The discovery of graphene in 2004^[1,2] has stimulated extensive studies on its novel property and potential applications. Graphene is formed by a single layer of carbon atoms bound together in a hexagonal lattice. Due to its unique structure, graphene has many superior properties, such as high Young’s modulus and fracture strength,^[3] high thermal conductivity,^[4] ultrafast dynamic optical properties,^[5] and high charge carrier mobility.^[1] These properties make graphene an attractive candidate for various applications, such as ultracapacitors,^[6,7] solar cells,^[8–11] photodetector,^[12] and low-power-consumption electronics.^[13,14] However, the lack of a bandgap limited its application in logic electronic devices. Furthermore, its relatively small optical absorbance is also a drawback for optoelectronic applications. Monolayer transition metal dichalcogenides (TMDs), on the other hand, have remarkably high absorbance in the visible range^[15] and a sizable bandgap.^[16,17] However, their charge carrier mobilities are relatively low. Hence, combining graphene and TMD can potentially produce bi-layer materials that can effectively absorb light and transfer charge carriers, which are two key elements for most optoelectronic applications.

Indeed, very recently, significant progress has been made in studies of graphene–TMD heterostructures. So far, most studies have focused on combining graphene with MoS₂. Initially, such heterostructures were fabricated by manually stacking graphene and MoS₂ monolayers together.^[18] Mechanical properties of graphene–MoS₂ were studied both theoretically and experimentally.^[19,20] The electronic structure of the formed heterostructure was calculated, measured, and controlled.^[21–26] For electronic applications, tunneling transistors have been demonstrated with MoS₂ serving as the tunneling barrier.^[18,27–33] Besides these investigations on graphene–MoS₂, heterostructure formed by graphene and tungsten based TMD monolayers has also been studied. For graphene–WS₂, spin–orbit interaction^[34] and various applications have been attempted, such as tunneling transistors,^[29,35] photovoltaics,^[36] light-emitting diodes,^[37] and photodetection.^[38] Measurements of its band alignment,^[39] photoluminescence properties,^[40] and light-emitting devices^[37] have been reported.

In contrast to these extensive efforts on developing heterostructures formed by graphene and MoS₂ and WS₂, MoSe₂ has been seldom used to form heterostructures with graphene. The only reports on graphene–MoSe₂ heterostructures are their molecular beam expitaxy^[41] and observation of photoluminescence quenching.^[42] MoSe₂ possesses several properties that make it an attractive member of TMDs. It has a direct optical bandgap of 1.55 eV,^[43] which is near the optimal bandgap of single-junction photovoltaic devices and photocatalysis.^[44–46]

Here we report fabrication of graphene–MoSe₂ heterostructures and ultrafast laser measurements on photocarrier dynamics. We observed efficient carrier transfer from MoSe₂ to graphene, and strong effect of carriers in graphene on optical properties of MoSe₂. These results indicate that graphene–MoSe₂ heterostructures are promising materials for optoelectronic applications.

Graphene and MoSe₂ flakes were fabricated by mechanical exfoliation. Adhesive tapes were used to mechanically exfoliate flakes from bulk crystals onto polydimethylsiloxane (PDMS) substrates. The monolayers were identified by optical contrasts with an optical microscope. Then a MoSe₂ monolayer flake was transferred to a Si substrate with a 90 nm SiO₂ layer and annealed for 2 h at 200 °C in an Ar (60 sccm) environment with a pressure of 3 Torr. Next, a graphene flake was transferred onto the MoSe₂ flake, followed by the same annealing procedure. The final optical microscope image of the sample is shown in Fig. 1(a), where the graphene–MoSe₂ heterostructure is in the triangle yellow area. Figure 1(b) illustrates the predicted band alignment^[47,48] of the heterostructure. We note that the band gap of MoSe₂ presented in Fig. 1(b) is a theoretical value, which is different from that of the experiment and has no influence on the measurement.

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. (color online) (a) Microscope images of the samples studied. (b) Band alignment of graphene and MoSe2 monolayers. (c) Experimental setup to measure differential reflectivity.

