Tuning the photo-response in monolayer MoS2 by plasmonic nano-antenna

Monolayer molybdenum disulfide (MoS2) has recently attracted intense interests due to its remarkable optical properties of valley-selected optical response, strong nonlinear wave mixing and photocurrent/photovoltaic generation and many corresponding potential applications. However, the nature of atomic-thin thickness of monolayer MoS2 leads to inefficient light-matter interactions and thereby hinders its optoelectronic applications. Here we report on the enhanced and controllable photo-response in MoS2 by utilizing surface plasmonic resonance based on metallic nano-antenna with characteristic lateral size of 40 × 80 nm. Our nano-antenna is designed to have one plasmonic resonance in the visible range and can enhance the MoS2 photoluminescence intensity up to 10 folds. The intensity enhancement can be effectively tuned simply by the manipulation of incident light polarization. In addition, we can also control the oscillator strength ratio between exciton and trion states by controlling polarization dependent hot carrier doping in MoS2. Our results demonstrate the possibility in controlling the photo-response in broad two-dimensional materials by well-designed nano-antenna and facilitate its coming optoelectronic applications.

Monolayer molybdenum disulfide (MoS 2 ) has recently attracted intense interests due to its remarkable optical properties of valley-selected optical response, strong nonlinear wave mixing and photocurrent/ photovoltaic generation and many corresponding potential applications. However, the nature of atomic-thin thickness of monolayer MoS 2 leads to inefficient light-matter interactions and thereby hinders its optoelectronic applications. Here we report on the enhanced and controllable photoresponse in MoS 2 by utilizing surface plasmonic resonance based on metallic nano-antenna with characteristic lateral size of 40 × 80 nm. Our nano-antenna is designed to have one plasmonic resonance in the visible range and can enhance the MoS 2 photoluminescence intensity up to 10 folds. The intensity enhancement can be effectively tuned simply by the manipulation of incident light polarization. In addition, we can also control the oscillator strength ratio between exciton and trion states by controlling polarization dependent hot carrier doping in MoS 2 . Our results demonstrate the possibility in controlling the photo-response in broad two-dimensional materials by well-designed nano-antenna and facilitate its coming optoelectronic applications.
As a new attractive material after graphene in the family of two-dimensional materials, monolayer MoS 2 has drawn intense interests due to its intriguing physical properties, including a direct optical band gap in the visible range 1,2 , a strong exciton binding energy 3,4 , and valley selective circular dichroism MoS 2 5-7 . Monolayer MoS 2 , together with its sibling materials of WS 2 , MoSe 2 , WSe 2 , MoTe 2 , WTe 2 , have shown great potential in the applications of nano-electronics 8 , photonics 9,10 , photovoltaics 11,12 , and valleytronics 13,14 . However, as an optoelectronic material, the light-matter interaction cross-section of monolayer MoS 2 is still too small for practical device applications due to its atomic-thin characteristic. Therefore, it is highly desirable to enhance the light-matter interaction strength in monolayer MoS 2 . Many efforts have been made to achieve this goal [15][16][17][18] , and surface plasmonic resonance (SPR) turns out to be very attractive due to its simplicity and high-efficiency 19,20 . It has been demonstrated that the metallic nanoparticles can enhance the photocurrent in MoS 2 field-effect transistors 21,22 and also the light emission 23 ; 100 × 1000 nm-sized metal nano-antenna can improve the photo-response to some extent 24 . Till now, the significant enhancement of optical response in a well-controlled scheme has not been realized yet.
Here we report on the greatly enhanced and polarization controllable photoluminescence (PL) in monolayer MoS 2 by fabricating metallic nano-antenna onto MoS 2 surface. The large photo-response enhancement is realized by designing the periodic nano-antenna with characteristic lateral size of sub-100 nm to produce a plasmonic resonance matching with the excitation laser energy; the polarization controllability comes from the asymmetric optical field enhancement at the direction parallel and perpendicular to the long axis of nano-antenna. We readily realized up to one order of magnitude enhancement of PL and this enhancement can be well-controlled by incident light polarization. In addition, we show that the shifting of the PL peak positions can be achieved by controlling the oscillator strength ratio between exciton and trion states using polarization manipulation. Our results demonstrate the possibility in controlling the photo-response in broad two-dimensional materials by well-designed nano-antenna structures and facilitate its coming potential applications in optoelectronics and photovoltaics.

