The transfer of charge carriers across the optically excited hetero-interface of graphene and semiconducting transition metal dichalcogenides (TMDCs) is the key to convert light to electricity, although the intermediate steps from the creation of excitons in TMDC to the collection of free carriers in the graphene layer are not fully understood. Here, we investigate photo-induced charge transport across graphene–MoS2 and graphene–WSe2 hetero-interfaces using time-dependent photoresistance relaxation with varying temperature, wavelength, and gate voltage. In both types of heterostructures, we observe an unprecedented resonance in the inter-layer charge transfer rate as the Fermi energy (EF) of the graphene layer is tuned externally with a global back gate. We attribute this to a resonant quantum tunneling from the excitonic state of the TMDC to EF of the graphene layer and outline a new method to estimate the excitonic binding energies (Eb) in the TMDCs, which are found to be 400 meV and 460 meV in MoS2 and WSe2 layers, respectively. The gate tunability of the inter-layer charge transfer timescales may allow precise engineering and readout of the optically excited electronic states at graphene–TMDC interfaces.

The van der Waals (vdW) heterostructures of graphene and transition metal dichalcogenides (TMDCs) are not only outstanding optoelectronic elements1–22 but also represent atomic scale prototypes of donor–acceptor (DA) complexes,23–25 where the conversion of photons to free charge carriers can be manipulated with excellent control. In optically excited bulk DA complexes, the excitons dissociate quasi-adiabatically via the transfer of either the electron (e) or the hole (h) across the interface and form a transient charge transfer state, which directly impacts the quantum efficiency. In the type-II TMDC–TMDC heterostructures, the charge transfer states manifest as inter-layer excitons, resulting in quenching of the intra-layer photoluminescence spectrum.26–28 In the graphene–TMDC heterostructures, the formation of such inter-layer bound states could not be observed,19–21 possibly due to strong screening by the graphene layer.29 The ultra-fast cross-interface selective transfer of the photo-excited delocalized charge carriers has been demonstrated in graphene–TMDC19–22 heterostructures, occurring with picosecond timescales. However, significant debate persists regarding the directionality of the charge transfer,22 as well as the role of Förster type energy transfer20 across the hetero-interface when the excitons are excited in the TMDC layer. An insight into the exciton dissociation process in the graphene–TMDC heterostructures can be obtained by tuning the Fermi energy (EF) of graphene, which is also expected to affect the rate of charge transfer. However, most ultra-fast pump–probe experiments probing the charge transfer kinetics are limited in the tunability of EF, especially those without field-effect transistor (FET) geometry. Here, we have explored the eh separation process at graphene–TMDC hetero-interfaces in the FET geometry using time (t)-dependent photoresistance relaxation. We quantitatively link the photoresistance relaxation to an inter-layer electron transfer process in which the photo-excited electron undergoes a phonon-assisted transfer from the excitonic state (EX) of the TMDC to the Fermi surface of the graphene layer. The rate (τi1) of this TMDC → graphene electron transfer process is observed to be sharply peaked around a characteristic value of EF (measured with respect to the Dirac point of graphene) in both graphene–MoS2 and graphene–WSe2 heterostructures. We attribute this to a resonance-like phenomenon when EF in graphene aligns with the EX state in the TMDC layer. We also obtain estimations of excitonic binding energies (Eb) ≈400 meV and 460 meV for the monolayers of MoS2 and WSe2, respectively, which closely match with the previous studies.30–33 Our time-dependent photoresistance measurements may not only be relevant to ultra-fast photodetection and thermalization34 in graphene–TMDC vdW heterostructures but also form a new and unique spectroscopic tool to probe the optical states in TMDCs.

The graphene–TMDC vdW heterostructures were created using the layer-by-layer transfer of individual mechanically exfoliated graphene and TMDC flakes to form the vertical heterostructures35,36 [Fig. 1(a)], which were then transferred onto SiO2/p++-Si substrates, where the heavily doped Si acts as the global backgate. We have performed experiments on one graphene–WSe2 and two graphene–MoS2 heterostructures (devices 1 and 2), where single molecular layers of both graphene and TMDC were used. Figure 1(b) shows an optical micrograph of a graphene–MoS2 heterostructure (device 1) turned FET. The Raman spectra of individual flakes, device fabrication and details, and optical source calibration are shown in Figs. S1–S3 of the supplementary material, respectively. Thermally evaporated Cr/Au contacts on the top surface are used to measure the electrical resistance of the graphene layer. For wavelength dependent optical illumination, we used a tungsten–halogen lamp (Horiba, LSH-T250), which acts as a continuous optical source over the wavelength (λ) of interest (≈550 nm–800 nm).

FIG. 1.

(a) Schematic of the graphene–TMDC heterostructure along with the circuit diagram for opto-electronic measurement. (b) Optical micrograph of a typical graphene–TMDC heterostructure device. (c) Relative band alignment showing the relevant energy scales for generic graphene–TMDC heterostructures. Here, W, χ, ED, EF, EX, Eg, and Eb are the work function of undoped graphene, electron affinity of the TMDC layer, energy at the Dirac point, Fermi energy of graphene, excitonic energy level inside the TMDC layer, optical band gap of TMDC, and the binding energy of the excitons, respectively. τi and τb are the timescales of photogenerated electron transfer in the TMDC → graphene and graphene → TMDC directions, respectively. (d) Photoluminescence (PL) spectra of graphene–WSe2 heterostructures showing the quenching of the PL signal in the heterostructure region in comparison to bare WSe2. A, B, and A−1 indicate the excitonic and negatively charged trionic transitions, respectively.

