Tailoring photoluminescence of WS₂-microcavity coupling devices in broad visible range

2.1 Static measurements

The WS₂-microcavity coupling devices were formed by embedding triangular monolayer WS₂ flakes into the center of the Ag microcavity as illustrated in Figure 1(a). The thickness of the Ag film at the bottom and the top of the WS₂-microcavity devices is 100 nm and 20 nm, respectively. The thickness of the top half SiO₂ film (L/2) was the same as that of the bottom half SiO₂ film, varying from 40 nm to 60 nm (the parameter details of devices a–h was shown in Table S1). The pure cavity mode can be tuned from 460 nm to 680 nm (dashed lines plotted at the top of Figure S1). The monolayer WS₂ flakes in the center of the microcavity were transferred onto the bottom half SiO₂ film layer via a wet transfer method (see Methods for details). The optical microscope image of the WS₂-microcavity coupled device is shown in Figure 1(b). Before evaporating the upper Ag film, the bare monolayer WS₂ in the open cavity can be regarded as the control sample, whose reflection spectrum is plotted in Figure S1 (gray solid line). The two peaks at 620 nm and 520 nm can be assigned to the A and B exciton, respectively. Then, after evaporating the upper Ag layer, compared with the pure microcavity, a noticeable red-shifted of the reflection spectra can be found by a first look (solid lines at the bottom of Figure S1). However, by further carefully analyzing these reflection spectra, combined with the PL emission spectra, the cavity-induced modulation on the A, B, and Rydberg excitons can be revealed more clearly.

Figure 1:

Modulation of the WS₂-microcavity devices. (a) Schematic diagram of the WS₂-microcavity device. (b) The optical image of WS₂-microcavity device. (c) and (d) the reflection and PL spectra of WS₂-microcavity devices. The excitation wavelength is at 473 nm.

First, as shown in Figure 1(c), the modulation of B exciton was examined in the WS₂-microcavity coupled devices a–c, in which the cavity mode swept over the B exciton resonance. It is found that the reflection spectra of the coupled devices are weakly disturbed, which can be attributed to the weak oscillator strength of the B exciton. Nevertheless, the modulation of the PL emission can be clearly observed as shown at the bottom of Figure 1(c). As a reference, the intrinsic PL from monolayer WS₂ in the open cavity was first measured and exhibited a dominated A exciton emission at 642 nm, while the emission from B exciton at 525 nm was extremely weak (gray line in Figure 1(c)). In sharp contrast, a significantly enhanced emission around B exciton was presented in the coupled devices a–c, which possesses an enhancement factor of 25, 17, and 14 respectively. The largest PL enhancement factor is observed in device a, which can be caused by the good overlap between the cavity mode and the B exciton resonance. Besides, the peak positions are also strongly modified in the coupled devices. The PL emission peak can be tuned as far as 545 nm in device c, which has the longest cavity mode of the three devices. Another it should be noted that the intensity of the PL emission of A exciton in devices a–c was weakened due to the confinement effect by the microcavity [16], while the peak position is almost unchanged. Thus, we can easily derive the PL emission originating from the A and B excitons featuring comparable intensity in the coupled devices a–c, which provides a good foundation for the development of white light-emitting diodes based on TMDCs [37].

Next, with the increase of the cavity length, the cavity mode redshift and sweep over the A exciton resonance of monolayer WS₂. Due to the much stronger oscillator strength of the A exciton, the reflection spectra of the coupled devices d-h show a more pronounced modulation. Notably, in Figure S3, the reflection spectra of the devices e–g slightly split into the upper polariton (UP) and the LP hybrid branch, whose peaks with the anti-crossing behavior showed the evidence of strong coupling regime. In addition, with the cavity length increasing, the PL emission peak of the coupled devices also shows a red-shift from 643 nm to 656 nm, which is corresponding to the conventional strong coupled system where the emission is from the LP state [38, 39]. Moreover, by further increasing the cavity length, the shape of the PL emission from the coupled device h is more strongly modified. As the pink line shown in Figure 1(d), a shoulder peak appeared at 720 nm, which shifts over 100 nm compared with the intrinsic A exciton emission.