Results
In our experiment, monolayer MoS 2 was controllably synthesized on mica through a low-pressure chemical vapor deposition process 25 and then transferred onto SiO 2 /Si substrate. PL (Fig. 1a) and Raman spectra (inset) confirmed the monolayer nature of our MoS 2 sample: the PL spectrum shows two peaks related to A exciton at ~ 1.85 eV and B exciton at ~ 2.00 eV, and the A 1g and E g 2 1 Raman modes are separated by about 26 19 cm −1 . Since the absorbance spectrum of MoS 2 is in the visible range, nanostructures with visible wavelength SPR are primarily considered in order to match the absorbance of MoS 2 efficiently. Here we employed standard electron beam lithography and pulse laser deposition to fabricate silver nano-antenna patterns on monolayer MoS 2 (Fig. 1b). The characteristic lateral size of single nano-antenna structure is ~ 40 × 80 nm with thickness of ~ 35 nm (Fig. 1c). The incident laser polarization has an θ angle relative to the long axis of nano-antennas. To have large photo-response, our nano-antennas are intentionally designed to have one resonance matching our excitation laser of 2.34 eV (532 nm). Our simulation results in Fig. 1d showed that in the 2.0-3.0 eV spectral region our nano-antennas have strong longitudinal mode (θ = 0°) absorption with peak at ~ 2.34 eV but negligible transverse mode (θ = 90°) absorption. Therefore, we can purposely choose the 532 nm linear-polarized incident laser to efficiently excite the longitudinal mode of nano-antenna SPR without exciting the transverse mode too much.
In our results, the PL intensity of monolayer MoS 2 with silver nano-antennas is significantly enhanced compared to the bare monolayer (Fig. 2a). The maximum enhancement is approaching 10 times when the incident polarization is of θ = 0°. Interestingly, the enhancement shows an obvious polarization angle dependence: the enhancement monotonically decreases when θ increases from 0°-90°. To better understand the polarization-dependent PL intensity enhancement, we draw the integrated PL peak intensity vs cos 2 θ (Fig. 2b) and obtain an nonlinear relation with offset of about 4 at cos 2 θ = 1. We also examined the PL peak position (Fig. 2c) with polarization angle θ . It is obvious that the PL peak position displays a monotonic dependence on the θ . The PL peak red-shifts about 20 meV when cos 2 θ increases from 0 to 1 and shows saturated behavior. This result shows that the incident polarization can not only tune the PL peak intensity but also the peak position.

