Open Access
Add to BibSonomy
Abstract
This study demonstrates that selective-area Si implantation performed on the GaN templates instead of conventional dielectric layers, such as SiO2 or SiNx, serves as the mask layer for the epitaxial lateral overgrowth (ELOG) process. Although the substantial mask layer is absent on the templates, selective growth initially occurs on the implantation-free area and then evolves a lateral overgrowth on the Si-implanted area during the regrowth process. This selective growth is attributed to that the crystal structure of the Si-implanted area subjected to the high doses of ion bombardment produces an amorphous surface layer, thereby leading to a lattice mismatch to the regrown GaN layer. Microstructural analyses reveal that the density of the threading dislocations above the Si-implanted regions is markedly lower than the GaN layer in the implantation-free regions. Consequentially, UV LEDs fabricated on the Si-implanted GaN templates exhibit relatively higher light output and lower leakage current compared with those of LEDs grown on ELOG-free GaN templates.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
GaN-based ultraviolet (UV) light-emitting diodes (LEDs) have been applied in water purification, the disinfection of medical tools, UV curing, document authentication, phototherapy and medical diagnostics [1–3]. High efficiency is one of the key issues for the LEDs used in the applications mentioned above. The efficiency issue of UV LEDs is mainly dependent on the internal quantum efficiency (IQE), but also on the light extraction efficiency (LEE). Many techniques have been used to enhance the LEE of GaN-based visible LEDs through light scattering from the roughened interfaces between the air and semiconductor layer or the growth substrate and semiconductor layer [4,5]. Considering the IQE issue of GaN-based UV LEDs, most efforts have been paid to alleviate the carrier overflow outside the active layer at high current injection and reduce the defect states in the epitaxial layers. The former can be partly achieved using complicated structural designs of epitaxial layers in the LEDs. The latter mainly originates from the threading dislocations (TDs) through the active layer of the LEDs and can be partly suppressed through the adjustment of epitaxial growth conditions. In contrast to UV LEDs, the IQE of GaN-based visible LEDs correlated with material quality issues is relatively insignificant because the high indium-containing active layer of visible LEDs could cause the formation of In-rich or quantum dot-like clusters to decrease the recombination events between injection carriers and structural defects [6]. In other words, the radiative recombination is insensitive to the structural defects. However, the indium-rich clusters do not seem to exist in the active layer of GaN-based UV LEDs due to the lack of indium or low indium content [6]. Therefore, the material quality issue becomes a crucial point for improving the IQE of GaN-based UV LEDs. Epitaxial lateral overgrowth (ELOG) is a well-known technique for reducing the TD density of GaN epitaxial layers grown on sapphire substrates [7]. For a typical ELOG process, the regrowth of GaN layers is performed on a GaN template layer with a stripe mask layer made of a thin dielectric film, e.g., SiO2 or SiNx. However, the impurity contamination during regrowth caused by out-diffusion of the mask material (e.g., SiO2 or SiNx) may lead to the degradation of crystal quality [8]. In this study, we demonstrate a novel mask-free technique to grow GaN-based LED structures on GaN templates with selective-area Si implantation. The Si implantation creates lattice distortion on the surface layer of the GaN template and this Si-implanted area serves as the mask area while the implantation-free area serves as the seed area during the regrowth process. In a typical regrowth process, the epitaxial layer initially grows on the implantation-free area toward the normal-plane direction to form a series of GaN stripes while the growth on the Si-implanted area does not occur because the lattice constant of the implanted area is slightly different from the implantation-free area [9–10]. After this growth stage, the epitaxial layer grows laterally and increasingly over the Si-implanted area. Consequently, the GaN stripes become wide and thereby merge laterally together to form a continuous film. These growth mechanisms are similar to the conventional ELOG process using dielectric materials as a mask layer. In addition, the average TD density in the GaN epitaxial layers can also be reduced compared with the GaN layers directly grown on the sapphire substrate. Therefore, GaN-based UV LED structures grown on GaN templates with a buried Si-implanted area exhibit improved performance compared with those who grown directly on the sapphire substrate. The details regarding the fabrication procedures and device characterization are discussed in the following sections.