Finally, a more interesting phenomenon can be found in device d, in which the cavity mode is tuned to the position between the A and B exciton resonance (green line in Figure 1(d)). In this case, a noticeable PL emission peak appeared at 578 nm on the higher energy side of the A exciton. Indeed, this emission peak is very close to the 2s Rydberg state of A exciton in the previous reports [13]. Thus, we suggest that emission from the excited Rydberg state of A exciton can be obviously enhanced by the Purcell effect when the cavity mode matches the Rydberg state resonance.

Overall, through the cavity-induced modulation on the A, B, and Rydberg exciton states, the emission from the coupled devices can be flexibly tuned in the range of 490–750 nm, which covers most of the visible spectra range. Whereas, the mechanism of the cavity-induced modification on the PL emission, especially in the coupled devices with B exciton and Rydberg states is still unclear. To deeper explore the dynamics of the coupling effect in the WS₂-microcavity devices, TA spectroscopy was further carried out.

2.2 Transient absorption experiments

The first series of TA experiments (see Methods for details) were performed by a 500 nm pump laser pulse on the WS₂-microcavity coupled devices a–c to examine the modulation of B exciton. As a reference, the TA spectra of monolayer WS₂ in the open cavity without the top Ag layer was first illustrated in Figure 2(a). The plots show a dominated negative ground state bleaching (GSB) signal at 625 nm associated with the A exciton of monolayer WS₂; while the GSB signal at 525nm associated with the B exciton is much weaker, which is consistent with the previous reports [40]. The positive signal at 650 nm was ascribed to the photon induced absorption. Then, the TA spectra of the WS₂-microcavity coupled devices a–c were measured (see Figure S4). As a representative shown in Figure 2(b), the TA spectra of device b showed very different characteristics with respect to the control sample. Especially in the B exciton region, A huge negative signal appears at 535 nm, which is possibly corresponding to the LP state. While the UP state is rarely present being it in the disturbed spectral region by the pump laser. Even though, from the less disturbed initial TA spectra at 0.4 ps (Figure 2(c)), a smaller GSB signal corresponding to the UP state can be found. With the cavity length increasing, both the UP and LP states show a trend of red shifting, which conformed to the signature of strong coupling: anti-crossing behavior. Thus, we identify that strong coupling regime can be achieved with B excitons in monolayer WS₂. The dispersion of the UP and LP branches in devices a–c is plotted in Figure S6a, in which a Rabi splitting of ∼160 meV can be measured. Moreover, the amplitude of the GSB signal of the LP states in coupled devices a–c shows a ∼10-fold enhancement compared with that of bare B exciton states, which indicates that much more population present in the LP state. Hence, a large PL emission enhancement factor is expected in devices a–c.

Figure 2:

Ultrafast modulation of B exciton in WS₂-microcavity devices. (a) and (b) Evolution of the TA spectra of bare monolayer WS₂ and device a at different delay times (0.4 ps, 0.8 ps, 1.5 ps, 47 ps, and 121 ps). The region between 490 and 520 nm was removed because of distortions due to the high optical density of the sample. (c) TA spectra of monolayer WS₂ and devices a–c at 0.4 ps. (d) Dynamics of the LP states of devices a–c and B excitons in monolayer WS₂. The excitation wavelength is at 500 nm.

On the other hand, such enhancement can be further explained by analyzing the dynamic process of the strong coupled system with B exciton. As shown in Figure 2(d), the relaxation rates of the LP states in devices a–c are all significantly slower than that of bare B exciton. Indeed, when the strong coupled hybrid states are formed with high-energy excited state, the internal conversion pathway, that is the population flows from the high-energy to the low-energy excited state, can be blocked [41]. So different from bare monolayer WS₂, in which the population in B exciton state would first rapidly relax to the lower energy level, namely A exciton state, and then to the ground state, the population in the LP state (coupled with B excitons) preferred to directly relax to the ground state without going through the A exciton state. At the same time, due to the population in the LP state being hugely increased as we have discussed above, hence the large PL emission enhancement factor relative to the LP states (coupled with B excitons) is expected in the coupled devices a–c, as we have observed in static PL emission measurements.

