1. Introduction

Ultraviolet solid-state lasers based on wide-bandgap semiconductors (WBG) have been implemented in a broad spectrum of applications in water purification, data storage, sterilization, and bioagent detection [1–3]. These conventional inorganic compound semiconductor lasers have many unbeatable advantages over organic material based lasers, including ultraviolet-wavelength emission, high quantum efficiency, high power output, strong robustness, small volume as well as long-term stability and reliability. However, for flexible optoelectronic applications, requiring mechanical flexibility and low-temperature processing, WBG inorganic materials are not ideally suited with the major obstacles of growing high-quality epitaxy of WBG semiconductors on large-area and flexible substrates. Despite numerous ultraviolet stimulated emissions and lasing actions have been extensively demonstrated in various randomly ordered ZnO and GaN nanostructures, the resultant lasing behaviors are notoriously varied and unreproducible as the threshold strongly depends on the geometry and dimension of nanostructures [4–9]. The difficulties in controllable doping and reproducibility of quantum structures in nanoscale, as well as complicated nano-fabrication processes seriously limit their practical applications in large-area flexible and stretchable optoelectronics [4–12]. Although organic semiconductor lasers have gained considerable attention owing to their color tunability, large active area, mechanically flexibility and low-cost processing, the shortcomings including the unacceptably high self-absorption loss, exciton quenching, thermal damage, ultraviolet photo-aging and operation instability hinder their practical applications in harsh environments [13–15]. Therefore, the challenge of a flexible ultraviolet laser lies in the integration of heterogeneous material platforms by combining the advantage of wide-bandgap inorganic semiconductors with excellent flexibility.

Alternatively, freestanding WBG membrane with simple architecture and unique mechanical flexibility is one of the promising routes to altering the physical properties from their bulk materials and enable demonstration of unconventional forms of electronic and optoelectronic devices [16–18]. High-quality WBG heterostructures including III-nitrides and ZnO have been demonstrated with commercially available ultraviolet LEDs and long-lifetime lasers. In particular, the state-of-the-art ZnO epitaxial structures have shown recorded high mobility, large-scale uniformity, and low threshold ultraviolet excitonic lasing, which exhibit even higher optical gains than that of AlGaN/GaN system [19–26]. However, these WBG heterostructures can be grown only on a lattice-matched single-crystal substrate at a high growth temperature, for which, glass and plastic substrates have no tolerance, and as amorphous materials, cannot support the epitaxial growth of a crystalline film. For this purpose, the combined techniques, most notably epitaxial growth and laser lift-off (LLO), have been developed to create relatively thick membranes (several tens of microns) and flip-chip devices by separating thin films from their grown substrates. For instance, large-area thick GaN free-standing substrates have been routinely employed to achieve high-performance lasers with long-lifetime operation stability and high-power electronic devices in vertical configurations [27–30]. Whereas, the crack induced by the relief of the stress limits the membrane thickness and thick membrane cannot be made in bendable or even stretchable formats and with extremely low specific weights [16,31]. As compared to GaN freestanding membrane devices, the mass-production of free-standing single crystalline ZnO membranes has serious technical issues such as the absence of sacrificial layer that is lattice matched and etching selective [16,32,33]. Particularly, the optoelectronic devices based on large size ZnO membranes remained challenging. To this end, we report on a simple laser lift-off technique to produce large-scale and ultra-light weight transferable ZnO epitaxial membrane with the thickness down to 2.6 μm in millimeter-scaled area that exhibits excellent mechanical flexibility with well-retained structural properties. The resultant ZnO thin membranes are operated freestanding in the air and readily transferable onto the foreign flexible polyethylene naphthalate (PEN) substrates, on which direct epitaxy may otherwise be impossible or impractical. As an example for UV laser applications, the ZnO thin membranes exhibit strong stimulated excitonic emissions with enhanced spontaneous decay rates, a low threshold and an improved quality factor. Theoretical simulations identify that it is a synergetic effect of the increased quantum efficiency via Purcell effect and the improved optical gain achieved by waveguiding in the free-standing membrane, which functions as a Fabry-Perot (F-P) photonic resonator due to the refractive index contrast at ZnO-air boundaries. These unique characteristics may facilitate the development of flexible and stretchable UV laser devices by using ZnO single crystalline free-standing membranes.

