Potential Profile of Stabilized Field-Induced Lateral p–n Junction in

Nov 21, 2017 - Electric field-induced p–n junctions are often used to realize peculiar functionalities in various materials. This method can be appl...
0 downloads 0 Views 3MB Size
Download PDF
www.acsnano.org

Potential Profile of Stabilized Field-Induced Lateral p−n Junction in Transition-Metal Dichalcogenides Yijin Zhang,*,†,‡,∥ Ryuji Suzuki,∥ and Yoshihiro Iwasa∥,§ †

The Institute of Scientific and Industrial Research, Osaka University, Osaka 067-0047, Japan Max Planck Institute for Solid State Research, Stuttgart D-70569, Germany ∥ Department of Applied Physics and Quantum-Phase Electronics Center (QPEC), The University of Tokyo, Tokyo 113-8656, Japan § Center for Emergent Matter Science (CEMS), RIKEN, Wako 351-0198, Japan ‡

S Supporting Information *

ABSTRACT: Electric field-induced p−n junctions are often used to realize peculiar functionalities in various materials. This method can be applied not only to conventional semiconductors but also to carbon nanotubes, graphene, and organic semiconductors to which the conventional chemical doping method is difficult to apply. Transition-metal dichalcogenides (TMDs) are one of such materials where the field-induced p−n junctions play crucial roles in realizing solar cell and light-emitting diode operations as well as circularly polarized electroluminescence. Although the field-induced p−n junction is a well-established technique, many of its physical properties are left to be understood because their doping mechanism is distinct from that of conventional p−n junctions. Here we report a direct electrical measurement of the potential variation along the field-induced p−n junction using multiple pairs of voltage probes. We detected the position of the junction, estimated the built-in potential, and monitored the effect of the bias voltage. We found that the built-in potential becomes negative under a forward bias voltage range where field-induced TMD p−n junctions have been operated as light-emitting diodes. This feature well reproduced the circularly polarized electroluminescence from the WSe2 p−n junction, indicating that the present observation provides a useful background for understanding and functionalizing field-induced p−n junctions. KEYWORDS: transition-metal dichalcogenides, field-induced p−n junction, built-in potential, light-emitting diode, valleytronics

T

gate voltage. This is a well-known technique for ambipolar transistors with organic materials22,23 and carbon nanotubes,24,25 to which the position-controlled chemical doping is difficult to be applied. The field-induced p−n junctions are also reported in conventional semiconductors such as GaAs/ AlGaAs quantum well for high-speed optoelectronic devices.26 The field-induced p−n junctions can be created by two means: driving the FET device into the ambipolar region22,23 or fabricating split-gate electrodes.24−26 The former p−n junctions are usually dynamic (the properties of p−n junctions change with biases), but can be stabilized by the cool-while-gating technique with liquid gate dielectrics.14 The latter p−n junctions can be regarded as stable junctions as long as the source-drain voltage is negligibly small compared to two gate voltages so that the carrier density variation caused by the bias voltage can be ignored. Although there may exist some small

ransition-metal dichalcogenides (TMDs) are layered semiconductors that attract a great number of researchers owing to their superior electronic device performance,1−4 to their enhanced light−matter interaction,5−7 and to their optical selection rule as well as potential application for next-generation technologies.8−13 The relatively small band gap of TMDs (1.2−2.2 eV)5 and the atomically flat surface structure without dangling bonds facilitate the realization of an ambipolar transistor operation (either electrons or holes can be the carrier, controlled by the gate voltage)1,4 and thus p−n junctions in size of TMD thin flakes.14 These p−n junctions bridge the superior electronic device performance and peculiar optical features of TMDs, realizing state-of-the-art optoelectronic devices.12,15−21 In particular, the circularly polarized electroluminescence (EL) with an electrical polarization controllability is a functionality that can only be realized with the peculiar band structure of TMDs.12,20 Although TMD p−n junctions have a wide variety of potential application, less attention has been paid to its fundamental device performance. TMD p−n junctions are electrostatically formed using the field-effect transistor (FET) structure by controlling the local © 2017 American Chemical Society

Received: September 22, 2017 Accepted: November 21, 2017 Published: November 21, 2017 12583

DOI: 10.1021/acsnano.7b06752 ACS Nano 2017, 11, 12583−12590

Article

Cite This: ACS Nano 2017, 11, 12583−12590

Article

ACS Nano

Figure 1. (a) Comparison of conventional and field-induced p−n junctions. (b) Bilayer WSe2 device with multiple voltage probes. Blue and red spheres represent W and Se atoms, respectively. (c) A schematic of the multiple voltage probes measurement of the field-induced p−n junction.

