Unveiling Defect-Related Raman Mode of Monolayer WS2 via Tip

Aug 24, 2018 - Unveiling Defect-Related Raman Mode of Monolayer WS2 via Tip-Enhanced Resonance Raman Scattering. Chanwoo Lee†‡ , Byeong Geun ...
0 downloads 0 Views 7MB Size
www.acsnano.org

Unveiling Defect-Related Raman Mode of Monolayer WS2 via Tip-Enhanced Resonance Raman Scattering Chanwoo Lee,†,‡ Byeong Geun Jeong,† Seok Joon Yun,†,‡ Young Hee Lee,†,‡,§ Seung Mi Lee,*,∥ and Mun Seok Jeong*,†,‡ †

Department of Energy Science, Sungkyunkwan University, Suwon 16419, Republic of Korea Center for Integrated Nanostructure Physics (CINAP), Institute for Basic Science (IBS), Suwon 16419, Republic of Korea § Department of Physics, Sungkyunkwan University, Suwon 16419, Republic of Korea ∥ Korea Research Institute of Standards and Science (KRISS), Daejeon 34113, Republic of Korea

ACS Nano 2018.12:9982-9990. Downloaded from pubs.acs.org by MIDWESTERN UNIV on 01/23/19. For personal use only.



S Supporting Information *

ABSTRACT: Monolayer tungsten disulfide (WS2) has emerged as an active material for optoelectronic devices due to its quantum yield of photoluminescence. Despite the enormous research about physical characteristics of monolayer WS2, the defect-related Raman scattering has been rarely studied. Here, we report the correlation of topography and Raman scattering in monolayer WS2 by using tip-enhanced resonance Raman spectroscopy and reveal defect-related Raman modes denoted as D and D′ modes. We found that the sulfur vacancies introduce not only the red-shifted A1g mode but also the D and D′ modes by the density functional theory calculations. The observed defect-related Raman modes can be utilized to evaluate the quality of monolayer WS2 and will be helpful to improve the performance of WS2 optoelectronic devices. KEYWORDS: tip-enhanced Raman scattering, tungsten disulfide, transition metal dichalcogenides, defects, sulfur vacancies electron microscopy, and dark-field optical microscopy.12,13 Among these methods, Raman spectroscopy is a simple and nondestructive technique compared to the other complicated methods.14 Furthermore, confocal Raman spectroscopy is not only widely applied to the thickness determination of few-layer WS215 but is also used to analyze the structural disorders in bulk WS2 by observing a B1u modes.16 Although the B1u mode has been known as a vibrational mode related to structural disorders in WS2, the observation of this mode is possible in few layers and nanotubes,17 because monolayer WS2 only has an A′1 (A1g) mode as a symmetric out-of-plane vibrational mode of two S atoms.18 In addition, the observation of B1u mode also requires a resonance excitation, because it is a Raman-inactive mode.17 Nevertheless, it may be possible that the D mode like B1u mode can be generated in monolayer WS2 because out-of-plane lattice vibrations can be affected by the existing defects, as similar to the generation mechanism of D

T

wo-dimensional (2D) transition metal dichalcogenides (TMD) materials have been widely investigated due to their optical properties such as the bandgap engineering depending on the number of layers.1,2 Among the various TMDs composed of layered structures, monolayer TMDs are known for emitting strong photoluminescence (PL) owing to their direct bandgaps.3,4 In particular, monolayer tungsten disulfide (WS2) shows intense PL at ∼1.95 eV with a relatively high quantum yield among TMDs.2,5 Thus, monolayer WS2 has attracted much interest as an outstanding material for 2D optoelectronic devices.4,6,7 To achieve the high performance of optoelectronic devices, the defect-free WS2 is highly required. Accordingly, it is necessary to establish a quality evaluation method by investigating the defects of monolayer WS2. In the case of graphene, a defect-induced (D) mode in the Raman scattering is widely used to evaluate the sample quality.8 However, most experimental and theoretical studies of defects in monolayer WS2 have rarely reported the Raman properties caused by the defect, except for shifts of Raman frequency.9−11 The defects of chemical vapor deposition (CVD)-grown monolayer TMDs have been investigated by using various analysis methods including Raman spectroscopy, transmission © 2018 American Chemical Society

Received: June 5, 2018 Accepted: August 24, 2018 Published: August 24, 2018 9982

DOI: 10.1021/acsnano.8b04265 ACS Nano 2018, 12, 9982−9990

Article

Cite This: ACS Nano 2018, 12, 9982−9990

Article

ACS Nano

Figure 1. TERS experiment of monolayer WS2. (a) Schematic of confocal Raman scattering experiment. Energy band diagram shows the work functions (Φ) and Fermi levels as the Au (tip and substrate) and WS2 make noncontact. (b) Schematic of STM-TERS experiment. A bias voltage of 50 mV is applied and the tunneling current is approximately 6.2 nA. The diagram shows the relative changes in Fermi levels when the Au and WS2 make contact. (c) Confocal Raman scattering image of A1g peak area intensity (exposure time per point: 7 s). (d) STM image of the sample. (e) TERS image of A1g peak area intensity (exposure time per point: 0.5 s). Green dotted square indicates the region of TERS mapping in Figure 3. (f) Confocal Raman spectrum of monolayer WS2. The spectrum is extracted from (c). (g) Confocal Raman spectrum (blue) and TERS spectrum (red) measured at the same spot and exposure time (5 s). The three images of (c−e) were measured at the same sample position.

