Fabrication of wafer-scale nanoporous AlGaN-based deep ultraviolet distributed Bragg reflectors via one-step selective wet etching

In this paper, we reported on wafer-scale nanoporous (NP) AlGaN-based deep ultraviolet (DUV) distributed Bragg reflectors (DBRs) with 95% reflectivity at 280 nm, using epitaxial periodically stacked n-Al0.62Ga0.38N/u-Al0.62Ga0.38N structures grown on AlN/sapphire templates via metal–organic chemical vapor deposition (MOCVD). The DBRs were fabricated by a simple one-step selective wet etching in heated KOH aqueous solution. To study the influence of the temperature of KOH electrolyte on the nanopores formation, the amount of charge consumed during etching process was counted, and the surface and cross-sectional morphology of DBRs were characterized by Scanning electron microscopy (SEM) and atomic force microscopy (AFM). As the electrolyte temperature increased, the nanopores became larger while the amount of charge reduced, which revealed that the etching process was a combination of electrochemical and chemical etching. The triangular nanopores and hexagonal pits further confirmed the chemical etching processes. Our work demonstrated a simple wet etching to fabricate high reflective DBRs, which would be useful for AlGaN based DUV devices with microcavity structures.

www.nature.com/scientificreports/ In this work, we have fabricated wafer-scale DUV NP-DBRs with a high reflectivity (> 95%) by the simple one-step vertically-laterally ECE in heated KOH aqueous solution. The influence of electrolyte temperature on the porosity of the n-AlGaN layer was studied in detail, and the formation mechanisms of nanopores under the synergic actions of ECE and chemical etching (CE) was clarified.

Experimental
As shown in Fig. 1a, the epitaxial structure consisted of a 1.5 μm-thick AlN buffer and 40 pairs of n-Al 0.62 Ga 0.38 N/ unintentional Al 0.62 Ga 0.38 N (28 nm/30 nm) DBRs layers, which were grown on 2-in. c-plane sapphire substrates by metal-organic chemical vapor deposition (MOCVD). During the growth, trimethylaluminum (TMA), trimethylgallium (TMG) and ammonia (NH 3 ) were used as Al, Ga and N sources, respectively. Silane (SiH 4 ) was the n-type dopant gases. Hydrogen was used as the carrier gas. The surface of AlGaN films was III-polar in this growth mode. The silicon doping concentration of n-Al 0.62 Ga 0.38 N was 4 × 10 18 cm −2 , and unintentional Al 0.62 Ga 0.38 N (u-Al 0.62 Ga 0.38 N) was about 1 × 10 16 cm −2 . Before electrochemical etching, the good electrical contact was formed by soldering indium at the sample edge. Then the AlGaN samples were anodization etched for 5 min, in 1 M KOH aqueous solution, with a platinum (Pt) plate as the counter electrode. The KOH aqueous solution was heated by an electric heating panel for constant temperature during etching. After etching, the samples were rinsed in deionized water and dried in N 2 . Scanning electron microscopy (SEM, Gemini300) and atomic force microscopy (AFM, SPM9700) were used to characterize the surface and cross-sectional morphologies of the samples. The reflectivity was measured with a UV-Vis spectrophotometer (SolidSpec-3700). The standard reference mirror (R > 99%, 200-400 nm) was used in the reflectivity spectra measurement. The finite element simulation of reflectivity was done using the Comsol Multiphysics software. To simplify the simulation, a bulk material with the same effective refractive index replaces the nanoporous layer. Keithley 2400C source meter provides DC constant bias and real-time monitoring and recording current through software. Figure 1B shows the reflectivity