1. Introduction

InGaN-based materials have been intensively investigated for various applications such as light-emitting diodes (LED) and laser diodes. The internal quantum efficiency (IQE) and the light extraction efficiency (LEE) should continue to be improved in high efficiency LEDs. Transferred anodic alumina template onto GaN layers [1,2], top p-type GaN:Mg roughened surfaces [3], shape transformation of nanoporous GaN layers [4], honeycomb nanoporous GaN thin film [5], air-void structures [6], photonic crystals structures [7,8], and nanoporous GaN structures [9–11], have all been reported to increase the light-extraction efficiency in nitride-based LED structures. Many researchers have reported about the lift-off processes on the GaN epitaxial layers by forming nanoporous GaN structures through a wet etching process on Si-doped GaN layers [12–14]. Bilousov et al. [15] reported on fully porous GaN p-n junction diodes through a chemical vapor deposition. Re-growth processes on treated GaN templates had been reported to fabricate the bottom roughened layers [16,17] in the InGaN LED structures. An embedded nanoporous GaN structure has been fabricated through a wet etching process on a GaN template replacing the more complex epitaxial regrown process [18].

Here, we report on a directional nanopipe (NP) GaN structure and its facile alignment over a 500μm width using an electrochemical (EC) wet etching process with an external electric field. A 0.85µm-thick Si-heavy-doped GaN:Si (n⁺-GaN:Si) layer embedded in the InGaN LED structure was transformed into a directional nanopipe structure. A high lateral etching rate and a wide etching width for the directional nanopipe structures were achieved by controlling the design of the epitaxial structures, the etching solutions, and the external bias voltage as an electrical driving force. A birefringence optical property for the directional NP-GaN structure was observed through a polarized optical microscopy and an ellipsometer measurement. The optical and electrical properties of the InGaN-based LED with the directional nanopipe GaN layer were analyzed and are discussed in detail.

2. Experiments

The epitaxial layers of the InGaN-based LED wafers were grown on 2 in. diameter prime grade c-plane sapphire substrates through a metal organic chemical vapor deposition (MOCVD) system. The LED structures consisted of a 30nm-thick GaN buffer layer, a 1.0μm-thick unintentionally doped GaN layer (u-GaN), a 0.5μm-thick n-type GaN:Si layer (4 × 10¹⁸cm⁻³), a 0.2μm-thick u-GaN layer, a 0.85μm-thick heavily Si-doped n-type GaN layer (8 × 10¹⁸cm⁻³), a 1.0μm-thick u-GaN layer, a 2.0μm-thick n-type GaN layer (2 × 10¹⁸cm⁻³), 10 pairs of InGaN/GaN multiple quantum wells (MQWs) active layers, and a 0.2μm-thick magnesium-doped p-type GaN:Mg layer. The active layers consisted of a 30Å-thick InGaN well layer and a 120Å-thick GaN barrier layer for the InGaN/GaN MQW LED structure. The p-type GaN:Mg mesa region was defined by photoresist materials, and then etched by an inductively coupled plasma (ICP) etcher to expose the bottom n-type GaN:Si region. The mesa region consisted of a 1.3μm-depth and 540 × 890 μm² in size. Subsequently, a 120nm-thick indium tin oxide (ITO) film was deposited on the mesa region as a transparent conductive layer. The open regions for n-type and p-type metal contact were defined by photoresist through the photolithographic process. The metal layers were deposited on the LED wafer by an electron beam evaporation deposition system. The Cr/Au (50nm/2500nm) metal layers were deposited on the indium tin oxide (ITO)/p-GaN and on the n-GaN layers at the same time to form ohmic contacts on both layers. The residual Cr/Au metal layer was removed through a photoresist lift-off process.

Then, a SiO₂ passivation layer was deposited on the LED wafer as a protective layer for the subsequent laser-scribing (LS) process. The LED chips were isolated through laser-scribing processes by using a triple frequency ultraviolet Nd:YVO₄ 355nm laser along one direction of the LED chips with a 560μm spacing. Laser scribing channels were also formed for the lateral wet etching process. The normal depth of the laser scribing line can be controlled to reach the n⁺-GaN layer by adjusting the laser pulse energy through the use of an optical variable attenuator. The pulse laser was used for the laser scribing process, the depth of laser beam center was deep enough to reach the n-GaN conductive layer. The n⁺-GaN and the bottom n-GaN conductive layers of the LED samples were submersed in a 0.5M nitric acid solution for wet electrochemical (EC) etching with an external dc bias voltage of + 10V for 30mins.

