Long-Term Stable 2H-MoS2 Dispersion: Critical Role of Solvent for Simultaneous Phase Restoration and Surface Functionalization of Liquid-Exfoliated MoS2.

Chemical exfoliation approaches such as Li-intercalation for the production of two-dimensional MoS2 are highly attractive due to their high yield of monolayer forms, cost-effectiveness, and mass-scalability. However, the loss of the semiconducting property and poor dispersion stability in solvent have limited the extent of their potential applications. Here, we report simultaneous phase recovery and surface functionalization for the preparation of a highly stable 2H-MoS2 dispersion in water. This study shows that high-yield restoration of the semiconducting 2H phase from a chemically exfoliated MoS2 (ce-MoS2) can be induced by a mild-temperature (180 °C) solvent thermal treatment in N-methyl-2-pyrrolidone (NMP). In addition to a phase transition, this solvent thermal treatment in NMP realizes concurrent surface functionalization of the 2H-MoS2 surface, which provides an outstanding dispersion stability to 2H-MoS2 in water for more than 10 months. Finally, we report the humidity sensor based on the functionalized 2H-MoS2, which shows a substantial response enhancement compared with a nonfunctionalized 2H-MoS2 or ce-MoS2.

Introduction

Over the past several years, two-dimensional (2D) transition metal dichalcogenides (TMDs) have received a large amount of attention due to their unique electrical, optical, and chemical characteristics compared with their bulk forms.[1−3] In particular, molybdenum disulfide (MoS2), which is the most widely investigated semiconducting TMD, is considered a potential substitute of Si for future electronics and optoelectronic devices due to its atomically thin nature and excellent performance such as a high on/off ratio (108), a large carrier mobility (200–500 cm2/V s), and a steep subthreshold swing (70 mV/dec)[1,4−6] as an active channel layer for field-effect transistors.

Various exfoliation methods for separating stacked layers of bulk MoS2 to prepare its 2D forms have been developed. Although mechanical exfoliation can produce a high-quality MoS2 with minimal defects,[7−9] extremely low throughput/yield, a limited flake size, and poor thickness control prevent its use for large-scale manufacturing. Thus, massive efforts have been devoted to develop a new method for producing a mass-scale and solution-processible exfoliated MoS2 through chemical approaches.[7,10−13] Among them, liquid-phase exfoliation in an appropriate solvent assisted by ultrasonication with or without a stabilizer and ion-intercalation method has been established. In particular, lithium-based intercalation is one of the most practical route due to the high yield of the monolayer MoS2 and mass productivity; meanwhile, the exfoliated MoS2 processed using the ultrasonication-assisted liquid exfoliation contains a high fraction of thick MoS2 even over nine layers.[8,14−19]

Li-intercalation is a promising route for the mass production of high-quality monolayer MoS2. However, several challenges still remain, although some of the issues of this method have been overcome by ammonia/amine intercalation.[20,21] Typically, during a lithiation process, MoS2 experiences loss of its semiconducting property caused by the structural phase transition from 2H (semiconductor) to 1T phase (metal).[14,18,22−26] This 1T phase is known as a metastable phase, and thus the 1T phase can be recovered to the 2H phase by annealing at high temperature typically over 300 °C. Eda and co-workers first reported the successful restoration of the semiconducting property of MoS2 via annealing after lithiation.[14] They systematically studied the phase restoration phenomenon with an elevation of the temperature above 100 °C and confirmed that the restoration to the 2H phase reached ∼100% at 300 °C. However, this annealing process was conducted on a substrate after forming a MoS2 thin film by a casting process, which has associated challenges related to the device stability because such high annealing temperature can result in thermal damage to other components in devices such as flexible substrate or organic semiconductor. Therefore, to widely extend the applicability of the semiconducting MoS2, it is necessary to develop a solution-phase restoration process. Furthermore, poor dispersion stability of the resultant Li-intercalated MoS2 upon restacking of the exfoliated MoS2 due to strong van der Waals interactions is a critical issue.[27−30] To improve the dispersion stability and avoidance of restacking, proper functionalization of the monolayer 2H-MoS2 surface is essential.[31−33] However, most functionalization methods are mainly focused on 1T-MoS2, and it is still challenging to produce a uniformly functionalized monolayer 2H-MoS2 with sufficient solution stability.