mode in graphene.19 However, the use of confocal Raman spectroscopy has limitations to observe nanoscale defects such as grain boundaries, cracks, and vacancies.4,20 This is because monolayer WS2 exhibits weak Raman scattering signals and the defects can be generated in areas smaller than the spatial resolution of confocal Raman spectroscopy. Despite these limitations, if a D mode could be detected in monolayer WS2, it would be helpful in determining the presence of defects and quality evaluation. Therefore, a correlation study with vibrational and topographical information in nanoscale such as tip-enhanced Raman scattering is crucially needed for confirming the exact origin of defect-related Raman mode.8 In this work, we perform tip-enhanced resonance Raman scattering (TERS) experiments and obtain highly enhanced Raman signals from monolayer WS2. The measurement of high-resolution TERS images and scanning tunneling microscopy (STM) in the same position shows strong correlations between Raman spectra and STM images. We also demonstrate that the red-shifted A1g mode accompanied by the D and D′ modes can be attributed to the defects in monolayer WS2. Furthermore, we identify that the emergence of the shifted modes can be introduced by S vacancies (VS) through our density functional theory (DFT) calculations.

of 2D materials.15 Monolayer WS2 shows the A1g mode at about 417 cm−1 regardless of wavelength of excitation laser.15,21 In addition, the A1g mode of bilayer WS2 appears at more than 418 cm−1 and a value of Raman shift of A1g peak increases as the number of WS2 layers increases.15 Therefore, we demonstrate that the synthesized WS2 is a monolayer sample because the A1g mode appears at about 417 cm−1 in Figure 1 (Supporting Information Figure S1). As shown in Figure 1, the Au substrate is necessary to enhance the Raman signals using the gap-mode from the nanogap between the tip and substrate.22−25 Furthermore, PL emission as a background signal in the Raman spectrum can be quenched by the charge transfer from WS2 to the substrate, as shown in Figure 1a,b. The electron transfer can be induced because of the work function difference between the Au and WS2,26 hot electrons at the tip apex,27 and the tunneling current.28 The TERS tips were fabricated by an electrochemical etching method29 (Supporting Information Figure S2). By using this TERS system, a correlation between structural defects and Raman modes can be investigated. We measured the confocal Raman mapping to examine the A1g peak position and intensity. In Figure 1c, the Raman intensity in the white dotted circle is weaker than that of the other positions, whereas the distribution image of the peak positions does not indicate a large difference depending on the sample position (Supporting Information Figure S1). In order to identify the peaks in the white dotted circle, confocal Raman spectrum was extracted and fitted with a Lorentzian curve, as shown in Figure 1f. Although the Raman spectra derived from the white-dotted circle present the shoulder peak of the A1g mode, the

RESULTS AND DISCUSSION TERS for Monolayer WS2. We synthesized monolayer WS2 on an Au substrate by using CVD,21 as illustrated in Figure 1. In order to identify whether a sample is monolayer or not, we measured the Raman spectra. Raman spectroscopy has been known as a powerful tool to count the number of layers 9983

DOI: 10.1021/acsnano.8b04265 ACS Nano 2018, 12, 9982−9990

Article

ACS Nano

conventional Raman microscope is ∼550 nm, which can be calculated by the optical diffraction limit.31 Therefore, the spatial resolution was improved by more than a factor of approximately 12. Moreover, the TERS and STM images measured at the same region show the strong correlation between the Raman intensity and surface roughness, as shown in Figure 2a,b. The Raman signals on the flat and clean surface are stronger than those on the rough surface. After synthesizing the monolayer WS2 on the Au substrate, the substrate surface beneath the sample can be rough (Supporting Information Figure S5).21 Therefore, the sample surface can be also affected by the substrate roughness, and this undulation may cause various disorders such as cracks and wrinkles in the sample. Owing to these defects which break the structural symmetry, the intensities of the Raman signals of monolayer WS2 can be also decreased. To ascertain the shoulder peak of A1g mode, we extracted the TERS spectra from the defective sites which show the weak Raman intensity in the TERS images, as shown in Figure 2 (Supporting Information Figure S6). The shoulder of A1g peak is denoted as a D peak and a peak at ∼432 cm−1 is denoted as a D′ peak which appears with the strong D peak. The shoulder peak of the A1g mode that presents at same Raman frequency with the B1u mode can be denoted as the D mode because the terminology of B1u mode has been usually used in case of more than bilayer.32,33 The TERS signals of monolayer WS2 were investigated by performing experiments at several positions (Supporting Information Figures S6, S7, S8, and S9). Moreover, we conducted TERS measurements for the other samples, which were positioned on the same substrate (Supporting Information Figures S10 and S11). Figure 3 shows the magnified STM image, multispectral TERS images, and the distribution image of peak position. These images were measured in the green dotted square in Figure 1e. The darkest line shapes in Figure 1e can be identified as cracks because no Raman scattering signals can be detected in Figure 3. If the lines were grain boundaries or wrinkles, the Raman scattering signals would be observed.8 The region between the crack and the sample shows strong TERS signals, which may seem that the sample is curled up or multilayer. If the sample were curled up or multilayer, the A1g mode in Raman spectrum would appear at more than 418 cm−1.15,29 Therefore, the edge region is also the monolayer WS2 because the Raman signals indicate well the characteristic of monolayer WS2. Moreover, in order to investigate the regions displaying the relatively low Raman intensity, we investigated an image that shows the Raman peak positions from 400 to 430 cm−1, as shown in Figure 3c. The distribution image indicates that the regions that have a low Raman intensity except for the cracks in Figure 2b exhibit the redshifted A1g peak. The shifted regions also correspond to the structural disorders in the STM image of Figure 3a (Supporting Information Figure S12). Furthermore, we extracted the TERS images for peak area intensity to investigate the correlation between the A1g peak and the other peaks. The A1g peak intensity in Figure 3b corresponds to the intensity of 2LA(M) peak, as shown in Figure 3d. In addition, the A1g and 2LA(M) peaks show the weak Raman intensity in the region where the A1g peak is red-shifted in Figure 3c. In contrast, the intensity images of D and D′ peaks in Figure 3e,f show the uniform intensity in the same region, as compared with the intensity images of A1g and 2LA(M) peaks. Correlation between Raman Modes of Monolayer WS2. For further investigation of the correlation between

