Rapid and Large-Area Visualization of Grain Boundaries in MoS2 on SiO2 Using Vapor Hydrofluoric Acid.

Grain boundaries in two-dimensional (2D) material layers have an impact on their electrical, optoelectronic, and mechanical properties. Therefore, the availability of simple large-area characterization approaches that can directly visualize grains and grain boundaries in 2D materials such as molybdenum disulfide (MoS2) is critical. Previous approaches for visualizing grains and grain boundaries in MoS2 are typically based on atomic resolution microscopy or optical imaging techniques (i.e., Raman spectroscopy or photoluminescence), which are complex or limited to the characterization of small, micrometer-sized areas. Here, we show a simple approach for an efficient large-area visualization of the grain boundaries in continuous chemical vapor-deposited films and domains of MoS2 that are grown on a silicon dioxide (SiO2) substrate. In our approach, the MoS2 layer on a SiO2/Si substrate is exposed to vapor hydrofluoric acid (VHF), resulting in the differential etching of SiO2 at the MoS2 grain boundaries and SiO2 underneath the MoS2 grains as a result of VHF diffusing through the defects in the MoS2 layer at the grain boundaries. The location of the grain boundaries can be seen by the resulting SiO2 pattern using optical microscopy, scanning electron microscopy, or Raman spectroscopy. This method allows for a simple and rapid evaluation of grain sizes in 2D material films over large areas, thereby potentially facilitating the optimization of synthesis processes and advancing applications of 2D materials in science and technology.

Introduction

Two-dimensional (2D) materials such as molybdenum disulfide (MoS2) have been studied for use in potential applications[1−3] such as transistors,[4] light emitters,[5] photodetectors,[6] modulators,[7] pressure sensors,[8] resonators,[9] biosensors,[10] gas sensing,[11] photocatalysis,[12] and electrochemical applications.[13] For using these 2D materials in device applications, it is critically important to realize the industrial-scale and reliable synthesis of high-quality monolayers of the 2D materials at low cost. Chemical vapor deposition (CVD) is a potential large-scale 2D material synthesis approach suitable for industrial applications and one of the most developed approaches for the preparation of large-area MoS2 of good quality.[14−17] For the development and optimization of CVD-grown single-layer MoS2, it is important to characterize the properties of the grown MoS2 layers that typically consist of irregularly shaped individual grains that connect the adjacent grains[18−21] through grain boundaries. As a result, the grain sizes and grain boundaries of the domains or continuous films of MoS2 have an important impact on its electrical,[18,22−24] optical,[18] optoelectronic,[25,26] mechanical,[27,28] and chemical properties[29] as well as on the characteristics of devices made of MoS2. Furthermore, the properties of grain boundaries in MoS2 might be beneficially used in specific applications by controlled defects engineering.[30−32] For the above reasons, fast and simple methods to directly observe the large-area distribution of grains and grain boundaries in MoS2 are of increasing importance.

Grains and grain boundaries in MoS2 films can be characterized by atomic resolution using transmittance electron microscopy (TEM)[18−20] and scanning tunneling microscopy (STM),[32,33] providing detailed information about the crystal structure of the grains and the grain boundaries. The MoS2 grain boundaries can also be identified using atomic force microscopy (AFM) by decorating a self-assembled octadecylphosphonic acid monolayer on the MoS2 surface.[34] However, these techniques are time-consuming, require complex sample preparation procedures, and/or are limited to the characterization of very small areas. Another approach to observe grain boundaries in MoS2 is the use of nonlinear optics.[35−38] For instance, the grain boundaries between the adjacent MoS2 grains can be distinguished by stacking MoS2 bilayers and using photoluminescence imaging based on second harmonic generation.[37] However, photoluminescence imaging is generally limited to MoS2 domains that feature a large rotation of the crystal axis as compared to the neighboring grains, requires sophisticated optical systems, and is typically slow.[39] Compared to photoluminescence imaging, a faster approach for visualizing grain boundaries in CVD-grown MoS2 is multiphoton microscopy based on third-harmonic generation, which is also independent of the degree of crystal axis rotation.[39] Yet, this approach still requires a relatively sophisticated optical system and therefore is not easily accessible. Grain boundaries in CVD-grown MoS2 layers can also be visualized by oxidizing MoS2 using UV irradiation in a moisture-rich environment and subsequently imaging the layer with scanning electron microscopy (SEM) or AFM.[40] However, with this approach, additional visible oxidized line defects within the MoS2 grains were produced alongside the grain boundaries because of easy oxidation of MoS2.[40] The grains and grain boundaries of large MoS2 layers were also visualized by depositing nematic liquid crystals on the MoS2 layers in combination with polarized optical microscopy[41] or by visualizing the differential diffusion of gold on the surface of MoS2 between the grain boundaries and the inner grain areas after depositing gold on the MoS2 surface using optical microscopy.[42] However, these techniques require elaborate and controlled preparation of the MoS2 samples.

