1. Introduction

Monolayer molybdenum disulfide MoS₂ has attracted a significant amount of attention, especially in the fields of optoelectronics and valleytronics research [1–4]. The direct band gap and 2D confinement of monolayer MoS₂ have led to its many novel applications and breakthroughs in ultrafast photodetection [5] and light harvesting [6]. Light absorption efficiency fundamentally limits device performance in atomically thin MoS₂, with a peak absorption of only about 10%–20% in the visible spectrum [6,7]. There have therefore been numerous recent efforts to enhance the light matter interactions in these 2D materials with techniques based around cavities, photonic crystals, and polaritonic resonances [8]. It is well-known that metallic nanoparticles could induce strong near-field enhancements and have been widely used in surface-enhanced Raman scattering and metal-enhanced fluorescence [9]. Moreover, nanoparticles have been reported to enhance optical properties in 2D structures, including exciton binding energy modulation via hot electron transfer, and nanoscale opto-electro modulation [10–13]. The tuning of these metallic arrays and particles in strong resonance modes could lead to enormous field enhancements and strong coupling between the surface plasmon mode and the exciton transitions [14]. In this letter, we show that enhancement of exciton population upon laser excitation can be achieved through a simple spin coating process using chemically synthesized silver nano-triangles. Compared to highly symmetric structures such as spheres and discs, nano-triangles have more plasmon resonance modes. The in-plane dipole resonance can also be changed by altering the nano-triangles’ edge length, with apexes providing large local field enhancement [15]. In our case, the nano-triangles are synthesized to have a peak absorption at around 520 nm (see Fig. 1a), giving a spectral overlap with the emission of the common green laser diode (doubled YAG). Despite much effort that has been put into showing that nanoparticles can enhance PL emission. Time-resolved studies are instrumental to the development of ultrafast MoS₂ devices which utilizing surface plasmon resonances.

 

Fig. 1. a) Microscope image of MoS₂ on substrate b) Typical photoluminescence spectrum of MoS₂ showing peak at ∼1.85 eV. c) Raman spectrum of MoS₂ showing a frequency split in the two vibrational modes of 21cm⁻¹.

Download Full Size | PDF

2. Experimental

Transient reflection was performed using a pump-probe spectroscopy system. A Ti:sapphire amplifier (Coherent Legend Elite 1 kHz, 150 fs, 800 nm) seeded by a Ti:sapphire oscillator (Coherent Mira 900) was used to provide 4mJ output pulses. The amplifier output was split, and a portion of the laser power was directed into an optical parametric amplifier (Light Conversion OPerA SOLO 200 nm–2500 nm) to provide a tunable pump beam. The remainder of the beam was passed through a filter and a motorized delay stage before being focused onto a sapphire crystal for white light generation. The sample was pumped at a normal angle and probed with a white light continuum (450–900 nm) at a small angle. A chopper wheel shuttered alternating pulses of the pump beam, providing the synchronization for the detector to perform a shot-by-shot measurement. Typical probe beam size is ∼ 0.1 mm in diameter and 2-3 times smaller than the pump beam size. The reflected probe beam was analyzed by a spectrometer (Acton SpectraPro 275) equipped with a 150 lines/mm grating dispersing the light onto a line scan CCD camera (Entwicklungsbuero Stresing CCD 2000), which was triggered in synchronization with the laser pulses. MoS₂ monolayer flakes were synthesized by seeding promoter assisted chemical vapor deposition (CVD) using S and MoO₃ as the precursor and perylene 3, 4, 9, 10-tetracarboxylic acid tetrapotassium salt (PTAS) as the seeding promoter on Si/SiO₂ substrate [16,17]. The large area flakes ranged are approximately 30 µm long and were easily observed optically (see Fig. 1 a)). Optical characterization was carried out to verify MoS₂ monolayers. Figure 1 b) shows the bright photoluminescence spectrum of the as-grown MoS₂, indicating direct transition for monolayer MoS₂. A Raman spectrum is shown in Fig. 1 c). The two vibrational modes in Fig. 1 c) labelled E_2g and A_1g is at 383 cm⁻¹ and 404 cm⁻¹ corresponds to a frequency difference of 21 cm⁻¹, similar to the monolayer Raman spectra reported on samples grown with a similar method [16,18]. The frequency difference between the two Raman modes are reported to range from 20 cm ⁻¹ to 25 cm⁻¹ for monolayer to bulk MoS₂ respectively [19].