FIG. 1.

(a) Schematic of the graphene–TMDC heterostructure along with the circuit diagram for opto-electronic measurement. (b) Optical micrograph of a typical graphene–TMDC heterostructure device. (c) Relative band alignment showing the relevant energy scales for generic graphene–TMDC heterostructures. Here, W, χ, ED, EF, EX, Eg, and Eb are the work function of undoped graphene, electron affinity of the TMDC layer, energy at the Dirac point, Fermi energy of graphene, excitonic energy level inside the TMDC layer, optical band gap of TMDC, and the binding energy of the excitons, respectively. τi and τb are the timescales of photogenerated electron transfer in the TMDC → graphene and graphene → TMDC directions, respectively. (d) Photoluminescence (PL) spectra of graphene–WSe2 heterostructures showing the quenching of the PL signal in the heterostructure region in comparison to bare WSe2. A, B, and A−1 indicate the excitonic and negatively charged trionic transitions, respectively.

Close modal

The band alignment at the graphene–TMDC interface, which is similar for both graphene–MoS29 and graphene–WSe237 heterostructures, suggests energy offset of ≈0.30 eV and ≈0.50 eV, respectively, between the Dirac point of graphene and the minimum (EC) of the quasi-particle conduction band of the TMDC layer [in Fig. 1(c)]. The photoluminescence (PL) spectra from our devices were obtained using a HORIBA LabRam HR tool under high vacuum condition (pressure ∼ 10−5 mbar). The PL from the TMDC layer underneath graphene is quenched compared to that from the bare TMDC region [Fig. 1(d) for the graphene–WSe2 heterostructure], confirming the significant decay of TMDC’s excitons in nonradiative pathways such as TMDC → graphene electron transfer.2,5,8,12,20,22 While the stronger quenching at room temperature than at 40 K can indicate a competition between TMDC → graphene electron transfer timescale (τi) and radiative lifetimes (τr),38,39 the PL quenching may also be due to the Förster-type energy transfer across the vdW interface in the graphene–TMDC heterostructures20 (see Fig. S4 of the supplementary material for PL quenching in the graphene–MoS2 heterostructure). The direct evidence of the charge transfer was established earlier when such heterostructures were implemented in the FET architecture.1,4,18

The transfer of charge following the dissociation of the excitons changes the resistance (R) of the graphene layer. Figure 2(a) illustrates the R vs back gate voltage (VG) characteristic of a graphene–WSe2 heterostructure, which is close to that of the pristine graphene because the WSe2 layer itself is highly resistive (>MΩ) [see Fig. S5(a) of the supplementary material for the graphene–MoS2 heterostructure]. When the optical illumination is turned on, R decreases (increases) from Roff to Ron in the electron (hole)-doped regime, indicating the transfer of electrons from the TMDC layer to graphene [Fig. 2(b) and Fig. S5(b) of the supplementary material for the graphene–MoS2 heterostructure]. Here, Roff and Ron are the steady state resistances of the graphene layer without and with the optical illumination, respectively. This observation is consistent with the earlier reports on the photoresistance in the graphene–TMDC heterostructures1,4 and can also be viewed as a photogating effect, where the net photoresistance ΔRsat = dR/dVG × eNg/Cox is the result of an effective change in VG. Here, Ng is the total change in the carrier density in the graphene channel by virtue of TMDC → graphene electron transfer once the system reaches the steady state, Cox is the capacitance of the 290 nm SiO2 per unit area, and e is the electronic charge. This is further confirmed by the observed proportionality of ΔRsat and dR/dVG (see Fig. S6 of the supplementary material). Here, ΔRsat is persistent, and the transferred electron (in graphene) and hole (in TMDC) do not recombine even after the illumination is turned off, unless a positive pulse of ≳20 V in VG is applied to reset the device.1 The persistence indicates a strongly suppressed electron backflow to the TMDC layer due to the paucity of available states in TMDC at the Fermi level of graphene. Importantly, ΔRsat [see Fig. 2(c)] is nonzero only for the photon energies (Eλ = hc/λ, where h and c are the Planck constant and velocity of light, respectively) at which the optical density-of-states (DoS) in the TMDC layer is nonzero, as confirmed from the comparison of the PL and ΔRsat [Fig. 2(c), also shown in Fig. S5(c) of the supplementary material for data from the graphene–MoS2 device]. The wavelength dependent photoresistance measurements were performed at T = 85 K, with an illumination power density (P) of 0.56 fW μm−2. The absence of the photoresponse at Eλ < optical bandgap (Eg) of the WSe2 allows us to ignore the photo-thermionic charge transfer in our devices.13,22

FIG. 2.