In the transient absorption microscopy setup shown as Fig. 1(c), a passively mode-locked Ti:sapphire oscillator was used to generate a 100 fs pulse with a central wavelength of 790 nm at 80 MHz. We used a beamsplitter to separate the pulse into two beams. One of the beams was coupled to a photonic crystal fiber to generate supercontinuum. A bandpass filter with a passing wavelength of 620 nm and a bandwidth of 10 nm was employed to select a 620 nm pulse from the supercontinuum,which served as the pump. Combined with the other beam probe which was outputted directly from the oscillator, the two beams were finally focused onto the sample by a microscope objective lens. The reflected probe was collimated by the objective lens and measured by one detector of a balanced detector. A portion of the probe beam was taken as a reference beam, which is sent to the other detector of the balanced detector. A lock-in amplifier was used to measure the voltage output of the detector. A mechanical chopper was placed in the pump arm to modulate the intensity of the pump beam at about 2 kHz. Hence, the balanced detector now outputs a voltage that is proportional to a differential reflectivity of the probe, R/R₀. It is defined as the relative change of the probe reflectivity caused by the pump, (R-R₀/R₀, where R and R₀ are the reflectivity of the probe with the pump presence and without it, respectively. All the measurements were performed at room temperature with the sample exposed in air.

We first studied a MoSe₂ monolayer sample. A pump pulse of 2.00 eV was used to inject photocarriers. A probe pulse of 1.57 eV, which is tuned near to the exciton resonance of MoSe₂, was used to monitor these photocarriers. The top panel of Fig. 2(a) shows the differential reflectivity signal as a function of the probe delay. In this measurement, the pump fluence is 4.9 μJ/cm². By using an absorption coefficient of 2 × 10⁵ cm⁻¹ for MoSe₂ monolayer at the probe photon energy,^[49] an injected carrier density of 2.1 × 10¹⁰ cm⁻² was established. A peak differential reflectivity signal of 1.14 × 10⁻⁴ was observed. Furthermore, the decay of the signal can be fitted by a bi-exponential function, with two time constants of 22 and 125 ps, respectively. The rest of Fig. 2(a) shows the measured signal at different pump fluences. By fitting these data, we found that as the pump fluence decreased, the fast decay component characterized by 22 ps becomes less pronounced. Based on this feature, we can attribute the long time constant of 125 ps to the photocarrier lifetime in MoSe₂. The fast decay channel at higher fluence can be attributed to the contribution of exciton–exciton annihilation.^[50]

Fig. 2.

Figure Option ViewDownloadNew Window

Fig. 2. (color online) Differential reflectivity measurement of monolayer MoSe2. (a) Differential reflectivity signal as a function of probe delay with pump fluences of (from top to bottom) 4.9, 4.29, 3.06, 2.45, 1.84, 1.23 and 0.61 μJ/cm2, respectively. The red curves are exponential fits. (b) Peak differential reflectivity signal as a function of pump fluence. (c) Peak differential reflectivity signal as a function of the probe photon energy.

Figure 2(b) summarizes the peak differential reflectivity signal as a function of the pump fluence. A linear relation is clearly observed, as confirmed by the linear fit (red line). Finally, with a fixed pump fluence of 1.23 μJ/cm², we repeated the measurement with different probe photon energies. Figure 2(c) shows the peak differential reflectivity signal as a function of the probe photon energy. The peak signal was observed at a probe photon energy of 1.57 eV, which is well consistent with the previously determined optical bandgap of MoSe₂ monolayers. This observation shows that the probe pulse senses the photocarriers via the change of the excitonic absorption peak induced by these carriers.

