1. Introduction

An exciton–polariton is a type of bosonic quasi-particle originating from the strong coupling between electron-hole pairs (excitons) and photons confined in an optical microcavity. Exciton–polaritons have attracted considerable attention in the past decades because of their bosonic nature and unique optical properties [1–6]. Electrically injected light-emitting devices, which emit polaritons originating from the spontaneous formation of a coherent population of the exciton—polaritons, can function as a coherent light source approximating conventional lasers. This type of polariton device can produce polariton lasing action with an ultra-low threshold, which is significantly lower than the pumping rates necessary for population inversion of the excitons [7–11]. Thus, it is possible to achieve high-density macroscopically occupied polariton states at relatively high temperatures because of the strong coupling interaction confined in optical cavities. Unlike the conventional photon lasing devices, the polariton lasers do not require population inversion. Further, it is predicted that the threshold for polariton lasers is several orders of magnitude smaller than that of the conventional semiconductor lasers [11–15]. Therefore, polariton lasers are appealing owing to their promising applications as coherent light sources, which can display extremely low thresholds when compared with the conventional semiconductor lasers [16–20].

In the fabrication of exciton–polariton emission devices, wide-bandgap semiconductors are more suitable for the construction of high-performance polariton optoelectronic devices, which are capable of operating at high temperatures. These wide band gap semiconductors are preferred because of their higher exciton binding energy, oscillator strength, and efficient carrier relaxation [21–24]. Over the past few years, the polariton emission devices have been fabricated mainly using bulk materials, quantum wells, and micro- or nanostructures (such as nanowires, nanotubes, microwires, etc.). In the case of micro- or nanostructures, many organic and inorganic semiconductors have been particularly favored as active media along with optical microresonators [25–28]. Among the various available wide-bandgap semiconductors, ZnO micro- or nanostructures are preferred for producing stable exciton–polaritons, owing to their significantly high exciton binding energy ($\sim$ 60 meV) and Rabi splitting energy at room as well as higher temperatures [4,29–34]. Although optically and electrically pumped polariton emissions and polariton-related lasing actions have been reportedly achieved, the fabrication of ZnO micro- or nanostructure-based polariton devices with higher exciton binding energy remains a challenging task [12,14,15,19,35]. Nanostructured metals possess unprecedented capabilities of optical mode confinement below the diffraction limit and production of strong localized electromagnetic fields. Consequently, they have been applied widely in tailoring devices capable of coupling the excitons with the electromagnetic field [36–39]. Therefore, searching for alternative ultraviolet plasmonic nanostructures is crucial since they may provide a tool to improve the quality factor of optical microresonators formed by ZnO micro- or nanostructures. Such alternative metal nanostructures can enhance the microresonator induced exciton–photon coupling strength to develop high-performance polariton devices [24,40–43].

In this study, electrically driven exciton–polariton light-emitting diodes (LEDs), composed of a single ZnO microwire (MW) and p-GaN substrate, which acts as the hole injector, were fabricated featuring the main electroluminescence (EL) peaks, centered at 410.5 nm, in the blue-violet region. A series of sharp peaks, arising from the exciton–polariton behavior, were resolved from the EL spectra via angle-resolved EL measurements. To improve the device performance, Ag nanowires (AgNWs) with desired plasmons were deposited on the hexagonal MW. Optical pumping can produce significant enhancement in the emission characteristics of the device, including the achievement of a reduced whispering-gallery-mode (WGM) lasing threshold, higher quality WGM microcavity, and remarkably enhanced light output. This enhanced mechanism can be attributed to the perfect overlap of the AgNW plasmons and ZnO excitons. Accordingly, the ZnO MW, covered by the AgNWs, was also used to construct a heterojunction LEDs. Using the AgNWs, a significant reinforcing effect on the EL features of the device, including a slight blueshift of the characteristic EL peaks and an enhanced light output was realized. In particular, the ultraviolet plasmons of the AgNWs enhanced the exciton–photon coupling strength in the as-fabricated LED. The characteristic subband anti-crossing features of the as-fabricated single MW-based LEDs indicated that a Rabi splitting energy $\sim$ 495 meV can be obtained, which is considerably higher than that of the single bare ZnO MW-based LED (Rabi splitting energy $\sim$ 370 meV). From the application point of view, plasmon-induced high exciton binding energy of an exciton–polariton LED can pave the way for the development of electrically driven exciton–polariton coherent light sources, which may have an ultra-low lasing threshold.