Here, we present a simple and rapid method for visualizing grain boundaries in large areas of CVD-grown single-layer MoS2 on a SiO2 surface using SEM, optical microscopy, or Raman spectroscopy. Similar to the previously reported method for observing grain boundaries in CVD-grown graphene placed on a SiO2 surface using vapor hydrofluoric acid (VHF) exposure,[43] in the proposed method, we first expose the MoS2 layer on the SiO2 surface to VHF, which causes VHF molecules to diffuse through the defects in the lattice structure of the MoS2 grain boundaries. The diffusion of the VHF through these defects results in etching of SiO2 underneath MoS2, with a difference between the etching speed of SiO2 directly at the grain boundaries and the etching speed of SiO2 in the areas below the grains away from the grain boundaries. The resulting etch pattern in the SiO2 layer along the MoS2 grain boundaries is then visible and can be imaged using optical microscopy, SEM, or Raman spectroscopy. Because the MoS2 and the underlying SiO2 layers are exposed to VHF and etched to some extent in our method, this is an invasive approach. SiO2 is one of the most commonly used growth substrates for large-area CVD-grown MoS2, and thus, our method will be useful in the development, characterization, and optimization of large-area MoS2 synthesis processes.

Experimental Section

In all experiments, we used CVD-grown single-layer MoS2 on a SiO2/Si substrate that was bought from 2D Semiconductors (USA), in which the SiO2 layer was 285 nm thick. MoS2 was composed of domains with continuous films and individual grains with visible grain boundaries that were not fully stitched to the adjacent MoS2 grains. For VHF etching, we exposed the SiO2/Si chips with the MoS2 films to VHF that was evaporated from a liquid HF solution with a HF concentration of 25%. The VHF reacts with SiO2 underlying the grain boundaries of the MoS2 layer by penetrating the defects of the grain boundaries. To control the amount of H2O that is present at the substrate surfaces, the substrate temperature was kept at 40 °C.[44,45] For details on the experimental setup of the VHF chamber, our previous publication on the visualization of grain boundaries in graphene could be referred.[43] The VHF etching of SiO2 involves reactions and 2.

Before and after exposing MoS2 on the SiO2/Si substrate to VHF for different times, we used optical microscopy (OLYMPUS, BX60, Japan) and SEM (Gemini, Zeiss, Ultra 55, Germany) to image the morphologies of the MoS2/SiO2 surfaces on the Si substrates with different magnifications, respectively. We used a Raman spectrometer (alpha300 R Microscope, WITec, Germany) to evaluate the quality of the MoS2 films on the SiO2 surface. For the Raman scans with areas of 15 μm × 15 μm (100 × 100 points per map), we used a laser with a wavelength of 532 nm, a laser power of 1.5 mW, and an integration time of 0.3 s. For the Raman scans with the areas of 5 μm × 5 μm (25 × 25 points per map), we used a laser with a wavelength of 532 nm, a laser power of 1 mW, and an integration time of 3 s. To analyze the topography of the MoS2/SiO2 surfaces before and after exposing MoS2 on the SiO2/Si substrates to VHF for different times, we used an AFM tool (Dimension Icon, Bruker) with a cantilever (Olympus AC240TM) and an AFM tip (tip radius = 15 nm) in tapping mode.