prominent change of the A1g peak position is not detected and the other spectra show weak intensity to distinguish Raman peaks (Supporting Information Figure S3). Moreover, as far as we know, there is no information about this shoulder peak in the Raman spectrum of monolayer WS2. Also, we performed STM imaging and scanning electron microscopy at the same position with confocal Raman image to precisely confirm the surface morphology, as shown in Figure 1d (Supporting Information Figure S4). Accordingly, a number of line shapes that resemble cracks or wrinkles in the sample and the undulation of the substrate are observed. We performed TERS imaging at the same area, as shown in Figure 1e. The fine structures that do not appear in the confocal Raman image but corresponds with STM image can be observed owing to the high spatial resolution. To calculate the contrast of Raman intensities and the enhancement factor (EF) of TERS,29 confocal Raman and TERS signals were obtained at the same spot. The contrast value in TERS is defined as an intensity difference between far-field (confocal Raman) and near-field (TERS) Raman scattering signals.29,30 The EF value is also calculated by using the contrast value, a radius of TERS tip (Rtip), and a radius of excitation laser spot (Rlaser). The equations for calculating the contrast and EF values are defined as follows Contrast =

ITERS − Iconfocal Raman Iconfocal Raman

ji R zy EF = contrast × jjjj laser zzzz j R tip z k {

2

The contrast value of the A1g peak intensities is ∼20.4 and the EF has a value of ∼2.7 × 104, as shown in Figure 1g. Defect-Related Raman Modes. We also conducted magnified TERS mapping to measure the spatial resolution and identify the shoulder peak of A1g mode as shown in Figure 2. The spatial resolution of TERS is estimated to be ∼45 nm, which is beyond the diffraction limit and is estimated within 10−90% of the optical contrast. The spatial resolution of the

Figure 2. Spatial resolution of TERS image and defect-induced Raman peaks of monolayer WS2. (a) TERS image of A1g peak area intensity. (b) STM image of the same position in (a). (c) Line profile of TERS intensity along the green bars, and the fitted Gaussian peak (red dashed line). (d) TERS spectrum of monolayer WS2. The spectrum is extracted from (a). 9984

DOI: 10.1021/acsnano.8b04265 ACS Nano 2018, 12, 9982−9990

Article

ACS Nano

Figure 3. Multispectral TERS images of defective monolayer WS2. (a) STM image measured at the defective area. (b) TERS image of A1g peak area intensity. The white dashed line indicates the investigated positions in Figure 3. (c) Distribution image of Raman peak position in TERS image of (a). TERS images of (d) 2LA(M), (e) D, and (f) D′ peak area intensity. All TERS images were measured at the same position in (a).

Figure 4. Defect-related Raman modes of monolayer WS2. (a) TERS spectral line trace of 2LA(M), D, A1g, and D′ peaks from Figure 3. The black dashed lines indicate each peak position and the white dashed lines indicate the extracted spectra in (d). (b) The 2D correlation coefficient map of the peak position depending on TERS intensity from (a). (c) Raman intensity correlation between the strongest peaks and defect-related peaks. (d) TERS spectra extracted from the spectral line traces. The spectra were fitted with a Lorentzian curve.

mode clearly appear between 400 and 415 cm−1 from 0 to 0.7 μm in the line trace. The line trace also shows the strong correlation between 2LA(M) and A1g peaks. Although the opposite behaviors for Raman intensities between the D peak and the 2LA(M) and A1g peaks are shown in the line trace, the relation between the D′ peak and the other peaks cannot be

2LA(M), A1g, D, and D′ peaks, we presented a TERS spectral line trace depending on the sample position. Figure 4a shows that the TERS spectral line trace derived from the white dashed line in Figure 3b. In the case of pristine monolayer WS2, the peak value of the A1g mode generally appears at ∼417 cm−1, while the red-shifted peaks and shoulder peaks of the A1g 9985

DOI: 10.1021/acsnano.8b04265 ACS Nano 2018, 12, 9982−9990

Article

ACS Nano

Figure 5. Raman mapping of monolayer WS2 by using a 473 nm excitation laser. Confocal Raman scattering image of (a) E12g peak area intensity and (b) A1g peak area intensity. Distribution image of (c) E12g peak position in (a) and (d) A1g peak position in (b). (e) Confocal Raman spectra extracted from the white dashed circle in (a). Each dashed line indicates the wavenumber of E12g and A1g modes, respectively.

