Use of a patterned current blocking layer to enhance the light output power of InGaN-based light-emitting diodes

1. Introduction

InGaN/GaN-based light-emitting diodes (LEDs) are of significant importance because of their use in various applications such as backlight unit (BLU) for display and solid-state lighting [1–3]. Regardless of substantial advances in their performance, however, the external quantum efficiency (EQE) of conventional lateral-geometry LEDs is still low and hence a further enhancement of EQE is essential [4,5]. Thus, in addition to the enhancement of current injection (by forming low contact resistances) [6,7], different techniques for increasing light extraction efficiency (LEE), e.g., surface roughening [8–10], patterned sapphire substrate (PSS) [11–13], and photonic crystal (PC) structures [14–16], have been actively investigated in order to increase EQE. These efforts notwithstanding, both electrical and optical losses that originate from current crowding (due to resistive p-GaN) and photon absorption around the opaque p-type pad still remain to be challenging issues. For lateral-configuration LEDs, most of current is crowded around the p-pad [17], so that some of photons generated from the active region are absorbed by the p-pad, decreasing LEE [18]. Thus, to spread current efficiently, current blocking layers (CBLs) were inserted underneath the p-pad [18–25]. For example, Huh et al. [19] reported that the introduction of a SiO₂ CBL beneath the p-pad dramatically improved the EQE of LEDs compared to conventional LEDs without a CBL. To insert the CBL, they were partially etched the surface region of the p-GaN layer beneath the p-pad until the n-GaN layer was exposed. The improvement was attributed to the fact that the insulating CBL effectively spread the current away from the p-pad, consequently reducing current crowding and optical loss. In this structure, however, some of photons emitted from the active region pass through the transparent SiO₂ CBL and then can be absorbed by the topmost opaque p-pad. To minimize the optical absorption, reflective Ag or Al layers were employed below p-pad, which effectively prevented the photon absorption [22]. Nonetheless, the insertion of such reflective layers degraded contact properties and caused poor adhesion. Furthermore, Kao et al. [23] proposed a reflective CBL combined with a distributed Bragg reflector (DBR) and showed that the use of the DBR CBL minimized both the current crowding and p-pad-induced photon absorption, leading to increase in the light output power. In the case of reflective CBLs, however, some of reflected photons can be also absorbed in the LED chips during propagation, causing internal absorption loss. Thus, in order to maximize the LEE, it is important to minimize photon absorption associated with the opaque p-pad and the LED chip, and current crowding.

Recently, nanoimprint lithography (NIL) has been widely used in optoelectronic devices for the fabrication of nano-/micro-scale patterns because of its advantages such as mass production, high resolution, and large area capability [26–28]. For instance, Kim et al. [28] investigated the effect of NIL-defined PC structures on the optical performance of green LEDs and reported that LEDs with 2-dimensional (2D) PC structures (with a period of 295 nm, a hole diameter of 180 nm, and a depth of 100 nm) yielded nine-fold higher photoluminescence intensity than LEDs without 2D PC structures. The enhancement was explained in part by Bragg scattering, which is directly related to the extraction efficiency. In this study, SiO₂ CBLs that were patterned by using NIL were inserted beneath the p-pad to increase the light output power of blue (440 nm) InGaN-based LEDs. As shown by Lee et al. [25], CBLs can be inserted not only under the p-pad but also over the active region. The CBL can improve current crowding, but reduces the emission area. Thus, the two must be optimized to maximize the output power, depending on the chip size. Unlike Lee et al.’ work (1150 × 600 μm²), our CBL was inserted only under the p-pad because our LEDs are small (800 × 300 µm²). To understand the patterned CBL-induced performance enhancement, the angular transmittance of LEDs with different CBLs was characterized.