2. Experimental section

2.1 Growth and preparation of a single AgNWs@ZnO MW

The individual ZnO MWs with hexagonal cross-sections were prepared successfully via a simple single-step chemical vapor deposition (CVD) method [44–47]. Further, Ag nanowires (AgNWs), with controlled sizes, were synthesized by the slow injection polyvinylpyrrolidone (PVP)-directed polyol method as [42,48]: (1) 15 mL of ethylene glycol (EG) was added into a flask under magnetic stirring in an oil bath at 160${}^{\circ }\mathrm {C}$ and heated for 5 min; (2) 5 mL of PVP (120 mg/mL in EG), NaCl (300 mM in EG), and NaBr (300 mM in EG) were pipetted sequentially into the flask, followed by fresh AgNO$_3$ (25 mg/mL in EG); the magnetic stirring of the solution was stopped after 5 min of the reaction; (3) the temperature of the reaction mixture was maintained at 160${}^{\circ }\mathrm {C}$ of about 60 min for the growth of the AgNWs. During this reaction, the diameter of the nanowires can be controlled by varying the amount of precursor. The residual organics were removed by centrifugal processing.

A single ZnO MW was selected for transfer on a cleaned quartz substrate. Next, the as-prepared AgNW solution was spin-coated on the entire MW, followed by annealing at 100${}^{\circ }\mathrm {C}$ for approximately 30 min. By controlling the volume of the spin-coated droplets, the surface coverage of the spin-coated AgNWs can be easily adjusted. Thus, a single ZnO MW, coated with AgNWs (AgNWs@ZnO), was successfully prepared [38,42,49].

2.2 Device fabrication

A heterostructured device, composed of a single MW and p-GaN substrate, was fabricated by the following process [42,46,50]: First, by using molecular beam epitaxy, a MgO thin film with a thickness of about 100 nm was deposited on a cleaned p-GaN substrate. Second, Ni/Au was evaporated on the p-GaN substrate using the electron-beam evaporation method. After thermal annealing in air, Ni/Au metal nanofilm forms an ohmic contact with the p-GaN substrate. Finally, a single MW was transferred on the p-GaN substrate and fixed on the MgO insulating layer with In, which served as an electrode. Thus, Ni/Au and In can function as contacting electrodes for current injection. Therefore, a single MW-based heterojunction light-emitting device was fabricated successfully.

2.3 Instrumentation

The as-synthesized single ZnO MW, both not covered and covered by AgNWs, were characterized by a scanning electron microscope (SEM). The extinction spectra of the AgNWs were recorded with a ultraviolet-6300 spectrophotometer. The photoluminescence (PL) measurements of the single ZnO MW, not covered and covered by AgNWs, were performed by a LABRAM-UV Jobin Yvon spectrometer equipped with a 325 nm He–Cd laser as the excitation source. The optically pumped lasing measurements of the single ZnO MWs, not covered and covered by AgNWs, were performed with a femtosecond (fs)-pulsed laser (excitation wavelength: 355 nm; repetition rate: 1 kHz; pulse length: 100 fs). The excitation laser pulse was generated from a Ti:sapphire laser by using an optical parametric amplifier (OPERA Solo, Coherent Inc.). This Ti:sapphire laser was focused on the body of the wire via a confocal micro-PL system (BX53, Olympus Corporation). Furthermore, the electrical properties with current–voltage ($I$–$V$) characteristics of the single ZnO MWs, not covered and covered by AgNWs, and the electronic transport properties of the single MW-based heterojunction device were determined by a semiconductor parameter device analyzer and measurement module (B1500A, Keysight technologies).