Results and Discussion

In our experiments, we used CVD-grown MoS2 on a 285 nm thick SiO2 layer on a silicon (Si) substrate, which was purchased from 2D Semiconductors Inc. (USA). The CVD-grown MoS2 was composed of domains with continuous MoS2 films without visible grain boundaries (Figure S1a), as well as domains in which there were presumed visible grain boundaries between the adjacent MoS2 grains that were incompletely stitched to each other (Figure S1b). First, we evaluated and confirmed the quality of the pristine MoS2 samples using Raman spectroscopy (Figure S2). The typical E2g1 band position (385.4 cm–1) and A1g band position (406 cm–1) depict the single-layer MoS2 (Figure S2e). Furthermore, we used Raman spectroscopy to indicate the presence of grain boundaries between the adjacent MoS2 grains that were incompletely stitched to each other (Figure S2b–d). In these measurements, we also found E2g1 and A1g bands in the Raman spectrum at the grain boundaries (Figure S2d,e), which can be attributed to the fact that the width of a MoS2 grain boundary is smaller than the diameter of the Raman laser spot (∼400 nm), as indicated in Figure S2d by the red and blue solid circles. Thus, the laser spot always overlaps either one or both sides of the adjacent MoS2 grains when exciting the boundary region, resulting in a nonzero Raman signal of MoS2. The surface topography of a sample of pristine MoS2 characterized by AFM further confirms that the MoS2 grains were incompletely stitched to each other (Figure S3).

To evaluate the proposed approach for visualizing grain boundaries in a single MoS2 layer on a SiO2 substrate, we exposed a sample with a large continuous MoS2 film without visible grain boundaries (Figure a) to VHF at 40 °C for 120 s (see Experimental Section). After VHF exposure, line patterns became visible and distinct within the continuous film in both SEM (Figure b) and optical microscopy images (Figure c,d). The SEM (Figure a,b) and optical microscopy images (Figure c,d) were taken at the same position of the same chip. In the images, it can be seen that the areas enclosed by a line pattern are of the order of 1 μm to several μm. We tentatively assign these line patterns to the grain boundaries in the MoS2 film. To further explore this, we selected a chip area in which the MoS2 film on the SiO2 surface contained distinguishable individual grains and grain boundaries prior to VHF exposure (Figure e), caused by incomplete growth and stitching of the adjacent MoS2 grains. After exposing this sample to VHF at 40 °C for 120 s, we found that the same type of line pattern in the SiO2 layer appeared at the locations of the previously visible grain boundaries (Figure f–h). Also, here, the SEM and optical microscopy images in Figure e–h were taken from the same position of the chip. Furthermore, we investigated a chip area where a continuous MoS2 film was located directly next to a MoS2 film, in which grain boundaries and individual grains were visible, caused by incomplete growth and stitching of the adjacent MoS2 grains. After exposing this sample to VHF at 40 °C for 120 s, we again observed that the same type of line pattern in the SiO2 layer appeared at the locations of the previously visible grain boundaries, with a comparable pattern appearing in the area of the continuous film (Figure S4a,b). Interestingly, there are thinner line patterns within the large-area grain boundaries of the MoS2 sample after exposure to VHF for 120 s, indicating that the visible larger grains are composed of several small grains (Figure S4). It should be noted that line defects other than grain boundaries, such as wrinkles in 2D materials, are easily introduced when a 2D material is transferred from its original growth substrate to a new target substrate. In contrast, the samples in our experiments consist of single-layer MoS2 that is directly grown on the SiO2 surface of the Si substrate by CVD, thus avoiding the transfer of the 2D material. Therefore, the MoS2 samples used in our experiments most likely do contain only few or no wrinkles that may interfere with the visualization of the grain boundaries in MoS2 by VHF exposure.