2. Experiment
The undoped GaN (u-GaN) epitaxial wafers were grown on c-face (0001) sapphire substrates in a vertical metal-organic vapor-phase epitaxy (MOVPE) system. Trimethylgallium (TMGa) and NH3 were used as the sources of gallium and nitrogen, respectively. The layer structure of the u-GaN epitaxial wafers consisted of a 30 nm-thick GaN nucleation layer and a 2 µm-thick u-GaN layer grown at 560 and 1050 °C, respectively. The growth pressure of the nucleation and u-GaN layers was 200 torr. Next, a 90-nm thick SiO2 layer prepared by plasma-enhanced chemical vapor deposition and a 6-µm thick photoresistor (PR) layer were sequentially deposited on the u-GaN wafers. The SiO2 layer can avoid the channeling effect during the Si ion implantation process and facilitate the accumulation of implanted Si ions near the GaN surface. The PR layer was then developed with an A µm-wide stripe opening through photolithography and hence of a B µm-wide PR-masked area, which served as a mask layer to shade the Si ion flux. Consequently, the Si-implanted area percentage (Aimp =A/(A + B)) varies with the different A and B values. In this study, the design had four different values of A, i.e., 8, 7, 6 and 3 µm, and four different values of B, i.e., 3, 4, 5 and 20 µm. These values correspond to four different Aimp, i.e., 73%, 64%, 55% and 13% on the u-GaN wafers. The Si-ion implantation, with dosage and energy of 1 × 1016/cm2 and 70 keV, respectively, was performed on the aforesaid u-GaN epitaxial wafers to create a series of Si-implanted striped areas on the u-GaN layer. The Si-implanted layer had an average depth of ∼60 nm from the u-GaN layer surface. This depth was estimated by simulation using the transport of ions in matter (TRIM) program and the inspection of transmission electron microscopy (TEM) images. After the Si implantation process, the PR and SiO2 layers were chemically removed using H2SO4-based (H2SO4:H2O2 = 3:1) and HF etchants, respectively. Figure 1(a) shows the schematic fabrication process of the Si-implanted u-GaN epitaxial layers and the subsequent regrowth of the GaN layer. The Si-implanted stripes were arranged along with the $[{\textbf{11}\bar{\textbf{2}}\textbf{0}} ]$ direction. The u-GaN wafers with different Aimp values were then loaded into the MOVPE system to regrow GaN-based epitaxial structures. Simultaneously, u-GaN wafers without Si implantation (i.e., Aimp = 0) were also loaded into the MOVPE system to grow GaN-based epitaxial structures for comparison.
Fig. 1. (a) Schematic fabrication process for the epitaxial wafers with Si-implanted stripe area on the u-GaN layer. (b) XRD spectra taken from the implantation-free and Si implanted (Aimp=100%) GaN templates at the (004) plane.
3. Results and discussion
Figure 1(b) shows the typical (004) X-ray diffraction (XR spectra taken from the implantation-free and Si-implanted (Aimp = 100%) GaN templates. The Si-implanted GaN exhibited a two-layer property with an additional broad peak located on the low-angle side. This result indicates that a lattice disorder (i.e. expansion) exists in the GaN layer. The lattice disorder was attributed to the fact that the implanted Si ions collide with the host atoms of the GaN target during ion implantation to cause the lattice distortion. These collisions occur both in- and normal-plane directions, thereby resulting in a three-dimensional Gaussian distribution to cause the expansion and/or disorder of lattice in the GaN layer. The distorted region (i.e. depth) depends on the incident energy of the ions.
