Nanoscale-Barrier Formation Induced by Low-Dose Electron-Beam

Oct 5, 2016 - Utilizing an innovative combination of scanning-probe and spectroscopic techniques, supported by first-principles calculations, we demon...
0 downloads 0 Views 2MB Size
Download PDF
Nanoscale-Barrier Formation Induced by Low-Dose Electron-Beam Exposure in Ultrathin MoS2 Transistors Masahiro Matsunaga,† Ayaka Higuchi,† Guanchen He,‡ Tetsushi Yamada,† Peter Krüger,†,§ Yuichi Ochiai,† Yongji Gong,∥ Robert Vajtai,∥ Pulickel M. Ajayan,∥ Jonathan P. Bird,†,‡ and Nobuyuki Aoki*,†,§,⊥ †

Graduate School of Advanced Integration Science and §Molecular Chirality Research Center, Chiba University, Inage-ku, Chiba 263-8522, Japan ‡ Department of Electrical Engineering, University at Buffalo, The State University of New York, Buffalo, New York 14260-1900, United States ∥ Department of Materials Science and Nano-Engineering, Rice University, Houston, Texas 77005, United States ⊥ Japan Science and Technology Agency, PRESTO, 4-1-8 Honcho, Kawaguchi 332-0012, Japan S Supporting Information *

ABSTRACT: Utilizing an innovative combination of scanning-probe and spectroscopic techniques, supported by first-principles calculations, we demonstrate how electron-beam exposure of field-effect transistors, implemented from ultrathin molybdenum disulfide (MoS2), may cause nanoscale structural modifications that in turn significantly modify the electrical operation of these devices. Quite surprisingly, these modifications are induced by even the relatively low electron doses used in conventional electron-beam lithography, which are found to induce compressive strain in the atomically thin MoS2. Likely arising from sulfur-vacancy formation in the exposed regions, the strain gives rise to a local widening of the MoS2 bandgap, an idea that is supported both by our experiment and by the results of first-principles calculations. A nanoscale potential barrier develops at the boundary between exposed and unexposed regions and may cause extrinsic variations in the resulting electrical characteristics exhibited by the transistor. The widespread use of electron-beam lithography in nanofabrication implies that the presence of such strain must be carefully considered when seeking to harness the potential of atomically thin transistors. At the same time, this work also promises the possibility of exploiting the strain as a means to achieve “bandstructure engineering” in such devices. KEYWORDS: molybdenum disulfide, field effect transistors, electron beam exposure, scanning probe microscopy, bandgap modification, potential barrier, first-principles calculations

A

First-principles calculations attribute this effect to compressive strain induced by the electron-beam exposure, most likely through a process involving sulfur-vacancy formation. The widespread use of electron-beam lithography (EBL) as a nanofabrication tool implies that the presence of such strain must be carefully addressed when seeking to harness the full potential of atomically thin transistors. At the same time, this work also suggests the possibility of exploiting the strain as a means to achieve “bandstructure engineering” in such devices.

tomically thin molybdenum disulfide (MoS2) is of increasing interest for use in nanoelectronics,1,2 with the large bandgap it exhibits in monolayer form allowing for the realization of field-effect transistors (FETs) with large on−off ratios.3 The ultrathin nature of this material moreover presents the opportunity to tailor its functionality by introducing appropriate structural modifications. Here, we provide an example of such modification using low-dose electron-beam exposure to induce nanoscale potential barriers in the conducting channel of MoS2 FETs. This phenomenon is revealed through an innovative combination of scanning-gate (SGM) and electrostatic-force (EFM) microscopy, Raman spectroscopy, and photoluminescence (PL), which collectively demonstrate how the electron-beam exposure locally modifies the MoS2 bandgap. © 2016 American Chemical Society

Received: September 3, 2016 Accepted: October 2, 2016 Published: October 5, 2016 9730

DOI: 10.1021/acsnano.6b05952 ACS Nano 2016, 10, 9730−9737

Article

www.acsnano.org

Article

ACS Nano

Figure 1. (a) AFM topographic image of the main components of a MoS2 FET fabricated on a SiO2 substrate. The bright areas to the left and right are the Cr/Au source and drain contacts, which are separated by a gap of 4.7 μm. The dotted lines form guides to the eye that denote the edges of the triangular MoS2 crystal. The open green square identifies the region that is indicated in expanded detail in Figure 5a−c. (b) Schematic setup of the SGM measurement. (c) SGM image obtained for the MoS2 FET of panel (a) during lift-mode operation using the tip voltage (Vtip‑dc) indicated. The right electrode was held at ground in these measurements, while a drain bias Vd = 0.2 V was applied to the left one. The color scale indicates the variation (ISGM) of Id obtained while rastering the tip above the FET. (d) SGM image obtained by reversing the biasing of the two FET electrodes. The scale bar in each of the panels (c) and (d) denotes a distance of 3 μm, and the back-gate was held at ground during all of the SGM measurements. The same color scale is used in both panels (c) and (d).

