A new approach to surface roughening was established and optimized in this paper for enhancing the light extraction of high power AlGaInP-based LEDs, by combining ultraviolet (UV) assisted imprinting with dry etching techniques. In this approach, hexagonal arrays of cone-shaped etch pits are fabricated on the surface of LEDs, forming gradient effective-refractive-index that can mitigate the emission loss due to total internal reflection and therefore increase the light extraction efficiency. For comparison, wafer-scale FLAT-LEDs without any surface roughening, WET-LEDs with surface roughened by wet etching, and DRY-LEDs with surface roughened by varying the dry etching time of the AlGaInP layer, were fabricated and characterized. The average output power for wafer-scale FLAT-LEDs, WET-LEDs, and DRY3-LEDs (optimal) at 350 mA was found to be 102, 140, and 172 mW, respectively, and there was no noticeable electrical degradation with the WET-LEDs and DRY-LEDs. The light output was increased by 37.3% with wet etching, and 68.6% with dry etching surface roughening, respectively, without compromising the electrical performance of LEDs. A total number of 1600 LED chips were tested for each type of LEDs. The yield of chips with an optical output power of 120 mW and above was 0.3% (4 chips), 42.8% (684 chips), and 90.1% (1441 chips) for FLAT-LEDs, WET-LEDs, and DRY3-LEDs, respectively. The dry etching surface roughening approach developed here is potentially useful for the industrial mass production of wafer-scale high power LEDs.
© 2014 Optical Society of America
Light-emitting diodes (LEDs) have been widely recognized as candidates for energy-saving and environment-friendly light sources. For the spectral region from yellow to red, the quaternary AlGaInP-based LEDs have been demonstrated as one of the promising material choices for LEDs devices used for exterior automotive lighting, traffic lights, full-color display signs and so on [1–3]. And, the internal quantum efficiency of high-quality AlGaInP-based LEDs can reach as high as 99% . However, the light extraction efficiency (LEE) of AlGaInP-based LEDs is not ideal, due to the absorption in the GaAs substrate and the large difference in refractive index between the AlGaInP-based LED material (n = 3.40 at λ = 650 nm) and the interfacing medium such as air (n = 1) or coupling epoxy layer (n = 1.5). Such large differences in refractive index results in small critical angles (θc) (e.g. θc = Sin−1 (nair /nAlGaInP) = ~17°) for light output and low LEEs as a result of the total internal reflection (TIR) effect . Therefore, there is a need for approaches to enhanced light extraction of AlGaInP-based LEDs, in order to make them energy-efficient light sources.
A variety of methods have been reported to increase the LEE of LEDs, including surface roughening , photonic crystals [6–8], colloidal-based microlens arrays [9,10], patterned substrates , optimized die shapes , and graded-refractive-index anti-reflection coating [13,14]. Among these methods, the surface roughening techniques have been extensively researched to improve the LEE of AlGaInP-based LEDs by wet and dry etching of the light-emitting surface, in most cases with the assistance of nanosphere lithography [15–18]. The introduced AlGaInP-saw-tooth structures , GaP nanopillars , AlGaInP-microbowls with nanorods , and nano-mesh ZnO layers , were found to substantially improve the LEE of chip-scale LEDs. However, for the application of these techniques in the industrial mass production, wafer-scale roughening and characterization of LEDs should be conducted.
One of the hindrances for uniform roughening of wafer-scale LEDs is the lack of an easy and reliable way to obtain the textured surface. Among the published surface roughening studies, many employ nanosphere lithography, with the monolayers of polystyrene or SiO2 spheres [16–19] or metal clusters  as a hard mask for either dry or wet etching to roughen the surface of chip-scale LEDs. Regardless of the demonstrated improvement in LEEs, however, the preparation of defect-free self-assembled monolayers of nanospheres in large areas with high throughput, to our knowledge, still remains challenging. Nanosphere packing defects such as voids and multilayers are not uncommon. Therefore, serious non-uniformity and poor repeatability could be present in the product LEDs. In contrast, ultraviolet-assisted nanoimprint lithography (UV-NIL), one of the recently developed unconventional lithography techniques, is simple to operate, suitable for large-area and high-throughput nanopatterning, and compatible with conventional nanofabrication facilities [21–23]. It would be worthwhile to apply UV-NIL as an alternative approach to the etch masks for LED surface roughening, and evaluate the feasibility of enhancing LEEs on wafer-scale LEDs with UV-NIL.