EL measurements of the as-fabricated single MW-based LEDs were performed using a charge-coupled device (CCD) camera (1024BR-PIXIS series CCD camera, Princeton instruments). To further analyze the exciton–polariton characteristics of the single MW-based LED, the far-field angle-resolved interference pattern of the corresponding EL signals were collected by the same objective and analyzed by the aforementioned spectrograph. In the angle-resolved EL measurement, the single MW utilizing in the emission device was rotated along its axis, which was perpendicular to the slit of the monochromator. The optical microscopy EL images of the light emission from the electrically illuminated single MW-based LEDs were recorded by a CCD camera (Olympus) using a high numerical aperture microscope objective. All measurements were performed at room temperature.

3. Results and discussion

Figure 1(a) shows a typical SEM image of the as-synthesized ZnO MW (diameter $\sim$ 10 $\mu$m), which possesses a perfect hexagonal, cylindrical structure. SEM image of the ZnO MW covered with the AgNWs is shown in Fig. 1(b). It is clearly seen that the AgNWs were deposited uniformly on the surfaces of the MW. Additionally, an enlarged SEM image of the AgNWs deposited on the ZnO MW is shown in Fig. 1(c). The optical absorption of these as-prepared AgNWs was characterized, and the main peak in the extinction spectrum was observed at 370 nm, as shown in Fig. 1(d). Besides, the SEM image of the AgNWs, with an average diameter $\sim$ 50 nm and average length $\sim$ 10 $\mu$m, is also exhibited in the inset of Fig. 1(d). The influence of the AgNWs on the electronic transport and optical properties of the single ZnO MW, not covered and covered by AgNWs, was studied using two droplets of ethanol solution containing 0.02 mg mL$^{-1}$ AgNWs [38,42]. As shown in Fig. 1(e), the $I$–$V$ characteristic curves exhibit a linear relationship, indicating the formation of a good ohmic contact between the In electrode and MW. It should be clearly noted that a distinct enhancement in the electronic transport characteristics of the as-synthesized MW is also obtained by incorporating the AgNWs on the single ZnO MW [42,49].

 

Fig. 1. (a) SEM image of a single ZnO MW with a perfect hexagonal cylindrical structure. (b) SEM image of a hexagonal ZnO MW covered by AgNWs. (c) Enlarged SEM image of the AgNWs deposited on the ZnO MW. (d) Extinction spectrum of the as-prepared AgNWs. Inset: SEM image of the as-synthesized AgNWs. (e) $I$–$V$ characteristic curves of the single ZnO MW not covered, and covered by AgNWs. (f) PL spectra of the single ZnO MW not covered, and covered by AgNWs.

Download Full Size | PDF

The PL measurements of a single ZnO MW, not covered and covered by AgNWs, were performed, and the recorded PL spectra are plotted in Fig. 1(f). The PL spectrum of the single bare ZnO MW reveals that the main PL peak, centered at 380.0 nm, lies in the ultraviolet region. This characteristic PL peak is attributed to the near-band-edge (NBE) emission of ZnO [25,31,47]. In addition, the other recorded weak broadband visible radiation, peaking around 510.0 nm, may be attributed to the defect-related emission. Thus, the as-synthesized individual MWs demonstrate high crystalline quality [31,46,47]. By introducing the AgNWs, a noticeable improvement in the ultraviolet emission of the ZnO MW was achieved. This enhancement can be attributed to the overlap between the extinction spectrum of the AgNWs and the typical NBE emission of the ZnO MW, yielding a resonant coupling between the ZnO exciton and localized surface plasmons of the deposited AgNWs [38,39,51].