During the EC-etching process, an external dc bias was fixed at a positive 10V, with an Indium metal as an anode contact applied on the GaN conductive layer at the edge of the LED sample, with the metal contact not immersed in the solution. A platinum electrode was used as the cathode for the wet etching process. Under the positive bias condition, the positive charges in the bottom n-GaN conductive layer provided an electric field and a driving force for the lateral etching process on the top n⁺-GaN:Si sacrificial layer. The n⁺-GaN layer was etched and transformed into the nanopipe GaN structure. The non-treated and the treated LED structures with the embedded nanopipe GaN structure were defined as the standard LED (ST-LED) and the nanopipe LED (NP-LED), respectively. A schematic of the NP-LED structure with the laser-scribing lines and the nanopipe GaN structure is shown in Fig. 1. In order to analyze their optical and electrical properties, the chosen ST-LED and NP-LED devices were located at the same 2” LED wafer center with similar epitaxial qualities. The geometric morphologies of the LED structures were observed using polarized optical microscopy (OM) and a field-emission scanning electron microscope (FE-SEM, JEOL 6700F). The light intensity profiles of the LED chips were measured using a beam profiler (Spiricon: number of effective pixels: 1600 × 1200 pixels). Light output power and far-field radiation patterns were measured on non-encapsulated LEDs in chip form. The photoluminescence (PL) spectra of both LED structures were obtained by an angle-resolved PL measurement through the use of a 405nm diode laser as an excitation laser source and a monochromator (JOBIN YVON iHR550) with a TE-cooled charge-coupled device (CCD) detector. The refractive index of the GaN and nanopipe GaN layers were measured by an Ellipsometer (J.A Woollam VASE^® Ellipsometer).

 

Fig. 1 A schematic of the NP-LED structure with fully EC-treated nanopipe GaN structure and the partially-etched top/bottom n-type GaN:Si layers. The dimension of the LED chip is 540 × 890 μm² in size.

Download Full Size | PDF

3. Results and discussion

The 45° tilted-view SEM micrograph of the NP-LED structure is shown in Fig. 2(a). The n-type/p-type metal pads, region-1 at the mesa edge region, region-2 (between p- and n- electrode finger patterns), and region-3 at the mesa center region are labelled in the SEM micrograph. In Fig. 2(b), the depth and width of the laser scribing line were measured at 3.43μm and 7.05μm, respectively, when contacting the bottom 0.85μm-thick n⁺-GaN:Si layer. The etchant reacted with the n⁺-GaN:Si layer through the laser scribing channels. In region 1 close to the LS line, the top n-GaN:Si layer, n⁺-GaN:Si layer, and bottom n-GaN conductive layer were treated as triple-porous GaN layers as shown in Fig. 2(b). The nanopipe GaN layer was formed through a selectivity EC-etching process. The doping concentration of the upper GaN layer is 2 × 10¹⁸cm⁻³ lower than the n⁺-GaN:Si layer (8 × 10¹⁸cm⁻³). The upper n-GaN layer, n⁺-GaN:Si layer, and bottom n-GaN conductive layer were etched as the first step of the EC-etching process. A high lateral etching rate was observed in the n⁺-GaN:Si layer when the bias voltage was applied from the bottom n-GaN:Si conductive layer (8 × 10¹⁸cm⁻³). The carrier concentration of the bottom n-GaN conductive layer is 4x10¹⁸ cm⁻³ lower than the n⁺-GaN:Si layer (8 × 10¹⁸cm⁻³) for improving the selectivity EC-etching process. When the n⁺-GaN:Si layer transformed into the nanopipe structure, the NP-GaN layer resistance increased and blocked the applied carriers from reaching the upper n-GaN layer. So, only the part of the upper n-GaN layer close to LS lines became a nanoporous structure. The cross-sectional SEM micrographs of the NP-LED structure were observed at perpendicular and parallel directions to the LS lines as shown in Figs. 2(c) and 2(d), respectively. In Fig. 2(c), parallel line patterns were observed at perpendicular directions to the LS lines. In Fig. 2(d), nano-hole patterns were observed on the cleave line which was parallel to the LS line and perpendicular to the lateral wet etching direction. High selectively lateral wet etching processes were observed at the u-GaN/n⁺-GaN:Si interfaces. In Fig. 2(e), a peel-off n⁺-GaN:Si layer was observed during the sample preparation process. The directional nono-pipe structure was observed clearly at the treated n⁺-GaN:Si layer perpendicular to the LS line. The direction of LS lines are perpendicular to the flat plane (11-20) of the sapphire, and the direction of LS line along the [1-100] GaN layer. In Fig. 2(f), in region-2, the mixture nanopipe and nano-hole structures were both observed in the bottom n-GaN conductive layer caused by the external bias voltage driving directly on this layer [19]. Only the nanopipe structure of the n⁺-GaN:Si layer was observed at the central mesa region (region-3). The n⁺-GaN:Si layer was isolated through the laser scribing process, then an external bias voltage was applied on the bottom n-GaN conductive layer. The positive charges were supplied from the bottom n-GaN conductive layer and mostly accumulated at the n⁺-GaN:Si layer. The positive charges attracted the etchants in the solution for the following EC etching process. The n⁺-GaN:Si layer transformed into the nanopipe GaN structure which was well controlled perpendicularly to the laser scribing lines.