(a) Resistance (R)–gate voltage (VG) characteristics of a graphene–WSe2 heterostructure shown in Fig. 1(b). (b) Change in R for a 120 s optical pulse at illumination power P = 0.56 fW μm−2 and wavelength λ = 600 nm. The increase and decrease in R are observed for the hole-doped (VG = −7 V) and electron-doped (VG = −2 V) regimes, respectively. (c) ΔRsat as a function of λ showing the suppression of photoresponse for photon energy Eλ > Eg (for monolayer WSe2, Eg ≈ 1.74 eV). (d) Exponential relaxation of R for the different power level (P) of optical excitation (at λ = 600 nm). The black lines are fits to the data. (e) Dependence of the relaxation rate (τ−1) on the excitation power (P) for both graphene–MoS2 and graphene–WSe2 heterostructures. (f) Comparison of the PL of bare WSe2 and τ−1 as a function of λ. All experiments are performed in vacuum at T = 85 K, except for PL in (c) and (f) performed at T = 40 K.

FIG. 2.

(a) Resistance (R)–gate voltage (VG) characteristics of a graphene–WSe2 heterostructure shown in Fig. 1(b). (b) Change in R for a 120 s optical pulse at illumination power P = 0.56 fW μm−2 and wavelength λ = 600 nm. The increase and decrease in R are observed for the hole-doped (VG = −7 V) and electron-doped (VG = −2 V) regimes, respectively. (c) ΔRsat as a function of λ showing the suppression of photoresponse for photon energy Eλ > Eg (for monolayer WSe2, Eg ≈ 1.74 eV). (d) Exponential relaxation of R for the different power level (P) of optical excitation (at λ = 600 nm). The black lines are fits to the data. (e) Dependence of the relaxation rate (τ−1) on the excitation power (P) for both graphene–MoS2 and graphene–WSe2 heterostructures. (f) Comparison of the PL of bare WSe2 and τ−1 as a function of λ. All experiments are performed in vacuum at T = 85 K, except for PL in (c) and (f) performed at T = 40 K.

Close modal

Figure 2(d) presents the time-dependent photoresistance relaxation data at different power densities (P) of the incident illumination with λ = 600 nm from the graphene–WSe2 device at VG = −7 V and T = 85 K (see Fig. S5 of the supplementary material for the data from the graphene–MoS2 device). In this case, the power calibrated LED was used as the source of optical illumination. R(t) = Roff + ΔRsat × (1 − exp(−t/τ)) [solid lines in Fig. 2(d)] behavior is observed at all values of P, where τ is the timescale of the photoresponse. τ is observed to be inversely proportional to P over the experimental range of P. τ−1P behavior from both graphene–MoS2 and graphene–WSe2 devices is presented in Fig. 2(e). The dashed lines indicate linear fits to the data.

The exponential relaxation can be understood in terms of charge in-flow and out-flow rates to/from graphene. Following the generation of the exciton with the radiative lifetime τr in TMDC, the electron makes transition to graphene with the inter-layer charge transfer timescale τi. This leads to a negative (positive) ΔR in graphene in the electron (hole) doped regime. The P independent ΔRsat [Fig. 2(d)] indicates a P independent number density (Ng) of electron transfer. Considering the electron transfer rate (Ngng)/(Ngτi) from the EX state (having energy EX) to graphene, the electron transfer dynamics under the optical illumination can be expressed as18 

dngdt=ne(Ngng)Ngτingτb.
(1)

Here, ng is the transferred electron density at time t and ne = ϕaτr is the photo-excited electron density in the TMDC. The ϕa = αλP/Eλ is the absorbed photon flux, where αλ is the absorption coefficient of the monolayer TMDC. Considering electron’s back transfer (graphene → TMDC) timescale τbτi at EFEX, we obtain the solution of Eq. (1) as ng(t) = Ng × (1 − exp(−t/τ)). This leads to a time-dependent photoresistance relaxation, where τ = (τi/τr)(Ng/ϕa) is the timescale of photoresistance relaxation (see Fig. S7 of the supplementary material for the τ calculation details). This agrees well with the observed τ−1P (τ−1ϕa) behavior. Intriguingly, τ−1 (∝αλ) follows closely the optical DoS of the TMDC underlayer [Fig. 2(f) (WSe2) and Fig. S5e of the supplementary material (MoS2)]. The αλ demonstrates maxima at the excitonic and trionic (A, A, and B) energies in TMDCs, leading to the maxima in τ−1 at those energies. The wavelength dependent τ−1 measurements were performed at T = 85 K, with an illumination power density (P) of 0.56 fW μm−2. Considering a typical value of τr = 1 ps38,40 at low T, Ng ∼ 1011 cm−2 (see Fig. S8 of the supplementary material), and αλ = 10% in the monolayer TMDC,41 and using the observed τ−1P relation [Fig. 2(e)], we estimate τi ≈ 4 ps in the graphene–WSe2 heterostructure at 80 K. This matches well with the charge transfer timescales observed using the pump–probe experiments.12,20,22 Although charge trapping can play an important role in the photoresponse of bare TMDC phototransistors42 (see Fig. S10 of the supplementary material), ultrafast TMDC → graphene charge transfer, facilitated by τi ≈ 4 ps, constitutes the primary source of photoresponse in our graphene–TMDC devices.