Figure 2. (a) (Upper panel) EFM image exhibited by the MoS2 FET of Figure 1 when a drain bias Vd = 1 V is applied to the left electrode and the tip bias Vtip‑dc = 8 V. (Lower panel) EFM signal detected along the scanning path indicated by the horizontal dashed line in the upper panel. The bright regions apparent around the edges of the device are artifacts of the observation and can be ignored. (b) Corresponding EFM image obtained by now applying a drain bias Vd = 1 V to the right electrode while keeping all other parameters as in panel (a). (c) Scanning PL map of the FET obtained at a wavelength of 680 nm. Red (blue) coloration corresponds to strong (weak) PL intensity. Reduction of the PL intensity is clearly apparent near the right electrode, corresponding to the strained region. Dashed lines are guides to the eye that indicate the edges of the FET channel and the source and drain electrodes.

electrical measurements. In SGM, the cantilever of a conducting atomic-force microscope (AFM) is used as a local gate that is rastered over some nanoscale channel to image its current

The key innovation of this study involves the correlated use of SGM and EFM to locally probe the nanoscale structure of MoS2 FETs, something that cannot be achieved in conventional 9731

DOI: 10.1021/acsnano.6b05952 ACS Nano 2016, 10, 9730−9737

Article

ACS Nano profile.4−10 EFM, in contrast, detects the electrostatic force between the AFM tip and the sample surface as a phase shift of the cantilever oscillations. Spatial information on both the potential and charge distribution at the surface is captured in such complementary measurements, providing a powerful method for understanding the electronic properties of semiconductor nanostructures. In this work, we apply these measurement techniques to probe the current and potential distributions in MoS2 FETs.

images denote regions where the channel current is enhanced by the electric field generated by the tip. Most importantly, this response is not observed at the edge of the (Cr/Au) electrodes but is rather located in the channel interior. The response becomes increasingly prominent when Vtip‑dc is increased, consistent with the n-type nature of the MoS2 (not shown here). In Figure 1c, the right electrode of the FET serves as the grounded source and the Vd is applied to the left electrode. In Figure 1d, this configuration is reversed, but it is clear that the SGM response remains pinned near the right contact. (See section 1 of the Supporting Information for a discussion of the dependence of the SGM response on drain bias.) This pinning effect suggests the formation of some fixed domain structure within the channel, in spite of the fact that the transistor was fabricated from a single-domain MoS211 crystal. In parts a and b of Figure 2, we show EFM images obtained simultaneously with the SGM mapping for opposite polarities of the drain bias. These images allow us to construct the evolution of the potential variation along the channel length, as we indicate in the plots shown underneath the EFM scans. The plots reveal two consistent features, the first of which is a workfunctioninduced potential difference of ∼0.4 eV between the MoS2 and the gold at the surface of each electrode. The second feature is a sudden drop in potential within the channel interior, occurring at

RESULTS AND DISCUSSION The phenomenon of interest here is demonstrated in Figure 1, in which we plot the SGM response of one of our back-gated transistors (see the Methods for further details on these devices and their fabrication). This response was obtained by continuously measuring the transistor drain current (Id) while rastering the AFM tip over the surface of the atomically thin channel. In this way, we were able to construct (for fixed tip bias, Vtip‑dc) the spatial variation of the current change (ISGM) caused by the presence of the tip. The electrical setup used during these measurements is shown in Figure 1b, and further details on this setup can be found in the Methods. Parts c and d of Figure 1 correspond to SGM scans taken over the channel region for different directions of the drain bias (Vd). Bright areas in these

Figure 3. (a) Optical image of a monolayer MoS2 crystal, taken just after EBL and development. The electron-beam exposure was performed on only the left side of the crystal so that the PMMA resist remains on the right side. (b) EFM image (Vtip‑dc = 4 V, lift height = 60 nm) of the MoS2 crystal shown in panel (a), obtained after removal of all remaining resist. Inset: magnified view of the area identified by the green square in the main image. The blue dashed line corresponds to the boundary between the exposed and unexposed regions. (c) Raman spectra of the exposed (red) and the unexposed (blue) areas of the crystal. (d) Raman scattering map of the MoS2 crystal, indicating the Raman shift of the E12g peak as a function of position. (e) PL spectra obtained at different positions within the unexposed (labeled here as (i)) and exposed (ii−iv) regions identified in panel (f). (f) PL map of the MoS2 crystal at a wavelength of 670 nm. 9732