In this work, three types of wafer-scale LEDs were fabricated and tested. Samples labeled as “FLAT-LED” were conventional LEDs, namely without any surface roughening. Samples designated as “WET-LED” were LEDs with their surface roughened by wet etching. Samples named “DRY-LED” were LEDs with their surface roughened by dry etching through UV-NIL patterns. Imprinted nanopatterns by UV-NIL were used as an intermediate etch mask to produce SiO2 patterns, which then served as a dry etching mask to roughen the LED surface. The optimal etching time was determined in order to achieve the best etched structures for enhancing the light extraction. The electrical properties and light output power of three types of wafer-scale LEDs were investigated for comparison. DRY-LEDs prepared with the optimal etching time were found to have an increase in light output power by 68.6% on average, without noticeable degradation in the electrical properties and output power distribution.
2. Fabrication and characterization
Wafer-scale LEDs were prepared from an AlGaInP epitaxial structure grown on 2-inch GaAs (100) substrates by low-pressure metal-organic chemical vapor deposition (MOCVD). The epitaxial structure, with its dominant emission wavelength at 650 nm, consists of a 80-nm-thick n+-GaAs ohmic-contact layer, a 3-µm-thick Si-doped n-(Al0.5Ga0.5)0.5In0.5P (n-AlGaInP) current spreading layer, a 500-nm-thick Si-doped n-Al0.5In0.5P (n-AlInP) cladding layer, a 700-nm-thick undoped active layer with 30-period (AlxGa1-x)In0.5P/(AlyGa1-y)0.5In0.5P multiple quantum wells (MQW), a 200-nm-thick undoped Al0.5In0.5P (u-AlInP) cladding layer, a 550-nm-thick Mg doped p-Al0.5In0.5P (p-AlInP) cladding layer, a 100-nm-thick Mg doped p-(Al0.5Ga0.5)0.5In0.5P (p-AlGaInP) cladding layer, a 10-nm-thick p-GaInP tensile strain barrier reducing layer (TSBR), a 5-µm-thick Mg doped p-GaP window layer, and a 100-nm-thick p++-GaP ohmic-contact layer, as depicted in Fig. 1(a).After MOCVD growth, epi-structurelift-off and metal-bonding processes were adopted to transfer the LED structure to a p-type Si substrate for vertical current conduction, followed by deposition of AuGe/Au n-contact leads by lift-off, resulting in n-side up (n-AlGaInP) LEDs .
In this work, three different types of LEDs, FLAT-LEDs, WET-LEDs, and DRY-LEDs, were prepared. A FLAT-LED has a planar surface profile without any roughening on the n-AlGaInP layer surface (n+-GaAs in illumination area was removed by dipping in a solution of phosphoric acid (H3PO4): hydrogen peroxide (H2O2): H2O = 1: 1: 10 for 30 s), as shown in Fig. 1(b). For the fabrication of a WET-LED, after the removal of n+-GaAs by the above-mentioned method, the n-AlGaInP layer was dipped in a solution of H3PO4: hydrochloric acid (HCl): H2O = 5: 1: 2 for 5 min to produce tilted saw-tooth structures as shown in Fig. 1(c) . A DRY-LED depicted in Fig. 1(d), having a surface with uniform cone-shaped etch pits, was produced by the combination of UV-NIL and dry etching.
The fabrication scheme for DRY-LEDs is shown in Fig. 2.A 300-nm-thick SiO2 layer by plasma enhanced chemical vapor deposition (PECVD) was first deposited on n+-GaAs. Micropatterning of the deposited SiO2 layer was then conducted by conventional photolithography with a positive photoresist (AZ7220, AZ Electronic Materials), followed by dry etching in a mixture of C4F8 and O2 (9: 1) using an inductively coupled plasma (ICP) etcher (ICP380, 7 mTorr, 2000 W, 75 s), for the passivation of the n+-GaAs contact line. This was followed sequentially by the blanket deposition of additional 200–nm-thick SiO2, and spin coating of 200–nm-thick polymethyl methacrylate (PMMA 950K A3, MicroChem) and 250-nm-thick UV-curable Si-containing resin (NIP-SC28LV400, ChemOptics). PMMA was spin-coated on the SiO2 layer at 1000 rpm for 60 s, followed by heating the substrate on a hot plate at 170 °C for 5 min. Then UV-curable resin was spin-coated on the PMMA layer at 2500 rpm for 60 s. The film assembly was then ready for UV-NIL.