spectra of NP-DBRs etched at 25 V DC bias at different electrolyte temperatures. The reflectivity of the unetched sample was low in the DUV region (< 30%), and significantly declined at 257 nm corresponding to the absorption edge of Al 0.62 Ga 0.38 N. After ECE, a reflection peak appeared near 278 nm, and the peak reflectivity increased as the electrolyte was heated up from 25 to 65 °C, then decreased at 85 °C. Thus, the highest reflectivity 95% was obtained at 65 °C. To study the influence of electrolyte temperature on reflectivity, the cross-sectional morphology of DBRs prepared at different electrolyte temperatures was characterized by SEM. As shown in Fig. 1c-e, the nanopores showed an obvious layered distribution, concentrated in n-Al 0.62 Ga 0.38 N  23 . As the electrolyte temperature increased, the nanopores size increased significantly, resulting in a larger refractive index contrast. This is the reason why the DBRs reflectivity increased with the temperature rising from 25 to 65 °C. The porosity of the n-Al 0.62 Ga 0.38 N layers etched at 65 °C was estimated to be 20% from the SEM image. Then, the n eff was calculated to be 2.35. The mismatch between simulation and experiment reflectivity is due to the finite nonuniformity of nanopores in n-Al 0.62 Ga 0.38 N layers and the scattering of DBRs surface. However, when the electrolyte temperature increased to 85 °C, the nanopores was too large and extend into the u-Al 0.62 Ga 0.38 N layer, as shown in Fig. 1f. Therefore, the periodic stacking of nanopores was destroyed, leading to the decrease of reflectivity compared with that at 65 °C. Subsequently, a 2-in. sample was etched at 25 V and 65 °C, and the reflectivity spectra were measured at the center, sub-center, and edge of the wafer, respectively, as shown in Fig. 2. The reflectivity at the three points was all over 95%, indicating that a wafer-scale DUV DBR was successfully prepared. The stopband of the reflectivity spectra was about 4.5 nm which was mainly limited by the short reflection wavelength and the low refractive index contrast. The peak wavelengths of the reflection spectra at the three points are 282 nm, 280.5 nm, and 279.5 nm, respectively, shifting slightly round 280 nm. The uneven doping and thickness can be the main cause of the slight shift. In general, the ECE process of the anode has three stages: (i) Avalanche breakdown or Zener tunneling to generate holes; (ii) Anode oxidation of n-AlGaN at the interface between the film and the electrolyte; (iii) The oxides dissolving in the electrolyte to form nanopores 24 . The reaction can be expressed as:

Results and discussion
Meanwhile, electrons (e − ) are transferred to the Pt electrode, participating in the reduction reaction: According to Faraday's law of electrolysis, the current density and the amount of charge passing through the source meter are proportional to the ECE rate and the mass of the AlGaN undergoing redox reaction, respectively. Figure 3a shows the J-t plots during ECE at 25 V. The current decreased to 0 after 23 s etching, indicating that the ECE process occurred within the first 23 s. When the temperature rose, due to the faster dissolution rate of oxides in the hotter KOH solution, the initial current density increased. Figure 3b shows the amount of charge consumed during ECE process, which was obtained by integrating J − t plots. The amount of charge decreased with the temperature rising, indicating that the mass of the AlGaN undergoing redox reaction decreased. Thus, it can be concluded that the increase of nanopores size with temperature is related to chemical etching other than ECE.