a large quantity of VS. In addition, the spectra show reduced intensities of LA(M) and 2LA(M) peaks in the defective region (Supporting Information Figures S8 and S13). The LA(M) mode has a lattice vibration motion of the collective inplane direction of atoms in the lattice structure, which is generated along the propagation direction of the vibration, similar to a sound wave.37 According to the characteristic of this mode, the 2LA(M) mode occurs with two electron− phonon scattering phenomena in the conduction band, including two electronic transitions from the conduction (valence) to valence (conduction) band.15 Therefore, the intensity of the 2LA(M) peak can be weakened when the VS induce the localized donor states near the conduction band inside the bandgap which can interrupt the vertical electronic transitions.36 In case of monolayer MoS2, Mignuzzi et al. reported that as the defect density increases, the LA(M) mode also increases and the 2LA(M) mode decreases.38 In the article, a pristine MoS2 does not show the LA(M) mode, thus they assumed that the LA(M) mode can not only be the defect-related mode of MoS2 but also that of the other TMD materials. However, in the case of monolayer WS2, the LA(M) mode appears clearly in Raman spectrum of a pristine sample.15 In addition, our TERS experiments also plainly reveal same trends, as shown in Supporting Information Figures S8 and S13. Raman Imaging for Strain Effect. We also conducted the confocal Raman scattering experiments by using a 473 nm laser to investigate the strain effect through alteration of the E12g mode, which appears at ∼356 cm−1. Figure 5a,b displays the confocal Raman mapping images of the E12g and A1g peak area intensity by using an excitation laser of wavelength 473 nm. The E12g peak emerges higher than the 2LA(M) peak when only using the 473 nm excitation laser,39 and the E12g peak is

easily observed because of weak Raman intensity. Thus, we calculated the correlation coefficient between 2LA(M), D, A1g, and D′ peaks depending on TERS intensities from the line trace, as shown in Figure 4b,c. The correlation coefficient can be obtained by dividing the covariance of each variable by the standard deviation of each variable, and the value is determined between −1 and 1.34 The 2D correlation coefficient map shows the reverse correlation between the representative Raman peaks of pristine monolayer WS2 and the defect-related Raman peaks. In addition, the correlation between D and D′ peaks can be discovered. We also extracted several TERS spectra from the line trace to precisely compare with Raman intensities as shown in Figure 4d (Supporting Information Figure S13) because the line trace has the background and noise signals of TERS image. TERS spectra 3 and 4 show high intensity A1g peaks at ∼417 cm−1 and also LA(M) and 2LA(M) peaks at ∼175 and ∼350 cm−1, respectively, in Supporting Information Figure S13. Although these spectra seem like the Raman peaks of pristine monolayer WS2, the A1g peaks that show very subtle asymmetry have shoulder peaks assigned as the D peak, as shown in Figure 4d. We notice that the spectra 1 and 2 show red-shifted A1g peaks with weak intensities at ∼413 cm−1. In addition, the red-shifted A1g peak occurs with the red-shifted D peak at ∼408 cm−1 and there is the D′ peak at ∼432 cm−1. It is known that the red-shift of the A1g peak depends on the number of VS.10 In particular, the VS in MoS2 and WS2 are known to cause the n-doping effect35 by creating localized states near a conduction band in the electronic band structure.36 Therefore, the features of the shifted spectra can be explained by the positions with different density of VS. As these shifted peaks also appear on the rough surface in Figure 3a, the structurally defective region can be estimated to contain 9986

DOI: 10.1021/acsnano.8b04265 ACS Nano 2018, 12, 9982−9990

Article

ACS Nano

Figure 6. Raman vibration modes of a monolayer WS2 through DFT calculations. (a) Atomic structure of an ideal model and a 15.625% VS density model of monolayer WS2. The red dashed circles in (a) indicate the positions of VS. (b) Symmetric (A1g) and (c,d) asymmetric vibrations (D and D′) of the ideal and defective (15.625% VS density) monolayer WS2. The blue arrows show the atom movement and orientation. (e) Raman peak shift as a function of VS density. (f) A1g peak shift and VS density depending on the TERS spectra.

nm excitation laser is much weaker than that of the A1g mode using the 633 nm laser and the confocal resonance Raman spectra using the 633 nm laser show the changes in the A1g peak (Supporting Information Figure S3). Therefore, we prove that the peak alterations in the Raman spectra of monolayer WS2 are not attributable to the strain effect, but rather to the defects, by comparing with the Raman peaks through the experiments depending on the laser wavelength. DFT Calculation for Defect-Related Raman Modes. In order to clarify the origin of the D and D′ modes, and the shift of the peak, we performed quantum mechanical calculations within DFT framework depending on various VS densities in monolayer WS2. Atomic orbital basis sets were used as implemented in DMOL3 code (as implemented in BIOVIA Materials Studio platform)42 in the double numerical with polarization form. All electrons, including those from the core part, were considered during calculations. The exchangecorrelation functions obtained using a local density approximation and the k-points with Monkhorst−Pack grid with a separation of 0.02/Å were used. The geometry optimization criteria were 0.005 Å for distance, 0.001 Ha/Å for force, and 10−5 Ha for total energy difference. We fully relaxed the bulk WS2 from ICSD with lattice parameters of a = b = 3.154 Å, c = 12.360 Å, α = β = 90°, and γ = 120°, including the cell optimization obtaining a = b = 3.1959 Å, c = 11.3908 Å, α = β = 90°, and γ = 120°. From the optimized geometry, we generated 4 × 4 unit cell and removed one monolayer in order to get the one monolayer WS2 geometry, then we fully relaxed