For UV-NIL, the silicon master used in this study consists of hexagonal arrays of holes that are 230 nm deep and 300 nm in diameter with a horizontal pitch of 500 nm and a diagonal pitch of 500 nm at 60 degree rotations, fabricated by deep ultraviolet (DUV) lithography and subsequent deep reactive ion etching (RIE). Structures on the silicon master were replicated onto a polyurethane acrylate (PUA) mold by UV imprinting to obtain a flexible polymer mold as described elsewhere [25,26]. Prior to use for imprinting, the PUA mold was treated by vapor phase deposition of trichloro(1H,1H,2H,2H-perfluorootyl)silane (97%, Sigma-Aldrich Co.) to improve the release between the PUA mold and the film being imprinted . When performing UV-NIL, the replicated PUA mold was pressed against the above-mentioned film assembly at a pressure of 20 bar at room temperature for 3 min, using a NIL-8 imprinter (Obducat, Sweden). The film was exposed to UV light (25 mW cm−2 with a major wavelength peak of 365 nm) through the PUA mold for 2 min to induce a photochemical reaction in the resin. By detaching the PUA mold from the irradiated film, hexagonal hole arrays were formed in the imprinted resin layer. The residual layer of the imprinted pattern was removed by etching with a mixture of CF4 and O2 (1: 7) in a plasma asher (ALA-0601E, 450 mTorr, 900 W, 15 s).
The imprinted pattern was subsequently transferred into the PMMA layer by O2 RIE (Versaline, 10 mTorr, 100 W, 60 s). To further transfer the imprinted pattern, the SiO2 layer was dry etched through the PMMA pattern in a 9: 1 mixture of C4F8 and O2 (Versaline, 10 mTorr, 100 W, 600 s), resulting in a SiO2 pattern with hole arrays. Finally, the etching of n+-GaAs and n-AlGaInP layers through the SiO2 pattern was conducted in a mixture of N2 and BCl3 (2: 3) in an ICP etcher (Multiplex, 5 mTorr, 900 W). By varying the etching time, 3.5 min (DRY1-LEDs), 4.0 min (DRY2-LEDs), 4.5 min (DRY3-LEDs), and 5.0 min (DRY4-LEDs), LEDs with various etch depths were prepared. The residual SiO2 pattern was removed by dipping in a buffered oxide etch solution (6: 1, NH4F/HF) for 3 min, followed by rinsing with de-ionized (DI) water (Milli-Q, Millipore Corp.) for 3 min. All LEDs prepared at wafer-scale were diced into 1 × 1 mm2 chips before characterization.
Morphology of the fabricated LEDs was examined using a field-emission scanning electron microscope (SEM) with a focused ion beam system (FIB, QUANTA 3D FEG, FEI, Netherlands). Electroluminescence (EL) spectra were measured by a StellarNet fiber optic spectrometer system with concave gratings. Measurements were performed under continuous wave (CW) at a constant heat sink temperature of 298 K. Wafer-scale characterization of the light output power was performed using a semi-auto LED prober on normal incidence (WPS3100, Opto System).
3. Results and discussion
Shown in Figs. 3(a)–3(d) are the tilted-view and cross-sectional SEM images obtained from the roughened n-AlGaInP surface of DRY-LEDs prepared with varied etching times. From these images, the depths of etched pits in the n-AlGaInP layer for DRY1-LEDs (3.5 min), DRY2-LEDs (4.0 min), DRY3-LEDs (4.5 min), and DRY4-LEDs (5.0 min) were measured to be 167, 191, 310, and 167 nm, respectively, and the etched angles of the cone-shaped pits were approximately 45°, 60°, 60°, and 110° respectively. At an etching time of 3.5 min, the diameter of etched pits is about 200 nm, two thirds of the hole diameter on the Si master. This indicates a substantial hole shrinking after pattern transfer by imprinting and multiple dry etch steps. With the etching time increasing from 3.5 to 4.0 min, the pit depth in the n-AlGaInP layer increased by 24 nm, while from 4.0 to 4.5 min the pit depth increased by 119 nm. This faster increase in pit depth is due to the widening of holes in the masking layer, as evident by the larger diameters of the etched pits in Figs. 3(b) and 3(c). The thinning of the n+-GaAs layer in Fig. 3(b) indicates the complete loss of masking SiO2 before 4.0 min etching time, and the n+-GaAs layer served as a masking layer at the moment. At 4.5 min the masking n+-GaAs layer was completely lost. The actual etch depth in the n-AlGaInP layer at this pointmight be larger than 310 nm, but the resulted pits are only 310 nm deep. Extended etching for 5.0 min simply led to much shallower etched pits (167 nm deep) and much widened etched angle (110 o) as shown in Fig. 3(d), due to the absence of any masking layer. To obtain the deepest etch pits in the n-AlGaInP layer, the dry etch should be stopped when the masking SiO2 and the n+-GaAs layer are both lost. Therefore, DRY3-LEDs (4.5 min etch) were prepared very close to this point.