p−n junction for general optoelectronics as well as for valleytronics. Here, we used a fully electrical transport measurement to investigate the band alignment and the built-in potential of the field-induced p−n junctions stabilized by the cool-while-gating technique. We fabricated an ionic liquid-gated FET with multiple voltage probes (V1−V8) using a bilayer WSe2 thin flake (Figure 1b). We selected bilayer rather than monolayer WSe2 because the previous bias-voltage-dependence of the circular polarization was recorded on a multilayer WSe2 instead of a monolayer.12 The simultaneous measurement of longitudinal voltages at different positions (Figure 1c) allows us to trace the formation and movement of the p−n junction at high temperature (220 K where the liquid gate dielectric is in the liquid phase) as well as provides the information on the potential variation of the stabilized p−n junction at low temperature (at 150 K where the liquid gate dielectric is frozen). In addition, the built-in potential can be evaluated based on the Hall effect measurement at low temperature. Furthermore, we estimated the strength of the in-plane electric field and simulated the circularly polarized EL spectra, which is in a good agreement with previous experimental results.12 In particular, our simulation reproduced the positive correlation between the bias voltage and the polarization, which was not explained previously.

differences between these two methods, both p−n junctions work as current rectifiers. More importantly, a much larger difference can be easily recognized when field-induced p−n junctions are compared with conventional p−n junctions formed by the chemical doping process: donors and acceptors are absent in field-induced p−n junctions (Figure 1a). Since donor and acceptor densities determine many of fundamental physical parameters of conventional p−n junctions, such as the built-in potential and the depletion region length,27 these parameters are basically unknown for field-induced p−n junctions. Although a number of diode-like rectification and optoelectric performance were reported from field-induced TMD p−n junctions (both by cool-while-gating technique12,14,20 or by split-gate technique15−17), their actual band alignment and the built-in potential have never been investigated. These parameters are of particular importance not only because they determine the overall device performance of p−n junctions but also because they are regarded as crucial parameters for circularly polarized EL when field-induced TMD p−n junctions are used as light emitters.12 The origin of this peculiar functionality is expected to be the broken valley symmetry caused by the in-plane electric field, 12,20,28 which is approximated to the built-in potential divided by the depletion region length. A previous literature indeed reported a bias voltage dependence of the circular polarization in EL from a stabilized field-induced WSe2 p−n junction,12 but neither qualitative nor quantitative explanation was presented. Therefore, investigation of the built-in potential and the corresponding electric field is highly demanded for the application of TMD

RESULTS AND DISCUSSION Ambipolar WSe2 Transistor and Field-Induced p−n Junction. Figure 2a shows the transfer curve, the gate voltage (VGS) dependence of the drain current (IDS) under a small drain voltage (VDS) with the source electrode connected to the 12584

DOI: 10.1021/acsnano.7b06752 ACS Nano 2017, 11, 12583−12590

Article

ACS Nano

Figure 2. (a) Transfer curve of the bilayer WSe2 device. (b) Temperature dependence of channel resistance under different VGS. (c) Temperature dependence of carrier density at VGS = 1.9 V and −2 V. (d) Temperature dependence of electron and hole mobility.

Figure 3. (a) Output curve of the WSe2 device. The black solid line represents a fit to the quadratic model. (b) Voltage drops measured simultaneously with the output curve measurement.

ground. Hereafter Vαβ represents the voltage (potential) of the contact β measured from the contact α. As shown in Figure 2a, our device clearly shows the ambipolar behavior. The off-state current is limited by the input impedance of the lock-in amplifier. The initial state is slightly shifted to the valence band, likely due to the unintentional hole doping during the microfabrication process. When a sufficiently large gate voltage (|VGS|) is applied, the intrinsic semiconducting state of WSe2 changes to metallic states. Figure 2b compares the temperature dependence of the longitudinal channel resistance (Rxx) under several values of VGS. At small |VGS|, where the Fermi energy still locates in the band gap, Rxx increases with reduced temperature (dRxx/dT < 0) representing a semiconducting (insulating) phase. When | VGS| is large enough so that the Fermi energy touches the band, on the other hand, dRxx/dT becomes positive, which is clear evidence of the metallic states. Figure 2c,d shows the temperature dependence of the carrier density and the mobility in such metallic states, respectively. The metallic behavior was observed for both positive and negative VGS, indicating that the Fermi energy can be shifted completely from the conduction band to the valence band controlled barely by the gate voltage. The ambipolar metallic transition was not clearly observed in liquid-gated bulk WSe2,29 which can be attributed to the roughness of the bulk surface. The ambipolar nature also manifests itself in the output curve, VDS dependence of IDS, as the upturn of IDS after the saturation region (Figure 3a). This upturn is the evidence that the FET device enters the ambipolar region, in which both holes and electrons are simultaneously accumulated at the channel surface and thus a field-induced p−n junction is formed.14 The black solid line in Figure 3a is a fit to the quadratic equation model.14,23,30 The band gap deduced from this fit is 1.5 eV, which is in agreement with the actual band gap of bilayer WSe2 (1.52 eV).31,32 In the ambipolar region, the position of the p−n junction strongly depends on the bias voltages (VGS and VDS). In the present configuration under a fixed VGS, the junction first appears close to the drain electrode and moves toward the