Next, to reveal the cavity-induced modulation on the PL emission of the A exciton, TA measurements were performed on devices d–h under 500 nm excitation (Figure S5). Since the cavity modes of devices e–g have a good overlap with the A exciton resonance, their initial TA spectra at 0.4 ps are first analyzed and shown in Figure 3(a). Compared with bare monolayer WS₂, two distinctive GSB signals associating with the hybrid cavity mode and A exciton states, appear with an anti-crossing behavior with the cavity length increasing, which is consistent with the signature of strong coupling more clearly. A Rabi splitting up to 110 meV can be extracted as shown in Figure S6b. Furthermore, by comparing the peak position of the LP states in TA spectra and the PL emission (Figure 1(d)) in devices e–g, the PL emission can be more clearly related to the LP state. Whereas, the lifetime of the LP states is not noticeably different from that of bare monolayer WS₂. As shown in Figure 3(b), the plots show that the LP states process a similar lifetime as the bleaching recovery of the A exciton, which indicates that the relaxation process of the LP states coupled with A excitons was not significantly modified. It is also the reason why the cavity-induced PL enhancement factor is lower in the strong coupling devices with A exciton. In addition, the GSB peak of device h is tuned to 720 nm shown in Figure S7, which corresponds to that the PL emission of device h is also tuned to 720 nm.

Figure 3:

Ultrafast modulation of A exciton in WS₂-microcavity devices. (a) TA spectra of monolayer WS₂ and devices e–g at 0.4 ps. (b) Dynamics of the LP states of devices e–g and A exciton of monolayer WS₂. The excitation wavelength is at 500 nm.

Finally, we turn to the TA study on device d, in which an unusual PL emission peak appeared on the higher energy side of the A exciton. In general, PL emission from excited states is difficult to obtain due to Kasha’s rule [42]. Specifically, for high-energy excited states, the population density is usually lower and the radiation dipole moment is weak, making it a difficult task to reveal these states in static optical measurements. Thus, TA experiments were further performed on device d and the result is shown in Figure 4(a). Compared with the control sample (Figure 2(a)), one obvious difference is that an additional GSB signal appears at 570 nm between the A and B exciton states.

Figure 4:

Ultrafast modulation of 2s Rydberg state  in WS₂-microcavity devices. (a) and (b) TA spectra of device d at different delay times under 500 nm (a) and 620 nm (b) excitation respectively. The insert picture in (a) shows internal exciton states of monolayer WS₂. (c) Comparison TA and PL spectra of device d. The PL spectrum of bare monolayer WS₂ was shown by the dark gray solid line.

Indeed, it has been demonstrated that there are a series of internal exciton states in monolayer WS₂ on the analogy of the Rydberg energy levels in the hydrogen atom, which can also be characterized by the principal quantum number (n = 1, 2, 3, …) and the s, p, d orbitals with quantized angular momentum as the insert picture shown in Figure 4(a) [9]. However, due to the weak oscillator strength and short lifetime of the high-energy Rydberg state, their emission is much weaker than that of 1s state. Even though for high-quality monolayer WS₂ encapsulated in h-BN film, the 2s, 3s, and 4s excitons have been well confirmed by utilizing photoluminescence excitation spectroscopy. In particular, the 2s excited state of A exciton in monolayer WS₂ is assigned at 2.17 eV [13]. Thus, we would like to associate the additional GSB signal at 570 nm with the 2s excited states, which arise from the cavity-induced enhancement. As a result, the unusual PL emission peak at 578 nm can also be explained by the enhanced population density of 2s states as revealed in the TA spectra. Another should be noted is that the energy level of 2s Rydberg state is unperturbed since the additional GSB signal peak coincided with the original position of 2s Rydberg state and not splitting. Hence, we suggest that the coupling between 2s Rydberg state and the cavity mode is in the weak coupling region, which is caused by the extremely weak oscillator strength of the Rydberg state exciton.

To ensure the additional GSB signal comes from the enhanced 2s Rydberg states, device d was further excited by a pump laser at 620 nm, in resonance with the A exciton. The TA spectra are reported in Figure 4(b). The additional GSB signal peak at 570 nm can still be observed and is in good agreement with that under 500 nm excitation, which demonstrated that the additional state and the A exciton shared the same eigenstate. Hence, we confirm the additional GSB signal is the Rydberg state of A exciton. Finally, the extracted TA spectra at 6 ps excited under 500 and 620 nm, as well as the corresponding PL emission are compared in Figure 4(c), in which the coincided additional GSB signal at 570 nm and the corresponding noticeable PL emission peak appeared at 578 nm clearly fingerprint of the cavity-enhanced 2s Rydberg states.