Optical output power at an injection current of 350 mA was measured for the DRY1-LEDs, DRY2-LEDs, DRY3-LEDs, and DRY4-LEDs, and histograms of the optical output power are graphed for different DRY-LEDs in Fig. 3(e). DRY3-LEDs apparently have the narrowest power distribution and much higher output power. The average values of optical output power for the DRY1-LEDs, DRY2-LEDs, DRY3-LEDs, and DRY4-LEDs were found to be 72, 84, 172, and 113 mW, respectively, as listed in Table 1.Also given in Table 1 are the thickness of the residual n+-GaAs contact layer and etched pit depth for various LEDs.
The surface nanopattern can be regarded as a medium with graded-refractive-index . The effective refractive index, neff, can be estimated based on the effective mediumapproximation: , where n0 is the refractive index of air, n1 is the refractive index of the patterns (3.40 for AlGaInP and 3.83 for GaAs at λ = 650 nm), and f(z) is the filling fraction occupied by the surface patterned layer . The surface nanopatterns for the DRY-LEDs are illustrated in Fig. 4(a).And Fig. 4(b) shows neff(z) of the patterned surface as a function of the depth from the top surface, z, calculated using the data from Table 1. DRY3-LEDs and DRY4-LEDs show gradual decrease in neff from 3.4 to lower values, both with the total internal reflection condition at the top surface effectively reduced. However, DRY3-LEDs have much higher optical output than DRY4-LEDs, presumably related to their higher surface area for light to escape due to deeper and steeper etch pits. The flat area at the top surface of DRY1-LEDs, meaning a still very large filling fraction at z = 0, leads to a neff(0) >> 1. The same is true for DRY2-LEDs, but its further reduced top surface area brings neff(0) smaller than that for DRY1-LEDs. Prior works have shown the importance of aspect ratio and material index of refraction for the microstructure arrays in modifying the far-field radiation pattern and optimum light extraction efficiency in LEDs [29,30]. Here in our case, by tuning the etching time, various etched angles, aspect ratios, and resultant total surface areas of the cone-shaped pits, were obtained, leading to interesting diverse results. The material index of refraction (e.g. n-AlGaInP/n+-GaAs vs n-AlGaInP), the top diameter and the etched depth of the cone-shaped pits are all critical to the enhancement in light extraction efficiency. The much higher optical output for DRY3-LEDs is due to the increase of the light extraction in the large angular direction, i.e. the diffuse light enhancement. All these analyses are in good agreement with the experimental results shown in Fig. 3(e), i.e. DRY3-LEDs have the maximum optical output.
Light will undergo reflection at the flat top surface as well as the inclined surface of the etched nanopatterns. The number of repeated reflection and optical reflectance depend on the geometric shape of the nanopatterns. Thus, reflectance, R, can be expressed as the following equation, , where S is the center-to-center spacing of the neighboring nanopits, Sf is the width of flat top surface, R0 is the optical reflectance at the flat top surface, and Rn is the optical reflectance at the inclined surface for the nth reflection . R of the actual patterned device requires massive calculations, considering all possible directions of wave vectors and light polarizations. Instead of such massive calculations, R of the one-dimensional (1D) grating structure, whose cross-section is identical to that of our patterned LED device cutting along the line through the centers of the neighboring nanopits, was estimated as a close approximation to the actual R. R0 and Rn were calculated for two kinds of linearly polarized light with incident normal to the surface (λ = 650 nm) and then averaged using the Fresnel equation. R of the 1D grating corresponding to DRY3-LEDs wasfound to be ~7%, the smallest, well explaining their highest LEEs. R of the grating corresponding to DRY4-LEDs was more than 20%, leading to lower LEEs. Due to the residual GaAs layer in the patterns of DRY1-LEDs and DRY2-LEDs, R values of these LEDs are even higher, resulting in much lower LEEs. GaAs has large refractive index, 3.83 at λ = 650 nm , and hence increases optical reflection at the flat top surface. In addition, the absorption of this residual n+-GaAs layer could further hinder the light extraction of DRY1-LEDs and DRY2-LEDs, making them even less efficient than the FLAT-LEDs (102 mW at 350 mA). The most efficient LEDs were obtained with the deepest pit depth in the n-AlGaInP layer but no residual n+-GaAs layer on top. Therefore, dry etching of the n-AlGaInP layer for 4.5 min (DRY3-LEDs) provides the optimal results under our experimental conditions.