 

Fig. 2 SEM images of an oblique view of NP-LED (a), its cross-section at a laser scribing (LS) line between 2 LED chips (b), cross-section perpendicular to the LS line (c) and cross-section parallel to the LS line (d), oblique view of residual n⁺GaN showing NP aligned along the etching direction after partial peeling of the LED layers on top (e), and cross-section showing nanopipes and nano-holes in the bottom conductive GaN:Si layer (f).

Download Full Size | PDF

In Fig. 3, the polarized microscopy images of the NP-LED structure at region-3, with the nanopipe GaN layer only, were observed. The OM images of the NP-LED structure were observed through polarized optical microscopy with an angle-rotated incident polarizer and a fixed analyzer. In Fig. 3(a), the incident light focused on the Au metal electrode lines (top p-metal line and bottom n-metal line) showing a darkened view at the cross of the polarizers condition. When the incident light was linearly polarized along the metal lines, the metal lines then appeared as dark lines since the reflected light was blocked by the analyzer. When the incident light was focused on the embedded EC-etched nanopipe GaN layer, a bright cross lines pattern was observed on the NP-LED structure as shown in Fig. 3(b). When the incident polarizer was rotated 45° clockwise, the line pattern along the direction of the polarizer was shown as a dark line pattern in Fig. 3(c). When, the polarizer was rotated 45° counter-clockwise as in Fig. 3(d), another dark line pattern along the direction of the polarizer appeared. A combined angle between these two rotated incident polarizers was 90° when observing these two series of bright lines. In Figs. 3(c) and 3(d), one side of the cross lines pattern was darkened by rotating the incident linear polarized light due to the phase retardation effect on the nanopipe layer. In Figs. 3(c) and 3(d), inserted polarized images of the NP-LED were observed by using a 450nm LED illuminated light source with the same polarization conditions. The phase retardation can be calculated as the formula: Γ = 2π·Δn·L/λ_o, in which the reflective light passes through the 0.85μm-thick nanopipe layer twice (L) with the 90° phase retardation (Γ), and wavelength of incident polarized light at 450nm (λ_o) of the blue LED lamp [20]. The birefringence Δn is the light refractive index difference parallel to and perpendicular to the nanopipe GaN structure was estimated to be 0.066. The OM images of the NP-LED structure with the different included angles between the incident polarizer and the analyzer shows obvious birefringence property in the anisotropic nanopipe GaN structure.

 

Fig. 3 Polarized optical microscopy images of the NP-LED structure were observed focused on (a) the metal lines and (b) the NP-GaN structure through the fixed analyzer (A). The incident polarizer (red, (P) and the analyzer (white, fixed, (A) are clearly labeled. OM images of the NP-LED were observed when the polarizer was rotated (c) 45° clockwise and (d) 45° counter clockwise. The inserted polarized images were observed by using a 450nm LED illuminated light source.