The quantitative relation between the photoresistance relaxation rate τ−1 and the charge transfer rate τi1 allows us to monitor the charge transfer process as a function of energy difference between graphene’s EF and the EX state of the TMDC layer. We have performed the VG dependent photoresistance relaxation experiments in both graphene–WSe2 and graphene–MoS2 devices. Converting the instantaneous photoresistance [ΔR(t)] to ng using ng = ΔRCox/(edR/dVG), we plot the normalized ng(t) in Fig. 3(a) at different values of EF [=±vFCox|(VGVD)|/e, where , VD, and vF are the reduced Planck constant, gate voltage at the charge neutrality point, and the Fermi velocity in graphene, respectively; the “+” and “−” signs are for VG > VD and VG < VD, respectively] from the graphene–WSe2 heterostructure at P = 0.56 fW μm−2 and λ = 600 nm (shifted vertically for clarity). The characteristic τ is clearly dependent on EF. To confirm this, we calculated τ−1 from the exponential fit and plotted it as a function of EF in Figs. 3(b) and 3(c) for graphene–WSe2 and graphene–MoS2 devices, respectively. The solid traces are guides to the eye. Using the experimentally observed τ and Ng (see Fig. S8 of the supplementary material for Ng vs EF data from both devices), we then calculate the TMDC → graphene electron transfer rate τi1=τr1τ1Ng/ϕa, which exhibits a sharply peaked [Figure 3(d), the solid traces are guides to the eye] variation with EF, with the peak positions around EF ≈ −100 meV (EM) and EF ≈ 70 meV (EW) in graphene–MoS2 and graphene–WSe2 devices, respectively. Notably, the FWHM (full width at half maxima) ∼100 meV to 50 meV of the peaks closely corresponds to the excitonic linewidth of TMDC observed in the PL spectra [Fig. 1(d) and Fig. S5 of the supplementary material]. This suggests a possible resonance of EF with the EX state in WSe2 (MoS2) at EF = EW(M). To verify this, we calculate the excitonic binding energy in WSe2 (MoS2), Eb = WEW(M)χW(M), considering the undoped graphene’s work function W ≈ 4.56 eV43 and the electron affinity of WSe2 (MoS2), χW(M) ≈ 4.06 eV37 (4.27 eV44). In the literature, the reported values of W and χW(M) show approximately few tens of meV variations [see Fig. S9(a) of the supplementary material for a detailed review], resulting in up to ∼50 meV difference between the actual Eb and its calculated value. The Eb of monolayer WSe2 and MoS2 were previously studied both experimentally and theoretically. The reported values show significant variations falling within the range of ≈300 meV–700 meV (see Fig. S9 of the supplementary material for a detailed review). Our estimated Eb ≈ 460 meV and 400 meV for WSe2 and MoS2, respectively, match closely with the experimentally reported Eb using PL,30,31 transient absorbtion,32 and photoresistance spectroscopy33 and also with the theoretical results using various approaches of the effective mass model.45–49 

FIG. 3.

(a) ng/Ng vs time data from the graphene–WSe2 heterostructure during the light off–on cycle with P = 0.56 fW μm−2. The data at different EF values are shifted vertically for clarity. The solid lines present exponential fits. [(b) and (c)] τ−1 (at P = 0.56 fW μm−2 using λ = 600 nm) from graphene–WSe2 and graphene–MoS2 heterostructures, respectively, are plotted as a function of the EF. (d) Extracted τi1 vs EF data (in units of τr1) from the graphene–WSe2 (right panel) and graphene–MoS2 (left panel) heterostructures. The solid lines in Figs. 3(b)3(d) are guides to the eye. (e) The resonance in τi1 is schematically presented. The energy band diagrams of graphene and TMDC are schematically shown. The DoS of the EX (ρx) state is marked in orange trace. EF values at different VG values are shown using dashed lines. (f) τi1 vs T1 data are presented.

FIG. 3.

(a) ng/Ng vs time data from the graphene–WSe2 heterostructure during the light off–on cycle with P = 0.56 fW μm−2. The data at different EF values are shifted vertically for clarity. The solid lines present exponential fits. [(b) and (c)] τ−1 (at P = 0.56 fW μm−2 using λ = 600 nm) from graphene–WSe2 and graphene–MoS2 heterostructures, respectively, are plotted as a function of the EF. (d) Extracted τi1 vs EF data (in units of τr1) from the graphene–WSe2 (right panel) and graphene–MoS2 (left panel) heterostructures. The solid lines in Figs. 3(b)3(d) are guides to the eye. (e) The resonance in τi1 is schematically presented. The energy band diagrams of graphene and TMDC are schematically shown. The DoS of the EX (ρx) state is marked in orange trace. EF values at different VG values are shown using dashed lines. (f) τi1 vs T1 data are presented.

