Monolayer transition metal dichalcogenides (TMDs) exhibit unique physical properties such as direct band gap, strong light−matter interaction, and quantum light emission. In addition, the ease of van der Waals hetero-integration makes them a promising platform for optoelectronics, such as light emitting sources, photodetectors, and solar cells [1–3]. However, according to reports, the impact of defects on TMDs remains a challenge [4–6]. The high surface area-to-volume ratio of these TMDs leads to a large portion of energy dissipated through non-radiative recombination via surface defects, which is in competition with the radiative recombination process, and limits the light emission/absorption efficiency in light emitting/detecting devices [7–9]. Plasmonic nano structures have been reported to dramatically modify light emission [10,11]. The structures resonant with corresponding light wavelengths concentrate the electromagnetic field in subwavelength scales far below the light diffraction limit, significantly enhancing the local density of electromagnetic states (LDOS) and thereby boosting the rate of spontaneous emission through the Purcell effect [12–15]. This emission enhancement, in conjunction with the flexible property of TMDs, makes plasmonic nanostructure coupled TMD systems promising candidates for advanced flexible lighting and detection applications, which could be utilized in the development of wearable technology, flexible display panels, biomedical imaging systems, and a range of other cutting-edge fields.

Recently, a variety of nanostructured designs, including plasmonic nanocavities [16,17], nanodimers [10,18], and nanodipole antennas [19,20], have been utilized to modulate the photoluminescence (PL) characteristics of TMDs. It has been demonstrated that the electric field enhancement within the cavities or gaps of these structures is significantly greater than that in the near-field of individual nanoparticle monomers. This enhancement is attributed to the extremely small mode volume, which effectively boosts the radiative recombination rate and, consequently, the PL intensity of TMDs. In particular, nanodipole antennas are analogous to traditional radio-frequency (RF) dipole antennas. Their radiation properties, such as efficiency and directivity, as well as design principles, can greatly benefit from the well-established theories of RF antenna design [12,21,22]. Half-wavelength (λ/2) and full-wavelength (λ) antennas are prevalent designs in the antenna field. The optical equivalent of the λ/2 antenna has been shown to effectively enhance photoluminescence in quantum emitters when the antenna gap is narrow [23]. Full-wavelength λ antennas offer superior radiation efficiency and improved directivity compared to λ/2 antennas, yet they are less commonly employed in the RF domain due to their high input impedance, which is mismatched with transmission lines. However, in the optical regime, impedance matching for λ antennas is not an issue, given the similar internal impedance of optical sources [22,24]. It is advantageous to harness the outstanding properties of λ antennas for optical device applications.

In this study, we investigate the interaction between optical λ antennas and monolayer molybdenum disulfide (MoS2), as well as the subsequent alterations to their radiative properties. A variety of antenna designs, including rods with bowtie and cross shapes, along with varying sizes, were developed to examine the spontaneous emission intensity and polarization dependence through finite-difference time-domain (FDTD) simulations and experimental measurements. The results demonstrated that significant PL enhancement can be achieved using λ antennas without the need for ultra-narrow antenna gaps as reported in λ/2 antennas. The PL enhancement of MoS2 on bowtie antennas is most pronounced when the polarization direction aligns with the antenna’s axis, with the enhancement effect intensifying as the gap width narrows and the antenna’s sharpness increases. The cross-shaped antennas exhibited less polarization sensitivity. By geometry and dimension optimization, a maximum PL enhancement of 17 was achieved.