Here, we demonstrate the one-step preparation of a highly stable 2H-MoS2 dispersion via solvent thermal treatment in polar solvents using Li-intercalated MoS2. During solvent thermal treatment under mild temperature (180 °C), efficient restoration of the 2H phase is accomplished with an approximately 85% conversion ratio from 1T-MoS2 obtained from a Li-intercalation method. Notably, when the solvent thermal treatment was conducted in N-methyl-2-pyrrolidone (NMP), 2H-MoS2 can be functionalized simultaneously, providing an outstanding long-term stability of a concentrated aqueous dispersion for over 10 months. Moreover, the functionalized 2H-MoS2, which has a high water absorption capability, presents a high sensitivity to humidity, realizing a humidity sensor with an outstanding response activity that is several times higher than that of the original 1T-MoS2 prepared by Li-intercalation method.

Results

The overall process for the preparation of a highly stable 2H-MoS2 dispersion is illustrated in Figure a. First, few and monolayer MoS2 nanosheets were prepared from the chemical exfoliation of bulk MoS2 via the Li-intercalation method[14,22,26] (see Experimental Section for details). After complete washing of the samples to remove the remaining Li ions, the obtained chemically exfoliated MoS2 (ce-MoS2 with majority 1T phase) was transferred to the following polar solvents with a sufficiently high boiling point and suitable dispersion capability for MoS2: ethylene glycol (EG), formamide (FA), propylene carbonate (PC), N-methyl-2-pyrrolidone (NMP), dimethyl sulfoxide (DMSO), tetraethylene glycol dimethyl ether (TGDE), and hexamethylphosphoramide (HMPA). During the transfer procedure, the resultant ce-MoS2 should be in a wet-clay state with a small quantity of water to prevent the restacking of MoS2 sheets.[3,34] A homogeneous dispersion and dispersion stability within polar solvents are also important for avoiding the agglomeration of ce-MoS2 during the solvent thermal treatment. Thus, after ultrasonication for 30 min to obtain a homogeneous MoS2 dispersion in each polar solvent, the dispersions were kept for 1 week to investigate their stability. Based on this result, EG, FA, PC, NMP, and DMSO were chosen as appropriate solvent media for the solvent thermal treatment (Figure S1 in the Supporting Information).

Figure 1

Schematic of one-step phase recovery and surface functionalization using solvent thermal treatment. (a) Overall process for the preparation of a highly stable 2H-MoS2 dispersion. To chemically exfoliate the MoS2 to a monolayer level, solution-based Li-intercalation using n-butyl lithium was conducted. The resultant ce-MoS2 was solvent thermal treated in various solvents such as N-methyl-2-pyrrolidone (NMP). (b) Photographs showing the dispersion stability of 2H-MoS2 (annealed in NMP) in water after 10 months.

Previously, several studies have reported that the thermally driven phase transition from 1T to 2H phase was about 100% when the temperature was elevated up to 300 °C.[14,22,33] However, structural and chemical defects such as sulfur vacancies, triangular etch pits, and oxidation of MoS2 were inevitable at such high temperature[22,35−38] (about 300 °C). Moreover, it should be noted that the majority (about 80%) of the phase transition has already occurred at 150 °C.[14,22] Thus, in this study, the thermal treatment was conducted at a relatively low temperature (180 °C) to minimize the formation of structural and chemical defects and to prevent the boiling of the solvent. During the solvent thermal treatment at 180 °C for 5 h, we observed the structural phase transition from 1T (metal) to 2H (semiconductor) phase. In addition, when the solvent thermal treatment was conducted in NMP, the resultant 2H-MoS2 (MoS2–NMP) dispersion showed high stability in the aqueous solvent, as shown in Figure b.