Consequentially, an amorphous-like layer can be produced by high-dose ion bombardment into a semiconductor layer, where the critical dose for GaN amorphization was ∼1016 cm−2. In addition, the difference of lattice constant between the Si-implanted GaN and the implantation-free GaN increased with the implantation dose and/or energy [11–13]. Although the lattice constant of Si-implanted GaN may restore to the implantation-free GaN due to the thermal annealing during the regrowth of GaN in the MOVPE system, the effect of thermal annealing on the lattice recovery of the Si-implanted GaN is minor, as proven by previous studies [13]. In other words, the double-peak XRD spectra were still observed from the annealed Si-implanted GaN, indicating that the annealed Si-implanted GaN still possesses a double-layer property. This result suggests that the lattice distortion of Si-implanted GaN would not be completely restored even after high-temperature thermal annealing. This contention was confirmed using selective-area electron diffraction under TEM performed on the GaN films grown on the Si-implanted GaN templates [9].
To evaluate the effect of the Si-implantation area on the selectivity of GaN regrowth, the u-GaN layer grown on the Si-implanted GaN templates with different Aimp values was investigated. For this evaluation process, the baseline growth rate of GaN was 2 µm/h, which was determined using the implantation-free wafers as the regrowth templates. The growth temperature, pressure and V/III ratio (gas-phase molar flow rates of NH3 to TMGa) were fixed at 1000 °C, 100 torrs and 610, respectively. The regrowth runs were carried out under identical conditions to the baseline runs but differed in growth duration. Figure 2(a) shows the top-view images taken using scanning electron microscopy (SEM) from the u-GaN layers grown on the selective-area Si-implanted templates after a growth duration of 5 min. There are a series of stripes and trenches on the surface of the wafers. The stripes and trenches corresponded to the implantation-free and the Si-implanted area, respectively, on the surface of the Si-implanted GaN templates. No regrown GaN layers could be observed in the Si-implanted area. However, the GaN growth occurred at the edge of the Si-implanted area, but it was not directly grown thereon [9]. This phenomenon was attributed to the fact that the reactant species (Ga and N) diffusing to the implantation-free area nucleate preferentially on GaN thereon because the lattice constant of the implantation-free surface layer is matched with the regrown GaN, as schematically shown in Fig. 2(b). In other words, the crystal structure of the Si-implanted area subjected to high doses of ion bombardment produces an amorphous surface layer, thereby leading to a significant lattice mismatch to the regrown GaN layer. Although the direct growth of GaN on the Si-implanted area did not take place during the initial stage of the regrowth process, resulting in uneven surface morphology, the surface will eventually become smooth due to the lateral overgrowth evolving to the whole Si-implanted area.
Fig. 2. (a) Typical top-view SEM images taken from the u-GaN layers grown on the selective-area Si-implanted templates after a growth duration of 5 min. (b) The schematic evolution process of regrown GaN islands nucleating at Si-implanted GaN template. (c) Typical cross-section SEM image taken from the samples grown on the Si-implanted GaN templates with Aimp = 73%.
The evolution process of regrown GaN on the selective-area Si-implanted GaN templates is schematically displayed in Figs. 1(a) and 2(b). Figure 2(c) shows typical cross-section SEM images taken from the samples grown on the Si-implanted GaN templates with Aimp = 73% after a growth duration of 10 min. In addition to the c-plane, there are two oblique planes with respect to the c-plane direction observed from the regrown GaN mesa (i.e. stripes). The angle between the oblique planes and the c-plane was ∼62°, indicating that the oblique facets were $\{{1\bar{1}01} \}$ planes [14–16]. The bottom width of the trapezoid mesa is ∼3.2 µm, which is larger than the window width defined by two adjacent Si-implanted stripes (i.e. 3 µm), as schematically shown in Fig. 2(b). This indicates that the lateral growth over the Si-implanted area occurred even at the initial stage of the regrowth process. In principle, the selective-area Si-implanted GaN templates used as regrowth substrates did not have a substantial mask layer on the surface of the templates. Therefore, the lateral growth over the Si-implanted area would occur at the very early stage during the