well-known as an indicator of the strain effect in the Raman spectrum of layered TMD materials.9,40,41 Li et al. reported that they can distinguish strained and unstrained regions of MoS2 induced by ∼200 nm size nanocones with confocal Raman and micro-PL systems.41 They used an objective lens with a 0.9 numerical aperture (NA) and an excitation laser with 532 nm wavelength, which indicates that a spatial resolution of confocal Raman is about 243 nm. In our confocal system, the spatial resolution is about 277 nm as using a 473 nm laser and an objective lens with a 0.7 NA. Therefore, the strain effect in monolayer WS2 can be demonstrated with our confocal Raman system. When comparing the Raman images in Figure 5, the intensity of the E12g peak is higher than that of the A1g peak and no peculiarities appear. To ascertain whether the shifted peaks are caused by the strain effect induced by the rough surface of the substrate, we compared the images of the E12g and A1g peak positions, as shown in Figure 5c,d. A white dashed circle in Figure 5a represents the white dotted circle in Figure 1c, which indicates the low intensity and shifted peaks. However, the E12g and A1g peak positions in the white dashed circle do not show any peak shifts. In addition, we extracted the Raman spectra from the white dashed circle to precisely identify the peak shift in the defective site, as shown in Figure 5e. These spectra also show the E12g and A1g modes which do not show the peak shift. The nonshifted E12g peak indicates that the observation of D mode in the defective site is not ascribed to the strain effect by the substrate. Although the A1g peak is not shifted, the intensity of the A1g mode using the 473 9987

DOI: 10.1021/acsnano.8b04265 ACS Nano 2018, 12, 9982−9990

Article

ACS Nano

a minimum value of −25 mV to a maximum value of 3 V with a frequency of 300 Hz and a duty cycle of 20%. The Pt ring was immersed in a mixture of 37% hydrochloric acid solution (SigmaAldrich) and 99.9% anhydrous ethanol (Sigma-Aldrich) at a depth of approximately 2 mm. The Au wire with a diameter of 0.25 mm was also immersed in the same mixture at a depth of approximately 0.5 mm. The etched tips were rinsed by various solutions such as acetone, ethanol, deionized water, and isopropanol. This etching process was a self-terminating experiment. Sample Preparation. We synthesized the monolayer WS2 on an Au foil.21 Ammonium metatungstate hydrate (AMT) used as a W source was dissolved in DI water. An Al2O3 plate was coated by the AMT solution. To synthesize monolayer WS2, the Au foil and coated Al2O3 plate were placed in a CVD chamber. The furnace was heated to 500 °C in a nitrogen atmosphere and was maintained for 20 min. In addition, the furnace was heated to 935 °C and H2S gas was then injected. Far- and Near-Field Raman Imaging. The TERS equipment is integrated with the STM and confocal Raman spectroscope (NTEGRA Spectra, NT-MDT Co., Zelenograd, Russia), thus it is possible to measure both the confocal Raman scattering and TERS. In TERS experiments, the laser of 633 nm in wavelength was polarized along the tip axis. In the confocal Raman measurements, the laser of 473 nm (only Figure 5) and 633 nm in wavelength was linearly polarized. The objective lens used in all the experiments has a 0.7 NA and 100× magnification (Mitutoyo, Japan). Accordingly, in case of using 473 nm laser, the spatial resolution of confocal Raman image is about 277 nm. In the case of using 633 nm laser, the spatial resolution is about 550 nm. The Raman scattering signals were obtained using a charge-coupled detector (CCD, Andor, U.K.) cooled to −80 °C and a spectrometer with an 1800 grooves/mm grating, blazed at 500 nm. The thermal drift of the system is about 15 nm/min/°C for XY-axis scanning and about 10 nm/min/°C for Z-axis scanning.

the geometry again to get the energetically stable reference system for further study. From the reference geometry, we generated the initial geometries with various VS by removing S atoms randomly and fully relaxed them again. The numbers of VS are 0, 1, 2, 3, 4, 5, and 6, which corresponds to 0% (reference), 3.125%, 6.250%, 9.375%, 12.50%, 15.625%, and 18.75%, respectively. We calculated several more than five different geometries per each VS. Also, the calculation details are described in Supporting Information. We found that the peaks of the vibration modes in the outof-plane directions are red-shifted in Raman spectra as the VS density in monolayer WS2 increases, as shown in Figure 6 (Supporting Information Table S1). Figure 6a shows the atomic structures of the pristine monolayer WS2 and the WS2 with a 15.625% VS density. Figure 6b−d shows the out-ofplane vibrational modes of the pristine and defective monolayer WS2. We identified the A1g (Figure 6b) and D modes (Figure 6c) red-shifted by 3.14 and 10.27 cm−1, respectively, by DFT calculations, which are close to the results of our TERS experiments. Furthermore, in Figure 6d,e the D′ mode occurs at a frequency shifted by 0.24 cm−1 in the calculations. From the analysis of vibrational direction in Figure 6, the D′ mode is similar to the A″2 mode known as infrared (IR) active vibrational mode.32,33 Although the D mode of monolayer WS2 without the VS has the same Raman frequency as the A1g mode, the D mode can be shown by splitting from the A1g mode with increasing VS, as shown in Figure 6e (Supporting Information Table S1). Although the IR-active D′ mode has no huge change in Raman shift in Figure 6e (Supporting Information Table S1), this mode can be detected in Raman spectrum with the presence of VS. Moreover, we can associate the each TERS spectrum with the VS density depending on the shift of A1g peak position, as shown in Figure 6e,f. Using DFT simulations, we identified the origin of the D and D′ modes in the TERS spectra in the defective area only.