Figure 1

SEM and optical microscopy observation of the top view of a single layer of CVD-grown MoS2 on a SiO2 surface before and after exposure to VHF for 120 s. (a) SEM image of a continuous CVD-grown MoS2 film on a SiO2 substrate without visible grain boundaries or line patterns. (b) SEM image, (c) optical image, and (d) optical dark-field image of MoS2 of the same area as (a) after exposure to VHF for 120 s. After exposure to VHF for 120 s, line patterns are visible within the continuous MoS2 film. (e) SEM image of CVD-grown MoS2 on a SiO2 substrate, with visible MoS2 grain boundaries between the adjacent grains that were not completely stitched to each other during CVD growth. (f) SEM image, (g) optical image, and (h) optical dark-field image of the same position as (e) after exposure to VHF for 120 s. After exposure to VHF for 120 s, the previously visible grain boundaries in the MoS2 film get more pronounced, indicating that the VHF interaction and resulting line patterns occur at the MoS2 grain boundaries.

To explore the impact of the VHF exposure time on the resulting line patterns in both the continuous CVD-grown MoS2 films and individual domains placed on a SiO2 substrate, we exposed such samples to VHF at 40 °C for different times, that is, 30, 60, and 120 s. We observed that the line patterns appear increasingly pronounced with the increasing exposure time to VHF (Figures and 3). We also observed that the patterns did not change in structure for the different evaluated VHF exposure times (Figures and 3). Generally, when the VHF exposure time was increased from 30 to 60 and to 120 s, thin line patterns appeared inside the large MoS2 domains and became increasingly distinct when the VHF exposure time was increased to 120 s (Figures –3 and S4). This indicates that a longer time (i.e., 120 s) of exposure to VHF reveals more information about the grain boundaries in MoS2 than shorter VHF exposure time (i.e., 30 and 60 s). This is in agreement with the observation of grain boundaries in CVD-grown graphene on SiO2/Si substrates after exposure to VHF.[43]

Figure 2

SEM images with the top view of CVD-grown MoS2 on SiO2 before and after exposure to VHF for 60 s. (a,b) SEM images of a chip area with a continuous MoS2 film without visible grain boundaries, before (a) and after (b) exposure to VHF for 60 s. (c,d) SEM images of a chip containing both an area with a continuous MoS2 film in which the grain boundaries are not visible and an area in which individual MoS2 grains are visible, before (c) and after (d) exposure to VHF for 60 s. (e,f) SEM images of a sample in which the MoS2 grains were not completely stitched to each other during CVD growth, before (e) and after (f) exposure to VHF for 60 s.

Figure 3

SEM images with the top view of CVD-grown MoS2 on SiO2 before and after exposure to VHF for 30 s. (a,b) SEM images of a sample in which the MoS2 grains were not completely stitched to each other during CVD growth, before (a) and after (b) exposure to VHF for 30 s.

The mechanism for observing grain boundaries in CVD-grown graphene on SiO2/Si substrates after exposure to VHF is based on the diffusion of VHF through the defects in the graphene lattice at the grain boundaries and the different speeds of VHF etching of SiO2 that is close to the graphene grain boundaries and that below the graphene grains.[43] In order to verify that the mechanism for the appearance of the line patterns in the SiO2 surface covered by the MoS2 films is the same as described above, we used AFM to image the surface topography of our samples and characterized the evolution of the MoS2 and SiO2 surface topography with increasing VHF exposure times (Figure ). Before exposing our samples to VHF, we did not observe significant topographical features in the MoS2 film on the SiO2 surface (Figure a,b). The high features appearing in white in Figure a might be associated with the multilayered growth or the formation of by-products occurring mainly at the grain boundaries during the synthesis of CVD-grown MoS2, which has been reported before.[46] After exposure to VHF for 30, 60, and 120 s, the surface topography near the assumed MoS2 grain boundaries progressively increased, with the line patterns being elevated from the other areas by about 8–10 nm (Figure c,d), 30–35 nm (Figure e,f), and 50–58 nm (Figure g,h), respectively. At the same time, the less pronounced lines within the initially larger domains were elevated from the other areas by about 15–20 nm after exposure to VHF for 60 s (Figure e,f). The three-dimensional (3D) representation of the AFM data, with the surface topography of the MoS2 samples after exposure to VHF for 30, 60, and 120 s, further illustrate that the line pattern is elevated with respect to the areas surrounded by the line pattern (Figure a–d). These results conclusively confirm that the VHF exposure of a MoS2 film on a SiO2 surface results in differential etching of SiO2. This can be explained by an increased etch rate of SiO2 underneath the MoS2 crystallites where liquid water with dissolved HF can accumulate, as compared to the SiO2 etch rate at the MoS2 grain boundaries where SiO2 is directly exposed to VHF and where liquid water does not accumulate (Figure e). Specifically, the net reaction of etching of SiO2 with VHF results in an excess of H2O molecules that get trapped underneath the MoS2 grains and accumulate in a water reservoir along with the easily water-dissolvable HF.[43] In contrast, in the areas along the grain boundaries, the excess water can evaporate through the grain boundaries, and thus SiO2 in these areas is directly exposed to VHF, resulting in a significantly lower SiO2 etch rate along nanoporous grain boundaries. This mechanism results in the distinct line pattern of the surface topography along the nanoporous grain boundaries in the MoS2 films that can be visualized by SEM or optical microscopy.