regrowth process. For a conventional ELOG process, a dielectric thin film, such as SiO2 or SiNx, patterned on the GaN layer serves as a mask layer. Therefore, the lateral growth over the masked area would only occur after the vertical growth thickness larger than the thickness of the mask layer [14–16]. In contrast, as shown in Fig. 2(c), the height of the trapezoid mesa is ∼890 nm, corresponding to a growth rate of 5.34 µm/h. The high growth rate of GaN stripes at the implantation-free area was attributed to the loading effect compared with the baseline growth rate [13,17]. To clarify whether the growth of GaN occurred in the Si-implanted area or not, TEM was performed on the samples, as shown in Fig. 2(a) with Aimp = 13%. The TEM image shows a trench with symmetrically oblique edges, as shown in Fig. 3(a). The TEM images also display that the TDs extended from the GaN/sapphire interface to the regrown GaN region, but they terminated at the surface of the trench area, which is the Si-implanted area. In addition, GaN did not directly grow in the Si-implanted area. However, the lateral overgrowth extending from the implantation-free area to the Si-implanted area was observed, as shown in Fig. 3(b), and the TDs were almost absent in the regrown region above the Si-implanted area. Therefore, the selective-area regrowth of GaN could be well defined by the Si-implanted patterns (i.e. stripes). This preliminary result indicates that the selective-area Si-implanted area on the GaN templates plays a similar role to those of patterned SiO2 or SiNx layers on GaN templates for the ELOG process. Figure 4(a) displays the typical cross-sectional images of TEM taken from GaN epitaxial layers grown on GaN templates with selectively Si-implanted stripes along with the $\left\langle {11\bar{2}0\,} \right\rangle \,$. The GaN layers used for the TEM analyses were grown on the selective-area Si-implanted templates with Aimp = 64% after a growth duration of 80 min. In the window region, i.e., implantation-free region, the TDs extend from the GaN/sapphire interface to the top surface of the regrown GaN. However, the TDs are almost absent in the overgrown layer, which is above the Si-implanted area. More microstructural analyses by TEM of the overgrown regions, which used the Si-implanted GaN templates with different Aimp values, indicated that the overgrown GaN layers contained dislocations far less than the implantation-free regions. Consequently, the overall TD density observed from the GaN epitaxial layers grown on the Si-implanted GaN templates was significantly lower than those of GaN epitaxial layers grown on the implantation-free GaN templates. For the conventional ELOG GaN films, surface pits are often observed right above the center of the mask regions (i.e. SiO2 stripes), which is due to the imperfect crystal growth occurring at the coalescence boundary of adjacent GaN mesa and at the window regions [17]. The crystal defects, such as TDs, extend from the GaN/sapphire interface along the growth direction (i.e. <0001 > direction) to the film surface, leaving dense pits on the surface. In addition, the coalescence of adjacent GaN mesa may leave air gaps at the coalescence boundaries and in the regrown GaN layer. Figure 4(b) shows the surface morphologies by atomic force microscopy (AFM) taken from the GaN epitaxial layers grown on Si-implanted GaN templates with different Aimp. All the samples used for the AFM analyses have with a thickness of 4.3 µm. The GaN epitaxial layers were chemically treated using the mixed etchant (H2SO3:H3PO4 = 3:2) before the AFM measurements. The surface pits, which are attributed to the termination of TDs at the surface, were clearly observed from AFM images, with the pit density correlated to the Aimp. The typical pit density obtained from the GaN epitaxial layers grown on the implantation-free templates (i.e. Aimp = 0) was ∼1×108/cm2. However, the pit density slightly deceased to be ∼5.6, 5.0 and 4.0 ×107/cm2 corresponding to the Si-implanted templates with Aimp of 55%, 64% and 73%, respectively. This result was similar to those of the GaN epitaxial layer grown on SiO2-patterned GaN templates [18].
Fig. 3. (a) TEM image taken from the samples around the Si-implanted area. (b) Enlarged TEM images taken from the edge of the trapezoid GaN mesa.
Fig. 4. (a) Typical cross-sectional images of TEM taken from GaN epitaxial layers grown on Si-implanted GaN templates (Aimp = 64%). The inset shows the enlarged image taken from the edge of Si-implanted stripe. (b) Typical AFM images taken from the GaN epitaxial layers grown on Si-implanted GaN templates with different Aimp.