ASSOCIATED CONTENT S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsnano.8b04265. Additional explanations about data and Figures S1−S14 (PDF)

CONCLUSIONS In summary, we have conducted STM-based TERS experiments to investigate the Raman modes caused by defects in monolayer WS2. By using an Au tip and substrate, we achieved a large contrast of ∼20 and a TERS EF of ∼2.7 × 104, induced by the highly enhanced near-fields in the nanogap between the tip and substrate, which originate from surface plasmons at the tip apex. Furthermore, we have measured successful TERS images that have a spatial resolution of ∼45 nm, which is beyond the optical diffraction limit. Owing to the TERS imaging at the nanoscale, we have demonstrated the defectrelated Raman vibrational modes in monolayer WS2, which indicate the D and D′ mode accompanied by the red-shifted A1g mode. According to the DFT calculations, the defect can be regarded as the VS. We expect that the defect-related Raman mode, found by the correlation study of Raman scattering and STM images through TERS technique, can be used to evaluate the quality of 2D nanomaterials.

AUTHOR INFORMATION Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. ORCID

Chanwoo Lee: 0000-0001-7630-2927 Byeong Geun Jeong: 0000-0003-0662-1820 Seok Joon Yun: 0000-0002-8695-7166 Young Hee Lee: 0000-0001-7403-8157 Seung Mi Lee: 0000-0002-5956-5853 Mun Seok Jeong: 0000-0002-7019-8089 Notes

The authors declare no competing financial interest.

ACKNOWLEDGMENTS This work was supported by IBS-R011-D1 and the National Research Foundation of Korea (NRF) Grant funded by the Korea government (MSIP) (2016R1A2B2015581). The work in KRISS was supported by Nano-Materials Technology Development Program (NRF-2016M3A7B4025406). The authors gratefully acknowledge to Dr. Kyoung-Duck Park for the valuable discussions.

METHODS TERS Tip Fabrication. The TERS Au tips were fabricated under an optimized etching process by using a homemade electrochemical etching system.29 An Au wire (0.25 mm, 99.95%, Nilaco, Japan) acted as the anode and a platinum (Pt) ring (0.20 mm, 99.98%, Nilaco, Japan) as the cathode. The Au and Pt were connected to a wave function generator, which supplied a pulsed square wave voltage from 9988

DOI: 10.1021/acsnano.8b04265 ACS Nano 2018, 12, 9982−9990

Article

ACS Nano

(19) Ferrari, A. C.; Meyer, J. C.; Scardaci, V.; Casiraghi, C.; Lazzeri, M.; Mauri, F.; Piscanec, S.; Jiang, D.; Novoselov, K. S.; Roth, S.; Geim, A. K. Raman Spectrum of Graphene and Graphene Layers. Phys. Rev. Lett. 2006, 97, 187401. (20) Lee, Y.; Park, S.; Kim, H.; Han, G. H.; Lee, Y. H.; Kim, J. Characterization of the Structural Defects in CVD-grown Monolayered MoS2 using Near-Field Photoluminescence Imaging. Nanoscale 2015, 7, 11909−11914. (21) Yun, S. J.; Chae, S. H.; Kim, H.; Park, J. C.; Park, J.-H.; Han, G. H.; Lee, J. S.; Kim, S. M.; Oh, H. M.; Seok, J.; et al. Synthesis of Centimeter-Scale Monolayer Tungsten Disulfide Film on Gold Foils. ACS Nano 2015, 9, 5510−5519. (22) Zhang, R.; Zhang, Y.; Dong, Z.; Jiang, S.; Zhang, C.; Chen, L.; Zhang, L.; Liao, Y.; Aizpurua, J.; Luo, Y. e.; et al. Chemical Mapping of a Single Molecule by Plasmon-Enhanced Raman Scattering. Nature 2013, 498, 82−86. (23) Zhong, J.-H.; Jin, X.; Meng, L.; Wang, X.; Su, H.-S.; Yang, Z.-L.; Williams, C. T.; Ren, B. Probing the Electronic and Catalytic Properties of a Bimetallic Surface with 3 nm Resolution. Nat. Nanotechnol. 2016, 12, 132−136. (24) Jiang, S.; Zhang, Y.; Zhang, R.; Hu, C.; Liao, M.; Luo, Y.; Yang, J.; Dong, Z.; Hou, J. Distinguishing Adjacent Molecules on a Surface using Plasmon-Enhanced Raman Scattering. Nat. Nanotechnol. 2015, 10, 865−869. (25) Bao, W.; Melli, M.; Caselli, N.; Riboli, F.; Wiersma, D.; Staffaroni, M.; Choo, H.; Ogletree, D.; Aloni, S.; Bokor, J.; et al. Mapping Local Charge Recombination Heterogeneity by Multidimensional Nanospectroscopic Imaging. Science 2012, 338, 1317− 1321. (26) Buscema, M.; Steele, G. A.; van der Zant, H. S. J.; CastellanosGomez, A. The Effect of the Substrate on the Raman and Photoluminescence Emission of Single-Layer MoS2. Nano Res. 2014, 7, 561−571. (27) Giugni, A.; Torre, B.; Toma, A.; Francardi, M.; Malerba, M.; Alabastri, A.; Zaccaria, R. P.; Stockman, M. I.; Di Fabrizio, E. HotElectron Nanoscopy using Adiabatic Compression of Surface Plasmons. Nat. Nanotechnol. 2013, 8, 845−852. (28) Zhang, Y.; Luo, Y.; Zhang, Y.; Yu, Y.-J.; Kuang, Y.-M.; Zhang, L.; Meng, Q.-S.; Luo, Y.; Yang, J.-L.; Dong, Z.-C.; et al. Visualizing Coherent Intermolecular Dipole-Dipole Coupling in Real Space. Nature 2016, 531, 623−627. (29) Lee, C.; Kim, S. T.; Jeong, B. G.; Yun, S. J.; Song, Y. J.; Lee, Y. H.; Park, D. J.; Jeong, M. S. Tip-Enhanced Raman Scattering Imaging of Two-Dimensional Tungsten Disulfide with Optimized Tip Fabrication Process. Sci. Rep. 2017, 7, 40810. (30) Stadler, J.; Schmid, T.; Zenobi, R. Developments in and Practical Guidelines for Tip-Enhanced Raman spectroscopy. Nanoscale 2012, 4, 1856−1870. (31) Betzig, E.; Trautman, J.; Harris, T.; Weiner, J.; Kostelak, R. Breaking the Diffraction Barrier: Optical Microscopy on a Nanometric Scale. Science 1991, 251, 1468−1470. (32) Yamamoto, M.; Wang, S. T.; Ni, M.; Lin, Y.-F.; Li, S.-L.; Aikawa, S.; Jian, W.-B.; Ueno, K.; Wakabayashi, K.; Tsukagoshi, K. Strong Enhancement of Raman Scattering from a Bulk-Inactive Vibrational Mode in Few-Layer MoTe2. ACS Nano 2014, 8, 3895− 3903. (33) Molina-Sanchez, A.; Wirtz, L. Phonons in Single-Layer and Few-Layer MoS2 and WS2. Phys. Rev. B: Condens. Matter Mater. Phys. 2011, 84, 155413. (34) Park, K.-D.; Muller, E. A.; Kravtsov, V.; Sass, P. M.; Dreyer, J.; Atkin, J. M.; Raschke, M. B. Variable-Temperature Tip-Enhanced Raman Spectroscopy of Single-Molecule Fluctuations and Dynamics. Nano Lett. 2016, 16, 479−487. (35) Salehi, S.; Saffarzadeh, A. Atomic Defect States in Monolayers of MoS2 and WS2. Surf. Sci. 2016, 651, 215−221. (36) Qiu, H.; Xu, T.; Wang, Z.; Ren, W.; Nan, H.; Ni, Z.; Chen, Q.; Yuan, S.; Miao, F.; Song, F.; et al. Hopping Transport through DefectInduced Localized States in Molybdenum Disulphide. Nat. Commun. 2013, 4, 2642.