source electrode with an increase of VDS. This motion can be captured by the voltage measurement with longitudinal combinations of two voltage proves (V4T = V21, V32, V43). The voltage measured at the regime closest to the source (drain) electrode is labeled as V21 (V43), and the one at the middle regime is V32 (see Figure 1c). Most of the voltage drop is expected to occur around the p/n interface (the contribution of the contact resistance is ignored) because the resistivity of this interface is much higher than that of the p- and the nregions. Therefore, when the interface enters the regime between the two neighboring voltage probes, the corresponding V4T shows a drastic increase. V4T subsequently drops steeply when the interface moves out from the regime, forming a very sharp peak as a function of VDS. As shown in Figure 3b, the peak in V4T first appears in V43 and then, with the increase of VDS, followed by peaks in V32 and V21. This result clearly indicates that the p−n junction moves from the drain electrode toward the source electrode. Stabilized Field-Induced p−n Junction. Based on the information from Figure 3b, we set VGS to 2 V and VDS to 3.5 V (this condition corresponds to the peak position of V32) in order to locate the junction at the middle of the channel and cooled the whole device down to 150 K to stabilize the p−n junction (cool-while-gating technique).14 Since the liquid is frozen at 150 K, field-induced carrier accumulation in the semiconductor channel is stabilized and becomes bias independent, making the field-induced p−n junction stable. At 150 K, VDS works as a conventional bias voltage to p−n junctions. In this particular study, a positive (negative) VDS corresponds to a forward (reverse) bias. Figure 4a shows a cyclic IDS−VDS curve of field-induced p−n junction at 150 K. The typical output curve of an ambipolar FET (Figure 3a) changes to a diode-like rectifying curve without hysteresis between the downward and the upward scans of VDS. Figure 4b presents bias dependence of V4T at 150 K measured simultaneously with the IDS−VDS curve shown in Figures 4a. V43 and V21 represent the potential drop in p-doped 12585

DOI: 10.1021/acsnano.7b06752 ACS Nano 2017, 11, 12583−12590

Article

ACS Nano

Figure 5. Hall effect measurement of stabilized field-induced p−n junction. Upper illustration shows the measurement configuration. Figure 4. (a) IDS−VDS curve of the stabilized field-induced WSe2 p− n junction. The inset shows the measurement configuration. (b) Voltage drop at three different regions of the p−n junction. (c) Bias dependence of the resistance in each region.

The built-in potential at equilibrium can be deduced from the carrier density obtained by the Hall effect. Here we emphasize that it is inadequate to adopt the conventional p−n junction models with finite donor density and acceptor density36−38 to evaluate the fundamental properties, such as the built-in potential, of the field-induced p−n junctions. The detailed assessment using the conventional model is placed in the Supporting Information (Section 3). As an alternative way, we evaluated the built-in potential based on the fact that the Fermi energy in the p-region (n-region) is touching the valence (conduction) band as demonstrated in Figure 2 and Figure S4 in the Supporting Information. In such a situation, the carrier density deduced from the Hall effect measurement can be converted to the Fermi energy by assuming a parabolic energy dispersion (ε(k) = ℏ2k2/(2m*)). For TMDs’ band structure with multiple possible Fermi pockets, it is difficult to precisely know the charge distribution in the momentum space under gate bias, because the band structure is modified by the gate electric field.29,39 Experimentally, EL spectra from multilayer TMDs confirm the occupation of both the indirect gap and the direct gap when p− n junctions are formed with liquid gate dielectrics.12,18,40 Thus, we equally took into account the ± T pockets (midpoint of Γ ±K line) and the ±K pockets for electrons. As well, the Γ pocket and the ±K pockets are equally considered for holes. With the effective mass from the density functional theory (DFT) calculation,41,42 the carrier density corresponds to the Fermi energy of 29 meV below the valence band top in the pregion and 32 meV above the conduction band bottom in the n-region. Therefore, as shown in Figure 6, we obtained the built-in potential at equilibrium as