Considering the analyses of TEM and AFM, one would like to suggest that the selective-area Si-implantation on the GaN templates can lead to the epitaxial laterally overgrowth when the GaN epitaxial layers were regrown on the Si-implanted templates, and the material quality could be improved compared with the ones grown on implantation-free templates. Figures 5(a), 5(b), and 5(c) show the top-view images by SEM and cathodoluminescence (CL) images taken from the GaN layers grown on the selective-area Si-implanted templates after a growth duration of 80 min. All the wafers exhibited a smooth surface after the 80 min-long regrowth process. The morphology of the regrown GaN changed from the uneven surface with a series of GaN mesa, as shown in Fig. 2, to a continuous film with a smooth surface. This result indicates that lateral growth evolves continuously and thereby the adjacent GaN mesa coalesced laterally into a film. The corresponding CL images are displayed on the right-hand side of Figs. 5(a), 5(b), and 5(c). Figure 5(d) shows the typical CL spectra taken from the samples under an electron beam current of 20 µA and an acceleration voltage of 20 kV. The CL spectra show a single tense peak at 365 nm, which originates from the band-edge emission of GaN. The defect-related emission at ∼550 nm is very weak compared with the band-edge emission.
Fig. 5. Top-view SEM and CL images taken from the surface of GaN epitaxial layers grown on Si-implanted templates with (a) Aimp = 73%, (b) Aimp = 55% and (c) Aimp = 0. (d) Typical CL spectra taken from the u-GaN layers grown on the Si-implanted templates. These CL images and spectra were taken at 300K.
The monochromatic emissive CL images at the peak wavelength of 365 nm are individually presented in the right-hand side of Figs. 5(a), 5(b), and 5(c). As shown in Figs. 5(a) and 5(b), the CL images show the periodic bright and dark bands, which correspond to the Si-implanted area and the implantation-free area, respectively. In addition, the narrow dark line displayed in the center of each bright band was attributed to the coalescence boundary of the growth fronts of GaN layers over the Si-implanted area. The bright bands stand for better optical property due to the relative lower TD density. These results are consistent with the TEM images shown in Fig. 4. In other words, the dark band means the relative poor material quality in the window region compared with the area of the bright band (i.e. Si-implanted area). Therefore, one would like to suggest that the optical property of the GaN epitaxial layers grown on the Si-implanted GaN templates can be improved compared with the ones grown on implantation-free templates.
To prove whether the GaN LED structure grown on the Si-implanted u-GaN templates would lead to an improved performance or not, UV LEDs were grown on the Si-implanted GaN templates with Aimp=A/(A ± B) values of 36%(A = 4 µm, B = 7 µm) and 13%(A = 3 µm, B = 20 µm), and the LEDs were labeled as LED-A36 and LED-A13, respectively. The UV LEDs grown on implantation-free GaN templates were also prepared for comparison and labeled as LED-A0. In this study, the layer structure of the UV LEDs grown on the GaN templates consisted of a 4 µm-thick undoped GaN (u-GaN) layer grown at 1000 °C at a chamber pressure of 100 torrs. To confirm whether the surface of the u-GaN layer grown on the Si-implanted GaN templates had become smooth or not, an in-situ interferometer was used to monitor the change of reflectivity of the regrown u-GaN layer. In general, the surface morphology becomes smooth enough for subsequent layer growth after the regrown GaN with a thickness of 4 µm. Thereafter, a 2 µm-thick Si-doped n-GaN, layer(n ≈ 5×1018 cm−3), a 10-pair InGaN/AlGaN multiple quantum well (MQW) containing a 3.5 nm-thick In0.01Ga0.99N well, with each pair separated by a 10 nm-thick Al0.05Ga0.95N barrier, were sequentially grown on the 4 µm-thick u-GaN layer. Subsequently, 20 nm-thick Mg-doped p-Al0.2Ga0.8N (p ≈ 1×1017 cm−3) and 0.1 µm-thick Mg-doped p-GaN (p ≈ 5×1017 cm−3) layers were sequentially grown on the MQW structure to serve as the electron blocking and contact layers, respectively. Afterward, a heavily Si-doped In0.23Ga0.77N (n ≈ 1×1020 cm−3) top layer with a thickness of 3 nm was grown on the p-GaN contact layer, which forms an- In0.23Ga0.77N/p-GaN tunneling junction for achieving a low-resistivity ohmic contact [19]. After the epitaxial growth, an indium tin oxide (ITO) layer with a thickness of 150 nm was deposited onto the n-In0.23Ga0.77N top layer to serve as a transparent contact layer [4]. Next, a Cr/Au (50/200 nm) bilayer metal contact was deposited on the n-GaN underlying layer, which was exposed using Cl2-based plasma etching, and the ITO layer to form the n-type ohmic contacts (cathode electrodes) [20] and anode electrodes, respectively. The chip area was 310×310 µm2. Figure 6 shows the schematic structure of the UV LEDs characterized in this study.