REFERENCES (1) Mak, K. F.; He, K.; Lee, C.; Lee, G. H.; Hone, J.; Heinz, T. F.; Shan, J. Tightly Bound Trions in Monolayer MoS2. Nat. Mater. 2013, 12, 207−211. (2) Hong, X.; Kim, J.; Shi, S.-F.; Zhang, Y.; Jin, C.; Sun, Y.; Tongay, S.; Wu, J.; Zhang, Y.; Wang, F. Ultrafast Charge Transfer in Atomically Thin MoS2/WS2 Heterostructures. Nat. Nanotechnol. 2014, 9, 682−686. (3) Huang, C.; Wu, S.; Sanchez, A. M.; Peters, J. J.; Beanland, R.; Ross, J. S.; Rivera, P.; Yao, W.; Cobden, D. H.; Xu, X. Lateral Heterojunctions within Monolayer MoSe2-WSe2 Semiconductors. Nat. Mater. 2014, 13, 1096−1101. (4) Duan, X.; Wang, C.; Shaw, J. C.; Cheng, R.; Chen, Y.; Li, H.; Wu, X.; Tang, Y.; Zhang, Q.; Pan, A.; et al. Lateral Epitaxial Growth of Two-Dimensional Layered Semiconductor Heterojunctions. Nat. Nanotechnol. 2014, 9, 1024−1030. (5) Peimyoo, N.; Shang, J.; Cong, C.; Shen, X.; Wu, X.; Yeow, E. K.; Yu, T. Nonblinking, Intense Two-Dimensional Light Emitter: Monolayer WS2 Triangles. ACS Nano 2013, 7, 10985−10994. (6) Wang, Q. H.; Kalantar-Zadeh, K.; Kis, A.; Coleman, J. N.; Strano, M. S. Electronics and Optoelectronics of Two-Dimensional Transition Metal Dichalcogenides. Nat. Nanotechnol. 2012, 7, 699− 712. (7) Lee, H. S.; Luong, D. H.; Kim, M. S.; Jin, Y.; Kim, H.; Yun, S.; Lee, Y. H. Reconfigurable Exciton-Plasmon Interconversion for Nanophotonic Circuits. Nat. Commun. 2016, 7, 13663. (8) Park, K.-D.; Raschke, M. B.; Atkin, J. M.; Lee, Y. H.; Jeong, M. S. Probing Bilayer Grain Boundaries in Large-Area Graphene with TipEnhanced Raman Spectroscopy. Adv. Mater. 2017, 29, 1603601. (9) Li, H.; Tsai, C.; Koh, A. L.; Cai, L.; Contryman, A. W.; Fragapane, A. H.; Zhao, J.; Han, H. S.; Manoharan, H. C.; AbildPedersen, F.; Norskov, J. K.; Zheng, X. Activating and Optimizing MoS2 Basal Planes for Hydrogen Evolution through the Formation of Strained Sulphur Vacancies. Nat. Mater. 2016, 15, 48−53. (10) Parkin, W. M.; Balan, A.; Liang, L.; Das, P. M.; Lamparski, M.; Naylor, C. H.; Rodríguez-Manzo, J. A.; Johnson, A. C.; Meunier, V.; Drndić, M. Raman Shifts in Electron-Irradiated Monolayer MoS2. ACS Nano 2016, 10, 4134−4142. (11) Kim, B. H.; Park, M.; Lee, M.; Baek, S. J.; Jeong, H. Y.; Choi, M.; Chang, S. J.; Hong, W. G.; Kim, T. K.; Moon, H. R.; et al. Effect of Sulphur Vacancy on Geometric and Electronic Structure of MoS2 Induced by Molecular Hydrogen Treatment at Room Temperature. RSC Adv. 2013, 3, 18424−18429. (12) Jeong, H. Y.; Lee, S. Y.; Ly, T. H.; Han, G. H.; Kim, H.; Nam, H.; Jiong, Z.; Shin, B. G.; Yun, S. J.; Kim, J.; et al. Visualizing Point Defects in Transition-Metal Dichalcogenides using Optical Microscopy. ACS Nano 2016, 10, 770−777. (13) Jeong, H. Y.; Jin, Y.; Yun, S. J.; Zhao, J.; Baik, J.; Keum, D. H.; Lee, H. S.; Lee, Y. H. Heterogeneous Defect Domains in SingleCrystalline Hexagonal WS2. Adv. Mater. 2017, 29, 1605043. (14) Oh, H. M.; Han, G. H.; Kim, H.; Bae, J. J.; Jeong, M. S.; Lee, Y. H. Photochemical Reaction in Monolayer MoS2 via Correlated Photoluminescence, Raman Spectroscopy, and Atomic Force Microscopy. ACS Nano 2016, 10, 5230−5236. (15) Berkdemir, A.; Gutiérrez, H. R.; Botello-Méndez, A. R.; PereaLópez, N.; Elías, A. L.; Chia, C.-I.; Wang, B.; Crespi, V. H.; LópezUrías, F.; Charlier, J.-C.; et al. Identification of Individual and Few Layers of WS2 using Raman Spectroscopy. Sci. Rep. 2013, 3, 1755. (16) Staiger, M.; Gillen, R.; Scheuschner, N.; Ochedowski, O.; Kampmann, F.; Schleberger, M.; Thomsen, C.; Maultzsch, J. Splitting of Monolayer Out-of-Plane A1′ Raman Mode in Few-Layer WS2. Phys. Rev. B: Condens. Matter Mater. Phys. 2015, 91, 195419. (17) Staiger, M.; Rafailov, P.; Gartsman, K.; Telg, H.; Krause, M.; Radovsky, G.; Zak, A.; Thomsen, C. Excitonic Resonances in WS2 Nanotubes. Phys. Rev. B: Condens. Matter Mater. Phys. 2012, 86, 165423. (18) Scheuschner, N.; Gillen, R.; Staiger, M.; Maultzsch, J. Interlayer Resonant Raman Modes in Few-Layer MoS2. Phys. Rev. B: Condens. Matter Mater. Phys. 2015, 91, 235409. 9989