DOI: 10.1021/acsnano.6b05952 ACS Nano 2016, 10, 9730−9737

Article

ACS Nano the boundary between the different regions identified in the SGM and EFM scans, rather than a steady variation along its length. Clearly, such behavior is consistent with the presence of some kind of potential barrier12 between these regions. In Figure 2c, we show the results of PL mapping performed at a wavelength (680 nm) close to the MoS2 bandgap. Red and blue regions correspond to strong and weak intensity, respectively, and the resulting image closely correlates to the SGM and EFM measurements; while strong PL is obtained over most of the channel, the domain formed near the right electrode clearly exhibits reduced luminescence. This suggests that the two domains are both semiconducting but characterized by different bandgaps. Since the devices used in this study were fabricated from single-domain crystals, the implication is that the domain structures apparent in Figures 1 and 2 were introduced during fabrication. One possibility, for example, is that the domain structure forms when the metal contacts of the device are deposited in a process that could involve chemical-bond formation or alloying between the metal and the MoS2 crystal.13,14 In order to explore such possibilities, we have therefore exclusively studied the influence of electron-beam exposure on the MoS2 crystals, independent of the other steps used in transistor fabrication. In Figure 3a, we show a monolayer (domain-free) MoS2 crystal that has been coated with 200 nm of polymethyl methacrylate (PMMA, 495 A4:180 nm; 950 A2:20 nm) electron-beam resist, and subjected to electron-beam exposure on only the left half of its area. The electron beam in the experiment was accelerated through a voltage of 15 kV, typical of that used in EBL, and the supplied areal charge dose was 280 μC/cm2. Figure 3a was taken after the standard development step used in electron-beam lithography, following which all remaining PMMA was removed and the EFM image of Figure 3b was obtained. While the unexposed region in this image exhibits a uniform EFM response, the exposed area shows a strongly nonuniform signal, reflecting the presence of corresponding inhomogeneity in the Fermi level within the MoS2 crystal. Most importantly, the boundary between the exposed and unexposed regions is not as sharp as that defined lithographically, but rather shows irregularities that are similar in scale to the domain structure apparent in Figures 1 and 2. In Figure 3c, we compare the Raman spectra of the exposed and unexposed regions, both of which exhibit the A1g and E12g peaks15 associated with MoS2. Importantly, however, the two peaks are blue-shifted in the exposed region, which can also be seen in the Raman map of Figure 3d. Here, we plot the Raman shift of the E12g peak as a function of position. According to this map, observation of the blue shift is well correlated to the region over which electron exposure was performed, and this conclusion is consistent with the PL spectra of Figure 3e. The four exemplary spectra shown in Figure 3 were recorded at different positions, corresponding to both the unexposed (i) and exposed (ii, iii, iv) regions of the device, and span the range of the color scale used in the PL map of Figure 3f. Most importantly, it should be noted that the spatial variations in intensity implied by Figure 3f are accompanied by a distinct shift in the PL peak energy; the observed blue shift of this quantity correlates well to an overall increase in PL intensity. The averaged PL spectrum from the unexposed region is characterized by a broad peak at 1.81 eV, consistent with the usual16 direct-gap emission for monolayer MoS2. The exposed region, in contrast, shows a narrower feature at ∼1.85 eV, corresponding to a blue shift of some 40−50 meV. This idea is further confirmed in the PL map of Figure 3f, recorded for a reference wavelength (670 nm) that corresponds to the blue-shifted bandgap of 1.85 eV.

Figure 4. (a) Calculated energy-level variations for a monolayer MoS2 crystal subjected to varying biaxial strain. Green triangles, red squares, and blue circles show the variation of Eg, Ec_min, and Ev_max, respectively. (b) Variation of the bandgap energy (ΔEg) of MoS2 as a function of applied biaxial strain. Positive and negative values of the abscissa correspond to tensile and compressive strain, respectively. In the latter range, the bandgap shift is 0.116 eV per percent of strain.

It is clear that high (low) intensity is obtained in the exposed (unexposed) region of the crystal, providing a good correlation with the results obtained from both the Raman mapping and the EFM scans (Figure 3b). In addition to confirming the role of electron-beam exposure in generating the different domain structures within the MoS2, these experiments therefore also exclude the possibility that metal deposition is responsible for their formation. Previously, it has been found17,18 that electron-beam exposure may be used to induce a transition between the distinct (2H- and 1T-) crystalline forms of MoS2. The dose required to achieve this is orders of magnitude larger than those employed here, however, allowing us to exclude such a mechanism in connection to our results. Instead, we focus on recent experimental19−21 and theoretical22−25 reports concerning the influence of strain on the monolayer MoS2 bandstructure. Around the equilibrium lattice constant, the bandgap is direct and increases linearly with increasing compressive strain.23 For an accurate estimate of this variation, we have performed density functional theory (DFT) calculations using two of the most common exchange-correlation potentials26 (see the Methods and section 3 of the Supporting Information for further details of these calculations). The calculated variations (obtained using the Perdew−Burke−Enzerhof (PBE) gradient approximation) of the conduction band minimum (Ec_min), the valence band maximum (Ev_max), and the energy gap 9733