Close modal

Resonant electron transfer is commonly observed in the tunneling diodes and the tunneling spectroscopy studies,50 which generally occurs via the phonon or the defect-assisted pathways.51 Here, the electron transfer mechanism is schematically described in Fig. 3(e). The EF of graphene is indicated at and away from the resonance (EF(1) and EF(1), respectively). The DoS of the EX state is indicated as ρx. Mi(EX, EF) is the TMDC → graphene transmission matrix element containing the wavefunction overlap integral. The charge transfer rate τi1 is proportional to |Mi|2. At EX = EF, the resonance causes large |Mi|2, which gives rise to the peaked behavior in τi1. Figure 3(f) presents the τi1 vs T1 data from the graphene–WSe2 heterostructure at VG = −7 V. τi1 increases with an increase in T, with an activation energy ≈7 meV, indicating a phonon-assisted electron transfer process, which is previously reported to occur across TMDC interfaces.52–55 In our heterostructures, τi1<τr1 behavior is observed, which can be related to the inter-layer coupling between graphene and TMDC layers. τi1τr1 behavior is previously reported in graphene–TMDC heterostructures, which requires an exceptionally coupled interface, where the separation (d) between the monolayers of graphene and TMDC is ≈3 Å–6 Å,20,22 leading to a large PL quenching (by factor of ∼250) in the overlap region.20 With an increase in d, |Mi|2 reduces and τi increases, which degrades the PL quenching effect,20,22,56 which is consistent with the weaker PL quenching in our heterostructures {up to factor of 5 and 1.5 in the graphene–WSe2 [Fig. 1(d)] and graphene–MoS2 (Fig. S4 of the supplementary material) heterostructure, respectively}.

Apart from the charge transfer rate, the magnitude of the charge transfer is also affected when EF is dynamically tuned with respect to the EX state. Here, we observe that Ng decreases rapidly when EF approaches increasingly closer to the EX state. Ng vs EFEX data [Fig. 4(a)] from the graphene–WSe2(MoS2) heterostructure at T = 85 K (180 K) show Ngexp(EFEXkBT) behavior (solid lines), confirming that the loss of Ng occurs via a thermally activated process. Such thermally activated graphene → MoS2 transfer of electrons has been discussed previously22 and represented by the −ng/τb term in Eq. (1), that cannot be ruled out in the |EXEF| ∼ kBT regime. At equilibrium (after Ron is reached), EX and EF can act as a two state system, where ne/τi = Ng/τb or Ngτb condition should be satisfied (ne is the number density of electrons in EX). The rate of the thermal activation of the electrons from the EF to the EX state is τb1exp(ΔBkBT), where ΔBEXEF, which gives rise to the activated behavior of Ng observed in Fig. 4(a). This is further verified by the T dependence of Ng [Fig. 4(b)] at EFEX ≈ −100 meV (VG = −7 V). In a sufficiently high T range (T > 100 K), Ng decreases with an activation energy ΔB ≈ 88 meV (solid black line) that closely matches with the corresponding EXEF and validates a thermally activated scenario of electron’s back transfer from EF to the EX state. At T ≤ 100 K, a much lower activation energy of ≈8 meV is observed, which closely matches with the phonon energies in monolayer WSe2, indicating a phonon-assisted pathway of electron transfer from EF to the EX state. Figures 3(f) and 4(b) suggest that the phonons52–55 can play a crucial role in graphene ↔ TMDC inter-layer charge exchange in our heterostructures in the low temperature range (T ≤ 100 K).

FIG. 4.

(a) Ng (recorded at P ≈ 0.56 fW μm−2), from the graphene–WSe2 (at T = 85 K) and graphene–MoS2 (at T = 180 K) heterostructures, is plotted as a function of EFEX. The solid lines present the Ngexp(EFEXkBT) fit. (b) Temperature-dependence of Ng. The black solid and dashed lines correspond to activation energies ≈88 meV and 8 meV, respectively.

FIG. 4.

(a) Ng (recorded at P ≈ 0.56 fW μm−2), from the graphene–WSe2 (at T = 85 K) and graphene–MoS2 (at T = 180 K) heterostructures, is plotted as a function of EFEX. The solid lines present the Ngexp(EFEXkBT) fit. (b) Temperature-dependence of Ng. The black solid and dashed lines correspond to activation energies ≈88 meV and 8 meV, respectively.

Close modal

In summary, using the time-dependent relaxation of photoresistance in the field-effect architecture, we have identified a new resonant electron transfer from the excitonic (EX) state of TMDC to the Fermi energy (EF) of graphene and a thermally activated back transfer electron from EF to the EX state in optically excited graphene–MoS2 and graphene–WSe2 heterostructures. Our experiments yield a reasonable estimation of the excitonic binding energies (Eb) in both MoS2 and WSe2. We have demonstrated precise controllability on timescales and magnitudes of charge transfer by tuning the temperature and gate voltage.

See the supplementary material for Raman spectroscopy on individual flakes, device fabrication and details, optical source calibration, photoluminescence (PL) spectra of the graphene–MoS2 heterostructure, photocurrent measurement of the graphene–MoS2 (device 1) hybrid device, photogating effect in graphene–WSe2 hybrid devices, procedure of transfer rate (τ−1) calculation, Ng vs EF in graphene–TMDC devices, estimation of exciton binding energy (Eb), and comparison of photoresponse in the graphene–MoS2 hybrid and the bare MoS2 underlayer.

R.K., J.K.M., A.P., and T.A. contributed equally to this work.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

R.K. acknowledges financial support from Dr. D. S. Kothari postdoctoral fellowship (UGC-DSKPDF), a program by the University Grant Commission (UGC), India. The authors thank NNFC, IISc Bangalore, India, and MNCF, IISc, Bangalore, India, for providing cleanroom fabrication and characterization facilities. The authors also acknowledge DST, Government of India, for funding the project.