and n-doped regions, respectively, and V32 represents the potential drop across the p/n interface. Similar to conventional p−n junctions, most of the voltage drop occurs at this interface (green line), in particular when VDS is small. It is also notable that V32 follows the line of V4T = VDS in this voltage region, indicating that almost all of VDS is applied to V32 and the contact resistance in the present device is negligibly small compared to the resistance at the p/n interface. Evaluation of the contact resistance is included in the Supporting Information, Section 1. The resistance across the p/n interface (R32) is shown in Figure 4c by the green curve. While R32 shows a large variation with VDS, R43 and R21 stay constant. This result indicates that the depletion region (the transition region from p-doping to n-doping) is confined between voltage probes V2 and V3, and thus the depletion region length (Ld) is shorter than the separation of two adjacent voltage probes (2 μm). These results are reproducible as shown in Figure S2 in the Supporting Information. Using transverse pairs of voltage probes, we can further assess the carrier density and the carrier polarity. Figure 5 shows the Hall effect measurement recorded with the voltage probe pairs of V7−V3 and V6−V2. As shown in the illustration, these two pairs are located in the p-region and the n-region, respectively. Clearly, the sign of the Hall coefficient is opposite between these two pairs, providing solid evidence of the formation of the p−n junction. The carrier density deduced from the Hall coefficient is 2.2 × 1013 /cm2 and 1.5 × 1013 /cm2 for the p- and the n-region, respectively. These carrier densities are sufficient to induce metallic states in TMDs.33−35 We note that these carrier densities are bias independent, because the depletion region in the present device is fully confined between V2 and V3 as discussed above.

Vbi,0 = εg + 29meV + 32meV = 1.58eV

(1)

where εg is the band gap (1.52 eV for bilayer WSe2). This value is similar to VTh of the IDS−VDS characteristics of the p−n junction (Figure 4a). We note that the present way to estimate the built-in potential is available only when the Fermi energy is touching 31,32

12586

DOI: 10.1021/acsnano.7b06752 ACS Nano 2017, 11, 12583−12590

Article

ACS Nano

large forward bias that far exceeds VTh. Here in our study, we revealed that the band bending is already suppressed at VTh so that Vbi is always negative when the light emission occurs from the field-induced TMD p−n junction. Simulating the Device Performance of Field-Induced p−n Junction. Finally, we show the usefulness of the present study for designing optoelectronic and valleytronic devices with TMDs by numerically demonstrating the circularly polarized light emission from WSe2 p−n junction. EL spectra from multilayer TMDs contain two peaks corresponding to the light emissions at the direct gap and the indirect gap,18,40 but here we focus on the EL component originated from the direct gap. Since the gate electric field breaks the inversion symmetry, the EL emission from ±K valley of multilayer TMDs is also circularly polarized and follows the same mechanism as monolayers.12,40 The circular polarization in EL is attributed to the in-plane electric field which breaks the symmetry of trigonally warped ±K valley by shifting the charge distribution in the momentum space, as illustrated in Figure 8.12,20,28 When

Figure 6. Schematic band bending across the p/n interface (the region between voltage probes V2 and V3) at equilibrium. The spatial dispersion of the conduction band (C.B.) and the valence band (V.B.) is a guide to eye.

the band, as evidenced by the metallic temperature dependence of resistivity shown in Figure S4 in the Supporting Information. Otherwise, the conducting holes and electrons are thermally excited ones, so that the carrier density deduced from the Hall effect cannot be directly converted to the Fermi energy. Also, split-gate devices15−17 are not suitable for investigating fieldinduced p−n junctions in TMDs, since the gate coupling through the solid dielectrics is so weak that either p-region and/or n-region remains insulating. The externally applied voltage to a p−n junction modifies the built-in potential from its equilibrium value.27 The multiple voltage probes measurement allows us to measure the actual value of the applied voltage excluding the voltage drop due to the contact resistance. In the present device configuration, the built-in potential under bias is Vbi = Vbi,0 − V32

Figure 8. Schematic band structure and the charge distribution shift caused by an in-plane electric field around ±K points.

the charge distribution shift is sufficiently large, a sizable difference appears in the radiative recombination rate between ±K valleys, which indicates the circularly polarized EL owing to the exclusive coupling between the valley degree of freedom and the circular polarization in light. The amount of the charge distribution shift is (m*μ/ℏ)E from the semiclassical Boltzmann equation with the relaxation time (τ) approximation.48 The electron distribution in the momentum space is

(2)

Figure 7a shows VDS dependence of Vbi. At VDS = VTh, Vbi ≈ 0 indicating that no band bending exists (band is almost flat). The schematic illustrations of the band bending at three different values of VDS are compared in Figure 7b. When VDS > VTh, Vbi becomes negative, which is usually not considered in conventional p−n junctions.27,36 A negative Vbi has been suggested for the ambipolar region of the organic43−45 and carbon nanotube46,47 ambipolar FETs. In these works, however, the field-induced p−n junctions are not stabilized and a large VDS is considered, which corresponds to a p−n junction with a

fe = f e0 (k) −

⎛ qτ ∂fe m*μ ⎞ ≈ f e0 ⎜k − E· E⎟ ⎝ ℏ ⎠ ℏ ∂k

(3)