Figure 7(a) shows the typical current-voltage (I-V) characteristics of UV LEDs grown on the Si-implanted GaN templates with different Aimp. With an injection current of 20 mA, the LEDs exhibited forward voltages (Vf) of 3.66, 3.70 and 3.72 V for LED-A36, LED-A13 and LED-A0, respectively, as shown in Fig. 7(a). The relative lower Vf observed from the LED-A36 could be attributed to the fact that the better material quality and hence the fewer compensation centers (e.g. nitrogen vacancy-related defects) in the LED-A36 would lead to higher hole concentration in the p-type layers. This contention could be indirectly evidenced by the characterization of the GaN layer grown on the Si-implanted GaN templates with different Aimp. The lower reverse currents obtained from the LEDs grown on the Si-implanted templates were consistent with the results of TEM and CL analyses. The reverse current of an LED is strongly dependent on the defect density in the epitaxial layers of the LED. For GaN/sapphire-based LEDs, the defect states mainly originate from dense TDs in the epitaxial layers grown on the sapphire substrate [21]. Although the inherent defects do not seem to significantly degrade the light output of InGaN-based blue and green LEDs, it has been evidenced that the light output power of UV LEDs decreases markedly with increasing structural defects (e.g. TDs) in the LEDs [7].
Fig. 7. (a) Typical forward and reverse current-voltage (I-V) characteristics of UV LEDs grown on the Si-implanted GaN templates with different Aimp. The inset shows the photograph taken from the LED-A0 with a driving current of 20 mA. (b) Typical light outputs measured from the LEDs with different driving currents. Typical EL spectra taken from the LEDs with a driving current of 20 mA are shown in the inset of Fig. 7(b).
Figure 7(b) shows the typical light output versus forward currents for the UV-LEDs grown on the Si-implanted GaN templates with different Aimp. This increased light output was mainly due to the increase in IQE. The increased IQE can be attributed to the fact that the underlying layers of LED-A36 should have lower TD density and hence fewer defect states, which act as nonradiative recombination centers, in the MQW. This contention could be indirectly evidenced by the reverse I-V characteristics, as shown in Fig. 7(a), and is consistent with results reported by Mukai et al., [7]. The inset of Fig. 7(b) shows the typical electroluminescence (EL) spectra taken from the LEDs with a driving current of 20 mA. The EL spectra exhibited an intense peak at 365 nm, which is attributed to the emission from the InGaN/AlGaN MQW in the LEDs. In addition, the EL spectra exhibited a weak yellow band peaked at ∼560 nm, which is due to defect- or impurity-assisted transitions. The yellow band has been previously attributed to a transition involving a complex consisting of a gallium vacancy and carbon on a nitrogen site [22–23].
4. Conclusion
In summary, we have demonstrated GaN-based epitaxial layers grown on GaN templates with selective-area Si implantation, exhibiting a growth mechanism like the conventional ELOG technique using dielectric films as the mask layer. The material quality of the regrown GaN-based layers was improved due to the suppression of up-award TDs, which originates from the sapphire/GaN interface, in the overgrown area. In addition, InGaN/AlGaN-based UV LEDs emitting a wavelength of ∼365 nm grown on the Si-implanted templates exhibited a relatively higher light output and lower reverse current compared with those of LEDs grown on the implantation-free GaN templates. This result was attributed to the fact that TD density in the GaN-based epitaxial layers grown on the Si-implanted templates was relatively lower than those of ones grown on implantation-free templates. In the light of the preliminary results, one would like to suggest that the GaN templates with selective-area Si implantation have a potential for replacing those of GaN templates with dielectric film as the mask layer for the ELOG growth.