To investigate the phase transition in MoS2, an X-ray photoelectron spectroscopy (XPS) analysis was conducted before and after the solvent thermal treatment. All of the XPS spectra were calibrated with reference to the C 1s peak at 284.5 eV to compensate for the charging effect. As depicted in Figure a, the XPS spectra of Mo 3d consist of three sets of peaks: (i) MoO3 peaks (yellow line), which can be deconvoluted into Mo6+ doublets 3d5/2 and 3d3/2 located at 232.6 and 235.3 eV, respectively; (ii) the Mo4+ 3d5/2 and Mo4+ 3d3/2 of 1T-MoS2 peaks (blue line) located at 228.2 and 231.3 eV, respectively, and (iii) the Mo4+ 3d5/2 and Mo4+ 3d3/2 of 2H-MoS2 peaks (red line) located at 229.0 and 232.1 eV, respectively.[14,22,39,40] The ce-MoS2 was identified to be metallic 1T phase (71.37%) partially mixed with oxidized MoS2 (11.35%) (Figure a), which is consistent with the findings of previous studies.[18,22] After the solvent thermal treatment in each polar solvent, the Mo 3d spectra were shifted toward the higher binding energy side because the Mo 3d peaks of 2H-MoS2 were intensified, as shown in Figure a. The ratio of each phase was calculated from the sum of the total area of each peak except the peak of the MoO3 phase, as presented in Figure b. (These data are also provided in Table S1 in the Supporting Information.) In addition, the O 1s spectrum also supports the quantitative analysis results of the MoO3 phase except the MoS2–NMP sample containing C=O groups, as shown in Figure S2.

Figure 2

Investigation of 1T to 2H phase transition. (a) XPS spectra of ce-MoS2 and 2H-MoS2 after the solvent thermal treatment in each polar solvent at 180 °C for 5 h. Deconvoluted red and blue lines represent the 2H-MoS2 and 1T-MoS2, respectively. The MoO3 peaks (yellow line) and S peaks (brown line) are also shown after deconvolution. (b) Calculated 2H phase and MoO3 ratio from Mo 3d XPS analysis results after the solvent thermal treatment in various solvents. (c, d) Transmission electron microscopy (TEM) images obtained before and after the solvent thermal treatment. (Mo atom—green dot, S atom—yellow dot). (e) Extinction spectra of ce-MoS2 and 2H-MoS2 after the solvent thermal treatment in each polar solvent. The A and B exciton peaks, which are located at 657.01 and 607.77 nm, respectively, originate from the monolayer MoS2.

For ce-MoS2, the 1T phase was dominant and the fraction of 2H phase was only about ∼30%. After the solvent thermal treatment, however, the 1T phase was restored to the 2H phase and the 2H phase ratio reached about 80% for all of the polar solvents used for the treatment. Similarly, the phase transition from 1T to 2H was also observed in the S 2p XPS spectra after the solvent thermal treatment, as depicted in Figure S3 in the Supporting Information. From the S 2p spectra, the 2H phase ratio was calculated to be approximately 85% for all of the polar solvents used (Table S1 in the Supporting Information), which is consistent with the calculation results based on the Mo XPS peaks.

Moreover, the Raman spectra of the ce-MoS2 and MoS2–NMP were compared, as shown in Figure S4, for a more in-depth characterization of the MoS2 nanosheets.[40] The Raman spectra from ce-MoS2 show several bands, J1, J2, J3, E1g, E2g, and A1g, indicating the 1T-phase MoS2, whereas only E2g and A1g peaks were observed for MoS2–NMP, confirming that it is the 2H phase. The spectrum for MoS2–NMP shows peaks at 385 and 407 cm–1. Also, from the difference (22.3 ± 1.49 cm–1) of E2g and A1g Raman peak positions, the average number of layers was calculated to be slightly over 2, which is consistent with the AFM measurement results shown in Figure S5.

Furthermore, the phase transition from 1T to 2H was identified by high-resolution scanning transmission electron microscopic (STEM) images, as depicted in Figure c,d. In the high-angle annular dark-field STEM image, the ce-MoS2 only reveals hexagonally oriented Mo atoms because the MoS2 was reconstructed to an ABC (S–Mo–S′) sequence stack and, as a result, the lattice points of S atoms were distorted.[41,42] After the solvent thermal treatment, the hexagonally arranged S atoms appeared between Mo atoms due to the restoration of the MoS2 structure to an ABA (S–Mo–S) sequence stack.[22,41−43]

Interestingly, when the solvent thermal treatment was conducted in NMP, the 2H phase ratio showed the highest value and the fraction of MoO3 was lowest (6.44%) among all of the samples, which is similar to the MoO3 fraction in natural bulk MoS2. This can be related to the fact that NMP is also a useful reduction agent for deoxygenating of other 2D materials such as graphene oxide due to its oxygen-scavenging property,[44] and thus these results may originate from the combination of thermal and chemical deoxygenating effects during the solvent thermal treatment. (This will be discussed in detail later.) The recovery from 1T to 2H phase by the solvent thermal treatment is further supported by the re-emergence of the A and B extinction peaks in the UV–vis absorption measurement results shown in Figure e. As presented in Figure e, no extinction peaks were observed in the extinction spectra of ce-MoS2 within the wavelength range of 550–750 nm due to the dominance of the 1T phase. After the solvent thermal treatment, however, the A and B extinction peaks appeared for the samples treated in all of the polar solvents. The A and B extinction peaks, which are located at 657.01 and 607.77 nm,[14,45] respectively, originated from the monolayer 2H-MoS2.