DOI: 10.1021/acsnano.6b05952 ACS Nano 2016, 10, 9730−9737

Article

ACS Nano

previously been investigated for MoS227−29 and graphene.30 The generation of sulfur vacancies has been highlighted in the former material and has been shown to cause point- and line-defect formation, both of which may generate compressive strain.28 While the electron dose utilized here is some 5 orders of magnitude smaller than that considered previously, a scenario of compressive strain arising from sulfur-vacancy could account for our observations. Based on this model for the induced strain, and using curve (i) in Figure 3e as a reference, the induced strain implied by the PL peak shifts is 0.25% (30 meV blue shift in curve ii), 0.43% (50 meV blue shift in curve iii), and 0.47% (55 meV blue shift in curve iv, which is the maximum difference). A simple average of these three values yields an induced strain of 0.37%, close to the value of 0.33% averaged in the whole exposed region. In Figure 5a, we show an AFM scan recorded over a region corresponding to a part of the boundary revealed in our EFM

(Eg) of single-crystal MoS2 are plotted as a function of applied biaxial strain in Figure 4a. Over a small range of the strain, Eg changes linearly as shown in the main text and shows a direct bandgap at the K-point. Eg reaches a maximal value for a strain of −1.6% (corresponding to compressive strain), at which point a crossover to an indirect bandgap (formed by the O-point at Ec_min and the K point at Ev_max) occurs. Under the influence of tensile strain, the transition to an indirect bandgap occurs for strain values larger than +0.3%, where an indirect gap is formed between the K-point at Ec_min and the Γ-point at Ev_max. Independent of the choice of the density functional potential, we find that the bandgap increases by ∼0.12 eV under compressive strain of 1% (Figure 4b). The experimentally inferred bandgap variation of 30−50 meV may thus be explained by the presence of an average compressive strain of ∼0.33% (40 meV). As to how this strain arises, the influence of electron-beam exposure has

Figure 5. (a) Magnified AFM scan of the area enclosed by the open green square in Figure 1a. (b) EFM image obtained by scanning over the same area as in panel (a). (c) Corresponding SGM image. The measurements in panels (b) and (c) were performed by applying a drain voltage Vd = 1 V to the left electrode, while the tip voltage was held at Vtip‑dc = 8 V. (d) AFM (upper panel), EFM (middle panel), and SGM (lower panel) line-scans obtained while rastering along the same path indicated by the light-blue dashed line in panels (a)−(c). (e) Suggested band profile for an FET channel in which an electron-beam induced barrier is present. The upper panel corresponds to thermal equilibrium, while the lower one is for a significant bias applied across the source−drain contacts. (f) Schematic illustration of the MoS2 FET. The lower part of the figure depicts the structural features of the device, while the upper part shows the variation of the MoS2 conduction-band edge at thermal equilibrium. The dasheddotted line represents the position of the Fermi level in the device. The MoS2 crystal is electrically connected to the Cr electrodes, forming ohmic contacts (or, more accurately, the “off state” of a Schottky-type contact, see section 2 of the Supporting Information) under ambient conditions and with no gate voltage applied. The portion of the crystal near the right electrode is assumed to be compressively strained and, therefore, to have a larger bandgap than the unstrained region to the left. Also shown is the potential barrier that develops at the boundary between the strained and unstrained regions due to their band offset. 9734

DOI: 10.1021/acsnano.6b05952 ACS Nano 2016, 10, 9730−9737

Article

ACS Nano

extends continuously across the entire channel width, a result that is also confirmed in panel (b) of the EFM image. This shows the EFM image obtained under the same biasing conditions as in panel (a) and is again consistent with the presence of a barrier that spans the channel width. In Figure 6c, the role of the two electrodes is now reversed, allowing very different behavior to be obtained. Now, the original barrier structure that was present near the left electrode in panel (a) is absent, and a new barrier appears near the right contact. The resulting EFM scan plotted in panel (d) is once again consistent with the presence of a barrier that runs along the channel width but in this case is formed near the right contact. Thus, we see that strain induced during electron-beam lithography can actually result in the formation of barriers at both ends of the channel. The suggested form of the band profile in an FET, with barriers formed at its source and drain ends, is indicated in Figure 6e, with the upper and lower parts of this figure corresponding, respectively, to conditions of thermal equilibrium and strong source−drain biasing. The corresponding spatial structure of the double-barrier system is suggested in Figure 6f. It should be mentioned that the straininduced barriers revealed in this work were present in just three out of a total of 14 investigated crystals, suggesting a strongly statistical character to the introduction of the strain in a process whose details are not yet fully understood. As we demonstrate in the Supporting Information, the electrical characteristics of devices with and without the barriers formed are qualitatively similar to one another. The key point, however, is that the effective channel geometry is strongly modified by the barrier introduction, since the potential drop within the channel is no longer simply defined by the source−drain separation. This could have critical implications