2. Experimental section

ZnO epilayers were grown on sapphire (0001) substrates through two-step growth method using Aixtron CCS 19 × 2-inch MOVPE system. A 0.4 μm-thick ZnO buffer layer was grown at a low temperature (LT) of 480°C and after an in situ annealing at 1000°C for 20 minutes the successive epitaxial growth of ZnO at high temperature of 920°C was performed, resulting in a total thickness of 2.6 μm. The growth processes have been described elsewhere [20,21]. The optimal and repeatable LLO processes were carried out in air using a 248 nm KrF pulsed excimer laser with a pulse duration of 20 ns and a power density of 750 mJ/cm². Si and PEN were chosen as the foreign substrates to sustain the ZnO membranes. As determined by the reflectance spectra, there is no obvious change in the thickness of ZnO membranes before and after LLO processes. High-resolution transmission electron microscopic (HRTEM) image and electron diffraction patterns were taken by TEVNAI F20 system. The surface morphology was investigated by atomic force microscopy (Asylum Research AFM). Raman scattering were recorded at room temperature in a backscattering configuration with a Jobin-Yvon spectrometer using 514 nm Ar⁺ laser as the excitation source. For temperature-dependent photoluminescence (PL) spectra, a continuous-wave He-Cd laser was used as the excitation source. Power-dependent PL was carried out at room temperature excited by a confocal micro-photoluminescence (μ-PL) setup (Olympus BX35) coupled with a 325 nm femtosecond pulsed laser (Coherent Libra-F-HE with the repetition rate of 1 kHz and a pulse duration of 150 fs). The PL emission was collected and resolved by a spectrometer (Princeton Instruments Acton SP2500i). Time-resolved PL spectra (TRPL) were performed by an optically triggered streak camera system (Optronis GmbH SC-10). FDTD simulations were performed using a Lumerical Solutions software. The absorption coefficient and refractive indices of ZnO and sapphire materials are obtained from the software database. Considering the limited penetration depth for 325 nm He-Cd laser, a single harmonic dipole source was placed at 200 nm beneath the ZnO surface. The dipole source was polarized in the direction parallel to the ZnO surface (i.e., electric field polarized perpendicular to the c-axis) with a broad spectral emission.

3. Results and discussion

The deposited Zn-faced ZnO epilayers are composed of a low-temperature (LT) grown ZnO buffer (thickness of 0.4 μm) and a high-temperature epitaxial layer (thickness of 2.2 μm). It exhibits excellent crystalline quality, layer-by-layer surface morphology and large scale uniformity [20]. In our previous work, by using these ZnO templates, state-of-the-art ZnMgO/ZnO heterostructures have been achieved, in which, high mobility two-/three-dimensional electron gases have been realized by the combination of polarization and bandgap engineering, evidenced by the observation of quantum Hall effects [21,34,35]. Figure 1(a) shows the schematics of the creation of ZnO thin membrane achieved by LLO process through the transparent sapphire substrate. The front side of the ZnO film was brought into contact with different carrier substrates including SiO₂/Si and flexible PEN. The optical appearance of 2.6 μm-thick ZnO layer after LLO process shown in Fig. 1(b) indicates that fracture-free millimeter-scaled uniform ZnO free-standing membranes are achieved without visible damages induced by such high-energy laser treatment. High-resolution transmission electron microscopic image and the corresponding electron diffraction pattern shown in Fig. 1(c) reveal the well-retained singly crystalline of ZnO membrane with an un-distorted lattice structure. The atomic force microscopic image in Fig. 1(d) indicates that atomic layers are observable while cracks or wrinkles are absent in the ZnO membrane. Only some pits and shallow holes damaged by the sublimation of ZnO under laser irradation are observed in the LLO-side of the membrane.

 

Fig. 1 (a) Schematics of laser lift-off process upon the ZnO epilayer grown on sapphire; (b) optical image of the produced ZnO free-standing membrane with a scale bar of 2 mm; (c) high-resolution TEM image and the insert is the selective area electron diffraction pattern; (d) atomic force microscopic image of a ZnO free-standing membrane.