Figure 5 shows the light output power-current-voltage characteristics of chip-scale FLAT-LEDs, WET-LEDs and DRY3-LEDs. With an injection current of 350 mA, the forward voltages of these LEDs all approximate at 2.27 V, implying no electrical degradation due to the surface roughening process. The average values of light output power at 350 mA injection current for FLAT-LEDs, WET-LEDs, and DRY3-LEDs were found to be 102, 153, and 192 mW, respectively. Therefore, the light output power at 350 mA of chip-scale WET-LEDs is 50.0% higher than that of FLAT-LEDs, while for DRY3-LEDs it is 88.2% higher. The improvement in light output power for WET-LEDs and DRY3-LEDs could be related to the enhanced light scattering from the nano-roughened surface of the n-AlGaInP layer. The key to the enhanced escape probability is to give photons multiple opportunities to find escape cones. The nano-roughened surfaces cannot only provide the photons multiple angled surfaces to escape but also redirect photons that cannot escape from the first surface, back into a new escape cone . With the WET-LEDs, the tilt saw-tooth structures offer enhanced escape probability. However, tilt structures lead to directionality and LED photons escape at different efficiencies in different directions. In contrast, DRY3-LEDs have uniform cone-shape etched pits in the n-AlGaInP layer and LED photons escape at similar efficiencies in different directions.
Room-temperature electroluminescence (EL) of various LEDs at a forward current of 350 mA was also measured and the spectra are shown in Fig. 6(a). All three types of LEDs have the similar spectral shape and peak position (650 nm). However, the EL intensity (integrated intensity) obtained from the DRY3-LEDs is larger than those achieved from the FLAT-LEDs and the WET-LEDs. Namely, the EL intensity for the DRY3-LEDs was about 23.5% and76.6% higher than that for the WET-LEDs and the FLAT-LEDs, respectively, which is consistent with the results of the light output power. Optical images of the FLAT-LEDs, WET-LEDs, and DRY3-LEDs with emission driven at 1 mA are shown in Figs. 6(b)–6(d), respectively. Compared to the FLAT-LEDs and the WET-LEDs, a uniform distribution of radiation can be observed in the emitting areas of the DRY3-LEDs. This reflects that theuniform surface texture in the AlGaInP layer enhances the escape of the generated light and improves the spatial uniformity of the light output.
To determine the effect of surface roughening on light extraction for wafer-scale LEDs, the mapping results and histogram graphs of optical output power at an injection current of 350 mA are presented in Fig. 7 for wafer-scale FLAT-LEDs, WET-LEDs, and DRY3-LEDs. In the histogram graphs, the data below 55 mW of optical output power are not presented for graph clarity. The average output power for wafer-scale FLAT-LEDs, WET-LEDs, and DRY3-LEDs was found to be 102, 140, and 172 mW, respectively. Compared to the wafer-scale FLAT-LEDs, the wafer-scale WET-LEDs and DRY3-LEDs have 37.3% and 68.6% light output power enhancement, respectively. The optical output power for majority of the FLAT-LEDs falls within 90 to 110 mW and that for majority of the DRY3-LEDs falls within 160 to 190 mW. As shown in Figs. 8(a) and 8(c), the uniform surface roughening by dry etching resulted in only a slight widening of output power distribution. However, as shown in Figs. 8(b) and 8(d), due to non-uniform surface roughening of the n-AlGaInP layer by wet etching, wafer-scale WET-LEDs have a much wider range of optical output power, from 80 to 170 mW.