Download Full Size | PDF

In Fig. 4(a), the light emission intensity of the ST-LED structure was distributed uniformly on the whole LED mesa region. In the NP-LED structure, high light emission intensity was observed on the mesa region when compared with the ST-LED structure as shown in Figs. 4(a) and 4(b). In Fig. 4(b), high EL emission intensity of the NP-LED structure was observed at the mesa edge region (region 1) which had the triple-porous GaN layers. In region-2, the nanopipe-GaN/u-GaN/nanoporous-GaN structure was formed to increase the light extraction efficiency. The lateral etching width and the etching rate of the bottom n-type GaN conductive layer (4 × 10¹⁸cm⁻³) were measured to be 155μm and 5.2μm/min from region-1 to region-2. The n⁺-GaN:Si layer was totally etched as a directional nanopipe GaN structure with 560μm-width between the two laser scribing lines. The lateral etching width and the etching rate on the n⁺-GaN:Si layer (8 × 10¹⁸cm⁻³) were measured to be 280μm and 9.3μm/min, respectively. The lateral wet etching rate was proportional to the Si doping concentration in the n-type GaN:Si layer. Park et al. [13] reported on the wet electrochemical etching process with a 12μm/min lateral etching rate on n-type GaN:Si layer under a 60V external dc bias voltage. A wide lateral etching width and a high lateral etching rate have been demonstrated in the NP-LED structure under a low bias voltage condition. This EC lateral wet etching process with directional nanopipe structure can also be applied into a conventional LED structure simply by comparing it with the more complex epitaxial re-grown process [18]. In this study, the lateral wet etching rates of the nanopipe GaN structure were 6.4 and 9.3 µm/min at bias voltage of + 8 and + 10V, respectively. The long-range directional nanopipe structure perpendicular to the laser scribing line was well controlled by the bottom n-type GaN conductive layer that is supplying the electrical field as a driving force.

 

Fig. 4 EL emission images of the (a) ST-LED and the (b) NP-LED observed via a beam profiler.

Download Full Size | PDF

The EL emission spectra of the ST-LED and the NP-LED were measured by varying the injection current from 1mA to 20mA as shown in Figs. 5(a) and 5(b), respectively. In Fig. 5(a), the Fabry–Pérot (FP) phenomenon was observed in the EL spectra of the ST-LED structure. Strong and smooth EL emission spectra were observed in the NP-LED structure compared to ST-LED structure because of the light scattering process on the multi-porous GaN layers. In Fig. 5(c), the EL peak wavelength and the EL light output power were measured by varying the injection current using an integrating sphere. The emission wavelengths of both LED structures had a slightly redshifted phenomenon at a low injection current and a blue-shifted phenomenon at a high injection current. The compressive strain induced piezoelectric field in the InGaN active layer was opposite to the build-in electric field in the PN junction. The compressive strain in the InGaN active layers was caused by the lattice mismatch at InGaN/GaN active layers and the bottom GaN/sapphire interface. At a low injection current, the peak EL wavelength has a redshift phenomenon. The intrinsic electric field in the InGaN active layer of the LED structure on the c-plane sapphire substrate consisted of a build-in electric field and a strain-induced piezoelectric (PZ) field that have opposite electric field directions. By slightly increasing the injection current, the build-in electric field was reduced and PZ field was fixed. The net electric field in the InGaN well layer increased and induced the energy band to become more tilted and the wavelength to slightly redshift in the EL spectra [21]. By further increasing the injection current, the blueshift phenomenon of the peak EL wavelength was observed to be caused by the band filling effect in the InGaN quantum well layers. The peak emission wavelengths at 5 and 20 mA for ST-LED were measured to be 463.8 and 463.2 nm, respectively, while those for the NP-LED were 463.5 and 463.0nm, respectively. The light output power of the NP-LED structure has a 27.0% enhancement compared to ST-LED structure at a 20mA operation current. In Fig. 5(d), the light output power of both LEDs was measured by varying the injection current from 1mA to 150mA. The light output power of the NP-LED structure had a 26.5% enhancement compared to the ST-LED structure at 120mA operation current (current density, 25A/cm²). High light extraction efficiency is observed in the NP-LED with the embedded directional nanopipe GaN structure. The light reflectance at the GaN/NP-GaN interface and the light scattering process on the multi-porous GaN layers can enhance the external quantum efficiency in the LED structure.

 

Fig. 5 EL emission spectra of ST-LED (a) and NP-LED (b), peak wavelength and light output power under driving current up to 20mA (c), and light output power under driving current up to 150 mA(d).

Download Full Size | PDF

The PL spectra of both LED structures were measured using a HeCd 325nm laser as an excitation laser source as shown in Fig. 6. The PL spectra of the ST-LED and NP-LED structures were measured as shown in Figs. 6(a) and 6(b), respectively. The relative internal quantum efficiencies (IQE) in the MQW active layers were calculated by dividing the integrated PL spectra measured at 300 K (I_300K, room temperature) by the it measured at 10K (I_10K, low temperature) [22,23]. The integrated PL intensity ratios (I_300K/I_10K) are defined as the relative internal quantum efficiency by assuming that the IQE of 100% was achieved at 10K when the photo-excited carriers were frozen and localized in the MQW active layer. The relative internal quantum efficiencies were estimated to be 46.4 and 46.9% for ST-LED and NP-LED, respectively. Therefore the relative IQE values were the same for both LED structures.