Figure 7. (a) VDS dependence of Vbi. (b)Schematic band bending across the p/n interface between voltage probes V2 and V3 under several representative VDSs. 12587

DOI: 10.1021/acsnano.7b06752 ACS Nano 2017, 11, 12583−12590

Article

ACS Nano with the relation qτ = m*μ. The hole distribution is f h(k) = 1 − fe(k). Then, the EL intensity from ±K valley at a photon energy of ℏω will be I±K(ω) =

∫ d k (ω − εc

±K (k)

junctions are used as light-emitting diodes as revealed in the present study (Figure 7a). It is also important toward application that the optoelectronic and valleytronic device performance was reproduced from the information obtained by the electrical transport measurement. Our results can be utilized to design various optoelectronic functional devices based on the simple electrical transport measurement.

δ f c (k)f hv, ±K (k) + ε±v K (k))2 + δ 2 e , ±K (4)

where the superscript “c” and “v” represent the conduction and the valence bands, respectively, and δ is the broadening factor. Knowing Vbi, the numerical calculation of EL spectra from ±K valleys becomes possible. Figure 9a shows a representative of simulated EL spectra with VDS = 6 V and Ld = 400 nm. The effective mass is taken

CONCLUSION In summary, we electrically investigated the stabilized fieldinduced p−n junction in TMD using multiple voltage probes. The obstacle in calculating the built-in potential of fieldinduced p−n junctions, due to the difference of the doping mechanism compared to conventional p−n junctions, was overcome by introducing the ionic liquid gate dielectric whose large carrier doping capability simultaneously creates degenerate electrons and holes. We found that the built-in potential is already zero at the threshold voltage of the p−n junction and stays negative for the entire voltage range where the light emission occurs from the p−n junction. Our method is quite powerful as it enables a numerical prediction of the peculiar valleytronic functionality of circularly polarized EL from fieldinduced WSe2 p−n junction. Since this method is applicable to a wide variety of materials, the present result provides a way to understand the device performance of field-induced p−n junctions and to design various optoelectronic and valleytronic functional devices.

Figure 9. (a) Simulated EL spectra from WSe2 p−n junction. (b) Bias voltage dependence of the degree of circular polarization in EL. Experimental data are taken from prior publication.12

from literatures,41,42 the mobility is deduced from the Hall effect measurement (Figure 2d), and the electron temperature is set to 100 K. Although the actual electric field at the p/n interface varies with the position (Ebi ≡ −dVbi/dr), the built-in potential divided by the depletion length (Vbi/Ld) is a good measure of the strength of the electric field (Ebi = |Ebi|). Ld deduced from the conventional p−n junction model, which is discussed in the Supporting Information (Section 2), does not reproduce the experimental result even qualitatively. Simulated spectra are shown in Figure S5 in the Supporting Information. From the viewpoint of the degree of circular polarization: η≡

I(σ+) − I(σ −) I − I −K = +K I(σ+) + I(σ −) I+K + I −K

EXPERIMENTAL SECTION Device Fabrication. WSe2 thin flakes were mechanically exfoliated from WSe2 single crystals grown by chemical vapor transport method49,50 and selected based on the surface flatness and the lateral dimension to allow multiple connection of voltage probes. The thickness was identified as bilayer by the optical contrast. The electrodes were fabricated using an electron-beam lithography system with PMMA as the resist, followed by the vacuum deposition of Ti (5 nm) as an adhesion layer and Au (80 nm). A large plate is simultaneously fabricated next to the sample, which works as the gate electrode for liquid gate dielectric. After the lift-off process to remove resist, a droplet of ionic liquid was placed to cover both channel WSe2 flake and the gate electrode. We used ionic liquid (N,N-diethyl-Nmethyl-N- (2-methoxyethyl)ammonium bis(trifluoromethylsulfonyl) imide) for the gate dielectrics. Measurement. The device was placed in Physical Property Measurement System (PPMS) from Quantum Design, Inc. All of the measurement was performed under high vacuum and below 220 K in order to reduce chemical reaction. The data shown in Figure 2 were recorded using an AC lock-in technique with a lock-in amplifier (SR830 from Stanford Research Systems). All other data were recorded using DC measurement with a two-channel source-meter (Keithley 2612b).