Funding
Ministry of Science and Technology, Taiwan (MOST-106-2221-E-006-164, MOST-107-2112-M-006-023-MY3, MOST-107-2221-E-006-187-MY3, MOST-107-2221-E-218-012-MY3).
Acknowledgments
The authors would like to thank Ms. Hui–Jung Shih with the Instrument Center of National Cheng Kung University for supporting the use of high-resolution SEM (Hitachi SU8000).
Disclosures
The authors declare no conflicts of interest.
References
1. M. Kneissl, T. Kolbe, C. Chua, V. Kueller, N. Lobo, J. Stellmach, A. Knauer, H. Rodriguez, S. Einfeldt, Z. Yang, N. M. Johnson, and M. Weyers, “Advances in group III-nitride-based deep UV light-emitting diode technology,” Semicond. Sci. Technol. 26(1), 014036 (2011). [CrossRef]
2. S. Vilhunen, H. Särkkä, and M. Sillanpää, “Ultraviolet light-emitting diodes in water disinfection,” Environ. Sci. Pollut. Res. 16(4), 439–442 (2009). [CrossRef]
3. L. Fenton and H. Moseley, “UV emissions from low energy artificial light sources,” Photodermatol., Photoimmunol. Photomed. 30(2-3), 153–159 (2014). [CrossRef]
4. J. K. Sheu, C. M. Tsai, M. L. Lee, S. C. Shei, and W. C. Lai, “InGaN light-emitting diodes with naturally formed truncated micropyramids on top surface,” Appl. Phys. Lett. 88(11), 113505 (2006). [CrossRef]
5. H. Jeong, R. Salas-Montiel, G. Lerondel, and M. S. Jeong, “Indium gallium nitride-based ultraviolet, blue, and green light-emitting diodes functionalized with shallow periodic hole patterns,” Sci. Rep. 7(1), 45726 (2017). [CrossRef]
6. H. Jeong, H. J. Jeong, H. M. Oh, C.-H. Hong, E.-K. Suh, G. Lerondel, and M. S. Jeong, “Carrier localization in In-rich InGaN/GaN multiple quantum wells for green light-emitting diodes,” Sci. Rep. 5(1), 9373 (2015). [CrossRef]
7. T. Mukai and S. Nakamura, “Ultraviolet InGaN and GaN Single-Quantum-Well-Structure Light-Emitting Diodes Grown on Epitaxially Laterally Overgrown GaN Substrates,” Jpn. J. Appl. Phys. 38(Part 1, No. 10), 5735–5739 (1999). [CrossRef]
8. N. Gmeinwieser, K. Engl, P. Gottfriedsen, U. T. Schwarz, J. Zweck, W. Wegscheider, S. Miller, H.-J. Lugauer, A. Leber, A. Weimar, A. Lell, and V. Härle, “Correlation of strain, wing tilt, dislocation density, and photoluminescence in epitaxial lateral overgrown GaNGaN on SiCSiC substrates,” J. Appl. Phys. 96(7), 3666–3672 (2004). [CrossRef]
9. J.-K. Sheu, P.-C. Chen, Y.-H. Yeh, S.-H. Kuo, M.-L. Lee, P.-H. Liao, and W.-C. Lai, “Mask-free regrowth of GaN p-i-n structure on selective-area Si-implanted n-GaN template layer,” Acta Mater. 108, 17–25 (2016). [CrossRef]
10. B. Kim, K. Lee, S. Jang, J. Jhin, S. Lee, J. Baek, Y. Yu, J. Lee, and D. Byun, “Epitaxial Lateral Overgrowth of GaN on Si (111) Substrates Using High-Dose, N+ Ion Implantation,” Chem. Vap. Deposition 16(1-3), 80–84 (2010). [CrossRef]