Nanoantenna shapes of bowties and crosses are designed. The structure of the bowtie antenna is illustrated in Fig. 1(a). The cross-shaped antenna consists of two vertically arranged bowtie structures. Due to near-field effects and the lightning rod effect, these structures can significantly enhance the intensity of the local electromagnetic field in the gap between two rod arms. Au was selected as the antenna material due to its high real part and low imaginary part of the dielectric constant at the wavelength of 665 nm [25]. Due to the low adhesion of gold on silicon substrates, gold is deposited after depositing a 5 nm thick titanium layer. The images under a scanning electron microscope (SEM) are displayed in Fig. 1(b), demonstrating nanoantennas of various sizes. To prevent interference from adjacent structures during testing, a spacing of 2 μm was maintained between each structure. MoS2 monolayer material is transferred onto a nanoantenna array by a wet method [26]. An image of the nanoantennas covered with MoS2 under an optical microscope is shown in Fig. 1(c). Raman spectra of MoS2 on a smooth silicon substrate and on a substrate with nanoantenna structures are depicted in Fig. 1(d). Monolayer MoS2 can be confirmed by comparing the frequency difference between the in-plane E2g1 peak and out-of-plane A1g peak [27–29]. In this experiment, the peak position difference between the E2g1 peak and A1g peak of MoS2 on a smooth substrate was 19.0cm−1, which is close to the reported monolayer MoS2 [27,28]. In addition, no shear or layer breath peaks unique to multilayer materials were observed within the range of 10−50cm−1 as shown in the inset [28], further confirming the monolayer property of MoS2. The introduction of nanoantenna will cause stress and near-field enhancement. When monolayer MoS2 is subjected to stress and near-field enhancement, the Raman spectra will show a shift in the E2g1 peak and a non-shift in the A1g peak [27,28], which is consistent with the Raman results in Fig. 1(d).

The physical image of the nanoantennas regulated light emission is depicted in Fig. 1(f). The nanoantenna is meticulously designed to resonantly interact with the 665 nm light emitted by the monolayer MoS2. The localized electromagnetic field that forms in the antenna gap enhances the local density of states (LDOS) for spontaneous emission. This enhancement increases the likelihood of radiative recombination over non-radiative processes, thereby significantly boosting the PL intensity. In addition, the coupling between excitons within MoS2 and plasmons within the antenna modifies the emission pattern, resulting in a greater proportion of the emitted photons being directed perpendicular to the surface.

The design principles for resonant structures in optical nanoantennas diverge from those in conventional radiofrequency (RF) dipole antennas. Metals in RF regions are good conductors with very subwavelength skin depth. RF antennas usually feature thin cross-sections and only make use of characteristic lengths L that are integer multiples of half the incident wavelength to attain resonance. While the conduction current is overwhelmed by the displacement current in the optical range, the non-negligible field penetration in the metal produces a much shorter effective wavelength λeff along the antenna. λeff has been reported to be positively related to the cross-section area [12,21]. The exact value of λeff is determined by the FDTD simulation. A simple rectangular rod structure is used to calculate λeff when it is resonant with the luminescence wavelength of MoS2. As shown in Figs. 2(a) and 2(b), the resonant wavelength of the rod increases with the rod length and decreases with the width when the height is set to 40 nm. The length values correspond to λeff/2 for their respective widths. By dividing the λeff/2 rod with a gap, an antenna with an effective length of λeff/2 is created. Conversely, a λeff antenna is constructed by spacing two λeff/2 rods with a gap between them.

Figure 2(c) shows the comparison of charge density distribution and electric field distribution between the λeff antenna and λeff/2 antenna under 665 nm resonance wavelength. The FDTD-simulated charge profiles indicate that the opposite charges are more densely accumulated at the ends of the gap in the λeff antenna compared to the λeff/2 antenna. Consequently, the electric field within the gap of the λeff antenna is much stronger than that in the gap of the λeff/2 antenna. The enhanced electric field will advantageously increase the spontaneous emission rate of the light source adjacent to the gap. The ratio of the enhanced spontaneous emission rate of a light source in the presence of the antenna to that in free space is known as the Purcell factor. Figure 2(d) demonstrates that the Purcell factor diminishes as the gap spacing widens. As shown in Fig. 2(e), the Purcell factor in the λeff antenna gap still maintains a high value of 25 at a gap of 50 nm, while it reduces to 11 for the λeff/2 antenna. Therefore, the λeff antenna configuration relaxes the fabrication difficulty for ultra-narrow antenna gaps needed in λeff/2, with the same Purcell factor.

The radiation patterns of a single oscillating dipole, a λeff/2 antenna integrated dipole, and a λeff antenna integrated dipole are compared in Fig. 2(f). The angle between the two half-power points in the main radiation direction of the λeff antenna is about 50º, which is better than the 79º of the λeff/2 antenna. Based on the simulation results presented above, it is evident that the λeff antenna offers substantial benefits in enhancing the radiation properties of light sources. In order to further enhance the electric field within the antenna gap using the lightning rod effect, the antenna is designed in a bowtie shape. When the sharpness of the inner side of the antenna increases, the Purcell factor of the radiation source at the antenna gap increases, and the resonance wavelength remains basically unchanged, as shown in Fig. 2(g).