As described above, the as-synthesized single ZnO MW exhibits a well-defined geometry, especially for the hexagon-shaped cross-section. Whilst possessing superior optical gain characteristics, the hexagon-shaped ZnO MWs can be used for achieving photon amplification in these naturally formed WGM microresonators [31,38,44,52]. A single ZnO MW not covered, and covered by AgNWs, are optically pumped by a fs-pulsed laser, as described in the experimental procedures. Figure 2(a) shows the power-dependent excitation emission spectra of a single bare ZnO MW. Under low pump fluence excitation (for example, 98.7 $\mu$W/cm$^2$, the black solid curve in Fig. 2(a)), a broad spontaneous emission band, peaking at 391.6 nm, with a full width at half-maximum (FWHM) of $\lambda _{FWHM}$ $\sim$ 12.5 nm (where $\lambda _{FWHM}$ represents the wavelength) is obtained. When the pump fluence increases to 175.4 $\mu$W/cm$^2$, several sharp peaks with an average linewidth $\lambda _{FWHM}$ of approximately 0.20 nm appear over the spontaneous emission band. The spectral linewidth of these sharp peaks is comparable to that reported for room-temperature hexagonal ZnO MWs, suggesting the occurrence of lasing action. Each emission peak corresponds to a single WGM microresonator [31,38,44]. The mode spacing between two neighboring peaks, that is, $\Delta \lambda$ $\sim$ 0.46 nm, is nearly the same, which suggests that the sharp emission modes originate from the same waveguide. As the pump fluence is increased further, the PL intensity increases correspondingly.

 

Fig. 2. Optically pumped lasing features of a single ZnO MW, not covered and covered by AgNWs: (a) PL spectra of a single ZnO MW as a function of the pumping fluence varying in the range of 98.7–259.8 $\mu$W/cm$^2$. (b) PL spectra of the ZnO MW covered by AgNWs, as a function of the pumping fluence varying in the range of 95.8–187.7 $\mu$W/cm$^2$. (c) Comparison of the PL spectra of the single ZnO MW, not covered and covered by AgNWs, under a pumping fluence of 175.4 $\mu$W/cm$^2$. (d) Comparison of the integrated PL intensity of the single ZnO MW, not covered and covered by AgNWs, versus various pumping fluences.

Download Full Size | PDF

By adding the AgNWs, the same MW was also optically pumped by a fs-pulsed laser (see details of the micro-PL measurement in the experimental procedures). Similarly, Fig. 2(b) shows the typical power-dependent PL spectra. As the pump fluence less than 136.4 $\mu$W/cm$^2$, the PL spectra were dominated by a group of spontaneous emission peaks. That is, the main PL peaks centered around 391.5 nm, and corresponding average FWHM of approximately 12.5 nm. When the pump fluence reaching 136.4 $\mu$W/cm$^2$, a series of sharp and evenly spaced peaks appear in the spontaneously emitted PL spectrum of the samples. Varying the pump fluence over 136.4 $\mu$W/cm$^2$, the PL spectra are dominated by a group of the sharp peaks. The appearance of these peaks indicates that the lasing action can also occur from the same ZnO MW, now covered by the AgNWs [31,38,44]. To further analyze the influence of the added AgNWs on the lasing performance of the single ZnO MW, a comparison of the lasing spectra of the same ZnO MW, under both bare as well as AgNW-covered conditions and same excitation power intensity of 175.4 $\mu$W/cm$^2$, is depicted in Fig. 2(c). We observed that an enhancement ratio—more than 15 times the lasing output intensity—was achieved by depositing the AgNWs on the ZnO MW [38,39,44,51]. In addition, the $Q$-factor of a single bare ZnO MW was calculated as approximately 2447, according to the formula $Q$ = $\lambda /\Delta \lambda$, where $\lambda$ is the lasing wavelength. By incorporating the AgNWs, a $Q$-factor of approximately 3918 is obtained, which is considerably higher than that of the bare MW-based WGM microcavity (or microresonator).