Notably, the extinction peaks of the sample annealed in NMP were almost identical to the A and B extinction peaks of pure monolayer MoS2, whereas those of other 2H-MoS2 were slightly red-shifted. (Individual peak positions are presented in Table S2 in the Supporting Information.)

Considering that the band gap of MoS2 decreases with increasing thickness of MoS2, these extinction data strongly suggest that the resultant 2H-MoS2 in EG, FA, DMSO, and PC were at least partially agglomerated and thickened during the solvent thermal treatment, whereas few or monolayer MoS2 flakes in NMP were preserved without agglomeration. In addition, to quantify the average number of atomic layers in the MoS2 flakes, an atomic force microscopy (AFM) analysis on the samples was conducted, as shown in Figure S5. MoS2–NMP sheets with an average diameter of 207.9 ± 82.7 nm can be clearly observed in the AFM images. A typical height of 1.2–1.4 nm was obtained using the AFM line profiles and the average number of layers in the flake samples was calculated to be 2.38 ± 0.92, confirming that mono- to few-layer MoS2–NMP was successfully preserved. These results are consistent with the Raman analysis results (Figure S4). Moreover, even after the long-term storage of MoS2–NMP for 10 months, the average diameter was maintained at 218.9 ± 85.7 nm (Figure S6). This is in contrast with the coagulation of MoS2 nanosheets reported previously.[46]

These observations imply the change in the surface chemistry of MoS2 during the solvent thermal treatment using NMP. As depicted in panel (a), the deconvoluted C 1s XPS spectra indicate the formation of C=O and C–N bonds in MoS2–NMP, which were not present in ce-MoS2. (The XPS spectra were acquired after fully washing the flakes with EtOH, acetone, and deionized (DI) water.) We also checked the N 1s spectra (Figure b), where the Mo4+ 3p3/2 and Mo6+ 3p3/2 peaks partially overlapped with the N 1s peak. The change in the XPS spectra indicates that, after the solvent thermal treatment in NMP, Mo–N bonding was formed and the fraction of Mo6+ 3p3/2 (from MoO3) decreased from 11.3 to 7.2%. This result is consistent with the XPS analysis results of the Mo 3d peaks shown in Figure a and indicates the reduction of MoS2 during the solvent thermal treatment in NMP. Due to the oxygen-scavenging property of NMP during the solvent thermal treatment, NMP can be partially oxidized or decomposed to organic compounds containing an amide group such as N-ethylacetamide, N-methylformamide, or acetamide,[46] which is implied by the color change of NMP from colorless to brown, as shown in Figure S7 in the Supporting Information. It should be noted that these species can be chemisorbed to the surface MoS2. Figure b shows that the N 1s XPS spectrum of the solvent-thermal-treated MoS2 in NMP has a noticeable shoulder peak at around 400 eV, which is known to be related to the Mo–N bond.[47−49]

Figure 3

Surface analysis of MoS2–NMP. High-resolution XPS spectra for (a) C 1s; (b) N 1s. Energy-dispersive X-ray spectroscopy (EDS) mapping of Mo, S, C, N, and O on the surface of (c) ce-MoS2 and (d) MoS2–NMP. Scale bar: 500 nm.

Surface analysis of n class="Chemical">MoS2–NMP. High-resolution XPS spectra for (a) C 1s; (b) N 1s. Energy-dispersive X-ray spectroscopy (EDS) mapping of Mo, S, C, N, and O on the surface of (c) ce-MoS2 and (d) MoS2–NMP. Scale bar: 500 nm.