(Figure 5b) and SGM (Figure 5c) measurements. The crucial point here is that no topographic structure whatsoever is apparent in the AFM scan (upper panel in Figure 5d). (This should be contrasted with results31 where memory behavior arising from the presence of multiple crystalline grains was demonstrated.) Furthermore, the peak in the SGM response (lower panel in Figure 5d) is located to the right of the boundary in the EFM map (middle panel in Figure 5d), where the cantilever tip is positioned just inside the strained region. Taking these features into account, along with the invariance of the SGM/EFM structure when reversing Vd (see Figure 1), we suggest that a barrier is formed between the strained and unstrained regions. The formation of such a barrier is a natural consequence of the presence of regions with different bandgaps. Indeed, similar barrier structures have also been observed at the crystalline boundary of MoS2/WS2 planar heterostructures32 and in scanning-tunneling-microscopy studies of the grain boundaries formed in MoS2 crystals.33 In Figure 5e,f we indicate schematically the suggested structure of the barrier and its dependence on drain bias. While the focus above was on the structural investigation of a single barrier, formed in close proximity to just one of the FET electrodes, it was found that, in some devices, such barriers may even appear near both electrodes. An example of this behavior is shown in Figure 6, where we present the results of measurements of an FET separate from that considered earlier. In panel (a), the left electrode is grounded and the presence of a barrier near this contact is revealed in the SGM measurement. At the same time, no such structure is evident at the opposite end of the channel. As the drain bias is increased, the intensity of this structure becomes more prominent (not shown here) and the barrier clearly

Figure 6. (a) SGM image obtained at Vd = 4 V. (b) EFM image obtained for the same conditions as in panel (a). (c) Corresponding SGM scans obtained after reversing the role of the source and drain contacts relative to panel (a). (d) EFM image obtained for the same conditions in panel (c). (e) Schematic illustration of suggested band profile for an FET channel in which electron-beam induced barriers are present at both the source and drain ends of the channel. The upper panel corresponds to thermal equilibrium, while the lower one represents the situation where a significant bias is applied across the source−drain contacts. (f) Suggested form of the strained regions realized near the opposite ends of the MoS2 channel. 9735

DOI: 10.1021/acsnano.6b05952 ACS Nano 2016, 10, 9730−9737

Article

ACS Nano for the capacity to reliably scale MoS2 transistors to nanoscale dimensions, while maintaining their yield at satisfactory levels. In addition to considering the role of contact properties13,14,34 and intrinsic defects,35 the suggestion of our work is therefore that the influence of electron-beam lithography must also be carefully assessed when attempting to maximize this yield.

density approximation to density functional theory as well as the Perdew−Burke−Enzerhof gradient approximation. The projectoraugmented-wave method, as implemented in the VASP code26 was used with an energy cutoff of 400 eV and a γ-centered 12 × 12 × 1 grid for the Brillouin zone sampling. MoS2 monolayers were separated by a vacuum layer of over 10 Å. The results of our calculations are summarized in section 3 of the Supporting Information.

CONCLUSIONS In conclusion, local bandgap modification, accompanied by nanoscale domain formation, has been shown to occur in the channel of CVD-grown monolayer MoS2 FETs, with this phenomenon being attributed to the low-energy electron beam exposure used during transistor fabrication by EBL. A combination of SGM and EFM has been used to reveal the existence of a significant potential barrier, arising from the presence of a large voltage drop, at the boundary between the exposed and unexposed domains. By making use of spatially resolved PL and Raman spectroscopy, a local increase of the bandgap was confirmed within the exposed areas. These results suggest that the electron-beam exposure introduces a local compressive strain in the MoS2 crystal, with first-principles calculations indicating that a strain of just 0.33% is needed to account for our observations. The likely origin of this strain is believed to be the associated formation of sulfur vacancies in the exposed areas. These results suggest an opportunity in which tailored electronbeam exposure may ultimately be exploited to realize “bandstructure engineering” of atomically thin transistors.

ASSOCIATED CONTENT S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsnano.6b05952. Further details of the SGM response, the simulation parameters, and the transistor characteristics of devices with different numbers of barriers (PDF)

AUTHOR INFORMATION Corresponding Author

*Email: [email protected]. Author Contributions

G.H. fabricated the MoS2 transistors, while M.M. and A.H. measured their electrical characteristics and performed the SGM and EFM studies. Y.G. performed PL mapping, and T.Y. and P.K. performed calculations of the strain-dependent MoS2 bandstructure. M.M., Y.O., J.P.B., and N.A. analyzed the data. Y.G. performed CVD growth of MoS2 crystals under the supervision of R.V. and P.M.A. M.M., N.A. and J.P.B. wrote the paper. Notes

METHODS

The authors declare no competing financial interest.