2. Experimental procedure

A metalorganic chemical vapor deposition (MOCVD) system was used to grow blue (440 nm) InGaN/GaN multi-quantum wells (MQWs) LED structures on (0001) sapphire substrates. The epitaxial structures were composed of a 2.0 µm-thick undoped GaN layer, a 4.0 µm-thick n-type GaN:Si (n_d = 5 × 10¹⁸ cm⁻³), a 2.0 µm-thick spreading layer, a 0.1 µm-thick active layer, a 20 nm-thick AlGaN electron blocking layer, and a 0.1 µm-thick p-type GaN:Mg (n_a = 5 × 10¹⁷ cm⁻³). To fabricate LEDs (chip size: 300 × 800 µm²), standard photo lithography, imprint lithography, and inductively coupled plasma reactive-ion etching (ICP-RIE) were employed. For instance, the fabrication procedure of LEDs with patterned SiO₂ CBL is as follows: To obtain lateral-geometry LEDs, a 1 µm-thick mesa was defined by ICP-RIE. Polydimethylsiloxane (PDMS) mold with concave patterns (a diameter of 300, a distance of 300, a period of 600, and a height of 240 nm) was coated with a hydrogen silsesquioxane (HQS, Fox-16) solution, and was imprinted on the mesa-etched sample. The imprinted sample was treated by UV-O₃ plasma for 10 min, followed by annealing at 450 °C for 1 h in order to transform the imprinted HSQ resin into SiO₂ [29,30]. CBL pattern was defined by buffered oxide etch (BOE) solution-etching for 30 s. A 200 nm-thick Sn-doped indium oxide (ITO) film was deposited on the p-GaN area including CBL by electron-beam evaporator and subsequently annealed at 550 °C for 1 min in an oxygen ambient. A Cr/Ni/Au (20 nm/25 nm/100 nm) film was deposited as both p-pad and n-pad. For comparison purpose, LEDs without and with conventional SiO₂ CBL were also fabricated. As for conventional SiO₂ CBL, a 150 nm-thick SiO₂ layer was deposited by plasma-enhanced chemical vapor deposition (PECVD). The surface morphologies were characterized using a scanning electron microscope (SEM). The light output-current-voltage (L-I-V) measurements were carried out by a Keithley 238 and a Newport dual channel powermeter. To obtain the typical electrical and optical characteristics of the samples, 10 LED chips for each sample were examined.

3. Results and discussion

Figure 1 shows schematic diagram of a LED with patterned SiO₂ CBL and an SEM image of NIL-patterned SiO₂ layer on p-GaN. The patterned layer consists of pillars with a diameter of 260 nm, a distance of 340 nm, a period of 600 nm, and a height of 180 nm. It is noted that compared to the PDMA mold, the diameter and the height were reduced, whereas the distance increased because HSQ resin was shrunk during annealing. The patterned SiO₂ is expected to serve as CBL as well as light scattering centers.Figure 2 exhibits typical current-voltage (I-V) characteristics of blue (440 nm) LEDs (chip size: 300 × 800 µm²) fabricated with and without CBLs. LEDs without CBL gave a forward voltage of 3.02 V at an injection current of 20 mA, while both LEDs with SiO₂ CBL and patterned SiO₂ CBL exhibited forward voltages of 3.1 V. The LEDs without CBL showed series resistance of 9.8 Ω, while the LEDs with SiO₂ CBL and patterned SiO₂ CBL produced 11.0 Ω.

 

Fig. 1 (a) Schematic diagram of an LED with a patterned SiO₂ CBL. (b) A plan-view SEM image of NIL-patterned SiO₂ layer on p-GaN. Inset in (b) shows a cross-section image.

Download Full Size | PPT Slide | PDF

 

Fig. 2 Current-voltage (I-V) characteristics of blue (440 nm) LEDs with and without CBLs.

Download Full Size | PPT Slide | PDF

Figure 3(a) shows typical light output-current (L-I) properties of LEDs fabricated with and without CBLs. The LEDs with patterned SiO₂ CBL yielded 39.6 and 11.9% higher light output powers at 20 mA, respectively, than the LEDs with no CBL and conventional SiO₂ CBL. In addition, electroluminescence (EL) spectra of the LEDs were characterized, Fig. 3(b). At 20 mA, the peak wavelength of LEDs with no CBL, CBL, and patterned CBL was 440.6, 440.5 and 442.1 nm, respectively, and their wavelength shifted by only Δ1 ± 0.1 nm when the current increased from 10 to 100 mA. Furthermore, at 20 mA, FWHMs were 19.03, 19.20, and 19.93 nm for LEDs with no CBL, CBL, and patterned CBL, respectively. The reason why the LEDs with different CBLs showed different behavior is presently not clearly understood.

 

Fig. 3 (a) Light output-current (L-I) properties and (b) EL spectra of LEDs fabricated with and without CBLs.

