*2.4. Characterization*

The structural shapes of the self-aligned ZNRs and hierarchical ZNR/NNS arrays were observed using a field emission-scanning electron microscope (FESEM, Hitachi S4300, Tokyo, Japan), and the hierarchical nanostructures were analyzed using an energy-dispersive spectroscopy (EDS) mounted on the FESEM. The electrical and optical properties were measured using a parameter analyzer (Keithley 2400, Tektronix, Beaverton, OR, USA), an optical power meter (Newport 1830C, Irvine, CA, USA), and

an optical microscope (Velcam CVC5220, Chun Shin Electronics Inc., Taipei, Taiwan). Finite-difference time-domain (FDTD) simulation was conducted to compare the light output of the fabricated photonic emitter with that of a C-emitter.

#### **3. Results and Discussion**

The schematic diagram of the ZNR/NNS arrays synthesis on the C-emitter by following the proposed experimental procedures is shown in Figure 1.

**Figure 1.** Fabrication process of C-emitter with hierarchical ZnO nanorod (ZNR)/NiO nanosheet (NNS) arrays.

The FESEM image in Figure 2a shows the ZNRs with an average diameter and length of 300 nm and 3.5 μm, respectively. To understand the growth mechanism of the ZNR/NNS arrays, the study of the morphology evolution of hierarchical ZNR/NNS arrays during the reaction time has been carried out, as shown in Figure 2b–e.

**Figure 2.** Field emission-scanning electron microscope (FESEM) images of (**a**) ZNRs and (**b**–**<sup>e</sup>**) hierarchical ZNR/NNS arrays grown for different times; (**a'**) tilt and (**b'**) cross-sectional FESEM images of (**d**). (**f**) an energy-dispersive spectroscopy (EDS) and (**g**) atomic composition of the hierarchical ZNR/NNS arrays.

Since the system tends to minimize the overall surface energy, the ZNRs grew preferentially along the [0001] direction [28,29]. Then, Ni-based nanoparticles (NPs) nucleated on the surface of the ZNRs to form active sites, which minimized the interfacial energy barrier to promote the subsequent growth of Ni-based NPs. The merge of these NPs reduced the overall energy by decreasing the surface energy, which was beneficial for adjacent Ni-based NPs to spontaneously self-organize together. The self-organized NPs shared a common crystallographic orientation and formed a planar interface, as shown in Figure 2b. When the reaction proceeded for 30 min, these NPs self-assembled to form large nanosheets (NSs) and finally generated the ZNR/NNS arrays. The above hypothesis is supported by examining the morphologies of NiO NSs at di fferent growth stages by controlling the reaction time, as shown in Figure 2c,d. After exceeding a growth time of 60 min, the NSs transformed into nanowalls (Figure 2e). Therefore, we selected 60 min as the optimum growth time for obtaining stable ZNR/NNS arrays. Figure 2a',b' show the tilted and cross-sectional FESEM images of the well-aligned NNS arrays grown for 60 min. The results of an EDS analysis along with the atomic and weight values of the hierarchical ZNR/NNS arrays are shown in Figure 2f. The atomic contents of oxygen, nickel, and zinc were 16.6%, 60.09%, and 23.32%, respectively. The lower percentage of oxygen was ascribed to the fact that oxygen is relatively lighter than nickel and zinc [30]. These results confirmed the successful synthesis of the ZnO and NiO nanostructures.

To evaluate the properties of the fabricated photonic emitters, we measured their current–voltage curves and optical output intensity. Under the injection current of 20 mA, the threshold voltages of the C-emitters without nanostructures, with ZNRs, and with hierarchical ZNR/NNS arrays ranged between 3.00 and 3.03 V, as shown in Figure 3a. This indicates that the ZnO nanostructures did not a ffect the electrical properties of the emitters because they were grown using a low-temperature growth process. At the injection current of 100 mA, the optical output power of the C-emitter with the hierarchical ZNR/NNS arrays was approximately 17% higher than that of the device with ZNRs and two times higher than that of the photonic emitter without any nanostructures (Figure 3b). The inset emission images in Figure 3c–e show that the hierarchical ZNR/NNS arrays on the C-emitter are brighter than the other emitters at the injection current of 0.05 mA. Compared to other studies on this topic, we realized the optimal refractive index between the ITO layer and air [31,32]. In addition, the higher optical output power can probably be attributed to the graded refractive index and the reduced Fresnel reflection.

**Figure 3.** (**a**) Current–voltage curves, (**b**) current–optical output power curves, and (**<sup>c</sup>**–**<sup>e</sup>**) emission images (at an injection current of 0.05 mA) of the C-emitter without nanostructures, with ZNRs, and with hierarchical ZNR/NNS arrays.