Figure 4

AFM characterization of a single layer of CVD-grown MoS2 on a SiO2 surface before and after exposure to VHF for 30, 60, and 120 s. (a) AFM image, with the surface topography of a MoS2 film on a SiO2 substrate before exposure to VHF. (b) Topographical scan of the area in (a). The corresponding data after the exposure of the sample to VHF for 30, 60, and 120 s are shown in (c,d), (e,f), and (g,h) respectively.

Figure 5

3D AFM views of the surface topography of a CVD-grown single-layer MoS2 film on a SiO2 surface before and after exposure to VHF for 30, 60, and 120 s as well as the schematics of etching processes during the exposure to VHF. (a–d) 3D views of the surface topography of Figure a–d. (e) Schematics of the reaction process when MoS2 on a SiO2 surface is exposed to VHF.

To evaluate the impact of VHF exposure on the quality of the MoS2 films on a SiO2 substrate, we exposed MoS2 films with presumed visible grain boundaries between the adjacent MoS2 grains to VHF for different exposure times, that is, 30, 60, and 120 s, and thereafter performed scanning micro-Raman spectroscopy on the samples. After exposing MoS2 to VHF for 30 s (Figure a), the Raman spectroscopy maps of the intensities of the E2g1 and A1g modes of MoS2 are still relatively strong (Figure a2,a3) with clear E2g1 and A1g band intensities (Figure a4,a5). This indicates that MoS2 is of relative high quality even after exposure to VHF for 30 s. As we increased the VHF exposure time to 60 s (Figure b), the Raman spectroscopy maps of the intensities of the E2g1 and A1g modes of MoS2 became weaker (Figure b2,b3), with significantly decreased E2g1 and A1g bands (Figure b4,b5). These results indicate that MoS2 was damaged to a larger extent during the exposure to VHF for 60 s. As expected, when we further increased the VHF exposure time to 120 s (Figure c), the Raman spectroscopy intensity maps of the E2g1 and A1g modes of MoS2 substantially weakened (Figure c2,c3), with very weak E2g1 and A1g bands (Figure c4,c5). This indicates that MoS2 was substantially damaged during the exposure to VHF for 120 s. Moreover, it can be seen from Figure a5–c5 that the E2g1 and A1g bands of MoS2 became weaker with increasing VHF exposure times, both within the grains and at the grain boundaries. This further indicates that long VHF exposure and more etching of SiO2 underneath the MoS2 film degrade the quality of MoS2. As an additional reference, we characterized samples with continuous MoS2 films using Raman spectroscopy before and after exposing them to VHF for 30, 60, and 120 s, respectively, and we found similar results (Figures S5–S8). Our Raman spectroscopy characterization shows that the exposure to VHF of MoS2 on a SiO2 surface for short times (i.e., 30 s) only marginally affects MoS2, whereas the exposure to VHF of MoS2 placed on SiO2 for longer times (i.e. 120 s) significantly affects MoS2.

Figure 6

Raman characterization of CVD-grown MoS2 on a SiO2 surface after exposing it to VHF for 30, 60, and 120 s. Raman spectroscopy map of CVD-grown MoS2 on a SiO2 surface on a Si substrate and Raman spectroscopy map of the corresponding Si substrate after exposure to VHF for 30 s (a): maps of the intensities of the E2g1 (a2) and A1g modes (a3) of MoS2 in the area of the sample that is marked by the red box in the optical image (a1). (a4) Close-up of (a3). (a5) Raman spectra of the two areas that are marked in (a4) by red and blue solid circles. (b,c) Corresponding data of MoS2 on a SiO2 surface after exposure to VHF for 60 s (b) and 120 s (c).