1.
K.
Roy
,
M.
Padmanabhan
,
S.
Goswami
,
T. P.
Sai
,
G.
Ramalingam
,
S.
Raghavan
, and
A.
Ghosh
,
Nat. Nanotechnol.
8
,
826
(
2013
).
2.
L.
Britnell
,
R. M.
Ribeiro
,
A.
Eckmann
,
R.
Jalil
,
B. D.
Belle
,
A.
Mishchenko
,
Y. J.
Kim
,
R. V.
Gorbachev
,
T.
Georgiou
,
S. V.
Morozov
,
A. N.
Grigorenko
,
A. K.
Geim
,
C.
Casiraghi
,
A. H. C.
Neto
, and
K. S.
Novoselov
,
Science
340
,
1311
1314
(
2013
).
3.
S.
Ghatak
,
A. N.
Pal
, and
A.
Ghosh
,
ACS Nano
5
,
7707
7712
(
2011
).
4.
K.
Roy
,
T.
Ahmed
,
H.
Dubey
,
T. P.
Sai
,
R.
Kashid
,
S.
Maliakal
,
K.
Hsieh
,
S.
Shamim
, and
A.
Ghosh
,
Adv. Mater.
30
,
1704412
(
2018
).
5.
Y.
Li
,
C.-Y.
Xu
,
J.-K.
Qin
,
W.
Feng
,
J.-Y.
Wang
,
S.
Zhang
,
L.-P.
Ma
,
J.
Cao
,
P. A.
Hu
,
W.
Ren
, and
L.
Zhen
,
Adv. Funct. Mater.
26
,
293
302
(
2016
).
6.
A.
Pradhan
,
A.
Roy
,
S.
Tripathi
,
A.
Som
,
D.
Sarkar
,
J. K.
Mishra
,
K.
Roy
,
T.
Pradeep
,
N.
Ravishankar
, and
A.
Ghosh
,
Nanoscale
9
,
9284
9290
(
2017
).
7.
S.
Islam
,
J. K.
Mishra
,
A.
Kumar
,
D.
Chatterjee
,
N.
Ravishankar
, and
A.
Ghosh
,
Nanoscale
11
,
1579
1586
(
2019
).
8.
Y.
Li
,
J.-K.
Qin
,
C.-Y.
Xu
,
J.
Cao
,
Z.-Y.
Sun
,
L.-P.
Ma
,
P. A.
Hu
,
W.
Ren
, and
L.
Zhen
,
Adv. Funct. Mater.
26
,
4319
4328
(
2016
).
9.
S.
Larentis
,
J. R.
Tolsma
,
B.
Fallahazad
,
D. C.
Dillen
,
K.
Kim
,
A. H.
MacDonald
, and
E.
Tutuc
,
Nano Lett.
14
,
2039
2045
(
2014
).
10.
D.
Pierucci
,
H.
Henck
,
J.
Avila
,
A.
Balan
,
C. H.
Naylor
,
G.
Patriarche
,
Y. J.
Dappe
,
M. G.
Silly
,
F.
Sirotti
,
A. T. C.
Johnson
,
M. C.
Asensio
, and
A.
Ouerghi
,
Nano Lett.
16
,
4054
4061
(
2016
).
11.
H.
Coy Diaz
,
J.
Avila
,
C.
Chen
,
R.
Addou
,
M. C.
Asensio
, and
M.
Batzill
,
Nano Lett.
15
,
1135
1140
(
2015
).
12.
M.
Massicotte
,
P.
Schmidt
,
F.
Vialla
,
K. G.
Schädler
,
A.
Reserbat-Plantey
,
K.
Watanabe
,
T.
Taniguchi
,
K. J.
Tielrooij
, and
F. H. L.
Koppens
,
Nat. Nanotechnol.
11
,
42
(
2016
).
13.
M.
Massicotte
,
P.
Schmidt
,
F.
Vialla
,
K.
Watanabe
,
T.
Taniguchi
,
K.-J.
Tielrooij
, and
F. H.
Koppens
,
Nat. Commun.
7
,
12174
(
2016
).
14.
J. Y.
Tan
,
A.
Avsar
,
J.
Balakrishnan
,
G. K. W.
Koon
,
T.
Taychatanapat
,
E. C. T.
O’Farrell
,
K.
Watanabe
,
T.
Taniguchi
,
G.
Eda
,
A. H.
Castro Neto
, and
B.
Özyilmaz
,
Appl. Phys. Lett.
104
,
183504
(
2014
).
15.
M.
Ghorbani-Asl
,
P. D.
Bristowe
,
K.
Koziol
,
T.
Heine
, and
A.
Kuc
,
2D Mater.
3
,
025018
(
2016
).
16.
C. E.
Giusca
,
I.
Rungger
,
V.
Panchal
,
C.
Melios
,
Z.
Lin
,
Y.-C.
Lin
,
E.
Kahn
,
A. L.
Elías
,
J. A.
Robinson
,
M.
Terrones
, and
O.
Kazakova
,
ACS Nano
10
,
7840
7846
(
2016
).
17.
W.
Zhang
,
C.-P.
Chuu
,
J.-K.
Huang
,
C.-H.
Chen
,
M.-L.
Tsai
,
Y.-H.
Chang
,
C.-T.
Liang
,
Y.-Z.
Chen
,
Y.-L.
Chueh