G. Tamulaitis et al. reported high-power LED based facility for plant cultivation using AlGaInP single-chip LEDs with about 120 mW optical output power (LuxeomTM type LXHL-MD1D of LUMILEDS LIGHTING, USA) . For the fabrication of red LEDs useful for applications such as plant cultivation, values of optical output power higher than 120 mW are desirable. Under our experimental conditions, the number of total chips for 2-inch wafer-scale LEDs is 1600 and the numbers of chips with more than 120 mW optical output power for wafer-scale FLAT-LEDs, WET-LEDs, and DRY3-LEDs are 4, 684, and 1441, respectively. Namely, the yield of LED chips with 120 mW optical output power or more for wafer-scale FLAT-LEDs, WET-LEDs, and DRY3-LEDs is 0.3, 42.8, and 90.1%, respectively. This makes the DRY3-LEDs potentially useful for the industrial mass production of wafer-scale high power LEDs.
We have demonstrated a UV-NIL assisted dry etching approach for wafer-scale surface roughening of high power AlGaInP-based LEDs. In this approach, the removal of n+-GaAs is integrated in the surface roughening. By fabricating hexagonal arrays of cone-shaped etch pits on the surface of LEDs with this approach, gradient effective–refractive-index, from n = 3.4 (AlGaInP) to gradually approaching to n = 1.0 (air), can be achieved to mitigate the emission loss due to total internal reflection and thus enhance the light extraction. Three types of wafer-scale LEDs were fabricated and studied in this work, FLAT-LEDs without any surface roughening, WET-LEDs with surface roughened by wet etching of the AlGaInP layer, and DRY-LEDs with surface roughened by dry etching of the AlGaInP layer through UV-NIL patterns. Wafer-scale DRY-LEDs with varied dry etching times, such as DRY1-LEDs (3.5 min), DRY2-LEDs (4.0 min), DRY3-LEDs (4.5 min), and DRY4-LEDs (5.0 min), were fabricated and evaluated. It was found that the DRY3-LEDs showed the highest average value of optical output power, 172 mW, at an injection current of 350 mA. Compared to the wafer-scale FLAT-LEDs, the wafer-scale WET-LEDs show 37.3% enhancement of light output power, and the wafer-scale DRY3-LEDs show 68.6% light output power enhancement. The yield of LED chips with 120 mW optical output power or more for wafer-scale FLAT-LEDs, WET-LEDs, and DRY-LEDs is 0.3, 42.8, and 90.1%, respectively. Based on the results obtained in this work, it is anticipated that our surface roughening method, combined UV-NIL patterning and dry etching, can be successfully applied to the development of wafer-scale high power LEDs for the industrial mass production.
This research was evenly supported by the Ministry of Small and Medium Business Administration, Republic of Korea, under research grants No. SL122760 and SV122720.
References and links
1. R.-H. Kim, M.-H. Bae, D. G. Kim, H. Cheng, B. H. Kim, D.-H. Kim, M. Li, J. Wu, F. Du, H.-S. Kim, S. Kim, D. Estrada, S. W. Hong, Y. Huang, E. Pop, and J. A. Rogers, “Stretchable, transparent graphene interconnects for arrays of microscale inorganic light emitting diodes on rubber substrates,” Nano Lett. 11(9), 3881–3886 (2011). [CrossRef] [PubMed]
2. A. Hayat, P. Ginzburg, and M. Orenstein, “Observation of two-photon emission from semiconductors,” Nat. Photonics 2(4), 238–241 (2008). [CrossRef]
3. T. Gessmann and E. F. Schubert, “High-efficiency AlGaInP light-emitting diodes for solid-state lighting applications,” J. Appl. Phys. 95(5), 2203–2216 (2004). [CrossRef]