In a previous study, we have demonstrated that grains and grain boundaries in a CVD-grown single-layer graphene on a SiO2/Si substrate can be visualized after exposing them to VHF.[43] In the study, we used the same processing conditions for exposing and visualizing the grains in graphene as in the experiments with MoS2 in the present study, and thus the two approaches are comparable. One advantage of the visualization of grains in MoS2 is that single layers of MoS2 can be directly grown on SiO2 surfaces by CVD, which is the required substrate material for our method. In contrast, for visualizing the grains and grain boundaries in CVD-grown graphene, graphene first has to be transferred from the original growth substrate (e.g., a copper substrate) to the SiO2/Si substrate. A disadvantage with this is that the transfer can easily introduce additional line defects such as wrinkles, folds, and cracks as well as possible contaminations such as poly(methyl methacrylate) residues in the graphene layer,[43] which can influence the results. In addition, in our previous work, we have not been able to observe the graphene surface on exactly the same position before and after exposure to VHF for different times, which, to some extent, limited the accuracy of the previous results.[43] In contrast to the previous work,[43] in the present investigation, we used single-layer MoS2 that was directly synthesized on the SiO2/Si substrate by CVD, thereby entirely avoiding the layer transfer process with its potential disadvantages of introducing wrinkles, folds, cracks, or polymer residues. In addition, we have been able to observe MoS2 on the SiO2/Si substrate at exactly the same position before and after the exposure to VHF for various times, which leads to the improved accuracy of our results. We speculate that our approach may be applicable to the large-area visualization of grains and grain boundaries in single-layer 2D materials other than MoS2 and graphene. The prerequisites for this would be that (a) the 2D material is placed on a SiO2 surface (either grown or transferred), (b) it is a single-layer 2D material with grains larger than approximately 0.5–1 μm, and (c) the 2D material is not permeable to VHF and does not get significantly attacked by VHF during the time period of exposure (approximately 30–120 s).

In summary, we demonstrate here a simple and efficient method to visualize grain boundaries over large areas in CVD-grown MoS2 films on a SiO2/Si substrate. Our approach only requires VHF etching for 30–120 s and subsequent optical microscopy or SEM inspection, which are all processes and tools that are commonly available in typical cleanrooms and semiconductor labs. Although our method is invasive, that is, the sample is permanently modified in the characterization process, it has advantages such as ease of use, speed, and simple large-area analysis, which could be very useful in the development and optimization of large-scale MoS2 synthesis processes. Our approach may also be useful for investigating and optimizing the mechanical, electrical, and chemical properties of CVD-grown MoS2, which are strongly influenced by grain boundaries and grain sizes, thereby ultimately promoting the utilization of MoS2 in research and its application in future 2D material devices.

Conclusions

We demonstrated that VHF can be used to rapidly visualize the location of grains and grain boundaries over large areas in CVD-grown single-layer MoS2 on a SiO2 surface by using optical microscopy, SEM imaging, or Raman spectroscopy. Our approach is based on the difference in the etching behavior of SiO2 near the MoS2 grain boundaries and of SiO2 below the MoS2 grains when exposed to VHF. The resulting microscale line patterns in the SiO2 surface are caused by higher etch rates of SiO2 in the areas underneath the MoS2 grains, thereby causing topographical differences between the areas where grains and grain boundary-based lattice defects in the MoS2 film are located. The higher etch rate of SiO2 in the areas underneath the MoS2 grains is ascribed to the accumulation of liquid H2O with dissolved HF underneath the MoS2 grains. By contrast, there is no trapped liquid H2O in the areas of SiO2 near the grain boundaries. Our approach will be useful for efficient and large-scale imaging of the MoS2 grains and grain boundaries, with utility in the development, optimization, and monitoring of the MoS2 growth by CVD and in the evaluation of MoS2 devices.