For flip-chip UV LEDs with a p-GaN layer, the well-designed thickness of the p-GaN layer plays a crucial role in improving light extraction efficiency since that the optical cavity is able to influence angle-dependent reflectivity through the adjustments in the direction of reflection light (Zhang et al. 2019). This effect can be confirmed through numerical simulations using a simplified model, as depicted in the insert of Fig. 1(a). The simulations involved UV light incidence at various angle α with a fixed wavelength of 280 nm. The lateral boundary conditions were set as periodic boundaries, while the top and bottom boundary conditions utilized PMLs.
Firstly, we investigated the angle-integrated reflectivity (Ref.) of a p-AlGaN/dielectric/metal structure by varying relevant parameters, i.e., electrode materials and the thickness of the p-AlGaN, using numerical simulation simulations, as shown in Fig. 1(a). To illustrate the positive influence of adopting Al electrode materials, we compared two cases: one with Ni and Ag electrodes without a dielectric layer (i.e., dielectric layer thickness td = 0 nm, metal thickness tm = 300 nm), and the other one with an Al electrode. The values of Ref. for the Ni and Ag cases were relatively low, reaching 38.8% and 36.4%, when the thickness of p-AlGaN was 140 nm. However, the Al electrode case exhibited considerable enhancement in reflectivity. In addition, an oscillation behavior induced by the optical cavity was observed in the p-AlGaN thickness range from 50 to 150 nm, though the fluctuating range was less obvious compared to the counterpart with p-GaN, which is attributed to the strong absorption of the p-GaN layer (Mondal et al. 2021). The maximum reflectivity of 90.0% is obtained when the p-AlGaN thickness was 140 nm. This indicates that the combination of a high-reflection Al electrode and an ultraviolet transparent p-AlGaN film can increase reflectivity by 53.6%. It is worth noting that the combined action of a high-reflectivity electrode and a p-AlGaN UV-transparent layer is crucial for enhancing reflectivity. This p-AlGaN/dielectric/metal structure can also be effectively utilized to design sidewall omni-direction reflection structures (ODRs) (Zheng et al. 2020; Kim et al. 2008).
Figure 1(b) provided a detailed illustration of the angle-integrated reflectivity (Ref.) of the sidewall ODR structure (i.e., the thickness of the dielectric layer td ≠ 0) as a function of the thickness of the dielectric layer. The reflectivity curves for different dielectric materials, such as MgF2 (nm = 1.39 at wavelength of 280 nm) ODR (red line), SiO2 (ns = 1.49 at wavelength of 280 nm) one (blue line), and HfO2 (nh = 2.05 at wavelength of 280 nm) one (black line), displayed similar periodic oscillation behaviors due to the effect of the optical cavity. It was observed that during the variation of td, the ODR structure with MgF2/Al acted as an excellent antireflection design, indicating that a smaller refractive index of the dielectric layer increases the angle-integrated reflectivity of the sidewall ODR structure (Oh et al. 2017). To elucidate the underlying reasons for the positive role of MgF2 in enhancing reflectivity, angle dependent reflectivity and spatial electric field distribution were investigated, as shown in Fig. 1(c)-(d). Figure 1(c) demonstrates angle dependent reflectivity (Ref.α) curves for the cases with/without MgF2 layer. It is evident that the Ref.α curve of case with MgF2 layer exhibits nearly omnidirectional enhancement compared to the other case, as shown in Fig. 1(c). The normalized distributions of electric-field intensities for 0° and 40° cases for p-AlGaN/MgF2/Al and p-AlGaN/Al structures are depicted in Fig. 1(d). The significant impedance differences at the interface can lead to serious internal reflection phenomena (Son et al. 2012). In the 40° case, the parasitic absorption in the Al electrode was prevented due to the existence of total internal reflection at the p-AlGaN/MgF2 interface.
