1. Introduction

GaN-based light-emitting diodes (LEDs) are attractive owing to their applications to solid-state lighting and high performance back lighting units in liquid crystal displays [1]. However, LED devices still require further improvements to efficiently extract photons from the multiple quantum wells (MQWs). Various research groups have used distributed Bragg reflectors (DBRs) to increase the light extraction efficiency (LEE) of LEDs [2–4]. Furthermore, the advances in GaN processing have led to high index contrast air-gap DBRs by wet etching of AlInN sacrificial layers [5], microcavity LEDs with air-gap DBRs beneath the InGaN quantum wells embedded within a GaN/AlGaN membrane [6], GaN-based light emitters with GaN/air 1D− and 2D−photonic crystal (PCs) [7], vertically oriented GaN-based air-gap DBR fabricated using band-gap-selective photoelectrochemical (PEC) etching [8], and GaN/AlN DBR on a Si substrate [9]. However, the demonstrated DBR structures rely on the PEC etching, inductive coupling plasma (ICP) etching process or complex process by flip-chip technique.

In this paper, application of an innovative air-gap/GaN DBR structure in GaN-based blue LEDs has been demonstrated. The air-gap/GaN DBRs, focused on obtaining a specific LED emission λ₀ (λ₀ = 460nm), usually consist of a periodic stack of two layers of thickness λ₀/4n_{(1, 2)} (with n₁ and n₂ are the optical reflective index of the layer-1 and layer-2, respectively). The main issue when dealing with highly reflective GaN-based DBRs is to find suitable pair of materials presenting a high index contrast n₁/n₂ over a wide spectral range. In addition, the lattice mismatch has to be kept as small as possible to avoid the formation of defects. Another problem when designing these DBRs is the requirement of a very small thickness of the GaN layer. Indeed, a thickness of λ₀/4n of GaN layer could cause the collapse of the structure after the etching process. Eventually, thicknesses of Nλ₀/4n_(1,2) (N = 1, 3, 5, …) of GaN and air-gap would lead to similar DBR effect [5]. The thicknesses of 138 nm for the GaN layer and 345 nm for the heavily doped n-GaN were estimated for the present study using n₍₁₎ = 2.5 (GaN), n₍₂₎ = 1 (air), and N = 3.

2. Experiments

The air-gap/GaN DBR structures were fabricated as follows. Firstly, a 100 nm thick SiO₂ layer was deposited on a c-plane (0001) sapphire substrate by using plasma-enhanced chemical vapor deposition. A mesa pattern window with periodicity of 20 µm of SiO₂ hexagonal dot pattern of 6 µm in diameter was carried out by conventional photolithography and buffered oxide-etching. Post-etching, the sample was set into metal-organic chemical vapor deposition (MOCVD) growth chamber for epilayer growth. Trimethylgallium (TMGa) and ammonia (NH₃) were used as the sources of Ga and N, respectively. Silane (SiH₄) was used as an n-type dopant source for the growth of n-GaN epilayer and heavily doped (n⁺) GaN sacrificial layers. A thin (30 nm) GaN nucleation layer was deposited at a low temperature of 560 °C. After that, an undoped GaN (u-GaN) epilayer was deposited at 1100 °C. For air-gap/GaN DBR structure, 4 periods of n⁺-GaN (about 345 nm, doping concentration of 8 × 10¹⁸ cm⁻³) and u-GaN (about 138 nm) were grown at 1100 °C a priori the growth of 1 µm thick u-GaN epilayer. The sample was taken out of MOCVD chamber and SiO₂ masks were removed using buffered oxide etch solution. The n⁺ layers were then removed using EC etching to get the desired air-gap/GaN DBR structures. The windows opened after etchings of the SiO₂ dots facilitated the etching of the n⁺-GaN layer. Cr/Au was deposited to form metal contacts for current injection into n⁺-GaN layers. H₃PO₄ was used as the etchant. The etching time was 3 h at a current of 3 mA. During the etching, no intentional UV light was illuminated onto the sample surface. On completion of the EC etching, the sample was rinsed in de-ionized water and methanol sequentially and then dried. The template thus obtained is referred to as DBR GaN template in the following text. Subsequently, InGaN-based LED structure was overgrown using MOCVD. The LED structure consisted of a 2 μm-thick u-GaN layer having 4 period DBR structures, a 2 μm-thick n-type GaN layer, five pairs of InGaN/GaN MQWs, and a 110 nm-thick Mg-doped p-type GaN layer. After the growth of the LED epitaxial layers, mesa-structure LEDs with a 315 × 315 μm² area were fabricated. Ni-Au was deposited as the transparent conductive layers, and annealed at 550 °C for 60 s in air. Then, the Cr/Au layers were deposited as metal contacts to both the n- and p-type layers. The LED wafers were diced into chips of 350 × 350 μm² size. Similar GaN epilayer and LEDs without SiO₂ pattern and DBR structure (referred to as conventional) were grown for comparison.

3. Result and discussion

Figure 1(a) and 1(b) shows the schematic bird-eye and side-views of the air-gap/GaN DBR GaN template. Figure 1(c) and 1(d) are the cross-sectional SEM images revealing the formation of well-aligned air-gap/GaN DBR structures following the electrochemical etching process at 3 mA.

 

Fig. 1 3D schematic diagrams of the air-gap/GaN DBR GaN template showing (a) bird-eye view, and (b) side-view. Cross-sectional SEM images taken near the center and the edge are shown in (c), and (d), respectively.

