24] A. Harley-Trochimczyk, J. Chang, Q. Zhou, J. Dong, T. Pham, M.A. Worsley, et al.,Catalytic hydrogen sensing using microheated platinum nanoparticle-loadedgraphene aerogel, Sens. Actuator B-Chem. 206 (2015) 399–406.

[25] A. Berkdemir, H.R. Gutierrez, A.R. Botello-Mendez, N. Perea-Lopez, A.L. Elias, C.I. Chia, et al., Identification of individual and few layers of WS2 using Ramanspectroscopy, Sci. Rep. 3 (2013).

[26] W.J. Zhao, Z. Ghorannevis, K.K. Amara, J.R. Pang, M. Toh, X. Zhang, et al., Latticedynamics in mono- and few-layer sheets of WS2 andWSe2, Nanoscale 5 (2013)9677–9683.

[27] S.L. Li, H. Miyazaki, H. Song, H. Kuramochi, S. Nakaharai, K. Tsukagoshi,Quantitative Raman spectrum and reliable thickness identification for atomiclayers on insulating substrates, Acs Nano 6 (2012) 7381–7388.

[28] X. Mao, Y. Xu, Q. Xue, W. Wang, D. Gao, Ferromagnetism in exfoliated tungstendisulfide nanosheets, Nanoscale Res. Lett. 8 (2013) 1.

[29] G. An, C. Lu, C. Xiong, Solid-phase reaction synthesis of mesostructuredtungsten disulfide material with a high specific surface area, Mater. Res. Bull.46 (2011) 1323–1326.

[30] Q. Zhou, A. Sussman, J. Chang, J. Dong, A. Zettl, W. Mickelson, Fast responseintegrated MEMS microheaters for ultra low power gas detection, Sens.Actuators A-Phys. 223 (2015) 67–75.

[31] G.A. Asres, A. Dombovari, T. Sipola, R. Puskas, A. Kukovecz, Z. Konya, et al., Anovel WS2 nanowire-nanoflake hybrid material synthesized from WO3

nanowires in sulfur vapor, Sci. Rep. 6 (2016) 25610.[32] M. Donarelli, S. Prezioso, F. Perrozzi, F. Bisti, M. Nardone, L. Giancaterini, et al.,

Response to NO2 and other gases of resistive chemically exfoliated MoS2-basedgas sensors, Sens. Actuator B-Chem. 207 (2015) 602–613.

[33] B. Cho, J. Yoon, S.K. Lim, A.R. Kim, D.H. Kim, S.G. Park, et al., Chemical sensing of2D graphene/MoS2 heterostructure device, Acs Appl. Mater. Interfaces 7(2015) 16775–16780.

[34] B. Cho, M.G. Hahm, M. Choi, J. Yoon, A.R. Kim, Y.J. Lee, et al., Charge-transfer-based gas sensing using atomic-layer MoS2, Sci. Rep. 5 (2015) 8052.

[35] F. Niu, J.-M. Liu, L.-M. Tao, W. Wang, W.-G. Song, Nitrogen and silica co-dopedgraphene nanosheets for NO2 gas sensing, J. Mate. Chem. A 1 (2013) 6130–6133.

[36] C. Sommerhalter, T.W. Matthes, J. Boneberg, C.M. Lux-Steiner, P. Leiderer,Investigation of accepters in p-type WS2 by standard and photo-assistedscanning tunneling microscopy spectroscopy, Appl. Surf. Sci. 144–45 (1999)564–569.

[37] J.J. Devadasan, C. Sanjeeviraja, M. Jayachandran, Electrodeposition of p-WS2thin film and characterisation, J. Crystal Growth 226 (2001) 67–72.

[38] S.K. Srivastava, B.N. Avasthi, Layer type tungsten dichalcogenide compounds -their preparation, structure, properties and uses, J. Mater. Sci. 20 (1985) 3801–3815.

[39] S. McDonnell, R. Addou, C. Buie, R.M. Wallace, C.L. Hinkle, Defect-dominateddoping and contact resistance in MoS2, ACS Nano 8 (2014) 2880–2888.

[40] B. Liu, L. Chen, G. Liu, A.N. Abbas, M. Fathi, C. Zhou, High-performance chemicalsensing using Schottky-contacted chemical vapor deposition grownmonolayer MoS2 transistors, ACS Nano 8 (2014) 5304–5314.

[41] C.J. Zhou, W.H. Yang, H.L. Zhu, Mechanism of charge transfer and its impactson Fermi-level pinning for gas molecules adsorbed on monolayer WS2, J. Chem.Phys. 142 (2015).

[42] J.D. Fowler, M.J. Allen, V.C. Tung, Y. Yang, R.B. Kaner, B.H. Weiller, Practicalchemical sensors from chemically derived graphene, Acs Nano 3 (2009) 301–306.

[43] J.A. Robinson, E.S. Snow, S.C. Badescu, T.L. Reinecke, F.K. Perkins, Role of defectsin single-walled carbon nanotube chemical sensors, Nano Lett. 6 (2006) 1747–1751.

[44] B.-R. Huang, T.-C. Lin, A novel technique to fabricate horizontally aligned CNTnanostructure film for hydrogen gas sensing, Int. J. Hydrog. Energy 36 (2011)15919–15926.

[45] Y.H. Kim, K.Y. Kim, Y.R. Choi, Y.-S. Shim, J.-M. Jeon, J.-H. Lee, et al.,Ultrasensitive reversible oxygen sensing by using liquid-exfoliated MoS2nanoparticles, J. Mater. Chem. A 4 (2016) 6070–6076.

[46] D.J. Late, R.V. Kanawade, P.K. Kannan, C.S. Rout, Atomically thin WS2nanosheets based gas sensor, Sens. Lett. 14 (2016) 1249–1254.

[47] A. Afzal, N. Cioffi, L. Sabbatini, L. Torsi, NOx sensors based on semiconductingmetal oxide nanostructures: progress and perspectives, Sens. Actuators B:Chem. 171 (2012) 25–42.