Figure 3 shows the results of the same measurement performed with the graphene–MoSe₂ heterostructure. If there was no interlayer photocarrier transfer or no interlayer coupling, the results should have been similar to those shown in Fig. 2. Due to the smaller absorption coefficient of graphene compared to MoSe₂, the carriers injected in graphene can be neglected for simplicity, and the pump pulse can be assumed to inject the same carrier density in the MoSe₂ of the heterostructure as the MoSe₂ monolayer. However, we observed two dramatic differences between the two measurements. First, the signal magnitude is about a factor of 10 larger in the heterostructure sample under the same conditions. Second, the signal decays rapidly compared to the MoSe₂ monolayer. Exponential fits (blue curves) produced a decay time constant of 8.5 ps. Meanwhile, similar dependences on the pump fluence and probe photon energy are observed.

Fig. 3.

Figure Option ViewDownloadNew Window

Fig. 3. (color online) Differential reflectivity measurement of graphene–MoSe2 heterostructure. (a) Differential reflectivity signal as a function of probe delay with pump fluences of (from top to bottom) 4.9, 4.29, 3.06, 2.45, 1.84 and 1.23 μJ/cm2, respectively. The blue curves are exponential fits. (b) Peak differential reflectivity signal as a function of pump fluence. (c) Peak differential reflectivity signal as a function of the probe photon energy.

We attribute these observed features to two physical mechanisms. First, the photocarriers excited in MoSe₂ rapidly transfer to graphene. Second, the carriers in graphene can induce a differential reflectivity signal of the probe tuned to the MoSe₂ resonance.

The strong dependence of the peak signal on probe photon energy indicates that the signal originates from a change of the absorption coefficient of MoSe₂. However, this change cannot be induced by the photocarriers in MoSe₂, since in the measurements on MoSe₂ monolayer (Fig. 2), we have established the magnitude of the signal for such photocarrier densities used in the measurements. The signal is too large to be attributed to photocarriers in MoSe₂. Furthermore, the decay of the signal is very fast. Since the lifetime of photocarriers in graphene was known to be on the same time scale, this further indicates that the signal monitors the carriers in graphene, instead of MoSe₂.

We assume that the mechanism for change of the absorption coefficient of MoSe₂ by carriers in graphene is via a screening effect of these carriers on the electric field of the excitons. It has been well established that the Coulomb interaction between electrons and holes in monolayer TMDs is significantly enhanced by the reduced dielectric screening. As shown in Fig. 4, the majority of the field lines are in the vacuum surrounding the monolayer. This effect has resulted in extremely large exciton binding energies in these materials. When combined with a graphene layer, the carriers in graphene can screen the fields in that layer, and hence change the interaction between the electrons and holes in excitons.

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. The electric field of the excitons in MoSe2 before and after forming the heterostruture.

Therefore, our results provide quantitative information on the physics mechanism of screening of graphene on many-body interactions in MoSe₂ monolayers. In particular, it is possible to control the electron–hole interaction in MoSe₂, as well as other 2D materials, by interfacing with graphene with a certain thickness. This opens up the opportunities of controlling electron–hole interactions in van der Waals materials.

Based on this mechanism and the fast decay of the signal observed in Fig. 3, as well as the lack of a long-lived signal, we conclude that photocarriers excited in MoSe₂ rapidly transfer to graphene. These carriers in graphene can alter the absorption of MoSe₂.

The observed effects have important implications on using these materials in optoelectronic devices. For example, the efficient transfer of photocarriers from MoSe₂ to graphene suggests that such bilayers can be used in photodetectors and solar cells. MoSe₂ has a large absorption coefficient at optimal wavelength for solar cells, while graphene possesses superior charge transport performance. The bilayer structure effectively combines these advantages. The demonstrated control of MoSe₂ absorption by carriers in graphene can be utilized in light modulation applications where gate controlled carriers in graphene can be used to modulate absorption of light by MoSe₂.

We have fabricated a less investigated graphene–MoSe₂ heterostructure, and studied its photocarrier dynamics. We found that photocarriers injected in MoSe₂ transfer to graphene on an ultrafast time scale. We also found that a carrier in graphene can change the excitonic absorption of MoSe₂, which can be potentially used for electric control of optical absorption of MoSe₂. Our results illustrate that graphene–MoSe₂ heterostructures can effectively combine the novel optical absorption property of MoSe₂ and charge the transport property of graphene, for potential applications in optoelectronic devices.