Download Full Size | PDF

A proper design of DBR structure with adequate number of air-gap/GaN pairs is important to maximize the reflectivity of the top surface. Figure 2(a) shows the normal incidence reflectivity versus wavelength for the grown air-gap/GaN DBR template as a function of the number of DBR pairs. For calculations, the optical transfer matrix method was used [10]. The optical parameters used in the calculation are given in the inset of Fig. 2(a). The reflectivity increases significantly as the number of DBR pairs increase. The normal incidence reflectivity obtained at λ_Bragg = 460 nm were 0.57, 0.904, 0.984, and 0.997 for one, two, three, and four pairs of air-gap/GaN DBR structures, respectively. This result indicates that the four pairs air-gap/GaN DBR has an excellent normal incidence reflectivity. Since the light from the active layer is emitted isotropically, the angle-dependent reflectivity should also be taken into consideration. Figure 2(b) shows the reflectivity of four pairs of air-gap/GaN DBR as a function of angle of incidence, calculated using the optical transfer matrix method [10]. Note that the reflectivity is very high across the range of incidence angles except around 22° where it has a sharp dip.

 

Fig. 2 (a) Normal-incidence reflectivity vs. wavelength spectra of air-gap/GaN DBR as a function of number of air-gap/GaN pairs. (b) The calculated reflectivity of 4 pairs air-gap/GaN DBR as a function of the angle of incidence at 460 nm.

Download Full Size | PDF

To investigate the microscopic electroluminescence (EL) properties of the air-gap/GaN DBR, confocal scanning electroluminescence microscopy (CSEM) with a spatial resolution of 200 nm (which is close to the spatial resolution 100 nm of a near-field scanning optical microscope) was employed. CSEM is known as an effective experimental tool for measuring optical characteristics such as light propagation and local light output. One of the most significant findings of our study is the distribution of EL light emission from DBR structures. Figure 3(a) and 3(b) show the SEM image and optical micrograph of air-gap/GaN DBR GaN templates, respectively. As shown in Fig. 3(c), strong light emission was observed from the region having air-gap/GaN DBR structures compared to the area with no DBR structure beneath. We found noticeable light emission with the contrast of luminescence intensity observed on air-gap/GaN DBR structures. Spatial distribution of light emission can be apprehended from Fig. 3(d) showing an intensity profile measured at an injection current of 3 mA across the line drawn in Fig. 3(b). The distribution reveals higher intensity at the LED surface located above DBR and a lower intensity elsewhere. In Fig. 3(d), the intensity spikes which are observed near the center of DBR structure are due to the effect of the deflector as described by Kim et al. [11]. Diffuse reflectance spectra were recorded using UV-VIS-NIR spectrophotometer (JASCO V-570), where the samples were illuminated from front side and the reflected light was detected under diffuse reflection geometry in order to understand the light reflection effect of air-gap/GaN DBR structures. Figure 3(e) shows the reflectance spectra of conventional and air-gap/Gan DBR GaN templates. As expected from the reflectivity behavior shown in Fig. 2(a), the diffuse reflectance is enhanced for wavelengths between 400 to 550 nm. The reflectance of GaN template is significantly improved compared with that of conventional GaN template; the reflectance of DBR being 1.7 times that of conventional at 460 nm. This result is analogous to the simulation results in Fig. 2(a). One of the causes for lower reflectance of the fabricated DBR GaN template compared to the simulated ones is the area of the DBR structures. The structure considered for the simulation has air-gap/GaN DBRs at whole sample area. However, the real device has DBRs fabricated at about 60% of the whole area.

 

Fig. 3 (a) SEM image of the air-gap/GaN DBR LED, (b) optical micrograph, (c) CSEM image, (d) line scan data of EL by CSEM, and (e) diffuse reflectance spectra.

Download Full Size | PDF

To demonstrate the true impact of air-gap/GaN DBRs, Fig. 4 shows a comparison of the measured light output power for conventional LED and air-gap/GaN DBR LED at different injection currents. As shown in Fig. 4(a), the optical power of the air-gap/GaN DBR LED is enhanced in the upper direction by 1.68 times at 20 mA. From the the L-I curves shown in Fig. 4(b), it is apparent that the light output power of air-gap/GaN DBR LED is 1.91 times higher than of the conventional LEDs at an injection current of 20 mA. As described above, a significant enhancement in light intensity can be attributed to an effective improvement in light extraction efficiency by DBRs. This improvement in light extraction efficiency is due to the photons reflected back in the upper direction from DBRs structure. The result of the light output is tantamount to that of the optical reflectance. Figure 4(d) depicts the beam profiles of the devices obtained at an injection current of 20 mA. The far−-field angles of conventional LED and air-gap/GaN DBR LED are measured to be 149.7° and 128.2°, respectively. The smaller far−-field angles obtained for air-gap/GaN DBR LEDs could be attributed to enhancement in light emission due to DBR effects.

 

Fig. 4 (a) L-I characteristics of air-gap/GaN DBR LED and conventional LED measured with a photo detector, (b) integral sphere. (c) The beam profiles of the two LEDs.

Download Full Size | PDF

4. Conclusion

The light reflecting behavior of air-gap/GaN DBR structures fabricated using MOCVD and selective lateral EC etching process was investigated. The measured reflectance of the 4 pairs air-gap/GaN DBR was approximately 1.7 times higher than that of the conventional GaN epilayer. Also, the light output power of the air-gap/GaN DBR was approximately 1.91 times higher than that of the conventional GaN LED. In addition, the DBR LED has a better light output in the upper direction than the lower direction.