MoS2 Transistor Fabrication. FETs were implemented using triangular-shaped, domain-free, monolayer MoS2 crystals grown directly onto SiO2/Si substrates by chemical vapor deposition (CVD).11 Cr/Au (5-/50 nm) source and drain electrodes were patterned by EBL (using PMMA as the resist and an areal dose of 280 μC/cm2) and lift-off, to form a channel with a typical length of 4 μm36 (see Figure 1a). The heavily doped Si substrate and the (300 nm thick) SiO2 served as the back-gate electrode and dielectric, respectively. Typical field-effect mobility at room temperature was ∼0.1 cm2/(V s) (∼10 cm2/(V s)), in measurements performed in air (vacuum). For further details on the basics of transistor operation, see ref 36. For a comparison of the transfer (Id−Vg) and transistor (Id−Vd) characteristics of devices, with or without one or two barriers present, see section 4 of the Supporting Information. Measurement Details. Fabricated devices were installed in an ambient scanning-probe microscopy system, in which a Pt/Ir-coated cantilever was used to apply the local electric field required in SGM, and to detect the electrostatic force between the tip and the sample due to the surface potential difference in EFM. In SGM, ac and dc biases are applied simultaneously to the AFM tip. By making use of a currentto-voltage converter, the ac-modulated component of the drain current is then converted into a voltage signal, which is detected by a lock-in amplifier to represent the SGM response. For further details on these techniques, see our previous publications.12,37 Raman and PL spectra were measured in a JASCO NRS-7100 Raman microspectrometer with an excitation wavelength of 532 nm and a 100× objective lens. In order to avoid undesirable thermal effects, the laser power was kept below 1 mW in these experiments. Studying the Influence of Electron-Beam Exposure. For our investigations of the influence of electron-beam exposure shown in Figure 3, the sample was subjected to an electron-beam dose nominally identical to that used in FET nanofabrication, after first coating the sample with 200 nm of PMMA (495 A4:180 nm; 950 A2:20 nm). The electron beam in the experiment was accelerated through a voltage of 15 kV, and the supplied areal dose of charge was 280 μC/cm2. Calculation of the Strain-Induced Bandgap Shift. Straindependent bandstructure calculations were performed using the local

ACKNOWLEDGMENTS Work at Chiba University was supported in part by the JSTPRESTO program under the project “Molecular technology and creation of new functions” and JSPS KAKENHI Grant No. JP16H00899 on Innovative Areas “Science of Atomic Layers”. Work at the University at Buffalo was supported by the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Materials Sciences and Engineering, under Award DE-FG0204ER46180. We thank JASCO Corp. for assistance with the Raman and PL mapping measurements plotted in Figure 3. REFERENCES (1) 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. (2) Schmidt, H.; Giustiniano, F.; Eda, G. Electronic Transport Properties of Transition Metal Dichalcogenide Field-Effect Devices: Surface and Interface Effects. Chem. Soc. Rev. 2015, 44, 7715−7736. (3) Radisavljevic, B.; Radenovic, A.; Brivio, J.; Giacometti, V.; Kis, A. Single-Layer MoS2 Transistors. Nat. Nanotechnol. 2011, 6, 147−150. (4) Topinka, M. A.; LeRoy, B. J.; Westervelt, R. M.; Shaw, S. E. J.; Fleischmann, R.; Heller, E. J.; Maranowski, K. D.; Gossard, A. C. Coherent Branched Flow in a Two-Dimensional Electron Gas. Nature 2001, 410, 183−186. (5) Brun, B.; Martins, F.; Faniel, S.; Hackens, B.; Bachelier, G.; Cavanna, A.; Ulysse, C.; Ouerghi, A.; Gennser, U.; Mailly, D.; Huant, S.; Bayot, V.; Sanquer, M.; Sellier, H. Wigner and Kondo Physics in Quantum Point Contacts Revealed by Scanning Gate Microscopy. Nat. Commun. 2014, 5, 4290. (6) Pioda, A.; Kičin, S.; Ihn, T.; Sigrist, M.; Fuhrer, A.; Ensslin, K.; Weichselbaum, A.; Ulloa, S. E.; Reinwald, M.; Wegscheider, W. Spatially Resolved Manipulation of Single Electrons in Quantum Dots Using a Scanned Probe. Phys. Rev. Lett. 2004, 93, 216801. 9736