,
J.-H.
He
,
M.-Y.
Chou
, and
L.-J.
Li
,
Sci. Rep.
4
,
3826
(
2014
).
18.
T.
Ahmed
,
K.
Roy
,
S.
Kakkar
,
A.
Pradhan
, and
A.
Ghosh
,
2D Mater.
7
,
025043
(
2020
).
19.
C.
Lu
,
C.
Quan
,
K.
Si
,
X.
Xu
,
C.
He
,
Q.
Zhao
,
Y.
Zhan
, and
X.
Xu
,
Appl. Surf. Sci.
479
,
1161
1168
(
2019
).
20.
G.
Froehlicher
,
E.
Lorchat
, and
S.
Berciaud
,
Phys. Rev. X
8
,
011007
(
2018
).
21.
J.
He
,
N.
Kumar
,
M. Z.
Bellus
,
H.-Y.
Chiu
,
D.
He
,
Y.
Wang
, and
H.
Zhao
,
Nat. Commun.
5
,
5622
(
2014
).
22.
L.
Yuan
,
T.-F.
Chung
,
A.
Kuc
,
Y.
Wan
,
Y.
Xu
,
Y. P.
Chen
,
T.
Heine
, and
L.
Huang
,
Sci. Adv.
4
,
e1700324
(
2018
).
23.
A. A.
Bakulin
,
A.
Rao
,
V. G.
Pavelyev
,
P. H. M.
van Loosdrecht
,
M. S.
Pshenichnikov
,
D.
Niedzialek
,
J.
Cornil
,
D.
Beljonne
, and
R. H.
Friend
,
Science
335
,
1340
1344
(
2012
).
24.
A. E.
Jailaubekov
,
A. P.
Willard
,
J. R.
Tritsch
,
W.-L.
Chan
,
N.
Sai
,
R.
Gearba
,
L. G.
Kaake
,
K. J.
Williams
,
K.
Leung
,
P. J.
Rossky
, and
X.-Y.
Zhu
,
Nat. Mater.
12
,
66
(
2013
).
25.
H.
Tamura
and
I.
Burghardt
,
J. Am. Chem. Soc.
135
,
16364
16367
(
2013
).
26.
X.
Hong
,
J.
Kim
,
S.-F.
Shi
,
Y.
Zhang
,
C.
Jin
,
Y.
Sun
,
S.
Tongay
,
J.
Wu
,
Y.
Zhang
, and
F.
Wang
,
Nat. Nanotechnol.
9
,
682
(
2014
).
27.
F.
Ceballos
,
M. Z.
Bellus
,
H.-Y.
Chiu
, and
H.
Zhao
,
ACS Nano
8
,
12717
12724
(
2014
).
28.
X.
Zhu
,
N. R.
Monahan
,
Z.
Gong
,
H.
Zhu
,
K. W.
Williams
, and
C. A.
Nelson
,
J. Am. Chem. Soc.
137
,
8313
8320
(
2015
).
29.
A.
Raja
,
A.
Chaves
,
J.
Yu
,
G.
Arefe
,
H. M.
Hill
,
A. F.
Rigosi
,
T. C.
Berkelbach
,
P.
Nagler
,
C.
Schüller
,
T.
Korn
,
C.
Nuckolls
,
J.
Hone
,
L. E.
Brus
,
T. F.
Heinz
,
D. R.
Reichman
, and
A.
Chernikov
,
Nat. Commun.
8
,
15251
(
2017
).
30.
K.
He
,
N.
Kumar
,
L.
Zhao
,
Z.
Wang
,
K. F.
Mak
,
H.
Zhao
, and
J.
Shan
,
Phys. Rev. Lett.
113
,
026803
(
2014
).
31.
H. M.
Hill
,
A. F.
Rigosi
,
C.
Roquelet
,
A.
Chernikov
,
T. C.
Berkelbach
,
D. R.
Reichman
,
M. S.
Hybertsen
,
L. E.
Brus
, and
T. F.
Heinz
,
Nano Lett.
15
,
2992
2997
(
2015
).
32.
L.
Wang
,
Z.
Wang
,
H.-Y.
Wang
,
G.
Grinblat
,
Y.-L.
Huang
,
D.
Wang
,
X.-H.
Ye
,
X.-B.
Li
,
Q.
Bao
,
A.-S.
Wee
,
S. A.
Maier
,
Q.-D.
Chen
,
M.-L.
Zhong
,
C.-W.
Qiu
, and
H.-B.
Sun
,
Nat. Commun.
8
,
13906
(
2017
).
33.
A.
Klots
,
A.
Newaz
,
B.
Wang
,
D.
Prasai
,
H.
Krzyzanowska
,
J.
Lin
,
D.
Caudel
,
N.
Ghimire
,
J.
Yan
,
B.
Ivanov
,
K.
Velizhanin
,
A.
Burger
,
D.
Mandrus
,
N.
Tolk
,
S.
Pantelides
, and
K.
Bolotin
,
Sci. Rep.
4
,
6608
(
2014
).
34.
Q.
Ma
,
T. I.
Andersen
,
N. L.
Nair
,
N. M.
Gabor
,
M.
Massicotte
,
C. H.
Lui
,
A. F.
Young
,
W.
Fang
,
K.
Watanabe
,
T.
Taniguchi
,
J.
Kong
,
N.
Gedik
,
F. H. L.
Koppens
, and
P.
Jarillo-Herrero
,