Furthermore, we also investigated the C 1s XPS spectra for the samples obtained with other polar solvents (EG, FA, PC, and DMSO). As shown in Figure S8, the overall C 1s peaks were similar to that of ce-MoS2. This result means that the surface functionalization of MoS2 did not take place in other solvents, which may explain the agglomeration of MoS2 during the solvent thermal treatment. In addition to the XPS results, we also investigated the TEM images and the corresponding energy-dispersive X-ray spectroscopy (EDS) mapping for Mo, S, and N elements on ce-MoS2 and MoS2–NMP. As shown in Figure c, the EDS mapping results of ce-MoS2 show the presence of only Mo and S, whereas the EDS results of MoS2–NMP additionally present well-distributed C, O, and N elements (Figure d). This result also supports the finding that the MoS2–NMP was uniformly functionalized after the solvent thermal treatment in NMP, although additional agents were not added to the solvent.

To further investigate the surface functionalization of MoS2–NMP, an attenuated total reflectance Fourier transform infrared (ATR-FTIR) analysis was performed and the results are shown in Figure a. After the solvent thermal treatment in NMP, we observed the appearance of strong peaks at 3423, 2917, 2846, 1702, 1433, and 1374 cm–1, which represent the N–H, C–H, C=O, and C–N bonds, respectively. As shown in Figure S9, these peaks and the overall FTIR spectra of MoS2–NMP are well matched with those of acetamide.

Figure 4

Characterization of functionalized MoS2. (a) FTIR spectra of the ce-MoS2 and MoS2–NMP. Strong new peaks appeared at 1433, 1708, and 2924 cm–1 after the solvent thermal treatment. (b) NMR analysis data. 13C NMR cross-polarization magic-angle-spinning (CP-MAS) spectra of acetamide (black), pristine 2H-MoS2 (red), and 2H-MoS2–NMP (blue). The green point designates the C signals corresponding to carbonyls and the blue point indicates α-C signals. (c) Thermal gravimetric analysis (TGA) results and (d) X-ray diffraction (XRD) patterns of ce-MoS2 and MoS2–NMP.

The surface-attached molecule was also characterized using the solid-state 13C cross-polarization magic-angle-spinning nuclear magnetic resonance (CP-MAS NMR) spectroscopy. As shown in Figure b, the NMR spectra for pure acetamide (black) show characteristic chemical shifts (δ) corresponding to carbonyl and aliphatic carbons (α-C in this case) at 179.9 and 25.7 ppm, which are marked with green and blue points, respectively. In the case of MoS2–NMP (blue), the peaks corresponding to both α-C (blue) at 173.5 ppm and carbonyl C (green) at 34.5 ppm, which are comparable with those of pure acetamide. The broad peak at 130.3 ppm was also observed for the nonfunctionalized bulk 2H-MoS2 (red), suggesting the peak is not related to the surface-functionalized layer. Overall, these NMR, FTIR, TEM, and XPS analysis results confirm that the ce-MoS2 was restored to 2H-MoS2 and simultaneously functionalized with acetamide during the solvent thermal treatment in NMP.

The amount of absorbed molecules was characterized by a thermogravimetric analysis (TGA) after completely drying the samples. Figure c shows the TGA weight loss curves of ce-MoS2 and MoS2–NMP measured from room temperature to 700 °C under an air atmosphere. The weight loss until 300 °C, which is caused by the evaporation of the adsorbed water molecules, was measured to be 2.15 and 3.68% for ce-MoS2 and MoS2–NMP, respectively. In the temperature range from 300 to 500 °C, a major weight loss of 14.35% occurred for ce-MoS2, which can be explained by the replacement of S by O due to the oxidation of MoS2 to MoO3. Overall, a comparison of the TGA curve of ce-MoS2 with that of MoS2–NMP indicates that the additional weight loss of MoS2–NMP of about 10% is due to the decomposition of acetamide attached on MoS2. In addition, the X-ray diffraction (XRD) analyses were conducted to confirm the interlayer spacing (Figure d). In ce-MoS2, we observed the sharp peak at 2θ = 14.06°, which corresponds to (002) planes of the Li-intercalated MoS2. In the case of MoS2–NMP, however, the (002) peak was found to be shifted to a significantly lower angle at 2θ = 9.04°. This result reveals that the expansion of the lattice spacing along the c axis from 0.629 nm (ce-MoS2) to 0.974 nm (MoS2–NMP) is due to acetamide molecules attached on MoS2.[13]