4. I. Schnitzer, E. Yablonovitch, C. Caneau, and T. J. Gmitter, “Ultrahigh spontaneous emission quantum efficiency, 99.7% internally and 72% externally, from AlGaAs/GaAs/AlGaAs double heterostructures,” Appl. Phys. Lett. 62(2), 131 (1993). [CrossRef]
5. T. Fujii, Y. Cao, R. Sharma, E. L. Hu, S. P. DenBaars, and S. Nakamura, “Increase in the extraction efficiency of GaN-based light-emitting diodes via surface roughening,” Appl. Phys. Lett. 84(6), 855 (2004). [CrossRef]
6. J. J. Wierer, A. David, and M. M. Megens, “III-nitride photonic-crystal light-emitting diodes with high extraction efficiency,” Nat. Photonics 3(3), 163–169 (2009). [CrossRef]
7. J. Jewell, D. Simeonov, S.-C. Huang, Y.-L. Hu, S. Nakamura, J. Speck, and C. Weisbuch, “Dobule embedded photonic crystals for extraction of guided light in light-emitting diodes,” Appl. Phys. Lett. 100(17), 171105 (2012). [CrossRef]
8. K. McGroddy, A. David, E. Matioli, M. Iza, S. Nakamura, S. DenBaars, J. S. Speck, C. Weisbuch, and E. L. Hu, “Directional emission control and increased light extraction in GaN photonic crystal light emitting diodes,” Appl. Phys. Lett. 93(10), 103502 (2008). [CrossRef]
9. W. H. Koo, W. Youn, P. Zhu, X.-H. Li, N. Tansu, and F. So, “Light extraction of organic light emitting diodes by defective hexagonal-close-packed array,” Adv. Funct. Mater. 22(16), 3454–3459 (2012). [CrossRef]
10. X.-H. Li, P. Zhu, G. Liu, J. Zhang, R. Song, Y.-K. Ee, P. Kumnorkaew, J. F. Gilchrist, and N. Tansu, “Light extraction efficiency enhancement of III-Nitride light-emitting diodes by using 2-D close-packed TiO2 microsphere arrays,” J. Display Technol. 9(5), 324–332 (2013). [CrossRef]
11. L. Li, T. Zhai, H. Zeng, X. Fang, Y. Bando, and D. Golberg, “Polystyrene sphere-assisted one-dimensional nanostructure arrays: synthesis and applications,” J. Mater. Chem. 21(1), 40–56 (2010). [CrossRef]
13. J. K. Kim, S. Chhajed, M. F. Schubert, E. F. Schubert, A. J. Fischer, M. H. Crawford, J. Cho, K. Kim, and C. Sone, “Light-extraction enhancement of GaInN light-emitting diodes by graded-refractive-index indium tin oxide anti-reflection contact,” Adv. Mater. 20(4), 801–804 (2008). [CrossRef]
14. C. H. Chiu, P. Yu, C. H. Chang, C. S. Yang, M. H. Hsu, H. C. Kuo, and M. A. Tsai, “Oblique electron-beam evaporation of distinctive indium-tin-oxide nanorods for enhanced light extraction from InGaN/GaN light emitting diodes,” Opt. Express 17(23), 21250–21256 (2009). [CrossRef] [PubMed]
15. Y. J. Lee, T. C. Lu, H. C. Kuo, S. C. Wang, T. C. Hsu, M. H. Hsieh, M. J. Jou, and B. J. Lee, “Nano-roughening n-side surface of AlGaInP-based LEDs for increasing extraction efficiency,” Mater. Sci. Eng. B 138(2), 157–160 (2007). [CrossRef]
16. R.-H. Horng, T.-M. Wu, and D.-S. Wuu, “Improved light extraction in AlGaInP-based LEDs using a roughened window layer,” J. Electrochem. Soc. 155(10), H710–H715 (2008). [CrossRef]
17. Y.-C. Lee, H.-C. Kuo, B.-S. Cheng, C.-E. Lee, C.-H. Chiu, T.-C. Lu, S.-C. Wang, T.-F. Liao, and C.-S. Chang, “Enhanced light extraction in wafer-bonded AlGaInP-based light-emitting diodes via micro- and nanoscale surface textured,” IEEE Electron Device Lett. 30(10), 1054–1056 (2009). [CrossRef]
18. J. J. Chen, Y. K. Su, C. L. Lin, and C. C. Kao, “Light output improvement of AlGaInP-based LEDs with nano-mesh ZnO layers by nanosphere lithography,” IEEE Photon. Technol. Lett. 22(6), 383–385 (2010). [CrossRef]
19. R. Windisch, B. Dutta, M. Kuijk, A. Knobloch, S. Meinlschmidt, S. Schoberth, P. Kiesel, G. Borghs, G. H. Bohler, and P. Heremans, “40% Efficient thin-film surface-textured light-emitting diodes by optimization of natural lithography,” IEEE Trans. Electron. Dev. 47(7), 1492–1498 (2000). [CrossRef]