DOI: 10.1021/acsnano.8b04265 ACS Nano 2018, 12, 9982−9990

Article

ACS Nano (37) Kim, M. S.; Yun, S. J.; Lee, Y.; Seo, C.; Han, G. H.; Kim, K. K.; Lee, Y. H.; Kim, J. Biexciton Emission from Edges and Grain Boundaries of Triangular WS2 Monolayers. ACS Nano 2016, 10, 2399−2405. (38) Mignuzzi, S.; Pollard, A. J.; Bonini, N.; Brennan, B.; Gilmore, I. S.; Pimenta, M. A.; Richards, D.; Roy, D. Effect of Disorder on Raman Scattering of Single-Layer MoS2. Phys. Rev. B: Condens. Matter Mater. Phys. 2015, 91, 195411. (39) Zhao, W.; Ghorannevis, Z.; Amara, K. K.; Pang, J. R.; Toh, M.; Zhang, X.; Kloc, C.; Tan, P. H.; Eda, G. Lattice Dynamics in Monoand Few-Layer Sheets of WS2 and WSe2. Nanoscale 2013, 5, 9677− 9683. (40) Wang, Y.; Cong, C.; Yang, W.; Shang, J.; Peimyoo, N.; Chen, Y.; Kang, J.; Wang, J.; Huang, W.; Yu, T. Strain-Induced DirectIndirect Bandgap Transition and Phonon Modulation in Monolayer WS2. Nano Res. 2015, 8, 2562−2572. (41) Li, H.; Contryman, A. W.; Qian, X.; Ardakani, S. M.; Gong, Y.; Wang, X.; Weisse, J. M.; Lee, C. H.; Zhao, J.; Ajayan, P. M.; et al. Optoelectronic Crystal of Artificial Atoms in Strain-Textured Molybdenum Disulphide. Nat. Commun. 2015, 6, 7381. (42) Delley, B. From Molecules to Solids with the DMol3 Approach. J. Chem. Phys. 2000, 113, 7756−7764.

9990

DOI: 10.1021/acsnano.8b04265 ACS Nano 2018, 12, 9982−9990