Download Full Size | PPT Slide | PDF

To characterize the light emission and current spreading behavior of the LEDs with different CBLs, the emission images of the chips were obtained at a forward-bias voltage of 0.3 mA. It is noted that the original monochromatic emission images were transformed into 256 different colors by using the MATLAB program. The plan-view emission images obtained from the LEDs with no CBL, conventional SiO₂ CBL, and patterned SiO₂ CBL are shown in Fig. 4. For the LED with no CBL (Fig. 4(a)), the photoemission is localized and not uniform. As shown in Fig. 4(b), the LED with conventional SiO₂ CBL exhibits much better light emission across the chip area than the LEDs without CBL, which can be attributed to the enhanced current spreading. As can be seen in Fig. 4(c), the LED with patterned SiO₂ CBL yielded much uniform light emission across the whole chip area than the LEDs with SiO₂ CBL. This implies that the patterned SiO₂ CBL effectively minimizes photon absorption by serving as efficient photon scattering centers and current crowding.

 

Fig. 4 Plan-view emission images obtained from LEDs with (a) no CBL, (b) conventional SiO₂ CBL, and (c) patterned SiO₂ CBL. The images taken at 0.3 mA.

Download Full Size | PPT Slide | PDF

Furthermore, to better understand the enhanced light output associated with patterned SiO₂ CBL, we obtained the angular transmittance of four structures (namely, a sapphire substrate, an ITO layer on a sapphire substrate, ITO/SiO₂ multilayers on a sapphire substrate, and ITO/patterned SiO₂ multilayers on a sapphire substrate) with a blue laser diode (λ = 455 nm) [31]. For the patterned SiO₂ structure, the same pattern parameters as those given in Fig. 1 were used. To characterize the internal reflection occurring in a high-refractive-index LED medium, the structures considered were bonded to a cylindrical BK7 prism with an index matching gel. In Fig. 5(a) is shown typical camera images that describe light incident into and reflected from planar and patterned SiO₂ structures at θ = 45°. This result suggests that the patterned structures are capable of diminishing reflection above a critical angle (i.e., optical tunneling) [31,32]. Figure 5(b) illustrates the angular transmittance of the four structures considered, for which an integrating sphere was used to collect both specular and diffuse transmission. The transmittance data exhibit two remarkable features: (i) at θ < 42° (i.e., a critical angle), the patterned structure yields transmittance lower than all of the planar structures and (ii) the patterned structure results in non-zero transmittance above the critical angle. Because CBL is placed right underneath the metal electrode, the transmittance around normal incidence is less significant. Therefore, the improved light output due to the patterned SiO₂ CBL can be ascribed to the increased transmittance above the critical angle.

 

Fig. 5 (a) Typical camera images showing light incident into and reflected from planar and patterned SiO₂ structures for θ = 45°. (b) Measured angular transmittance of four structures.

Download Full Size | PPT Slide | PDF

It was found that the LEDs with CBLs exhibited slightly higher forward-bias voltage than the LEDs without CBL, as shown in Fig. 2. This can be explained by a reduction in the Ohmic contact area between p-GaN and ITO caused by the presence of CBLs. Note that the LEDs with both the CBLs gave the same forward-bias voltages. As shown in Fig. 2, the LEDs with conventional SiO₂ CBL exhibited higher light output than the LEDs without CBL. This can be attributed to reduced current crowding effect. On the one hand, the LEDs with patterned SiO₂ CBL produced higher light output than LEDs with conventional CBL. This improvement can be explained by the fact that the patterned SiO₂ structure can scatter the photons, resulting in an increase in the light extraction. This is in good agreement with the measured angular transmittance (Fig. 5). Consequently, the improved light output power of the LEDs with the patterned SiO₂ CBL can be attributed to the enhanced light extraction and current spreading.

4. Conclusion

We investigated the effect of patterned SiO₂ CBL on the light output power of InGaN-based LEDs. The periodic pillar-pattern array was defined by means of NIL. The LEDs fabricated with the patterned SiO₂ CBL yielded higher light output and more uniform emission across chip area than the LEDs with and without conventional SiO₂ CBLs. The improved performance was ascribed to the combined effect of better photon extraction and current spreading. The result shows that the use of patterned CBL could be a promising processing tool for the fabrication of high-performance InGaN-based LEDs.