Figure 2(a) illustrates the schematic diagram of the simulated device with a NPSS or/and PC structure. The thicknesses of the sapphire substrate, AlN buffer layer, n-AlGaN electron transport layer, MQWs (multiple quantum wells), p-AlGaN hole transport layer, Al metal layer were set to 2 µm, 3 µm, 2.2 µm, 58 nm, 140 nm, 300 nm, respectively. The typical parameters of NPSSs or PCs were fixed at 1.5 µm (Period p) and 1.1 µm (height h), respectively. Notably, the UV LED featured an inclined sidewall with a 250 nm-thick MgF2 dielectric interlayer and an Al metal cladding layer. Figure 2(b) displays the radiation pattern emitted by a single electric dipole. The left side of Fig. 2(b) clearly shows that: 1) The direction of the dipole moment is along the y-axis parallel to the direction of film growth. 2) Emission along the x-axis is much stronger than that along the y-axis, indicating anisotropic emission (Kim et al. 2015). 3) Most of the emitted light travels along the x-axis. In contrast, the main propagation direction of TE (transverse-electric) polarized light is perpendicular to TM polarized light on the right side of Fig. 2(b). For UV LEDs featuring an inclined sidewall structure, adopting a rational sidewall angle can significantly improve LEEs (Tian et al. 2021). Therefore, the results in Fig. 2(c)-(d) provide further insights into the influence of the sidewall angle θ, ranging from 25° to 85°, on LEEs in UV LEDs with an inclined sidewall coated with an ODR structure. Figure 2(c) shows the effect of various structure designs, i.e., inclined sidewall (IS), composite design of inclined sidewall and NPSSs (IS & NPSS), and composite design of inclined sidewall and PCs (IS & PC) on TM polarized light extraction in detail. It is evident that LEEs for all three cases display similar trends as the sidewall angle θ changes. Specifically, LEEs increase to a maximum for a sidewall angle θ of up to 40°, and then decrease as the sidewall angle θ increases from 40° to 85°. Furthermore, at a sidewall angle θ of 40°, the UV LED with IS exhibits the lowest LEE value of 40.8% among the three structure designs. Notably, the latter two cases show significant LEEs enhancement compared to UV LEDs with an IS structure, indicating that the addition of NPSSs or PCs can tap into the greater potential of the inclined sidewall ODR structure. Similarly, Fig. 2(d) demonstrates the light extraction of TE polarized light for the IS, IS & NPSS, and IS & PC structure designs at different sidewall angles. Compared to the IS type structure, the IS & NPSS type and IS & PC type structure designs exhibit higher LEEs for all sidewall angles θ, suggesting that the LEEs enhancement for the latter two structure designs can be attributed to the existence of a roughened interface. Figure 2(e) summarizes the LEEs at a sidewall angle θ of 40° for the IS, IS & NPSS, and IS & PC structure designs. The IS & NPSS structure design demonstrates the highest LEE value. On the other hand, TE polarized light is more sensitive to the IS & PC structure among the three typical structures. It is observed that the IS & NPSS type and the IS & PC structure can improve total LEEs by 21.3% and 18.3%, respectively, compared to the IS structure. The ratio of TM polarized and TE polarized light follows the reference (Mondal et al. 2021), and further discussion on this aspect will be presented in the later section. Notably, Fig. 2(f) depicts the LEEs response with polarization angles of the dipole moment. The information contained in Fig. 2(f) suggests that: 1) The LEE of UV LED with the IS & NPSS structure is the least sensitive to the orientation of the dipole moment among the three structure cases. 2) Typical orientations of the dipole moment, i.e., TE and TM polarized light cases, align with the results shown in Fig. 2(c)-(e). 3) The LEE values from this figure provide valuable reference for MQWs.
A deeper understanding of the light-extraction mechanism of UV LEDs with IS, IS & PC, IS & NPSS structures at a sidewall angle θ of 40° is revealed through the spatial electric field distribution in Fig. 3. In the case of TM polarized light, the radiation patterns of the IS and IS & PC structures were basically similar. However, the incorporation of NPSSs at the sapphire/AlN interface disrupted these patterns, preventing the generation of numerous hot spots in the AlN region. These hot-spots are a result of high-orders mode resonance (Yang et al. 2019). As depicted in Fig. 3(a) and Fig. 3(c), the IS and IS & PC structures showed clear hot spots in regions such as AlN and AlGaN, indicating that light was trapped in these regions. This observation further underscores the superior LEEs value of the IS & NPSS structure compared to the IS & PC structure at a sidewall angle θ of 40°. In contrast, as shown in Fig. 3(f), although NPSSs help in reducing the formation of hot-spots in the AlN region to extract a higher proportion of light, the extracted light propagates laterally in the sapphire region, which is not conducive to further extraction into the air region. Conversely, the presence of PCs in the IS & PC structure alters the radiation pattern of TE polarized light compared to the IS case, as shown in Fig. 3(b) and 3(d), allowing a higher ratio of light to leak into the air region. This improvement in light extraction in the air region can be attributed to the reduction in high-mode resonance generation within the device (Qian et al. 2023).
Furthermore, Fig. 4 presents angle-dependent transmissivity (Tra.) curves and the corresponding electric field distributions for PC and NPSS substrates at β = 0° and 10°. This additional information provides further insights into Fig. 2(e). The results reveal that a significant portion of TM polarized light experiences reflection by inclined ODR structures and redirects towards the vicinity of β = 10° (Lee et al. 2016; Zheng et al. 2020). As a consequence, UV LEDs employing the IS & NPSS design outperform these with IS & PC design in terms of extracting TM polarized light. Notably, the simulated results from Fig. 4(b) demonstrate complete agreement with Fig. 4(a).