Chemically synthesized silver nano-triangles suspended in acetone were spin coated onto the MoS₂ sample surface. The SEM micrograph in Fig. 2 a) shows that the nano-triangles deposited are about 150 nm in length. The absorption spectrum of the silver nano-triangles deposited on a glass slide displayed a plasmon absorption band at ∼520 nm (see Fig. 2 b)).

 

Fig. 2. a) SEM image of the Ag nano-triangles on the MoS₂ surface. The triangles vary in size of around 150 nm base length. b) Absorption spectrum of Ag nano-triangles coated on a glass slide. c) Transient reflection spectra intensity plot of control Ag nano-triangles on glass slide (covering the A- and B- exciton region of the MoS₂) carried out under the same experimental conditions as in MoS₂ case.. The integrated signal probed at 620-600 nm (i.e., B-exciton region of MoS₂) is plotted in the black line overlaid on the colored intensity plot, showing no appreciable transient reflection signal in that wavelength window.

Download Full Size | PDF

3. Results and discussion

Figure 3 a) shows the typical transient reflection spectra of the bare MoS₂ monolayers grown on SiO₂/Si. The transient reflection was plotted with the wavelength at different probe beam delays. Dispersion in the system was corrected in software by a shift in the delay time at the different wavelengths, giving estimated reliable time resolutions of ∼300-400 fs. The sample was pumped at a wavelength of 420 nm, where MoS₂ has a relatively high absorption [20]. Two large positive signals at ∼1.87 eV and at ∼2.02 eV were observed, separated by ∼0.15 eV in accordance with transient absorption studies reported in literature [21,22]. These positive bands correspond to the A- and B-excitons that originate from the spin-orbit valence band splitting present in monolayer MoS₂. In contrast, Fig. 3c) indicates that the MoS₂ sample coated with Ag nano-triangles exhibited a much stronger bleaching signal at the same pump power and wavelength. The enhancement in the pump-probe signal was stronger on the B-exciton (i.e. ∼2 eV). Both the A- and B-exciton bleaching peaks responded to the pump pulse simultaneously in all the samples, with little or no detectable delay in rise time.

 

Fig. 3. Intensity plot of the broadband transient reflection spectra of MoS₂ (a, c), and MoS₂/Ag nano-triangles (b, d) at pumping wavelengths of 420 and 520 nm. The scale is the same across the intensity plots for comparison. Clear bleaching and induced absorption bands can be seen.

Download Full Size | PDF

On initial absorption of the ultrafast pump photon flux, carriers are created and subsequently thermalized via carrier-carrier scattering [23]. The thermalized hot carriers cool to occupy lower energy levels through scattering with the crystal lattice, and excitons are formed [24]. Both these effects cause a rise in the band edge or exciton bleaching signal and are reported to be ultrafast (i.e. <100 fs) [22]. The rise time is beyond the time resolution limit of our system for both the bare MoS₂ and MoS₂ with Ag-triangles samples [22,23]. A positive photo-bleaching signal accompanied by a negative induced absorption (in deep blue) differential signal is seen at the exciton positions in the spectrum, due to exciton self-energy renormalization [21]. We believe that the positive signal shown in Fig. 3 a-d) is largely due to the change in occupation of the electronic states (Pauli blocking) [25–27] and stimulated emission [23], related directly to the carrier population [26]. The bleach magnitude is therefore indicative of the amount of absorption of pump photons.