Download Full Size | PDF

To evaluate the evolution of strain relief, Raman scattering measurements were performed at room temperature. Figure 2(a) presents the normalized Raman scattering spectra taken from the ZnO epilayer grown on sapphire (denoted as sample S0) and the frontside (S1) and LLO-side (S2) of ZnO free-standing membranes, respectively. In terms of Raman selection rules, Raman-allowed sharp E₂ (high), A₁(TO) and second-order 2E(M) modes are repeatable, confirming that high epitaxial quality is well retained in the ZnO membrane. Broad vibrational bands in the range of 570-590cm⁻¹ are the disorder-activated longitudinal optical modes (DALO), which are resulted from the relaxation of Raman selection rules and enhanced strength of electron-phonon Fröhlich interaction via the localized electronic states bound to defects [36]. The appearance of these modes indicates the disorder near the surface region of membrane LLO-side. Especially, the position of strain-sensitive non-polar E₂(high) modes are steady at 440 cm⁻¹ before and after the LLO process. It reveals that, differed with the case of ultra-thin buffer layer in MOVPE grown GaN on sapphire, the 0.4 μm thick LT-grown ZnO buffer layer functions as an interfacial layer to release the compressive strain induced by lattice mismatch, but also a sacrificial layer to be decomposed during LLO process to avoid crack to the epitaxial film above.

 

Fig. 2 (a) Normalized Raman scattering spectra at room temperature; (b) photoluminescence spectra at room temperature; (c) normalized PL spectra at 10K; (d) time resolved PL spectra for the ZnO epilayer grown on sapphire (S0), and the ZnO free-standing membrane measured from front (S1) and LLO-side (S2).

Download Full Size | PDF

Figure 2(b) depicts the photoluminescence spectra of the ZnO epilayer (front-side and LLO-side) and membrane recorded at room temperature. A dominant narrow near-band-edge (NBE) emission at 376 nm with a weak and broad green emission is observed for ZnO epilayer (sample S0), corresponding to the excitonic recombination and the radiative recombination through deep-level intrinsic defects, respectively [37]. The fine oscillation structure observed in the inset of Fig. 2(b) correspond to longitudinal Fabry-Perot waveguide modes with the mode spacing of about 4.5 nm, satisfies the relationship$\Delta \lambda \text{=}{\lambda }_0^2/2L\left(n-{\lambda }_{0}dn/d\lambda \right)$, where the cavity length L is equal to the film thickness, n is the dispersion-corrected refraction index and the dispersion parameter is $dn/d\lambda \approx -0.012n{m}^{-1}$ [38]. Notably, the broad deep level emission (DLE) is distinct in ZnO epilayer (S0) but is almost absent in the ZnO membrane regardless being excited from the front-side (S1) or LLO-side (S2). In addition, as compared to sample S0, the ultraviolet NBE emission of sample S1 exhibits a 1.84-fold enhancement in intensity while it decreases for the LLO-side of ZnO membrane. The complicated variation of luminescent properties is correlated with the redistribution of deep-level defects (such as oxygen vacancies and zinc vacancies) involved in the membrane bulk and LLO-side surface due to the combined effect of the laser abladation and acompanied annealing. In particular, the contrast of NBE emission for frontside and LLO-side of ZnO membranes is resulted from a competition of Purcell effect and non-radiative recombination. The enhancement of excitonic emission via Purcell effect will increase quantum efficiency while the non-radiative decay channels induced by surface defects will degrade the near-band edge emission. The normalized PL spectra recorded at 10 K in Fig. 2(c) exhibits fine features of excitonic emissions, which were identical in both ZnO epilayer and membrane, including principle lines of free exciton (FX) at 3.376 eV, FX + 1LO at 3.425 eV, neutral donors (D₀X) at 3.372, 3.364 and 3.360 eV and their longitudinal optical phonon replicas at 3.289, 3.216 and 3.146 eV, as well as two-electron satellites (TES) at 3.321 eV [39]. The spectral position and width of excitonic emissions in membrane show no measurable changes, confirming that the membrane does not experience residual stress relief and thus maintains superior optical properties after substrate removal.