 

Fig. 6 PL spectra of the ST-LED (a) and NP-LED (b) structures were measured at room temperature (300K) and 10K. The relative internal quantum efficiencies of both LED structures were calculated from the ratio of integrated PL intensity at the two measurement temperatures.

Download Full Size | PDF

In Fig. 7, the PL spectra of both LED structures were measured through an angle-resolved PL measurement by using a 405nm diode laser as an excitation laser source. The PL emission spectra were measured as shown in Fig. 7(a) for the ST-LED and in Fig. 7(b) for the NP-LED structures by varying the detection angles. The far-field radiation patterns were measured from the PL light collected from the front side of the LED chips. The line patterns of the angle-resolved PL spectra were observed in both LED structures indicating the Fabry–Pérot (FP) interference effect. The clearly observed FP phenomenon in the ST-LED was caused by the light reflectance between the top air/GaN and bottom GaN/sapphire interfaces as shown in Fig. 7(a).

 

Fig. 7 Angle-resolved PL spectra of the (a) ST-LED and (b) NP-LED structures were measured using a 405nm diode laser as an excitation source. The far-field radiation patterns were measured from the PL light collected from the front side of the LED chips.

Download Full Size | PDF

The thickness of the ST-LED can be calculated at about 5.9μm from the interference line pattern. In Fig. 7(b), the PL emission intensity of the NP-LED structure was higher than that of the ST-LED structure. The NP-GaN layer with the air-gap nanopipe structure has a lower effective refractive index compared to a non-treated u-GaN layer that can enhance the light reflectivity at the NP-GaN/top-GaN interface. The low FP phenomenon and the strong PL emission intensity of the NP-LED structure caused by the light reflection and the scattering processes occurred on the multi-porous GaN layers. The mixture line patterns of the NP-LED structure were caused by the light interference effect at the LED/NP-GaN and the GaN/sapphire interfaces. The non-clear FP line patterns of the NP-LED structure indicated that the light scattering process occurred on the multi-porous GaN layers.

To analyze the optical refractive index of the GaN and the fabricated NP GaN layers, a 1.11μm-thick n⁺-GaN:Si (8 × 10¹⁸cm⁻³)/2.0μm-thick u-GaN/sapphire structure was grown for the EC-etching process. The photoresist was coated on the top n-GaN:Si layer to prevent the EC-etching process proceeding in the vertical direction. After the etching process, the lateral etching directions were perpendicular to the laser scribing lines to form the directional nanopipe structure as shown in Fig. 8(a). The porosity of the GaN nanopipe structure at the perpendicular direction is calculated at about 16.5%. In Fig. 8(b), the optical refractive index of the bulk GaN layer (non-treated) and the GaN nanopipe layer at parallel and perpendicular directions was measured though an ellipsometer measurement. At 450nm, the refractive index values were measured at 2.46 for the non-treated GaN, 2.30 for the nanopipe at a parallel direction, and 2.21 at a perpendicular direction, respectively. then dry etched nanopipe structure, the effective refractive index of the nanopipe GaN layer was lower than that of the bulk GaN layer. The birefringence Δn, light refractive index difference, parallel and perpendicular to nanopipe GaN structure, was measured to be 0.09. The birefringence values of these two nanopipe GaN layers were slightly different due to the different porosity in the n-type GaN:Si layer with the different Si doping concentrations. By forming the nanoporous GaN layer, the refractive index contrast is small but a large refractive index contrast difference is located the surface of NP and air. The scattering happened inside the NP GaN layer is more effective when such reflective interface is available in random directions as the NP/air surface does. The low refractive index and the birefringence optical properties were measured in the directional nanopipe GaN structure through the lateral EC etching process.

 

Fig. 8 (a) An SEM image of the nanopipe-GaN/u-GaN/sapphire structure was prepared for optical analysis. (b) The refractive index as a function of the wavelength is measured for the n-type GaN and the fabricated NP GaN structure.

Download Full Size | PDF

4. Conclusion

The n⁺-GaN:Si layer inserted into the bottom u-GaN layer had been totally etched into nanopipes with lateral direction perpendicular to the etching side wall. Wide lateral etching width (>500μm) and high lateral etching rate were achieved in the directional nanopipe GaN layer. High light extraction efficiency of the NP-LED structure was caused by the high light scattering process on the nanopipe GaN layer and the multi-porous GaN layers. A birefringence optical property for the directional nanopipe GaN structure is observed clearly in the polarized OM images and in the ellipsometer measurement. The wide nanopipe GaN structure with the low effective refractive index and the birefringence properties can provide high efficiency optoelectronic devices applications.