Nat. Phys.
12
,
455
(
2016
).
35.
C. R.
Dean
,
A. F.
Young
,
I.
Meric
,
C.
Lee
,
L.
Wang
,
S.
Sorgenfrei
,
K.
Watanabe
,
T.
Taniguchi
,
P.
Kim
,
K. L.
Shepard
, and
J.
Hone
,
Nat. Nanotechnol.
5
,
722
(
2010
).
36.
P. J.
Zomer
,
S. P.
Dash
,
N.
Tombros
, and
B. J.
Van Wees
,
Appl. Phys. Lett.
99
,
232104
(
2011
).
37.
K.
Kim
,
S.
Larentis
,
B.
Fallahazad
,
K.
Lee
,
J.
Xue
,
D. C.
Dillen
,
C. M.
Corbet
, and
E.
Tutuc
,
ACS Nano
9
,
4527
4532
(
2015
).
38.
M.
Palummo
,
M.
Bernardi
, and
J. C.
Grossman
,
Nano Lett.
15
,
2794
2800
(
2015
).
39.
C.
Robert
,
D.
Lagarde
,
F.
Cadiz
,
G.
Wang
,
B.
Lassagne
,
T.
Amand
,
A.
Balocchi
,
P.
Renucci
,
S.
Tongay
,
B.
Urbaszek
, and
X.
Marie
,
Phys. Rev. B
93
,
205423
(
2016
).
40.
G.
Moody
,
J.
Schaibley
, and
X.
Xu
,
J. Opt. Soc. Am. B
33
,
C39
C49
(
2016
).
41.
W.
Zhao
,
R. M.
Ribeiro
, and
G.
Eda
,
Acc. Chem. Res.
48
,
91
99
(
2015
).
42.
M. M.
Furchi
,
D. K.
Polyushkin
,
A.
Pospischil
, and
T.
Mueller
,
Nano Lett.
14
,
6165
6170
(
2014
).
43.
R.
Yan
,
Q.
Zhang
,
W.
Li
,
I.
Calizo
,
T.
Shen
,
C. A.
Richter
,
A. R.
Hight-Walker
,
X.
Liang
,
A.
Seabaugh
,
D.
Jena
 et al,
Appl. Phys. Lett.
101
,
022105
(
2012
).
44.
C.
Gong
,
H.
Zhang
,
W.
Wang
,
L.
Colombo
,
R. M.
Wallace
, and
K.
Cho
,
Appl. Phys. Lett.
103
,
053513
(
2013
).
45.
T. C.
Berkelbach
,
M. S.
Hybertsen
, and
D. R.
Reichman
,
Phys. Rev. B
88
,
045318
(
2013
).
46.
I.
Kylänpää
and
H.-P.
Komsa
,
Phys. Rev. B
92
,
205418
(
2015
).
47.
M. Z.
Mayers
,
T. C.
Berkelbach
,
M. S.
Hybertsen
, and
D. R.
Reichman
,
Phys. Rev. B
92
,
161404
(
2015
).
48.
D. K.
Zhang
,
D. W.
Kidd
, and
K.
Varga
,
Nano Lett.
15
,
7002
7005
(
2015
).
49.
T.
Olsen
,
S.
Latini
,
F.
Rasmussen
, and
K. S.
Thygesen
,
Phys. Rev. Lett.
116
,
056401
(
2016
).
50.
J. P.
Sun
,
G. I.
Haddad
,
P.
Mazumder
, and
J. N.
Schulman
,
Proc. IEEE
86
,
641
660
(
1998
).
51.
U.
Chandni
,
K.
Watanabe
,
T.
Taniguchi
, and
J. P.
Eisenstein
,
Nano Lett.
16
,
7982
7987
(
2016
).
52.
Y.
Wang
,
Z.
Wang
,
W.
Yao
,
G.-B.
Liu
, and
H.
Yu
,
Phys. Rev. B
95
,
115429
(
2017
).
53.
K.
Wang
,
B.
Huang
,
M.
Tian
,
F.
Ceballos
,
M.-W.
Lin
,
M.
Mahjouri-Samani
,
A.
Boulesbaa
,
A. A.
Puretzky
,
C. M.
Rouleau
,
M.
Yoon
 et al,
ACS Nano
10
,
6612
6622
(
2016
).
54.
Q.
Zheng
,
W. A.
Saidi
,
Y.
Xie
,
Z.
Lan
,
O. V.
Prezhdo
,
H.
Petek
, and
J.
Zhao
,
Nano Lett.
17
,
6435
6442
(
2017
).
55.
H.
Zhu
,
J.
Wang
,
Z.
Gong
,
Y. D.
Kim
,
J.
Hone
, and
X.-Y.
Zhu
,
Nano Lett.
17
,
3591
3598
(
2017
).
56.
B.
Yang
,
E.
Molina
,
J.
Kim
,
D.
Barroso
,
M.
Lohmann
,
Y.
Liu
,
Y.
Xu
,
R.
Wu
,
L.
Bartels
,
K.
Watanabe
,
T.
Taniguchi
, and
J.
Shi
,
Nano Lett.
18
,
3580
3585
(
2018
).

Supplementary Material