To evaluate the change in spontaneous emission rate and carrier dynamics, time-resolved PL (TRPL) measurements monitored at NBE emission of 380 nm were performed and recorded in Fig. 2(d). The spectral features of ZnO epilayers (S0) and membrane (S1) can be reproduced by a bi-exponential function of $I(t)=A_f\exp \left(-t/{\tau }_f\right)+A_s\exp \left(-t/{\tau }_s\right)$ where ${\tau }_f\left({\tau }_s\right)$ and $A_f\left(A_s\right)$ represent the lifetime and weight factor of the fast (slow) decay processes for various excitonic recombination channels, respectively [40]. The fast decay channel is attributed to the free exciton emission whereas the slow component is associated to the surface related exciton recombination. The fast decay rates $\left(1/{\tau }_f\right)$ and slow decay rates $\left(1/{\tau }_s\right)$ increase from 1/250 to 1/1053 ps⁻¹ and from 1/175 to 1/624 ns⁻¹, respectively. The average spontaneous decay rate ${\gamma }_0\left(\gamma \right)$ obtained from the relationship of$〈\gamma 〉=\left(1/〈\tau 〉\right)=\left(A_f{\tau }_f+A_s{\tau }_s\right)/\left(A_f{\tau }_f{}^2+A_s{\tau }_s{}^2\right)$ is calculated to be 1.05 and 1.87 ns⁻¹ for ZnO epilayer and free-standing membrane, respectively. Although non-radiative recombination will possibly cause the lifetime shorterning but also cause the decreasing of quantum efficiency and the NBE emission. It is contrary to the observed experimental facts and hence allows to exclude the dominant non-radiative decay in the case of sample S1. The Purcell enhancement factor, defined as the ratio of spontaneous emission rate ($\gamma /{\gamma }_0={\tau }_0/\tau$), is yielded to be 1.78, which is consistent with the intensity enhancement ratio of NBE emission under the same excitation power. It is resulted from the improved optical density of states with more confinement in ZnO free-standing membrane, which serves as a F-P photonic cavity due to a large contrast of refractive indices of ZnO and surrounding air. Indeed, the Purcell factor is governed by $\gamma /{\gamma }_0=3Qg{\left(\lambda /2n\right)}^3/2\pi V_{eff}$, where Q is the quality factor, g is the electromagnetic mode degeneracy, and V_eff is mode cavity volume [41]. The ZnO membranes with the observed high quality factors are expected to show cavity quantum electrodynamics effects such as modified spontaneous emission rates that would also reduce the threshold of stimulated emissions.

Figures 3(a)-3(d) show the power-dependent PL spectra of the ZnO epilayer (S0), and the frontside (S1) and LLO-side (S2) membrane on silicon, and membrane on PEN flexible substrate (S3), respectively. NBE spontaneous emissions associated with free exciton recombinations are observed around 379 nm (3.272 eV) for all samples at low pump power [39]. The broadening feature of NBE is observed in the case of exitation on the LLO-side of membrane (S2), where a residual LT-grown ZnO buffer layer is present and nonradiative recombination gives rise to relatively low quantum efficiency. Above a certain excitation intensity, population inversion starts building up, leading to an intense and sharp amplified spontaneous emission (ASE) emerging at about 389.5 nm (3.183 eV), which then slightly redshifts to a saturation value of 390.5 nm (3.175 eV) at higher pumping power [42]. Despite a relatively low quantum efficiency for LLO-side of ZnO membrane, strong and sharp ASE emissions are still dominant in the spectra of sample S2. The peak narrowing combined with super-linear dependence at high excitation density is a typical sign of the onset of stimulated emission, a process in which spontaneously emitted photons are amplified by stimulated emission in a single pass through a gain medium [43,44]. It is attributed to the inelastic exciton-exciton scattering process, which leaves one exciton scattered to an excited state given by$E_n=E_{ex}-E_B^{ex}\left(1-1/n^2\right)-3k_{B}T/2$, where $E_{ex}$is the free exciton energy, $E_B^{ex}$ is the exciton binding energy of 60 meV, n is the quantum number of the excited exciton, and $k_{B}T$ is the thermal energy [45]. Thus, the energy difference between the scattered excitons and the free excitons is 84 meV for $n=2$and 99 meV for$n=\infty$, in good agreement with the experimental observations and thus identifying the nature of ASE emissions. Since exciton-exciton scattering induced ASE only dominates in high quality ZnO, it indicates that high epitaxial quality is well retained in ZnO membranes. While for ZnO membrane on PEN substrate, the stimulated emissions exhibit a further redshift to 393 nm with a broadening feature. It could be possibly attributed to the development of degenerate electron-hole plasma (EHP) forms with a distinct band gap renormalization or the heating effect due to the poor heat disippation of PEN carrier substrate with a low thermal conductivity of PEN under high pumping power [42,46]. Overall, the comparison of dark field optical images in the insets of Figs. 3(a) and 3(b) confirms again that the distinct stimulated emission from ZnO membrane is much stronger and more directional than that of ZnO epilayer on sapphire.