Further, to illustrate the optically pumped lasing action of the wires, the integrated PL emission intensity of the single ZnO MW, not covered and covered by AgNWs, as a function of the pumping fluence is shown in Fig. 2(d). These results imply that a single MW covered by the AgNWs yields a more efficient output than a bare MW. In particular, the lasing threshold ($P_{th}$) of a single ZnO MW was estimated to be approximately 171.4 $\mu$W/cm$^2$, which is much higher than $P_{th}$ = 135.1 $\mu$W/cm$^2$ of the same MW covered by the AgNWs. For the AgNW-coated ZnO MW, significant improvement of the WGM lasing characteristics, including the observed enhancement of lasing output, a lower threshold and an improved performance optical microresonator, can be achieved. This improves the external luminous efficiency of a single ZnO MW having a hexagonal cross-section [36,38,39]. As the primary PL peak of the single ZnO MW matches well with the localized surface plasmon resonance of the AgNWs, the PL emissions of the AgNW covered ZnO MW is enhanced. Therefore, the excitation of the ultraviolet plasmons of AgNWs can open new avenues to achieve enhanced WGM lasing characteristics of single ZnO MWs having hexagonal cross-sections [36,38,39].

Considering the influence of the deposited AgNWs on the electrical and optical properties of a single ZnO MW, another single ZnO MW covered without and with AgNWs (the deposition of the AgNWs can be referred to in the experimental procedures), was also selected to construct heterojunction devices. The basic architecture of the heterojunction emission device is schematically illustrated in Fig. 3(a). The device consists of a single ZnO MW, not covered and covered by AgNWs, together with a p-GaN substrate, which acts as the hole supplier. Figure 3(b) shows the $I$–$V$ characteristic curves of a single MW-based heterostructured LED. By comparison, the turn-on voltage of the single AgNWs@ZnO MW-based LED was estimated to be approximately 3.0 V, which is slightly smaller than that of the bare ZnO MW-based LED ($\sim$ 3.5 V). More importantly, the device incorporated with AgNWs demonstrates better diode-like rectifying characteristics. The prominent enhancement of the electronic transport properties of a single ZnO MW can facilitate the carrier injection process via the addition of the AgNWs [43,53,54].

 

Fig. 3. (a) Schematic diagram of the heterojunction device structure composed of a p-GaN substrate and single ZnO MW covered by AgNWs. (b) $I$–$V$ characteristic curves of a single ZnO MW-based LED, not covered and covered by AgNWs. (c) EL emission spectra of the single bare ZnO MW-based LED under various injection currents ranging from 0.5 to 5.0 mA. (e) EL emission spectra of the single AgNWs@ZnO MW-based LED under injection current ranging from 0.2 to 4.3 mA. (e) EL emission spectra of the single ZnO MW-based LEDs, with and without AgNWs, under the same injection current of 2.69 mA. (f) Integrated EL intensity of the single MW-based LEDs versus the injection current.

Download Full Size | PDF

Under electrical illumination, the single MW-based LED emits blue-violet light. These emitted photons were recorded. Figure 3(c) shows the EL spectra of the single bare ZnO MW-based LEDs with injection currents in the range of 0.5–5.0 mA. The main EL peaks are positioned at 410.5 nm with a spectral linewidth of approximately 50 nm. Similarly, the light output of the single AgNWs@ZnO MW-based heterojunction LED was also recorded, and the corresponding main EL peaks are located at 397.5 nm with a narrower spectral linewidth of approximately 40 nm, as shown in Fig. 3(d). To further examine the influence of the deposited AgNWs on the modulation of a single ZnO MW-based emission device, the EL spectra from the same MW, not covered and covered by AgNWs, are compared for the same injection current of 2.69 mA, as illustrated in Fig. 3(e). The enhancement ratio of the EL intensity is more than five-fold for the AgNW-coated ZnO MW. In particular, a slight blueshift of the emission peak envelope was also observed after depositing the AgNWs on the ZnO MW. The modulation of AgNWs on the EL features of a single ZnO MW based LED can be explained by the plasmon-mediated coupling interaction between the ZnO MW and AgNWs, as mentioned previously [38,39,42]. Consequently, by incorporating nanostructured metals exhibiting ultraviolet plasmonic behavior, the performance of the single ZnO MW-based optoelectronic devices can be immensely enhanced. This enhancement is attributed to the coupling between the localized plasmons of the deposited metallic nanostructures and the semiconductor micro- or nanostructures [38,39,51]. Furthermore, Fig. 3(f) demonstrates the variation of the integrated EL intensity with different driving currents of the single MW-based emission devices. For the bare ZnO MW-based LED, the emission intensity increases slowly under the injection current. Interestingly, by introducing the AgNWs, the output light intensity of a single MW-based diode increases significantly compared to that of the bare one. Thus, the deposition of AgNWs on ZnO MW can lead to a significant improvement in the light output of the as-fabricated LEDs [42,43,54].