AlGaN film with III-polar plane was grown along the [0001], while that with N-polar plane grown along the opposite [0001 ]. Different from the III-polar plane, the N-polar plane is very easy to be etched in hot KOH solution 25 , and the etching rate is temperature and crystal face dependent 26,27 . The CE reaction can be expressed as: www.nature.com/scientificreports/ Here we propose the formation process of nanopores under the combined action of ECE and CE. The ECE process makes n-Al 0.62 Ga 0.38 N layers porous and CE can be ignored at 25 °C, as shown in Fig. 3c i . The small nanopore size can be explained by the depletion model 28 . The thickness of space charge region (SCR) around nanopores is expressed as: d SCR = 2εε 0 U SCR qN D , where ε is the dielectric constant, ε 0 is the permittivity of AlGaN, U SCR is the voltage drop across the depletion region, q is the charge of an electron, and N D is the n-doping concentration. Due to the relatively low doping concentration of high Al composition AlGaN, the SCR around nanopores were thick. When the SCR overlapped, the space between nanopores was completely depleted, so the pore diameter cannot increase due to the lack of holes required for ECE. Moreover, ECE was isotropic without crystal plane selectivity, thus the nanopores are approximately circular. However, CE can no longer be ignored as the electrolyte temperature increases to 45 °C. The nanopores formed by ECE exposed the N-polar plane of the AlGaN film. Subsequently the CE of the N-polar plane takes place and the nanopores size increase, as shown in Fig. 3c ii . Undoubtedly, the effect of chemical etching is more obvious at 65 °C. The nanopores presents a triangular profile composed of {0001} and {1101} family of planes, as shown in Fig. 3c iii , which is due to the relatively stable energy of the {1101} family of planes 19,26 . Since CE has no silicon doping selectivity, when the solution temperature was 85 °C, the triangular nanopores could expand to u-Al 0.62 Ga 0.38 N layers and destroy the periodic stacking of the nanopores, as shown in Fig. 3c iv . This is the key reason why the reflectivity of DBR prepared in 85 °C KOH electrolyte decreases compared with that at 65 °C. www.nature.com/scientificreports/ The surface morphology of DBRs was also characterized, as shown in Fig. 4. The evenly distributed nanopores, about 35 nm in diameter, were formed on the DBR surface after etching at 25 V and 25 °C (Fig. 4a). As the temperature rose, the pore size increased at 45 °C (Fig. 4b), and the hexagonal pits appeared at 65 °C (Fig. 4c). When the temperature continued to rise to 85 °C, the surface was completely etched, which was ascribed to the coalescence of adjacent hexagonal pits (Fig. 4d). Due to the extremely low carrier concentration of the surface u-Al 0.62 Ga 0.38 N layer and the overlap of SCR, the nanopore size was small at 25 °C. However, in the depletion model, the SCR was not influenced by the electrolyte temperature, thus the formation of hexagonal pits cannot be attributed to ECE. Figure 4e is the enlarged image of the yellow box in Fig. 4d. It can be observed that the dendritic lines appeared around the nanopores. These lines were the paths for the diffusion and etching of KOH electrolyte in the n-Al 0.62 Ga 0.38 N layer. Due to the overlap of SCR, the dendritic lines around adjacent nanopores cannot be connected, thus forming a clear boundary, as marked by the red line. Therefore, it can be concluded that the nanopores acted as vertical downward channels for electrolyte in ECE process and the electrolyte spread laterally in n-Al 0.62 Ga 0.38 N layers 21,29 . In addition, the porous n-Al 0.62 Ga 0.38 N layer allowed the hot KOH solution to chemically etch the u-Al 0.62 Ga 0.38 N from the N-polar plane, which accelerated the etching of the surface u-Al 0.62 Ga 0.38 N layer. The hexagonal profile of the surface pits was formed due to the intersection of {0001} and {1 1 01} family of planes. The surface morphology of the DBR etched at 85 °C and the depth of the hexagonal pits were also characterized and measured by AFM, as shown in Fig. 4f and g, respectively. The selected hexagonal pit exposed five step surfaces and seven step heights (∆h x ). Except for Δh 1 and Δh 2 , the other step heights are between 55 and 59 nm, consistent with the thickness (58 nm) of a pair of n-Al 0.62 Ga 0.38 N/u-Al 0.62 Ga 0.38 N (28 nm/30 nm). This confirms that a pair of n-Al 0.62 Ga 0.38 N/u-Al 0.62 Ga 0.38 N was etched as a whole where the n-Al 0.62 Ga 0.38 N layer was etched by both ECE and CE while the u-Al 0.62 Ga 0.38 N etched by CE only, thus forming hexagonal pits with uniform depth.
The influence of sample conductivity (doping) and anodization potential have been researched by Han et al. 28 . We have tried to play with the voltage for porosity tuning. However, the reflectivity of the DBR prepared at 30 V was lower than that at 25 V, as shown in Fig. 5. The relatively low Si doping concentration of n-Al 0.62 Ga 0.38 N (4 × 10 18 cm −2 ) resulted in poor ECE selectivity at the higher voltage. Some vertical nanochannels (as marked by the red arrows) instead of periodically stacked nanopores were observed on the cross-sectional SEM images of the DBRs etched at 30 V. It was not feasible to achieve high reflectivity just by playing with the voltage.