(5)

we found that Ld in the present field-induced p−n junction is around several hundreds of nanometers, by comparing the experimental spectra12 and the simulated spectra with various values assumed for Ld. A smaller Ld leads to a larger |η|, as shown in Figure 9b, since the electric field and the charge distribution shift become larger. The experimental data of VDS dependence of η are taken from the literature12 and overplotted in the figure by blue solid symbols. We note that these experimental results were not obtained with a monolayer,12 thus it is fair to compare with the present simulations. The deviation of the experimental data from a single simulated curve can be attributed to the increase of the effective carrier temperature with the increasing VDS, which broadens the charge distribution in the momentum space and reduces the contribution of the trigonal warping and hence |η|. Nonetheless, the experimentally observed positive correlation between |η| and VDS is qualitatively reproduced by the simulation. Such a correlation originates from the fact that Vbi becomes and stays negative when field-induced TMD p−n

ASSOCIATED CONTENT S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsnano.7b06752. Estimation of the contact resistance of the field-induced p−n junction, additional data representing the stability of the field-induced p−n junction, assessment of the built-in potential using the conventional p−n junction model, and simulated EL spectra (PDF) 12588

DOI: 10.1021/acsnano.7b06752 ACS Nano 2017, 11, 12583−12590

Article

ACS Nano

(17) Ross, J. S.; Klement, P.; Jones, A. M.; Ghimire, N. J.; Yan, J.; Mandrus, D.; Taniguchi, T.; Watanabe, K.; Kitamura, K.; Yao, W.; Cobden, D. H.; Xu, X. Electrically Tunable Excitonic Light-Emitting Diodes Based on Monolayer WSe2 p−n Junctions. Nat. Nanotechnol. 2014, 9, 268−272. (18) Jo, S.; Ubrig, N.; Berger, H.; Kuzmenko, A. B.; Morpurgo, A. F. Mono- and Bilayer WS2 Light-Emitting Transistors. Nano Lett. 2014, 14, 2019−2025. (19) Ponomarev, E.; Gutierrez-Lezama, I.; Ubrig, N.; Morpurgo, A. F. Ambipolar Light-Emitting Transistors on Chemical Vapor Deposited Monolayer MoS2. Nano Lett. 2015, 15, 8289−8294. (20) Onga, M.; Zhang, Y.; Suzuki, R.; Iwasa, Y. High Circular Polarization in Electroluminescence from MoSe2. Appl. Phys. Lett. 2016, 108, 073107. (21) Kozawa, D.; Pu, J.; Shimizu, R.; Kimura, S.; Chiu, M.-H.; Matsuki, K.; Wada, Y.; Sakanoue, T.; Iwasa, Y.; Li, L.-J.; Takenobu, T. Photodetection in p−n Junctions Formed by Electrolyte-Gated Transistors of Two-Dimensional Crystals. Appl. Phys. Lett. 2016, 109, 201107. (22) Swensen, J. S.; Soci, C.; Heeger, A. J. Light Emission from an Ambipolar Semiconducting Polymer Field-Effect Transistor. Appl. Phys. Lett. 2005, 87, 253511. (23) Zaumseil, J.; Friend, R. H.; Sirringhaus, H. Spatial Control of the Recombination Zone in an Ambipolar Light-Emitting Organic Transistor. Nat. Mater. 2006, 5, 69−74. (24) Misewich, J.; Martel, R.; Avouris, P.; Tsang, J.; Heinze, S.; Tersoff, J. Electrically Induced Optical Emission from a Carbon Nanotube FET. Science 2003, 300, 783−786. (25) Lee, J. U.; Gipp, P.; Heller, C. Carbon Nanotube p−n Junction Diodes. Appl. Phys. Lett. 2004, 85, 145−147. (26) Dai, V.-T.; Lin, S.-D.; Lin, S.-W.; Wu, J.-Y.; Li, L.-C.; Lee, C.-P. Lateral Two-Dimensional p−i−n Diode in a Completely Undoped GaAs/AlGaAs Quantum Well. Jpn. J. Appl. Phys. 2013, 52, 014001. (27) Sze, S. M.; Ng, K. K. Physics of Semiconductor Devices; John Wiley & Sons: Hoboken, NJ, 2006. (28) Yu, H.; Wu, Y.; Liu, G.-B.; Xu, X.; Yao, W. Nonlinear Valley and Spin Currents from Fermi Pocket Anisotropy in 2D Crystals. Phys. Rev. Lett. 2014, 113, 156603. (29) Yuan, H.; Bahramy, M. S.; Morimoto, K.; Wu, S.; Nomura, K.; Yang, B.-J.; Shimotani, H.; Suzuki, R.; Toh, M.; Kloc, C.; Xu, X.; Arita, R.; Nagaosa, N.; Iwasa, Y. Zeeman-Type Spin Splitting Controlled by an Electric Field. Nat. Phys. 2013, 9, 563−569. (30) Dimitrakopoulos, C. D.; Malenfant, P. R. Organic Thin Film Transistors for Large Area Electronics. Adv. Mater. 2002, 14, 99−117. (31) Zeng, H.; Liu, G.-B.; Dai, J.; Yan, Y.; Zhu, B.; He, R.; Xie, L.; Xu, S.; Chen, X.; Yao, W.; Cui, X. Optical Signature of Symmetry Variations and Spin-Valley Coupling in Atomically Thin Tungsten Dichalcogenides. Sci. Rep. 2013, 3, 1608. (32) Zhao, W.; Ghorannevis, Z.; Chu, L.; Toh, M.; Kloc, C.; Tan, P.H.; Eda, G. Evolution of Electronic Structure in Atomically Thin Sheets of WS2 and WSe2. ACS Nano 2013, 7, 791−797. (33) Ye, J.; Zhang, Y.; Akashi, R.; Bahramy, M.; Arita, R.; Iwasa, Y. Superconducting Dome in a Gate-Tuned Band Insulator. Science 2012, 338, 1193−1196. (34) Shi, W.; Ye, J.; Zhang, Y.; Suzuki, R.; Yoshida, M.; Miyazaki, J.; Inoue, N.; Saito, Y.; Iwasa, Y. Superconductivity Series in Transition Metal Dichalcogenides by Ionic Gating. Sci. Rep. 2015, 5, 12534. (35) Costanzo, D.; Jo, S.; Berger, H.; Morpurgo, A. F. Gate-Induced Superconductivity in Atomically Thin MoS2 Crystals. Nat. Nanotechnol. 2016, 11, 339−344. (36) Shockley, W. The Theory of P-N Junctions in Semiconductors and p-n Junction Transistors. Bell Syst. Tech. J. 1949, 28, 435−489. (37) Achoyan, A. S.; Yesayan, A.; Kazaryan, E.; Petrosyan, S. TwoDimensional p−n Junction under Equilibrium Conditions. Semiconductors 2002, 36, 903−907. (38) Petrosyan, S.; Yesayan, A.; Reuter, D.; Wieck, A. The Linearly Graded Two-Dimensional p−n Junction. Appl. Phys. Lett. 2004, 84, 3313−3315.

AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected]. ORCID

Yijin Zhang: 0000-0003-1127-1124 Notes

The authors declare no competing financial interest.

ACKNOWLEDGMENTS The authors thank M. Nakano and S. Z. Bisri for insightful comments. Y.J.Z. and R.S. are supported by Japan Society for the Promotion of Science (JSPS) through the research fellowship for young scientists. R.S. is also supported by Materials Education program for the future leaders in Research, Industry, and Technology (MERIT). This research was supported by Grant-in-Aid for specially promoted research (no. 25000003) from JSPS. REFERENCES (1) Podzorov, V.; Gershenson, M. E.; Kloc, C.; Zeis, R.; Bucher, E. High-Mobility Field-Effect Transistors Based on Transition Metal Dichalcogenides. Appl. Phys. Lett. 2004, 84, 3301−3303. (2) Radisavljevic, B.; Radenovic, A.; Brivio, J.; Giacometti, V.; Kis, A. Single-Layer MoS2 Transistors. Nat. Nanotechnol. 2011, 6, 147−150. (3) Radisavljevic, B.; Whitwick, M. B.; Kis, A. Integrated Circuits and Logic Operations Based on Single-Layer MoS2. ACS Nano 2011, 5, 9934−9938. (4) Zhang, Y.; Ye, J.; Matsuhashi, Y.; Iwasa, Y. Ambipolar MoS2 Thin Flake Transistors. Nano Lett. 2012, 12, 1136−1140. (5) Wilson, J. A.; Yoffe, A. D. The Transition Metal Dichalcogenides Discussion and Interpretation of the Observed Optical, Electrical and Structural Properties. Adv. Phys. 1969, 18, 193−335. (6) Splendiani, A.; Sun, L.; Zhang, Y.; Li, T.; Kim, J.; Chim, C.-Y.; Galli, G.; Wang, F. Emerging Photoluminescence in Monolayer MoS2. Nano Lett. 2010, 10, 1271−1275. (7) 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. (8) Xiao, D.; Liu, G.-B.; Feng, W.; Xu, X.; Yao, W. Coupled Spin and Valley Physics in Monolayers of MoS2 and Other Group-VI Dichalcogenides. Phys. Rev. Lett. 2012, 108, 196802. (9) Cao, T.; Wang, G.; Han, W.; Ye, H.; Zhu, C.; Shi, J.; Niu, Q.; Tan, P.; Wang, E.; Liu, B. L.; Feng, J. Valley-Selective Circular Dichroism of Monolayer Molybdenum Disulphide. Nat. Commun. 2012, 3, 887. (10) Zeng, H.; Dai, J.; Yao, W.; Xiao, D.; Cui, X. Valley Polarization in MoS2 Monolayers by Optical Pumping. Nat. Nanotechnol. 2012, 7, 490−493. (11) Mak, K. F.; He, K.; Shan, J.; Heinz, T. F. Control of Valley Polarization in Monolayer MoS2 by Optical Helicity. Nat. Nanotechnol. 2012, 7, 494−498. (12) Zhang, Y. J.; Oka, T.; Suzuki, R.; Ye, J. T.; Iwasa, Y. Electrically Switchable Chiral Light-Emitting Transistor. Science 2014, 344, 725− 728. (13) Mak, K. F.; McGill, K. L.; Park, J.; McEuen, P. L. The Valley Hall Effect in MoS2 Transistors. Science 2014, 344, 1489−1492. (14) Zhang, Y. J.; Ye, J. T.; Yomogida, Y.; Takenobu, T.; Iwasa, Y. Formation of a Stable p−n Junction in a Liquid-Gated MoS2 Ambipolar Transistor. Nano Lett. 2013, 13, 3023−3028. (15) Pospischil, A.; Furchi, M. M.; Mueller, T. Solar-Energy Conversion and Light Emission in an Atomic Monolayer p−n Diode. Nat. Nanotechnol. 2014, 9, 257−261. (16) Baugher, B. W.; Churchill, H. O.; Yang, Y.; Jarillo-Herrero, P. Optoelectronic Devices Based on Electrically Tunable p−n Diodes in a Monolayer Dichalcogenide. Nat. Nanotechnol. 2014, 9, 262−267. 12589