Then, we performed FDTD simulations to compare the three C-emitter types. Figure 4a,b show schematic diagrams of the overall structures of the devices with the ZNR and the hierarchical ZNR/NNS.

**Figure 4.** Layouts of the C-emitters with (**a**) ZNR and (**b**) hierarchical ZNR/NNS. (**b'**) FESEM and transmission electron microscope (inset) images of the hierarchical ZNR/NNS arrays. (**c**) Calculated refractive indices and escape efficiencies. (**d**–**g**) Electric field propagation for various C-emitters.

In these simulations, we used the following parameters: ZNR diameter and height of 300 nm and 3.5 μm, respectively, based on the corresponding FESEM image (Figure 2a); NNS width and height of 195 nm and 103 nm, respectively; and consistent interval of 88 nm between adjacent NNSs based on the FESEM and transmission electron microscope results (Figure 4b'). These structures were simulated on the 200 nm thick ITO layer deposited on the 430 nm photonic emitters. The refractive indices of various materials were calculated using EMA [15–17]:

$$m\_{\rm eff} = \left[ n\_{\rm ZnO}^2 V\_{\rm ZnO} + n\_{\rm A\dot{x}}^2 (1 - V\_{\rm ZnO}) \right]^{\frac{1}{2}} \tag{1}$$

where *n*eff is the effective refractive index of the ZNRs; *n*ZnO and nAir are the refractive indices of ZnO and air, respectively; and *V*ZnO is the volume fraction of ZnO in the effective medium. The refractive indices and the volume fraction were determined from the FESEM image, as shown in Figure 2a. The average refractive index of the ZNRs was significantly lower than that of the ZnO film because of the inclusion of air in the effective medium. Therefore, according to EMA, ZNRs have a lower refractive index, even though the refractive index of the ZnO film was similar to that of the ITO film.

Furthermore, the refractive index of NNSs was lower than that of the ZNRs because the refractive index of NiO is lower (1.68) than that of the ZnO film, even though NiO has the lower air volume, as can be inferred from the FESEM image shown in Figure 2d. Therefore, we can confirm that the calculated refractive indices of the GaN, ITO, ZNRs, NNSs are 2.49, 2.1, 1.47, and 1.44, respectively.

The optimal refractive index of the antireflection layer between the ITO film and air can be computed using the formula: [33,34]

$$m\_1 = \sqrt{n\_0 n\_\\$}.\tag{2}$$

The formula yielded a value of 1.449, which is very similar to the refractive index of the NNSs for a high antireflection effect.

The escape e fficiencies of GaN, ITO, ZNRs, and NNSs (Figure 4c), based on the corresponding calculated refractive indices, can be given as follows [35]:

$$\frac{P\_{\text{escape}}}{P\_{\text{source}}} \approx \frac{1}{4} \frac{n\_{\text{air}}^2}{n\_{\text{CaN}}^2} \tag{3}$$

where *nGaN* is the refractive index of the GaN layer. *Pescape* and *Psource* are the escape and source powers of the photonic emitter, respectively. According to the equation, the photonic emitter without the ITO electrode has an escape e fficiency of only 4%. This e fficiency can be increased to 12.2% by controlling the refractive index through NNSs growth. However, because this value represents the escape e fficiency of photon extraction obtained by considering only the refractive index, it will decrease when considering the e fficiency lost through GaN, ITO, and ZNRs. Moreover, the photon e fficiency of the device based on various nanostructures is not discussed here, because this calculation considers only layered thin films.

Figure 4d–g illustrate the simulated electric field propagation for the various photonic emitters. As can be inferred from the simulation results shown in Figure 4d,e, the two emitters have similar propagation images. However, the C-emitter shows a higher field extraction toward the outside compared to the emitter without the ITO electrode because of Fresnel reflection. In the case of Figure 4f, although the ZNR on the C-emitter shows a remarkably higher propagation of electric field toward the outside because of the refractive index control and wave guide e ffect of ZNR, it is limited by the refractive index di fference between the ZNR and air and the flat end of the ZNR. This can be explained based on the strong intensity of the electric field inside the ZNR because of interference of the reflected electric field. However, the simulation results of ZNR/NNS on the C-emitter, which are presented in Figure 4g, shows the stronger electric propagation image than that of ZNR on the C-emitter. This di fference can be explained by the fact that the NNSs have a smoother graded e ffective refractive index change and lower TIR loss from the GaN-based photonic emitter to the outside.