Discussions
From our design, only the longitudinal mode has resonance with the excitation laser and the transverse mode has negligible effect. So when our polarized excitation is with θ angle to the long axis of nano-antennas, the effective electric field will have a scaling factor of cosθ . Intuitively, the PL integrated intensity should be proportional to the electric field squared and show a quadratic dependence on cos θ , i.e. the enhancement should have a linear relation with cos 2 θ without any offset. The observed offset is attributed to the non-perfect shape of nano-antenna, where the transverse mode contributes about 30% absorption (Fig. 3). The nonlinear shape in Fig. 2b is believed to relate to the PL profile evolution and will be discussed in the latter part.
Several mechanisms may explain the polarization dependent PL peak position evolution: 2H-1T phase transition 27 , thermal effect 28 , and many-body effects 29,30 .
For 2H-1T phase transition, the crystalline structure will change from non-centrosymmetric to centrosymmetric. Essentially, second-harmonic-generation (SHG) signal is very sensitive to the crystalline symmetry of a materials [31][32][33] . As a result, SHG generation will be strong and negligible for non-centrosymmetric 2H monolayer MoS 2 phase (ordinary phase) and centrosymmetric 1T phase (possible phase induced by light excitation), respectively. We carried out SHG experiment (800 nm 100 fs) with and without SPR excitation (Fig. 4a), and the SHG  signal doesn't disappear under light excitation (Fig. 4b). Moreover, we didn't clearly observe the three characteristic 1T Raman peaks 27 in Fig. 4c. All these results above implied that the crystal structure and the original asymmetry were not changed. Therefore, the 2H-1T phase transition could be excluded in our experiment.
For thermal effect, the observed PL shifts of nearly 0.02 eV in experiments would be caused when temperature was heated by ~ 100 K 28 , while temperature rise by SPR was estimated to be ~ 10 K 27 under the current illumination intensity in our measurements. Hence, the thermal effect is neither the dominant factor for PL shifts.
SPR-excited-hot-electrons [34][35][36] with energy higher than the Schottky barrier between metal structure and MoS 2 can skip from metal into MoS 2 and dope it effectively. The optical response in two-dimensional MoS 2 are excitonic in nature and PL peak of intrinsic MoS 2 are related to the lowest energy 1 s exciton. However, when hot electron injects into the MoS 2 , it effectively dopes the MoS 2 and will lead to a new quasi-particle states of trion (or charged exciton). The trion effects were first observed in electrically gated MoS 2 29 .
To obtain more quantitative information of the many-body trion effects, we decompose the PL spectra into A, B excitons and trions X-under different θ (Fig. 5a,b) by employing Lorentzian fitting, following the method in refs 29,37. The optimal fitting center energy of A, B excitons and trions X-are 1.883 eV, 2.007 eV and 1.857 eV, respectively. The fitting FWHM of A, B excitons and trions X-are 0.0548 eV. 0.2120 eV and 0.0918 eV, respectively. It is observed that, under cos 2 θ = 0, or the smallest field enhancement, exciton accounts for the largest proportion in the PL peak. However, when cos 2 θ increases (Fig. 5c), the trion portion increases gradually and finally reach its maximum at cos 2 θ = 1, while exciton portion decreases to zero. Since the trion energy was about 20 meV, a number lower than that of the exciton energy, the combined peak position redshifts will show saturation behavior, agreeing well with our experimental observation in Fig. 2c.
These exciton-trion competition mechanism is also the origin for the nonlinear shape in Fig. 2b. When cos 2 θ increases, although the longitudinal field enhancement will contribute an increased weighting factor of cos 2 θ in PL intensity, the hot carrier doping becomes stronger at the same time. The hot carrier induced trion will result in a decreasing of the PL intensity simultaneously. These two opposite effects compete with each other and the net effect is that the PL enhancement is monotonically increasing with cos 2 θ , but smaller than the predicted linear relation by pure field calculation.
In summary, we have demonstrated an enhanced photo-response in MoS 2 by utilizing surface plasmonic resonance (SPR) based on metallic nano-antenna with lateral characteristic size of 40 × 80 nm. The enhanced photo-response is realized by the matching of resonance of nano-antanna with our incident laser energy. The photo-response shows obvious polarization dependence: the PL intensity will be enhanced from 10 to 4 times, and the PL peak position will blueshift 20 meV when polarization changes from being parallel to vertical to the long axis of the nano-antenna. This polarization dependent behavior originates from angle dependent field enhancement along long axis of the nano-antenna and is associated with exction-to-trion evolution. Our method should be versatile to enhance and control the optical response of other 2D materials, such as MX 2 , and therefore be useful for the coming optoelectronic applications of 2D materials.

Methods
Sample preparation. Monolayer MoS 2 samples, typically 30 um 2 , were controllably synthesized on mica through a low-pressure chemical vapor deposition (LPCVD) process and then transferred onto SiO 2 (300 nm)/Si substrate. The sample thickness of monolayer was confirmed by photoluminescence measurements and atomic force microscopy (NTEGRA SNOM, NT-MDT), respectively. Ag nano-antenna arrays were fabricated through a traditional electric beam lithography (e-LINE plus, RAITH) method and then followed by laser molecular beam epitaxy (LMBE450, KYKY).
Optical measurements. The photoluminescence measurements were performed using a single mode semiconductor laser centered at 532 nm. The laser was kept at 4.5 mW power and focused onto monolayer MoS 2 with and without Ag nano-antenna (40 × 80 × 35 nm) arrays by a 60 × objective (NA = 0.85, Edmond). The SHG experiment (phase transition test) was performed by using an 800 nm femtosecond laser (80 fs, 82 MHz, Tsunami) focused onto monolayer MoS 2 coated with Ag nano-antenna arrays by a 60 × objective (NA = 0.75, Edmond). The power of 800 nm femtosecond laser was kept at 5 mW. The set-ups of SHG experiment were elaborated in the supporting material. The PL and SHG signals were collected and analyzed with a grating spectrometer (SP2500i, PIActon) equipped with a liquid nitrogen cooled CCD. The Raman spectrum was measured by UV Raman Spectrograph (invia 6365, Renishaw).