DOI: 10.1021/acsnano.7b06752 ACS Nano 2017, 11, 12583−12590

Article

ACS Nano (39) Brumme, T.; Calandra, M.; Mauri, F. First-Principles Theory of Field-Effect Doping in Transition-Metal Dichalcogenides: Structural Properties, Electronic Structure, Hall Coefficient, and Electrical Conductivity. Phys. Rev. B: Condens. Matter Mater. Phys. 2015, 91, 155436. (40) Zhang, Y. J.; Yoshida, M.; Suzuki, R.; Iwasa, Y. 2D Crystals of Transition Metal Dichalcogenide and Their Iontronic Functionalities. 2D Mater. 2015, 2, 044004. (41) 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, 033305. (42) Ramasubramaniam, A. Large Excitonic Effects in Monolayers of Molybdenum and Tungsten Dichalcogenides. Phys. Rev. B: Condens. Matter Mater. Phys. 2012, 86, 115409. (43) Slinker, J. D.; DeFranco, J. A.; Jaquith, M. J.; Silveira, W. R.; Zhong, Y.-W.; Moran-Mirabal, J. M.; Craighead, H. G.; Abruña, H. D.; Marohn, J. A.; Malliaras, G. G. Direct Measurement of the ElectricField Distribution in a Light-Emitting Electrochemical Cell. Nat. Mater. 2007, 6, 894−899. (44) Pingree, L. S.; Rodovsky, D. B.; Coffey, D. C.; Bartholomew, G. P.; Ginger, D. S. Scanning Kelvin Probe Imaging of the Potential Profiles in Fixed and Dynamic Planar LECs. J. Am. Chem. Soc. 2007, 129, 15903−15910. (45) Matyba, P.; Maturova, K.; Kemerink, M.; Robinson, N. D.; Edman, L. The Dynamic Organic p-n Junction. Nat. Mater. 2009, 8, 672−676. (46) Guo, J.; Alam, M. A. Carrier Transport and Light-Spot Movement in Carbon-Nanotube Infrared Emitters. Appl. Phys. Lett. 2005, 86, 023105. (47) Tersoff, J.; Freitag, M.; Tsang, J. C.; Avouris, P. Device Modeling of Long-Channel Nanotube Electro-Optical Emitter. Appl. Phys. Lett. 2005, 86, 263108. (48) Ashcroft, N. W.; Mermin, N. D. Solid State Physics; Holt, Rinehart and Winston: Austin, TX, 1976. (49) Prasad, G.; Rao, N.; Srivastava, O. Growth, Structural Characteristics and Photoelectrochemical Behaviour of Tungsten Diselenide (WSe2) Single Crystals. Cryst. Res. Technol. 1986, 21, 1303−1311. (50) Solanki, G.; Gujarathi, D.; Deshpande, M.; Lakshminarayana, D.; Agarwal, M. Transport Property Measurements in Tungsten Sulphoselenide Single Crystals Grown by a CVT Technique. Cryst. Res. Technol. 2008, 43, 179−185.

12590

DOI: 10.1021/acsnano.7b06752 ACS Nano 2017, 11, 12583−12590