We investigated the long-term stability of the various aqueous dispersions of solvent-thermal-treated MoS2. The concentration of MoS2 in the dispersions was estimated using the Beer–Lambert law, A = αCl, where A is the measured absorbance at a particular wavelength, α is the extinction coefficient, C is the concentration of the dispersion, and l is the thickness of the solution layer.[10,19] We first obtained the extinction coefficient (at λ = 672 nm) of each sample by fitting the slope of A/l as a function of MoS2 concentration (Figure S10 in the Supporting Information). As shown in Figure S10, similar concentrations of MoS2 dispersions with 0.75 mg/mL were then prepared, and the absorbance of MoS2 aqueous dispersion was monitored regularly during 50 days. To enhance the accuracy of this measurement, all of the MoS2 dispersions were diluted 50-fold before the measurement to minimize the light-scattering effect. As shown in Figure a, the MoS2–NMP aqueous dispersion showed a superior long-term dispersion stability compared with other dispersion samples. However, during the early-stage storage up to 1 week, ce-MoS2 and the solvent-thermal-treated MoS2 in EG (MoS2–EG) presented a stable dispersion state, the absorbance of ce-MoS2 in the supernatant was sharply reduced and reached almost 0 within 20 days, which can be attributed to the fact that the negative surface charge resulting from the Li-intercalation was gradually transferred to the water molecules.[50] Other aqueous dispersions obtained by the solvent thermal treatment in EG, PC, FA, and DMSO also showed poor dispersion stability, and a significant portion of MoS2 was precipitated before 20 days, whereas the aqueous dispersions containing MoS2–NMP were still stable even after 50 days. The absorbance of dispersed MoS2–NMP (0.2337) is about 39.0 and 16.5 times higher than that of the dispersed ce-MoS2 (0.006) and MoS2–EG (0.0142) after 50 days, respectively. Additional analysis data regarding the dispersion properties of the samples are provided in Figure S10 and Table S3. A photograph of highly stable MoS2–NMP is depicted in Figure b.

Figure 5

Long-term stability test of MoS2 samples in aqueous dispersion. (a) Absorbance of 2H-MoS2 in the supernatant of each dispersion as a function of storage time. The aqueous dispersion of MoS2–NMP shows a superior long-term dispersion stability compared with the other dispersions due to the hydrophilic acetamide surface functionalization. (b) Photographs of MoS2 dispersions as prepared and after 50 days.

The outstanding dispersion stability of the MoS2–NMP nanosheets is consistent with the measured ζ potential values (Table S4 in the Supporting Information). The ζ potential of the chemically exfoliated nanosheets of MoS2 containing a large fraction of 1T phase was −43.1 mV due to excess surface charge.[33,50,51] In general, the surface of bare 2H-MoS2 is not charged, and therefore it cannot be stabilized in common organic solvents because of a low ζ potential (−20.2 mV).[24,33] However, the ζ potential (−41.2 mV) of the MoS2–NMP functionalized with polar molecules was sufficiently high and can explain the good dispersion stability in water.[51,52]

Considering the dispersion stability in an aqueous solvent (Figure ), it is expected that the resultant MoS2 after the solvent thermal treatment in polar solvents will have a different surface hydrophilicity. We compared the water contact angle (WCA) of ce-MoS2, MoS2–EG (without functional group), and MoS2–NMP (with acetamide functional group). As shown in Figure S11a, the WCAs were 34.63° for ce-MoS2 and 71.96° for MoS2–EG. These WCAs of ce-MoS2 and MoS2–EG are comparable with the recently reported values of Li-intercalated 1T-MoS2 and mechanically exfoliated 2H-MoS2 from bulk.[18,53] The substantially lower WCA (20.21°) of MoS2–NMP than that of other samples indicates that the MoS2–NMP has a much higher hydrophilicity, which is consistent with the outstanding dispersion stability in water.