11. J. K. Sheu, M. L. Lee, C. J. Tun, C. J. Kao, L. S. Yeh, S. J. Chang, and G. C. Chi, “Characterization of Si implants in p-type GaN,” IEEE J. Sel. Top. Quantum Electron. 8(4), 767–772 (2002). [CrossRef]
12. J. C. Zolper, H. H. Tan, J. S. Williams, J. Zou, D. J. H. Cockayne, S. J. Pearton, M. Hagerott Crawford, and R. F. Karlicek, “Electrical and structural analysis of high-dose Si implantation in GaN,” Appl. Phys. Lett. 70(20), 2729–2731 (1997). [CrossRef]
13. K. Lorenz, E. Wendler, A. Redondo-Cubero, N. Catarino, M.-P. Chauvat, S. Schwaiger, F. Scholz, E. Alves, and P. Ruterana, “Implantation damage formation in a-, c- and m-plane GaN,” Acta Mater. 123, 177–187 (2017). [CrossRef]
14. Y. Kato, S. Kitamura, K. Hiramatsu, and N. Sawaki, “Selective growth of wurtzite GaN and AlxGa1−xN on GaN/sapphire substrates by metalorganic vapor phase epitaxy,” J. Cryst. Growth 144(3-4), 133–140 (1994). [CrossRef]
15. K. Hiramatsu, K. Nishiyama, M. Onishi, H. Mizutani, M. Narukawa, A. Motogaito, H. Miyake, Y. Iyechika, and T. Maeda, “Fabrication and characterization of low defect density GaN using facet-controlled epitaxial lateral overgrowth (FACELO),” J. Cryst. Growth 221(1-4), 316–326 (2000). [CrossRef]
16. S. Kitamura, K. Hiramatsu, and N. Sawaki, “Fabrication of GaN hexagonal pyramids on dot-patterned GaN/sapphire substrates via selective metalorganic vapor phase epitaxy,” Jpn. J. Appl. Phys. 34(Part 2, No. 9B), L1184–L1186 (1995). [CrossRef]
17. W. V. Lundin, A. F. Tsatsulnikov, S. N. Rodin, A. V. Sakharov, S. O. Usov, M. I. Mitrofanov, I. V. Levitskii, and V. P. Evtikhiev, “Selective epitaxial growth of III–N structures using ion-beam nanolithography,” Semiconductors 52(10), 1357–1362 (2018). [CrossRef]
18. T. S. Zheleva, O.-H. Nam, M. D. Bremser, and R. F. Davis, “Dislocation density reduction via lateral epitaxy in selectively grown GaN structures,” Appl. Phys. Lett. 71(17), 2472–2474 (1997). [CrossRef]
19. J. K. Sheu, J. M. Tsai, S. C. Shei, W. C. Lai, T. C. Wen, C. H. Kou, Y. K. Su, S. J. Chang, and G. C. Chi, “Low-operation voltage of InGaN-GaN light-emitting diodes with Si-doped In0.3Ga0.7N/GaN short-period superlattice tunneling contact layer,” IEEE Electron Device Lett. 22(10), 460–462 (2001). [CrossRef]
20. M. L. Lee, J. K. Sheu, and C. C. Hu, “Nonalloyed Cr∕AuCr∕Au-based Ohmic contacts to n-GaN,” Appl. Phys. Lett. 91(18), 182106 (2007). [CrossRef]
21. E. F. Schubert,“Light-Emitting Diode,” Cambridge University Press, New York, USA, 2006, Ch.9
22. J. K. Sheu and G. C. Chi, “The doping process and dopant characteristics of GaN,” J. Phys.: Condens. Matter 14(22), R657–R702 (2002). [CrossRef]
23. P. Kamyczek, E. Placzek-Popko, V. Kolkovsky, S. Grzanka, and R. Czernecki, “A deep acceptor defect responsible for the yellow luminescence in GaN and AlGaN,” J. Appl. Phys. (Melville, NY, U. S.) 111(11), 113105 (2012). [CrossRef]