20. C. Huh, K. S. Lee, E. J. Kang, and S. J. Park, “Improved light-output and electrical performance of InGaN-based light emitting diode by microroughening of the p-GaN surface,” J. Appl. Phys. 93(11), 9383–9385 (2003). [CrossRef]
21. H.-H. Park, D.-G. Choi, X. Zhang, S. Jeon, S.-J. Park, S.-W. Lee, S. Kim, K.-D. Kim, J.-H. Choi, J. Lee, D. K. Yun, K. J. Lee, H.-H. Park, R. H. Hill, and J.-H. Jeong, “Photo-induced hybrid nanopatterning of titanium dioxide via direct imprint lithography,” J. Mater. Chem. 20(10), 1921–1926 (2010). [CrossRef]
22. H.-H. Park, X. Zhang, S.-W. Lee, K.-D. Kim, D.-G. Choi, J.-H. Choi, J. Lee, E.-S. Lee, H.-H. Park, R. H. Hill, and J.-H. Jeong, “Facile nanopatterning of zirconium dioxide films via direct ultraviolet-assisted nanoimprint lithography,” J. Mater. Chem. 21(3), 657–662 (2010). [CrossRef]
23. H.-H. Park, W. L. Law, X. Zhang, S.-Y. Hwang, S. H. Jung, H.-B. Shin, H. K. Kang, H.-H. Park, R. H. Hill, and C. K. Ko, “Facile size-tunable fabrication of functional tin dioxide nanostructures by multiple size reduction lithography,” ACS Appl. Mater. Interfaces 4(5), 2507–2514 (2012). [CrossRef] [PubMed]
24. L. J. Yan, C. C. Yang, M. L. Lee, S. J. Tu, C. S. Chang, and J. K. Sheu, “AlGaInP/GaP heterostructures bonded with Si substrate to serve as solar cells and light emitting diodes,” J. Electrochem. Soc. 157(4), H452 (2010). [CrossRef]
26. K. Y. Suh, H. E. Jeong, D.-H. Kim, R. A. Singh, and E.-S. Yoon, “Capillarity-assisted fabrication of nanostructures using a less permeable mold for nanotribological applications,” J. Appl. Phys. 100(3), 034303 (2006). [CrossRef]
27. J. Y. Kim, D.-G. Choi, J.-H. Jeong, and E.-S. Lee, “UV-curable nanoimprint resin with enhanced anti-sticking property,” Appl. Surf. Sci. 254(15), 4793–4796 (2008). [CrossRef]
28. F. W. Mont, J. K. Kim, M. F. Schubert, E. F. Schubert, and R. W. Siegel, “High-refractive-index TiO2-nanoparticle-loaded encapsulants for light-emitting diodes,” J. Appl. Phys. 103(8), 083120 (2008). [CrossRef]
29. X.-H. Li, R. Song, Y.-K. Ee, P. Kumnorkaew, J. F. Gilchrist, and N. Tansu, “Light extraction efficiency and radiation patterns of III-nitride light-emitting diodes with colloidal microlens arrays with various aspect ratios,” IEEE Photonics J. 3(3), 489–499 (2011). [CrossRef]
30. P. Zhu, G. Liu, J. Zhang, and N. Tansu, “FDTD analysis on extraction efficiency of GaN light-emitting diodes with microsphere arrays,” J. Display Technol. 9(5), 317–323 (2013). [CrossRef]
31. J. Zhao and M. A. Green, “Optimized antireflection coatings for high-efficiency silicon solar cells,” IEEE Trans. Electron. Dev. 38(8), 1925–1934 (1991). [CrossRef]
32. V. Roppo, C. Cojocaru, F. Raineri, G. D. Aguanno, J. Trull, Y. Halioua, R. Raj, I. Sagnes, R. Vilaseca, and M. Scalora, “Field localization and enhancement of phase-locked second- and third-order harmonic generation in absorbing semiconductor cavities,” Phys. Rev. A 80(4), 043834 (2009). [CrossRef]
33. W. C. Peng and Y. S. Wu, “Improved luminance intensity of InGaN–GaN light-emitting diode by roughening both the p-GaN surface and the undoped-GaN surface,” Appl. Phys. Lett. 89(4), 041116 (2006). [CrossRef]
34. G. Tamulaitis, P. Duchovskis, Z. Bliznikas, K. Breivė, R. Ulinskaitė, A. Brazaitytė, A. Novičkovas, and A. Žukauskas, “High-power light-emitting diode based facility for plant cultivation,” J. Phys. D Appl. Phys. 38(17), 3182–3187 (2005). [CrossRef]