Nanoantenna structures were fabricated on silicon substrates by electron beam lithography (EBL). A customized confocal microscopy system was used for micro-PL testing. The excitation laser was focused through a 100× objective lens (NA = 0.9), to form a spot of approximately 1μm2. PL is collected through the same objective lens whose light collection angle is ±64° covering the MoS2 radiation angle calculated through simulation. Samples were placed on a high-precision electric stage with a minimum movement step of 50 nm, and the relative position of the laser is observed in real time through a microscope. The system can accurately align the laser with the structural position.

Figure 3(a) illustrates the PL mapping of MoS2 deposited on a variety of nanoantenna structures featuring distinct shapes and sizes, excited by a 532 nm pump laser with power of 300 μW measured under the objective lens. Antenna structures that share the same resonant wavelength, but vary in gap spacing and the sharpness of the inner ends of their arms, are arranged in the same column. A column of structures, as marked by a dashed rectangular box, resonates at the emission wavelength of MoS2 and exhibits significantly enhanced PL intensity compared to the PL of bare MoS2. This is consistent with the Purcell effect theory and simulation outcomes, which indicate that the resonant structure alters the LDOS in the surrounding environment of the light source, thereby enhancing the spontaneous emission rate and consequently leading to a stronger PL intensity. Figure 3(b) displays the most significantly enhanced PL test results for different-shaped nanoantennas under resonance conditions. The data reveal that the MoS2 integrated with the λeff bowtie antenna exhibits the highest luminescence intensity. Attributed to the lightning rod effect, it surpasses that of MoS2 integrated with the λeff rectangular rod antenna, the λeff/2 rod antenna, as well as bare MoS2. The strongest PL enhancement of up to 11 times is achieved in comparison to the intrinsic emission of a bare MoS2 flake. In addition, the bowtie antenna has a wider resonant bandwidth due to varying distances along the long axis. The full width at half maximum (FWHM) value of the resonant peak in the PL spectrum for the bowtie antenna is 61 nm, larger than that of the rectangular rod antenna with FWHM of 52 nm. As the MoS2 B exciton is in the range of 600–640 nm, the enhancement of B exciton radiation was also observed in the experiment owing to the wide FWHM.

Time-resolved PL measurements of MoS2 on different nanoantenna structures were conducted to verify the PL enhancement mechanism. A pulse laser with a wavelength of 420 nm was used as the excitation source, with a pulse width of 48 ps and a repetition rate of 80 MHz. The excitation power measured under a microscope is 50 μW. The photoluminescence of MoS2 is detected by an avalanche photodiode (APD) with a time resolution of 60 ps. Time-correlated single photon counting (TCSPC) with a time accuracy of 4 ps is used to record the occurrence times of a large number of avalanche events, thereby constructing the fluorescence decay curve. The results are shown in Fig. 3(c) with the rising edges of the curves aligned and the amplitudes normalized. Double exponential decay functions y=y0+A1exp(−t/t1)+A2exp(−t/t2) were used to fit the curves (see the dotted line). Two decay time constants t1 and t2 were extracted, which corresponded to the fast and slow decay component times [11,30]. Obtained from the fitting results, the radiation lifetime of bare MoS2 at room temperature is 1.45 ns. The introduction of nanostructures reduces the radiation lifetime of MoS2 to 0.59 ns, 0.28 ns, and 0.18 ns for the λ/2 antenna, rectangular- shaped λ antenna, and bowtie shaped λ antenna, with corresponding lifetime shortening multiples of 2.45, 5.17, and 8.06, respectively. The lifetime shortening factor is slightly smaller than the PL enhancement factor, which is due to the enhanced excitation rate of MoS2 by the nanoantenna [30]. It is confirmed that the enhancement of PL intensity mainly comes from the shortened spontaneous emission time.