 

Fig. 3 (a)-(d) Photoluminescence spectra of the ZnO epilayer grown on sapphire (S0), and ZnO membrane measured from front (S1) and LLO-side (S2), and ZnO membrane on PEN substrate (S3) at various pumping intensities, respectively. The corresponding dark-field optical images are shown in the right column.

Download Full Size | PDF

Figures 4(a) and 4(b) show the integrated intensity and the full-width-at-half-maximum (FWHM) of PL emissions as a function of excitation power density. For all samples, a conclusive demonstration of threshold marks the transition between a superlinear region-characteristic of ASE, and a linear region-characteristic of spontaneous emission. The threshold of 0.45 MW/cm² for the ZnO epilayer (S0) is comparable to that in the case of backside illumination of membrane (S2), while the thresholds for front illumination of ZnO membranes (S1 and S3) are reduced to 0.41 and 0.35 MW/cm², respectively. The threshold reduction of the stimulated emission together with intense and narrower features have been also observed in ZnO microdisks [47]. The threshold condition for stimulated emission in a planar F-P cavity is achieved when the round-trip gain for a waveguided mode equals to the round-trip losses. The optical losses in an ideal F-P cavity is mainly resulted from the mirror losses occuring at the two end faces of the ZnO epilayer or membrane. Under the ideal F-P cavity approximation with no internal absorption loss is taken into account, the gain threshold, $G_{th}$, can be estimated from $G_{th}={\left(2\Gamma L\right)}^{-1}\ln {\left(R_1{R}_2\right)}^{-1}$, where Γ is the mode confinement factor, L is the cavity length and R₁ and R₂ are the mode reflectivity of the two end faces [6]. In terms of the dielectric waveguide effect, the confinement factor for each supported TE and TM modes in ZnO epilayer and membrane is almost close to unity due to the larger refractive indices of the gain media (ZnO, n = 2.03 at 390 nm) than sapphire substrate (n = 1.76) and air (n = 1). As there is still an ultrathin air gap between the membrane and carrier substrates, it is reasonable to treat the transferred ZnO membrane as a F-P resonator with two opposite ZnO/air boundaries. The mode reflectivity (R₂) at boundary of ZnO/sapphire and ZnO/air (R₁) can be calculated to be 0.004 and 0.11, respectively, by considering the Frensel loss at the different interfacial boundaries, thus yielding an threshold ratio of 0.58 for stimulated emission between ZnO membrane and epilayer, roughly consistent with the experimental value of 0.77 for ZnO membrane on PEN.

 

Fig. 4 (a) Integrated intensity and FWHM of the emission spectra as a function of pumping power density for different ZnO epilayer and membranes (S0-S3), respectively. (c) and (d) shows the FDTD simulated electric field distribution for ZnO epilayer and membrane, respectively. The dipole with a radiation wavelength of 390 nm is located at 200nm beneath the ZnO topmost surface.

Download Full Size | PDF

Figure 4(b) illustrates that the intense ASE from ZnO membranes (S1 and S3) exhibit smaller FWHM down to only 3.1 and 2.4 nm, and the corresponding quality factors ($Q=\lambda /\Delta \lambda$) are 125 and 160, respectively. A low Q value of 89 is obtained if excited on the LLO-side of ZnO membrane (S2), comparable to that of ZnO epilayer (83.5). If treating the ZnO thin film as a planar F-P optical cavity, the quality factor can be calculated to be 87.4 for ZnO epilayer in terms of $Q=2\pi nL/\lambda \left(1-\sqrt{R_1{R}_2}\right)$ [47], consistent well with the experimental value (83.5). By ignoring the negligible reflectance difference of frontside and LLO-side of ZnO membrane, the mode reflectivity is $R_1=R_2=0.11$for both ZnO-air boundary, and hence the quality factor is calculated to be 95, which is lower than the experimental values in the case of excitation through the frontside of ZnO membranes (S1 and S3) while in verse for the case of excitation on the buffer layer of ZnO membrane (S2). Together with the combination of ASE threshold, it suggests that the stimulated emission properties in an F-P cavity depend on a delicate interplay between the radiative and nonradiative recombination rates as well as the parasitic light absorption rates. Overall, for ZnO membranes (S1 and S3), the reduced threshold with improved quality factor is resulted not only from the improved quantum efficiency of the active emitting region but also from the vertical confinement provided by refractive index guiding in F-P cavity which increases nonlinear phenomena.