The optical characterization of the blue-violet light emitted from the single MW-based LEDs was performed under the applied forward-bias voltage, beyond the turn-on value. Optical micrographs of the EL light emissions from the single MW-based LEDs were acquired with a CCD camera (details given in the experimental procedures). Figure 4(a) shows the optical microscopy EL images of a single bare ZnO MW-based LED with an injection current varying in the range of 2.0–5.0 mA. As the injection current is increased, the observed blue-violet light becomes increasingly brighter with emission regions distributed along the wires. By incorporating the AgNWs, an obvious enhancement in the light brightness and emission regions is observed, as shown in Fig. 4(b). Further, a clear change in the wavelength of the emitted light, from blue-violet to a typical violet, can also be observed.

 

Fig. 4. Optical micrographic of the EL: (a) bright blue-violet light emissions from the single bare ZnO MW based LED (scale bar: 50 $\mu$m), and (b) dazzling violet light emissions from the same single ZnO MW based LED covered by AgNWs under the same injection current ranging from 2.0 to 5.0 mA (scale bar: 50 $\mu$m).

Download Full Size | PDF

Further to exploit EL characteristics, the emission spectra of the single MW-based LEDs can be spatially resolved into a series of sharp emission peaks over a broadband spectral range in the blue-violet region [35,55,56]. As an example, for an injection current of 2.69 mA for the single bare ZnO MW-based LED, the NBE-type emission of ZnO MW dominates the blue-violet emission band, peaking at 410.5 nm, for the as-fabricated LED, as shown in Fig. 5(a). It is evident from the figure that the average mode spacing of approximately 7.27 nm can be resolved from the spontaneous emission background. In particular, by reducing the photon energy at the lower-energy side of the EL spectrum, the mode spacing of the interference peaks exhibited a significant broadening from 5.50 to 11.05 nm, which cannot be explained by pure photonic modes. Moreover, the spectral FWHM of each sharp peak ($\lambda$ $\sim$ 410.5 nm) is approximately $\delta \lambda$ $\sim$ 3.5 nm, and the $Q$–factor is estimated to be approximately 115, according to the formula: $Q=\lambda /\delta \lambda$ (see Fig. 5(a)). This $Q$-factor is considerably smaller than that obtained in the hexagonal ZnO MW case by ultraviolet excitation. The low $Q$–factor of the electrically biased waveguide emission may be caused by light leakage at the GaN/ZnO interface as well as the waveguide propagation loss along the axial direction of each MW. Therefore, the uneven mode spacing of the WGM modes is attributed to the formation of the exciton–polaritons.

 

Fig. 5. (a) Energy–wavevector dispersion curve of the single bare ZnO MW based LED. (b) Detected EL intensity of the single bare ZnO MW-based LED with changing degree of detection polarization at an injection current of 2.69 mA. In the angle-resolved EL measurement, the polarization of the electric field parallel to the axial direction, $E_{\|}$, is denoted as 0$^\circ$. (c) Corresponding degree of polarization of the individual WGM modes ($mode_{i+2}$, $mode_{i+3}$, and $mode_{i+4}$) exhibit polarization ratios similar to those of a single bare ZnO MW-based LED. (d) Energy–wavevector dispersion curve of a single MW-based LED incorporated with AgNWs. (e) Detected EL intensity of the single AgNWs@ZnO MW-based LED with changing degree of detection polarization at the same injection current of 2.69 mA. (f) Corresponding degree of polarization of the individual WGM modes ($mode_{i+2}$, $mode_{i+3}$, and $mode_{i+4}$) exhibit polarization ratios similar to those of the single AgNWs@ZnO MW-based LED.