We observed a large increase in the pump probe signal on the samples with Ag-triangles coated on the MoS₂ across different pump wavelengths. Figure 3 b) shows the transient reflection pumped at 520 nm, far away from the absorption peak of MoS₂. For the bare MoS₂ sample with no Ag nano-triangle coating, the signal was weakened, in accordance with the smaller absorption cross-section of MoS₂ at this wavelength [7]. However, Fig. 3 d) represents the same experiment on sample coated with the Ag nano-triangles, showing an obvious large increase in the transient reflection positive signal. We note that no transient reflection signal can be seen in the 600-680 nm range in an equivalent control Ag nano-triangle coated glass slide as shown in Fig. 2 c).

The nano-triangles have a broad absorption peak at 520 nm. We believe that the increased bleaching signal is due to an enhanced absorption of the MoS₂ with Ag nano-triangles via localized surface plasmon resonance supported by the nano-metallic triangles [28,29]. To verify that the increase in transient reflectivity is a plasmonic-enhanced effect, the pump wavelength was tuned from 420 nm to 560 nm, through the broad plasmon resonance peak at ∼520 nm (see Fig. 2 b)). Figure 4 a) and b) summarizes the excitation photon energy-dependent transient reflectivity of the B- and A- exciton respectively. The transient reflectivity plotted is of the two exciton-related bleaching signals, and it was taken by integrating temporally around the delay time of 1 ps after t = 0. In Fig. 4 a) the black squares indicate that, as the photon energy of the pump was decreased from 420 nm, a change in the bleach signal was observed in both samples that was consistent with the absorption of MoS₂ [5]. The bare MoS₂ sample did not increase in bleach intensity until around 560 nm. As for the samples coated with Ag nano-triangles (red circles), the signal increased as soon as the pump wavelength reached 500 nm and continued to increase as the pump reached 560 nm. Through a simple division of the two samples’ bleaching signals for the two excitons, we roughly estimated that the increase of pump photon absorption due to the nano-triangle coating can be up to a factor of 5 shown with the green triangles. This effect was seen when the samples were probed in both the A- and B- exciton bands.

 

Fig. 4. a) and b) show the dependence of maximum transient reflection signal (integrated around 1 ps) on varying pump photon energy, probed at 2.0 eV (i.e., B-exciton), and 1.87 eV (i.e., A-exciton) respectively. Black squares (red circles) denote the transient reflectivity signal of the MoS₂ sample without (with) nano-triangle coating. The green triangles represent the enhancement factor, which was seen to rise at <2.7 eV pump energy. c) and d) depict the transient reflectivity spectra at different delay times of MoS₂ without the nano-triangle coating and with the nano-triangle coating respectively, pumped at 520 nm. The dashed line shows the position of the bleaching peak at delay times of 1-3 ps. It can be clearly seen that the A- and B- exciton bleach has similar magnitude, and that the high energy feature displays a slightly higher intensity similar to the monolayer study in in Ref. [30]

Download Full Size | PDF

The absorption spectra of the silver nano-triangles dispersed on a glass substrate in Fig. 2 b) are consistent with the wavelength dependence of the transient reflectivity of the Ag nano-triangle samples on MoS₂ (Fig. 4 a) and b)). The absorption spectra of this system should be approximately a convolution between the absorption spectra of the nanoparticles and the semiconductor [29]. Our results from the wavelength-dependent studies are also consistent with this claim. We note that plasmonic resonance peaks would undergo a small spectral shift and broadening when in contact with MoS₂, due to changes in effective refractive index of the medium surrounding the nanoparticle [12]. However, our Ag nano-triangles have a much broader resonance compared to the exciton absorption features. Hence the comparison is still valid and is in good agreement with our experimental results.

For both samples with and without the Ag nano-triangles, the exciton features (both induced absorption and bleach) underwent a blueshift simultaneously after the pump pulse. A clear bend in all the spectra can be seen in Fig. 3. We attribute this shift to the ultrafast renormalization effects discussed in detail by Pogna et al. [27,31,32]. Figure 4 c) and d) depict the spectra at different time delays of the coated and uncoated MoS₂ pumped at 520 nm. The spectra in both these samples gradually blueshifted and relaxed to a higher energy over the first 20 ps, as indicated by the dashed line (approximately 10 meV).