DOI: 10.1021/acsnano.6b05952 ACS Nano 2016, 10, 9730−9737

Article

ACS Nano (7) Aoki, N.; Brunner, R.; Burke, A. M.; Akis, R.; Meisels, R.; Ferry, D. K.; Ochiai, Y. Direct Imaging of Electron States in Open Quantum Dots. Phys. Rev. Lett. 2012, 108, 136804. (8) Freitag, M.; Johnson, A. T.; Kalinin, S. V.; Bonnell, D. A. Role of Single Defects in Electronic Transport through Carbon Nanotube FieldEffect Transistors. Phys. Rev. Lett. 2002, 89, 216801. (9) Hunt, S. R.; Fuller, E. J.; Corso, B. L.; Collins, P. G. Distinguishing Carbon Nanotube Defect Chemistry Using Scanning Gate Spectroscopy. Phys. Rev. B: Condens. Matter Mater. Phys. 2012, 85, 235418. (10) Connolly, M. R.; Chiou, K. L.; Smith, C. G.; Anderson, D.; Jones, G. A. C.; Lombardo, A.; Fasoli, A.; Ferrari, A. C. Scanning Gate Microscopy of Current-Annealed Single Layer Graphene. Appl. Phys. Lett. 2010, 96, 113501. (11) Zhan, Y.; Liu, Z.; Najmaei, S.; Ajayan, P. M.; Lou, J. Large-Area Vapor-Phase Growth and Characterization of MoS2 Atomic Layers on a SiO2 Substrate. Small 2012, 8, 966−971. (12) Aoki, N.; Sudou, K.; Matsusaki, K.; Ochiai, Y. Visualization of Bias-Dependent Potential Barriers Using Scanning Gate Microscopy in Copper-Phthalocyanine Field-Effect Transistor. J. Vac. Sci. Technol. B 2009, 27, 785. (13) Popov, I.; Seifert, G.; Tománek, D. Designing Electrical Contacts to MoS2: Monolayers: A Computational Study. Phys. Rev. Lett. 2012, 108, 156802. (14) Smyth, C. M.; Addou, R.; McDonnell, S.; Hinkle, C. L.; Wallace, R. M. Contact Metal−MoS2 Interfacial Reactions and Potential Implications on MoS2 -Based Device Performance. J. Phys. Chem. C 2016, 120, 14719−14729. (15) Lee, C.; Yan, H.; Brus, L. E.; Heinz, T. F.; Hone, J.; Ryu, S. Anomalous Lattice Vibrations of Single- and Few-Layer MoS2. ACS Nano 2010, 4, 2695−2700. (16) Mak, K. F.; Lee, C.; Hone, J.; Shan, J.; Heinz, T. F. Atomically Thin MoS2: A New Direct-Gap Semiconductor. Phys. Rev. Lett. 2010, 105, 136805. (17) Lin, Y.-C.; Dumcenco, D. O.; Huang, Y.-S.; Suenaga, K. Atomic Mechanism of the Semiconducting-to-Metallic Phase Transition in Single-Layered MoS2. Nat. Nanotechnol. 2014, 9, 391−396. (18) Katagiri, Y.; Nakamura, T.; Ishii, A.; Ohata, C.; Hasegawa, M.; Katsumoto, S.; Cusati, T.; Fortunelli, A.; Iannaccone, G.; Fiori, G.; Roche, S.; Haruyama, J. Gate-Tunable Atomically Thin Lateral MoS2 Schottky Junction Patterned by Electron Beam. Nano Lett. 2016, 16, 3788−3794. (19) Hui, Y. Y.; Liu, X.; Jie, W.; Chan, N. Y.; Hao, J.; Hsu, Y.-T.; Li, L. J.; Guo, W.; Lau, S. P. Exceptional Tunability of Band Energy in a Compressively Strained Trilayer MoS2 Sheet. ACS Nano 2013, 7, 7126− 7131. (20) Conley, H. J.; Wang, B.; Ziegler, J. I.; Haglund, R. F.; Pantelides, S. T.; Bolotin, K. I. Bandgap Engineering of Strained Monolayer and Bilayer MoS2. Nano Lett. 2013, 13, 3626−3630. (21) Liu, Z.; Amani, M.; Najmaei, S.; Xu, Q.; Zou, X.; Zhou, W.; Yu, T.; Qiu, C.; Birdwell, A. G.; Crowne, F. J.; Vajtai, R.; Yakobson, B. I.; Xia, Z.; Dubey, M.; Ajayan, P. M.; Lou, J. Strain and Structure Heterogeneity in MoS2 Atomic Layers Grown by Chemical Vapour Deposition. Nat. Commun. 2014, 5, 5246. (22) Scalise, E.; Houssa, M.; Pourtois, G.; Afanas’ev, V.; Stesmans, A. Strain-Induced Semiconductor to Metal Transition in the TwoDimensional Honeycomb Structure of MoS2. Nano Res. 2012, 5, 43−48. (23) Yun, W. S.; Han, S. W.; Hong, S. C.; Kim, I. G.; Lee, J. D. Thickness and Strain Effects on Electronic Structures of Transition Metal Dichalcogenides: 2H- MX2 Semiconductors (M = Mo, W; X = S, Se, Te). Phys. Rev. B: Condens. Matter Mater. Phys. 2012, 85, 33305. (24) Lu, P.; Wu, X.; Guo, W.; Zeng, X. C. Strain-Dependent Electronic and Magnetic Properties of MoS2 Monolayer, Bilayer, Nanoribbons and Nanotubes. Phys. Chem. Chem. Phys. 2012, 14, 13035−13040. (25) Park, S.; Kim, S. Y. The Effect of Strain on the Electronic Properties of MoS2 Monolayers. Multiscale Multiphysics Mech. 2016, 1, 77−86. (26) Kresse, G.; Furthmüller, J. Efficient Iterative Schemes for ab initio Total-Energy Calculations Using a Plane-Wave Basis Set. Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 54, 11169−11186.