Download Full Size | PDF

As explained above, the series of sharp peaks can be attributed to the exciton–polaritons, which being obtained from the strong coupling between the optical cavity modes (with different mode numbers) and excitons of the ZnO. This suggests that a strong light–matter interaction can occur in a single ZnO MW-based LED. The polariton effects in a single ZnO MW-based LED can be described well by the coupling oscillator model [25,33,35,57]. The solid lines, shown in Fig. 5(a), represent the calculated results for the polariton dispersion in $k$ space, which are resolved by numerical solutions of the eigenvalue equation of a dielectric hexagonal prismoid [25,31,33]. Further, the black and blue solid lines represent the exciton energy of ZnO and pure photon dispersion in WGM modes, respectively. The EL spectra with corresponding resonance energies were extracted and marked in the form of scatter data points situated at the lower polariton branch ($LP$). In addition, the upper polariton branch ($UP$) is severely damped in the wavevector space, leading to fast nonradiative depopulation. Following this, the reserved excitons further scatter into the lower polariton states through light emission. Therefore, the experimental results coincide well with the polariton dispersion curves induced by the strong coupling between the optical WGM modes and excitons [6,25,26]. Note that, from the figure, the Rabi splitting energy, $\Omega$, is evaluated to be approximately 370 meV, which is given by the minimum vertical distance between the lower and upper polariton branches.

Angle-resolved EL measurements for the single MW-based polariton LEDs were performed (details can be referred to in the experimental procedures) [7,26,58]. As described before, a single ZnO MW with a hexagonal cross-section is a potential candidate that can achieve total internal reflection of light, which is favorable for the fabrication of WGM microresonators [31,38]. Thus, the angle-resolved spectroscopy can be used to study the coupling interaction between the WGM modes and excitons (cavity polaritons) in a single hexagonal MW-based WGM microresonator [10,32,35]. To describe the exciton–photon dispersions, the total wavevector of the cavity modes in the WGM microresonator is expressed as: $\vec k ={\vec k}_\perp +{\vec k}_\parallel$, where ${\vec k}_\perp$ is determined by the diameter ($D$) of the WGM microresonator. The energy $E_\perp$ (for TE-polarization) is given by [20,33,57]:

Figure 5(b) illustrates the EL spectra recorded from a single bare ZnO MW-based LED for different collection angles, ranging from 0$^\circ$ to 360$^\circ$, at an input current of 2.69 mA [26,58,59]. To identify the exciton–polariton features, a series of narrow peaks with corresponding resonance modes denoted as $mode_{i}$ to $mode_{i+5}$ were marked (See Fig. 5(b)). Interestingly, the output intensity exhibits a visible periodic change as the detection angle rotates; however, the mode spacing shows negligible change. The details about the degree of polarization of the individual modes are also given in Fig. 5(c). By changing the detection angle, the electric field perpendicular ($E_\perp$) to the substrate plane can be separated from the parallel electric field ($E_\parallel$). Then, the polarization of the light emission modes shows a dominated $E_\parallel$ component, accompanied by a minor contribution of the $E_\perp$ component [7,19,58].