The greater water affinity of MoS2–NMP suggests its potential application to humidity sensor devices. We fabricated a resistance-type humidity sensor using an interdigitated electrode with 4 μm channel width by drop-casting of MoS2–NMP (Figure a) to investigate the humidity-sensing properties (see the Experimental Section for details). As shown in Figure b, we observed a remarkably different response behavior depending on the functionalization at 80 relative humidity (RH)%. The response of the sensor was calculated by RH/Ro, where RH is the resistance value in the presence of humidity and Ro is the resistance value when dry air was passed through the chamber. ce-MoS2 and MoS2–EG show a positive response (increase in resistance), and the response of MoS2–EG was higher than that of ce-MoS2 due to the transition from 1T (metal) to 2H (semiconductor) phase. In contrast, the response of MoS2–NMP shows a negative response (decrease in resistance) and its response was much higher than that of other devices. This result can be attributed to the difference in water affinity and the hydrogen bond formation between the carbonyl group and water molecules (Figure a, right), as already reported.[54] The sensor performance was further characterized by adjusting the humidity from 25 to 95 RH%, and the corresponding responses of the three sensors are shown in Figure c. Different sensing behaviors depending on functionalization were also detected over the whole range of humidity. The response of ce-MoS2 and MoS2–EG increased linearly with increasing humidity levels. However, MoS2–NMP shows a more rapid increase in response with increasing humidity levels due to hydrophilic functionalization. The sensitivity of MoS2–NMP was 5 times higher than that of ce-MoS2 at 95 RH% (Figure S11b in the Supporting Information). These results indicate that an appropriate modification of MoS2 surface combined with a phase transition can improve the device performance to a great extent.

Figure 6

Fabrication and characterization of MoS2 humidity sensor. (a) Illustrations of the humidity sensor device (left) and the molecular interaction between carbonyl group and water molecules (right). (b) Resistance response of the humidity sensor devices measured at 80 relative humidity (RH)%. (c) Resistance vs relative humidity for humidity sensors. Fct-2H MoS2 indicates the MoS2–NMP.

Conclusions

In summary, we have demonstrated a practical method for a high-yield phase transition from 1T to 2H phase and simultaneous functionalization of MoS2 via the solvent thermal treatment in a polar solvent. We found that the solvent thermal treatment in a polar solvent at a mild temperature (180 °C) is appropriate for restoration from 1T (metal) to 2H (semiconductor) phase. In particular, solvent thermal treatment of MoS2 in NMP was simultaneously and uniformly functionalized with acetamide—a thermal decomposition product of NMP, which was confirmed on the basis of XPS, FTIR, NMR, TGA, and XRD analysis results. This surface functionalization by hydrophilic molecules can provide an excellent dispersion stability of 2H-MoS2 in water for more than 10 months. Finally, the humidity sensor fabricated using MoS2–NMP showed an outstanding sensing response in a wide relative humidity range compared with nonfunctionalized samples due to the water affinity of acetamide molecules. We expect that this strategy can also be applied to other 2D materials to precisely control their phases and surface properties, and thereby tailor their fundamental properties and extend the application areas.

Experimental Section

Materials

MoS2 powder (∼6 μm) was purchased from 2D Semiconductor, Inc. A 1.6 M n class="Chemical">n-butyl lithium solution, ethylene glycol (EG), formamide (FA), propylene carbonate (PC), N-methyl-2-pyrrolidone (NMP), dimethyl sulfoxide (DMSO), tetraethylene glycol dimethyl ether (TGDE), and hexamethylphosphoramide (HMPA) were purchased from Sigma-Aldrich. All of the chemicals were used as purchased without any further purification.

Preparation of MoS2 Nanosheets

MoS2 nanosheets were prepared using the lithium-intercalation/exfoliation method. First, the MoS2 powders were immersed in 1.6 M n-butyl lithium solution (0.1 g/mL) and stirred for 48 h in a N2-filled glovebox at 60 °C. The Li-intercalated MoS2 (LiMoS2) was filtered and washed repeatedly with hexane to remove excess lithium and organic residues. The collected LiMoS2 powder was then dispersed in deionized (DI) water (18.3 MΩ). Upon contact with water, the evolution of gas was observed and the MoS2 powder formed a highly opaque suspension. Subsequently, LiMoS2 was exfoliated via bath ultrasonication (200 W, JAC Ultrasonics 2010; KODO Technical Research Co., Ltd.) for 90 min. After exfoliation, the suspension was centrifuged several times (five to six times) at 14 000 rpm for 1 h with DI water to remove LiOH and unexfoliated MoS2. This washing process was completed until the pH of the solution reached a neutral value.