Li. et al. have reported evidence of hot carrier transfer and doping of MoS₂ with Au nanoparticles under plasmonic resonance excitation that similarly causes a redshift of up to 15 meV in the differential absorption spectrum, due to modifications to A- and B- exciton binding energies [11]. Comparing the plots in Fig. 3 and in Fig. 4 c) and d), we do not find such a significant difference in peak energies of the Ag nano-triangle coated sample. This implies that the relaxation of this hot electron transfer, if present in our system, is ultrafast and beyond our experimental resolution.

Figure 5 shows the transient reflection signals pumped at 420 nm and 520 nm and probed at the two exciton energies. Pumping at 420 nm, Fig. 5 a) and b) display the decay traces of the bleaching peaks probed at 1.87 eV (A-exciton) and 2.0 eV (B-exciton) respectively for the two samples. The lifetimes from fittings are shown together with the decay curves on Fig. 5, with τ_fast and τ_slow being the fast and slow time constants from bi-exponential fits. The B-exciton on the plain MoS₂ exhibited much less of an initial fast decay, and the long component was fitted with a mono-exponential after t = 1 ps (Fig. 5 b. black solid line). The B-exciton for the MoS₂ with Ag nano-triangles shows a longer τ_slow than the plain MoS₂ sample (43 ± 3 ps for the plain MoS₂ compared to 63 ± 4 ps for the MoS₂ + Ag), despite the much stronger bleach. Figure 5 a) demonstrates qualitatively slower recombination dynamics for the MoS₂ with Ag nano-triangles in the A-exciton as well, albeit to a lesser degree.

 

Fig. 5. Decay traces of the transient reflection with (red dotted line) and without (grey solid line) Ag nano-triangle coating; pumped at 420 nm, probed at a) 1.87 eV, b) 2.0 eV, and pumped at 520 nm probed at c) 1.87 eV, d) 2.0 eV. a), c), and d) were fit with bi-exponential fittings with the time constants indicated. b) was fit with a mono-exponential for the MoS₂ film, and a bi-exponential for MoS₂ with Ag coating.

Download Full Size | PDF

However, as the pump wavelength was tuned closer to the plasmonic resonance of the nano-triangles, the transient reflection dynamics were similar for both samples with and without the nano-triangles, shown in Fig. 5c) and d). The low signal of the plain MoS₂ pumped at 520 nm served as an obstacle for a more quantitative comparison. Nonetheless, it can clearly be seen that the decay is comprised of two parts for both samples pumped at 520 nm. Fittings to the bleaching decays of the MoS₂ with coated Ag had a fast component of 3 ± 0.5ps (7 ± 1ps) and a slow component of 49 ± 6 ps (59 ± 4ps) for the A- (B-) exciton.

The multi-component decay shown in Fig. 5 is commonly observed in transient absorption spectroscopy for 2D dichalcogenides [22,33]. In the first 1-2 ps, the fast recovery is due to non-radiative relaxation of free carriers. Origins of this fast recovery include carrier-carrier scattering, carriers captured by fast traps due to defects assisted by Auger scattering [34], and exciton formation from the excited free carriers [22]. The slower component is due to the recombination of the excitons, mostly through both radiative and non-radiative recombination. This has previously been extensively discussed as Auger recombination and carrier capture via “slow” defects described by Wang et al. [33–35].

Furthermore, an increased carrier generation from the plasmonic enhancement on our samples should lead to faster rates of decay due to the many-body effects that play a significant role in 2D systems. The decay dynamics, on the contrary, do not vary significantly, and in the 420 nm pump case, there is even a slight increase in the slow component of the decay lifetime. A possible explanation is that some energy transfer between the localized surface plasmons on the Ag nano-triangles and the MoS₂ excitons is in effect. Photons created by the radiative recombination of the excitons could re-excite the nano-triangle plasmonic modes, which, in turn, could transfer energy back to the exciton from the plasmon resonance. This process could prolong the bleaching of these energy bands. A similar increase in the bleaching decay of the exciton bands due to plasmon-exciton coupling was also observed by Lin et al. [36] and Yu et al. at low excitation intensity [10]. On the other hand, excitation at the plasmon resonance wavelength of 520 nm yielded much stronger plasmon-exciton coupling. We observed that this process is much faster, and we noted an increase in exciton recovery time approaching that of the bare MoS₂.