To investigate the impact of the sharpness of the inner ends of the antenna arm rods and the gap spacing on the PL spectrum, Fig. 3(d) presents the PL spectra of MoS2 with antennas having varying end widths and gap spacings, corresponding to the area marked by the black box in Fig. 3(a). The length and outer end width of these six nanoantennas are the same, ensuring that their resonant wavelengths are close. It can be seen that as the tips of the bowtie antenna become increasingly sharp, the nanoantenna’s enhancement of the PL intensifies. Concurrently, as the gap spacing between the antenna arms diminishes, the electric field strength intensifies, leading to a more pronounced enhancement of the PL by the nanoantenna. Based on the comprehensive test results across all structures, we observed that the peak of the MoS2 PL spectrum fluctuates within a range of 20 nm. This fluctuation is attributed to the coupling between excitons in MoS2 and plasmons in antennas, which have varying natural frequencies across different structures [see the scattering spectrum in Fig. 3(e)]. As depicted in Fig. 3(f), the peak wavelength of the PL spectrum experiences a redshift as the antenna length increases and the outer width of the antenna arm narrows.

Drawing from the operating principle of nanoantennas, the disparity in the lengths of the long and short axes leads to inconsistent resonance wavelengths along different orientations. Consequently, for any given single wavelength, a significant resonance effect is only achievable in one specific direction. This demonstrates that the electromagnetic response of nanoantennas is anisotropic. Given that the light emitted by MoS2 lacks distinct polarization properties, employing nanostructures with a single resonant direction is not effective in enhancing light with a polarization orientation orthogonal to the resonant direction. As a result, the PL modulation by the nanoantenna exhibits a pronounced polarization dependence [20,30,31]. Figures 4(a) and 4(b) depict the simulated electric field distribution around the nanoantenna under 665 nm resonant wavelength with orthogonal polarization directions. The results demonstrate that the bowtie-shaped nanoantenna exhibits strong polarization dependence, resonating intensely with light polarized parallel to the long axis, and generating a strong local electric field between the antenna arms. To address the limitation of the bowtie antenna resonating in a single direction, we have developed a dual-polarized cross-shaped nanoantenna that resonates in two orthogonal directions, thereby diminishing the nanoantenna’s reliance on the polarization state of the incident light.

The performance of the bowtie and cross antennas, both resonating at a wavelength of 665 nm and integrated with MoS2, has been compared experimentally. By altering the polarization direction of the excitation laser, the polarization properties of the MoS2 PL can be influenced, thereby assessing the polarization characteristics of the nanostructures [20,30]. The normalized polarization dependent PL measurement results are shown in Figs. 4(c) and 4(d). The results demonstrate that the PL intensity is more substantially enhanced along the resonance direction of the antenna, corroborating theoretical predictions. Consequently, for unpolarized light, the cross antenna configuration can provide a higher overall enhancement effect. Subsequently, PL tests were carried out on both the cross antenna and the bowtie antenna using unpolarized laser light. The results, as depicted in Fig. 4(e), reveal that the PL enhancement achieved by the cross antenna surpasses that of the bowtie antenna. This phenomenon can be attributed to the fact that cross antennas can resonate with light in two directions.

We also investigated the effect of resonant nanoantennas on the Raman spectra of monolayer MoS2. The results of Fig. 4(f) indicate that the resonant nanoantenna significantly enhances the Raman signal of monolayer MoS2, which can be attributed to the surface plasmon enhanced Raman scattering [32]. Compared with the Raman signal of bare MoS2 on silicon, the E2g1 peak of MoS2 on the nanoantenna has shifted. This phenomenon provides direct evidence for the strong electron-phonon coupling between MoS2 and a nanoantenna [33,34].

In conclusion, we have investigated the PL characteristics of monolayer MoS2 utilizing λ nanodipole antennas through a combination of simulations and experimental studies. Compared to the commonly employed λ/2 antennas, we discovered that the λ antenna not only yields a stronger electric field within the gap, leading to higher LDOS and more significant PL enhancement, but also possesses superior directivity. Through meticulous optimization of the antenna’s geometry and dimensions, we have achieved a PL enhancement factor of 17 times for MoS2 on a bowtie-shaped antenna in comparison with bare flakes. Moreover, to address the limitation of the bowtie antenna’s resonance being unidirectional, we have developed a dual-polarized cross-shaped nanoantenna that resonates in two orthogonal directions, thereby reducing the dependence of the nanoantenna’s performance on the polarization state of the incoming light. These findings suggest that significant enhancement of light emission can be attained using λ nanoantennas, obviating the need for an ultra-narrow antenna gap, as previously reported in the literature. This streamlines the fabrication requirements, thereby lowering the complexity of the manufacturing process.

Acknowledgment. The authors world like to thank the Advanced Semiconductor Laboratory of Beijing University of Posts and Telecommunications for their support.