Solvent Thermal Treatment

The purified wet sediment was dispersed in several polar solvents with a high boiling point such as ethylene glycol, formamide, propylene carbonate, dimethyl sulfoxide, tetraethylene glycol dimethyl ether, hexamethylphosphoramide, and N-methyl-2-pyrrolidone. MoS2 dispersions in polar solvents were sonicated for 30 min to produce a homogeneous solution before the solvent thermal treatment. After sonication, the MoS2 dispersion was transferred to a two-neck flask and refluxed at 300 rpm under an Ar flow. A heating mantle was used to heat the MoS2 dispersion up to 180 °C for 15 min and the temperature was maintained for 5 h without boiling over. The resultant suspension was filtered over a poly(tetrafluoroethylene) membrane (0.2 μm, JGWP04700; Millipore) and rinsed several times with ethanol, acetone, and DI water. The obtained product was redispersed into the aqueous solvent via bath sonication for 30 min. Finally, the MoS2 dispersion was centrifuged with a low rpm (2000 rpm, 5 min; 1000 rpm, 10 min) to remove the agglomeration of MoS2 during the solvent thermal treatment.

Characterization of Phase Transition and Functionalization of MoS2

For confirming the phase transition, an X-ray photoelectron spectroscopy (XPS) analysis was conducted using a multipurpose X-ray photoelectron spectrometer (Sigma Probe; Thermo VG Scientific). The absorbance of the MoS2 dispersion was measured by a UV–vis spectrophotometer (Optizen POP; Mecasys). The FTIR spectra were collected with an attenuated total reflectance FT-infrared spectroscopy (ATR-FTIR, ALPHA-P; Bruker) in the range from 500 to 4000 cm–1. High-resolution-TEM images and EDS mapping were obtained using a double Cs-corrected Titan TEM (Titan G2 60–300; FEI) with 80 kV accelerating voltage and a 200 kV TEM (JEM-2100F; JEOL), respectively. The X-ray diffraction (XRD) measurement was carried out using a multipurpose thin-film X-ray diffractometer (D/Max 2500; Rigaku) and weight loss was measured by thermogravimetric analysis (TGA, Setsys 16/18; Setaram Instrumentation). The AFM characterization was conducted in a noncontact mode using the atomic force microscope (AFM) from Asylum Research Cypher. The solid-state 13C (100 MHz) NMR spectra were measured using an Agilent 9.4 T (400 MHz 1H NMR frequency) spectrometer with a 1.6 mm magic-angle spinning (MAS) probe. All of the reported chemical shifts were referenced to adamantane at 298 K. For cross-polarization (CP) MAS, 13C{1H} NMR spectra were attained with 8 h of signal averaging at 20 kHz MAS, with a constant 50 kHz radio frequency field strength applied to the 13C NMR channel during the 1H–13C NMR CP (contact time of 3 ms).

Stability Test of MoS2 Aqueous Dispersion

We first fabricated each MoS2 dispersion with the saturation state and then the absorbance of each MoS2 dispersion was measured with a UV–vis spectrophotometer (Optizen POP; Mecasys) with a 1 cm path cuvette after 24 h to exclude the aggregated MoS2. To obtain the concentration of each MoS2 dispersion, we used the Lambert–Beer law. The concentration changes were monitored during 50 days. To enhance the accuracy, all of the MoS2 dispersions were measured after dilution to obtain the absorbance below 1. ζ Potentials of the samples were measured using Malvern Instrument zeta sizer Nano-ZS at pH ≈ 7.4. The samples were prepared as follows: the representative solid sample (1 mg) was first dispersed in DI water (10 mL) and sonicated for 30 min. One milliliter of that solution was taken out and put into the ζ potential cell for the measurement of ζ potential. The values are reported as an average of three measurements.

Fabrication and Measurement of Humidity Sensor

Water contact angles (WCAs) were obtained by a dynamic contact angle meter (Phoenix 300; SEO) equipped with a digital camera. To fabricate a interdigitated electrode with 4 μm width on a Si wafer, we used a typical photolithography with AZ5142E photoresist and a mask aligner (MDA-8000B). After the development of the patterns, Au with a thickness of 100 nm was deposited as a source and drained by thermal evaporation. The MoS2 dispersion was then drop-cast on the electrode patterns. The electrical characteristics of the sensors were measured by a digital multimeter (34970A; Keysight Technologies). The humidity-sensing properties were examined by controlling the relative humidity from 25 to 95%. To modify the atmosphere inside the chamber, the mixtures of dry air and water vapor with different mixing ratios were used. In addition, a commercial humidity sensor (testo 608-h2) was installed inside the chamber to obtain the real-time humidity data inside the chamber. The sensing experiments were carried out at an ambient temperature of 25 °C and the total gas flow rate was fixed at 500 cm3/min.