To better understand the enhanced absorption of MoS₂ by the localized surface plasmons of Ag nano-triangles, we performed electromagnetic calculations using a finite-difference time-domain (FDTD) method (FDTD Solutions, Lumerical). The Ag triangle was modelled as an equilateral triangle with a side length of 150 nm and a thickness of 16 nm. Each corner was rounded by a radius of 10 nm. Figure 6 a) shows a schematic of the different layers of materials used in our model. The simulation area was set to 600 nm × 600 nm, and the nano-triangle was positioned above a SiO₂ substrate with a spacing layer of MoS₂. The thickness of monolayer MoS₂ was 0.615 nm, and the dielectric constant of MoS₂ was taken from measurements performed by Li et al. [37]. A plane wave was used to illuminate the Ag nano-triangle. Since the structure of silver nano-triangles is highly symmetrical around the core, the polarization of the plane wave was set to be parallel to one side of the Ag triangle. The square scatter in Fig. 6 b) shows the simulated absorption spectrum of the monolayer MoS₂ without the Ag triangle, which is in good agreement with previous experiments. As expected, the two exciton absorption peaks are apparent, and the absorption is higher in the shorter wavelength range. The circle scatter in Fig. 6 b) displays the absorption spectrum in the presence of a single silver nano-triangle, indicating a clear enhancement throughout the whole spectral range. The square scatter in Fig. 6 c) presents the enhancement factor of the absorption. The calculated absorption shows a small enhancement at 420 nm, which gradually increases with increasing wavelength. The peak of the enhancement is at approximately 560 nm, which is consistent with our experimental result in Fig. 4. After integrating the electric field intensity at the monolayer MoS₂ in the presence of an Ag triangle, we found that the enhancement factor as a function of wavelength followed the line shape of the integrated electric field intensity at different illumination wavelengths (circle scatter line in Fig. 6 c). Figure 6 d) shows the simulated electric field intensity in the layer of MoS₂ at the wavelength of 550 nm. Clearly, a strong near-field enhancement around the perimeter of a silver nano-triangle is observed. The calculations indeed support the experimental results, strongly suggesting that the observed enhanced absorption is caused by the Ag nano-triangle-induced near-field enhancement.

 

Fig. 6. The FDTD simulation of the near-field enhancement near the Ag nano-triangle. (a) The structure of FDTD simulation. (b) The absorption spectra of the MoS₂ with and without the Ag nano-triangle. (c) The absorption enhancement factor (square scatter symbols) and integrated electric field intensity at the MoS₂ layer (circle scatter symbols) as a function of incident wavelength. (d) The electric field intensity in the MoS₂ layer at an incident wavelength of 550 nm.

Download Full Size | PDF

4. Conclusion

We have observed a large enhancement of femtosecond pump pulse laser absorption of large-area CVD-grown MoS₂ by spin coating Ag nano-triangles on the surface of the sample. The peak transient reflection signal was observed to increase by a factor of about 5. Broadband pump-probe transient reflection spectroscopy was used to show that the exciton bleach is significantly enhanced near a 520 nm pump wavelength. We attribute this enhancement to the change in absorption due to the broad plasmon resonance supported by the Ag nano-triangles, supporting the claim with a pump laser wavelength-dependent study on the transient reflectivity. We observed that both excitonic bleaches happen simultaneously, but the enhanced carrier generations affecting the MoS₂ excitonic dynamics are slightly different for resonance and off-resonance excitation at the plasmonic band. A further understanding of the metal nanoparticles and how their strong near field-enhancements can change the ultrafast electronic properties of 2D metal dichalcogenides is necessary for enhancing optical performance and the eventual realization of ultrafast 2D optoelectronic devices.