By incorporating the AgNWs, the exciton–polariton coupling and coupling strength of the single AgNWs@ZnO MW-based LED were further measured via the angle-resolved EL spectroscopy. The dispersion curves of the WGM microresonator (navy blue solid line) and excitons (black dashed line) are also shown in Fig. 5(d) for comparison (the left side of the figure) [24,60]. The EL spectrum of the single AgNWs@ZnO MW-based LED was also collected at the same injection current of 2.69 mA (the right side of Fig. 5(d)). From the figure, a series of narrow peaks can also be clearly resolved above the blue-violet emission background [11,33,35]. It is also obvious that the narrow peak spacing between the adjacent oscillation modes, called the free spectral range (FSR), becomes smaller as the photon energy approaches the exciton energy of ZnO (ranging from 10.2 to 4.5 nm), suggesting an increase in the group index and thus, significant dispersion. Accordingly, the nonuniformity in the high-energy region, beyond 3.15 eV, indicates that the recorded EL signals become damped in this region. In addition, the spectral FWHM of the sharp peak ($\lambda$ $\sim$ 398.5 nm) is approximately $\delta \lambda$ $\sim$ 2.65 nm. The $Q$-factor is estimated to be approximately 150, which is slightly higher than that of the single bare ZnO MW-based LED. Therefore, the $Q$-factor of the single MW-based microresonator can be improved by introducing the AgNWs [24,36,39].

The remarkable decrease in the FSR has been widely recognized as the evidence of a strong coupling between the excitons and photons; thus, the EL emissions with a series of narrow peaks for the single AgNWs@ZnO MW-based LED can also be attributed to the exciton–polariton interactions [7,35,56]. The measured photon energy as a function of the wavevector ($k$) dispersion was further investigated to probe the exciton–polariton interaction effects in the as-fabricated single AgNWs@ZnO MW-based LED. As shown in Fig. 5(d) (left side), clear anti-crossing behavior was observed in the experimental results, suggesting the occurrence of a strong exciton–photon coupling [10,25,35]. Thereby, the exciton–photon coupling strength, which is expressed as the Rabi splitting energy, $\Omega$ = 495 meV, can be extracted. Consequently, the measured exciton–photon coupling strength is significantly enhanced in a single AgNWs@ZnO MW-based LED, which is higher than that of the bare MW-based LED.

To further verify the exciton–photon coupling enhancement, angle-resolved EL measurements were performed and the corresponding results were analyzed to evaluate the energy–wavevector dispersion in a single AgNWs@ZnO MW-based LED. This was done by using a home-built spatial resolution spectrometer with a forward-biased current of 2.69 mA. The measured EL spectra are plotted in Fig. 5(e), which were obtained by rotating and varying the collection angles in the range of 0–360$^\circ$. As seen in the figure, the polarization of the different modes, varying from $mode_{i}$ to $mode_{i+5}$, remain comparable for the changing polarizer angles. In particular, the $E_\perp$ component dominates the polarization of all the modes, as illustrated in Fig. 5(f). Further, the degree of polarization was fitted with a value of approximately 0.25. The waveguide light features of the single MW-based LED were guided in the single MW-based WGM cavity, which might influence the measured degree of polarization to some extent. Moreover, it is also indicated that the marked emission modes can be assigned to identical transverse modes, that is, the fundamental waveguide mode that propagating circularly in the hexagonal cross-section of the as-synthesized MWs can be improved significantly [24,60,61].

4. Conclusions

To summarize, current-driven exciton–polariton LEDs were fabricated by assembling a single ZnO MW on a p-GaN substrate, which working as their hole suppliers. The multipeak emission characteristics that were resolved from the EL spectra. It suggested that strong light–matter interaction can occur in a single ZnO MW-based heterojunction LED, leading to the exciton–polariton emissions. In the as-fabricated LED, a single ZnO MW with a hexagonal cross-section can support an optical microcavity to manipulate light–matter interaction, which is crucial for fabricating high-performance exciton–polariton devices. Furthermore, the strength of the exciton–photon coupling in the LED can be modulated by incorporating AgNWs on the MW. An enhanced coupling strength, characterized by a vacuum Rabi splitting energy of 495 meV, can be achieved, which is much higher than that of a single bare MW-based LED ($\sim$ 370 meV). Owing to the benefits of AgNWs with desired ultraviolet plasmons and excellent electrical properties, the experimental findings are helpful for facilitating exciton–photon coupling for the development of electrically driven polariton devices, such as polariton lasers, slow light devices, coherent light sources, and nonlinear optical devices.