For an unpatterned thin film, the external quantum efficiency, ${\eta }_{ext}$, can be given by ${\eta }_{ext}=\left(1/{\tau }_{fr}\right)/\left(1/{\tau }_{fr}+1/{\tau }_{nr}+1/{\tau }_{loss}\right)$, where $1/{\tau }_{fr}$is the radiative spontaneous emission rate into free space modes, $1/{\tau }_{nr}$is the nonradiative recombination loss and $1/{\tau }_{loss}$is parasitic absorption loss rate (the sum of internal absorption loss trapped within ZnO and radiative leakage into substrates) [41]. In terms of Fermi’s Golden rule, the spontaneous emission rate is proportional to the square of the local field of the mode [41], and thus the variation of ASE quantum efficiencies could be analyzed with the help of finite-difference-time-domain (FDTD) numerical simulations. Figures 4(c) and 4(d) decipt the calculated electromagnetic field distribution of TE modes in ZnO membrane and ZnO epilayer, respectively, at the wavelength of 390 nm where ASE emissions located at. The use of an intensity log scale allow a clear extraction of all the resonances expected for such cavitiy structures. In the absence of sapphire substrates, the longitudinal F-P resonance guiding modes distribut throughout the ZnO membrane are more distincted with enhanced electric field distribution than that in ZnO epilayer with the sapphire substrate. It is attributed to the enhanced total reflection at the ZnO-air bottom interface and improved vertical confinement due to significant index contrast in terms of the Snell’s law. The multiple reflection feedback loop between two facets of ZnO membrane makes more light energy back to the cavity, leading to an increase in the stimulated emission rates causing population inversion. On the contrary, in the case of ZnO/sapphire (S0), the similar refractive index increases the energy loss via the leaky mode from ZnO into sapphire substrates. Therefore, in combination of Purcell enhancement of spontaneous emission rate ($1/{\tau }_{fr}$) and the reduction of leakage loss into substrate ($1/{\tau }_{loss}$), the relatively higher external quantum efficiency is expected, as confirmed by the observations of improved quality factor and reduction of threshold in ZnO membrane (S1 and S3) than ZnO epilayer with sapphire substrate (S0). Moreover, the electric field distributed above the front surface of ZnO membrane also shows comparably enhanced intensity and improved directionality of vertical propagation in free space regime, which is beneficial to improve the light extraction capability with enhanced external quantum efficiency and consequently the reduction of the stimulation emission threshold. We also notice that, under excitation on the LLO-side of ZnO membrane (S2), the presence of nonradiative recombination loss ($1/{\tau }_{nr}$) is larger due to the existence of residual low-temperature grown ZnO buffer layer. The Purcell effect in membrane configuration combats to the nonradiative loss and is not favorable for external spontaneous emission. Therefore, a competition between the Purcell enhancement versus nonradiative recombination leads to comparable quality factor and threshold of ASE with respect to the ZnO epilayer (S0). In the context of phase-transition theory, in a microcavity with the dimension equals to the wavelength of the cavity-confined radiation, spontaneous emission will be merged into sitmulated emission at exceedingly low-level excitation [48,49]. Yang et al. have experimentally demonstrated that by choosing proper substrates with greater contrast in refractive index and optimzing the film thickness, the stimulated emission could be further enhanced in terms of Purcell effect [47]. A lateral optical confinement has been also realized by properly designing a two-dimensional photonic crystal pattern or integrating microsphere array, the mixed transverse whispering gallery modes (WGMs) and longitudinal F-P modes are expected with enough quality factor to support lasing. It provides a new way to realize ZnO ultraviolet lasing array with a low threshold and an ultra-high optical gain.

4. Conclusions

In conclusion, large-sized single crystalline ZnO free-standing membranes have been produced by the optimal LLO process and demonstrate well-retained structural properties and improved excitonic emissions. Serving as a well-defined F-P photonic cavity, the resultant ZnO membrane demonstrates the enhanced stimulated emission induced by exciton-exciton scattering with a reduced threshold and high quality factor, which is resulted from the improved quantum efficiency and reduced optical loss in terms of waveguiding. Therefore, large sized ZnO flexible membranes provide a unique strategy to fully realize their potential for use in flexible optoelectronic and advanced UV photonic devices.