(27) Komsa, H.-P.; Kotakoski, J.; Kurasch, S.; Lehtinen, O.; Kaiser, U.; Krasheninnikov, A. V. Two-Dimensional Transition Metal Dichalcogenides under Electron Irradiation: Defect Production and Doping. Phys. Rev. Lett. 2012, 109, 035503. (28) Komsa, H.; Kurasch, S.; Lehtinen, O.; Kaiser, U.; Krasheninnikov, A. V. From Point to Extended Defects in Two-Dimensional MoS2: Evolution of Atomic Structure under Electron Irradiation. Phys. Rev. B: Condens. Matter Mater. Phys. 2013, 88, 35301. (29) Parkin, W. M.; Balan, A.; Liang, L.; Das, P. M.; Lamparski, M.; Naylor, C. H.; Rodríguez-Manzo, J. A.; Johnson, A. T. C.; Meunier, V.; Drndić, M. Raman Shifts in Electron-Irradiated Monolayer MoS2. ACS Nano 2016, 10, 4134−4142. (30) Murakami, K.; Kadowaki, T.; Fujita, J. Damage and Strain in Single-Layer Graphene Induced by Very-Low-Energy Electron-Beam Irradiation. Appl. Phys. Lett. 2013, 102, 43111. (31) Sangwan, V. K.; Jariwala, D.; Kim, I. S.; Chen, K.-S.; Marks, T. J.; Lauhon, L. J.; Hersam, M. C. Gate-Tunable Memristive Phenomena Mediated by Grain Boundaries in Single-Layer MoS2. Nat. Nanotechnol. 2015, 10, 403−406. (32) Kobayashi, Y.; Yoshida, S.; Sakurada, R.; Takashima, K.; Yamamoto, T.; Saito, T.; Konabe, S.; Taniguchi, T.; Watanabe, K.; Maniwa, Y.; Takeuchi, O.; Shigekawa, H.; Miyata, Y. Modulation of Electrical Potential and Conductivity in an Atomic-Layer Semiconductor Heterojunction. Sci. Rep. 2016, 6, 31223. (33) Huang, Y. L.; Chen, Y.; Zhang, W.; Quek, S. Y.; Chen, C.-H.; Li, L.-J.; Hsu, W.-T.; Chang, W.-H.; Zheng, Y. J.; Chen, W.; Wee, A. T. S. Bandgap Tunability at Single-Layer Molybdenum Disulphide Grain Boundaries. Nat. Commun. 2015, 6, 6298. (34) English, C. D.; Shine, G.; Dorgan, V. E.; Saraswat, K. C.; Pop, E. Improved Contacts to MoS2 Transistors by Ultra-High Vacuum Metal Deposition. Nano Lett. 2016, 16, 3824−3830. (35) McDonnell, S.; Addou, R.; Buie, C.; Wallace, R. M.; Hinkle, C. L. Defect-Dominated Doping and Contact Resistance in MoS2. ACS Nano 2014, 8, 2880−2888. (36) He, G.; Ghosh, K.; Singisetti, U.; Ramamoorthy, H.; Somphonsane, R.; Bohra, G.; Matsunaga, M.; Higuchi, A.; Aoki, N.; Najmaei, S.; Gong, Y.; Zhang, X.; Vajtai, R.; Ajayan, P. M.; Bird, J. P. Conduction Mechanisms in CVD-Grown Monolayer MoS2 Transistors: From Variable-Range Hopping to Velocity Saturation. Nano Lett. 2015, 15, 5052−5058. (37) Wei, X.; Aoki, N.; Yahagi, T.; Maeda, K.; Bird, J. P.; Ishibashi, K.; Ochiai, Y. Analysis of Operation Mechanism of Field Effect Transistor Composed of Network of High-Quality Single Wall Carbon Nanotubes by Scanning Gate Microscopy. Jpn. J. Appl. Phys. 2012, 51, 04DN05.

9737

DOI: 10.1021/acsnano.6b05952 ACS Nano 2016, 10, 9730−9737