Journal of Crystal Growth 395 (2014) 9–13 Contents lists available at ScienceDirect Journal of Crystal Growth journal homepage: www.elsevier.com/locate/jcrysgro AlGaN-based deep ultraviolet light-emitting diodes grown on nano-patterned sapphire substrates with significant improvement in internal quantum efficiency Peng Dong a,n, Jianchang Yan a, Yun Zhang a, Junxi Wang a, Jianping Zeng a, Chong Geng b, Peipei Cong a, Lili Sun a, Tongbo Wei a, Lixia Zhao a, Qingfeng Yan b, Chenguang He c, Zhixin Qin c, Jinmin Li a a b c Research and Development Center for Semiconductor Lighting, Institute of Semiconductors, Chinese Academy of Sciences, Beijing 100083, China Department of Chemistry, Tsinghua University, Beijing 100084, China State Key Laboratory for Artificial Microstructure and Mesoscopic Physics, School of Physics, Peking University, Beijing 100871, China art ic l e i nf o a b s t r a c t Article history: Received 20 October 2013 Received in revised form 17 January 2014 Accepted 25 February 2014 Communicated by R.M. Biefeld Available online 3 March 2014 We report high-performance AlGaN-based deep ultraviolet light-emitting diodes grown on nanopatterned sapphire substrates (NPSS) using metal organic chemical vapor deposition. By nanoscale epitaxial lateral overgrowth on NPSS, 4-μm AlN buffer layer has shown strain relaxation and a coalescence thickness of only 2.5 μm. The full widths at half-maximum of X-ray diffraction (002) and (102) ω-scan rocking curves of AlN on NPSS are only 69.4 and 319.1 arcsec. The threading dislocation density in AlGaN-based multi-quantum wells, which are grown on this AlN/NPSS template with a lightemitting wavelength at 283 nm at room temperature, is reduced by 33% compared with that on flat sapphire substrate indicated by atomic force microscopy measurements, and the internal quantum efficiency increases from 30% to 43% revealed by temperature-dependent photoluminescent measurement. & 2014 Elsevier B.V. All rights reserved. Keywords: A1. Defects A3. Epitaxial lateral overgrowth A3. Metalorganic chemical vapor deposition B1. Nitrides B2. Semiconducting aluminum compounds 1. Introduction Because of their direct and wide band-gap, AlGaN alloys have attracted considerable attention in fabricating light-emitting diodes (LEDs) in deep ultraviolet (DUV) waveband [1], for disinfection, sensing, water/air purification, bio-medical and non-line-of-sight communication [2]. These applications require an emitting wavelength (λ) less than 300 nm, which means that the Al content is at least 40% in LED's quantum wells and even higher in other layers. However, the growth of high-aluminum (Al)-content AlGaN on sapphire substrates is a big challenge due to the large lattice mismatch and thermal expansion mismatch, as well as the low surface mobility of aluminum species. Although an AlN buffer layer has been widely adopted between flat sapphire substrates (FSS) and AlGaN epi-layers for better material quality, the typical threading dislocation density (TDD) in the AlGaN layers grown on the AlN/FSS template is still as high as 1010–1011 cm 2 [3]. Threading dislocations in multi-quantum wells (MQWs) act as non-radiative n Corresponding author. E-mail addresses: [email protected] (P. Dong), [email protected] (J. Yan), [email protected] (Y. Zhang). http://dx.doi.org/10.1016/j.jcrysgro.2014.02.039 0022-0248 & 2014 Elsevier B.V. All rights reserved. recombination centers, thereby resulting in low internal quantum efficiency (IQE) [4,5]. Various methods have been reported to suppress the TDD in AlN and the upper epi-layers grown on sapphire substrates, including migration-enhanced metal–organic chemical vapor deposition (MEMOCVD) [6,7] and pulsed-flow multilayer AlN buffers growth technique [8]. Particularly, epitaxial lateral overgrowth (ELO) techniques on micro-stripe patterned sapphire or AlN/FSS template have significantly enhanced light-output power (LOP) and reliability of DUV LEDs by reducing the TDD [9–12]. However, the major issue of the AlN ELO on micro-stripe patterned sapphire is the large space between micro-patterns that needs a coalescence thickness almost 10 μm for AlN and greatly increases epitaxy time and cost. 2. Experimental In this study, we fabricated nano-patterned sapphire substrates (NPSS) by an optimized nanosphere lithography (NSL) process for nanoscale ELO of AlN. Furthermore, we investigated the material quality, stress state and TDD decrease of nanoscale ELO–AlN on 10 P. Dong et al. / Journal of Crystal Growth 395 (2014) 9–13 NPSS as well as its effect on the quality and IQE of AlGaN based MQWs. Fig. 1(a) shows the schematic diagram of NSL for fabricating NPSS. First, positive photo resist (PR) was spin-coated on a 2-in. (001) sapphire substrate with a 200-nm-thick SiO2 film predeposited by plasma-enhanced chemical vapor deposition. The wafer was then dip-coated with a highly ordered self-assembled monolayer of polystyrene (PS) nanospheres with uniform diameters of 600 nm [13] and followed by flood UV-exposure. After PS nanospheres were removed by DI water, the PR was developed to form nano-holes. The pattern was transferred to the SiO2 film by inductively coupled plasma (ICP) etching. Finally, the sapphire substrate was etched for 10 min in a mixture of H2SO4 and H3PO4 solution (H2SO4:H3PO4 ¼3:1) at 280 1C, and the SiO2 mask was removed by HF. Top-view SEM images of the developed PR and the fabricated NPSS are shown in Fig. 1(b) and (c), respectively. The patterns on the NPSS surface are concave triangle cones caused by the anisotropic etching of the sapphire crystal. The period of patterns is 900 nm that is determined by the PS diameter. The depth of cones and the width of the unetched regions are approximately 250 nm and 400 nm, respectively. A home-made low-pressure metal–organic chemical vapor deposition (LP–MOCVD) system with a vertical shower-head reactor was used to process epitaxial growth. Trimethylaluminum (TMAl), trimethylgallium (TMGa) and ammonia (NH3) were aluminum, gallium and nitrogen sources, respectively. The AlN template growth on the NPSS started with a 25-nm AlN buffer layer that was grown at 550 1C, with a V/III ratio of 3000, a TMAl flux of 7.5 μmol/min and a growth rate of 12 nm/min. Then the growth temperature rose to 1200 1C to finish the whole 4-μm AlN growth using a V/III ratio of 1000, a TMAl flux of 40 μmol/min and a growth rate of 1 μm/h. The reactor pressure during AlN growth was kept at 50 Torr. 3. Results and discussion A cross-sectional SEM image of 4-μm AlN on NPSS is shown in Fig. 2(a). Thanks to the nano-scale substrate patterns and the AlN lateral growth, AlN completely coalesces after 2.5-μm growth, which is much shorter than the reported coalescence thickness of nearly 10 μm [9–12,14,15]. Fig. 2(b) presents a 5 5 μm2 atomic force microscopy (AFM) image of the surface morphology of the AlN on NPSS, demonstrating an atomically flat surface with a stepflow growth mode and a root-mean-square (RMS) roughness of 0.19 nm. Fig. 3(a) shows the typical X-ray rocking curves (XRCs) of 4-μm AlN film on NPSS. The full width at half-maximum (FWHM) values of (002) and (102) reflections are 69.4 and 319.1 arcsec, respectively. The corresponding screw and edge dislocation densities are calculated to be 1.0 107 cm 2 and 1.2 109 cm 2 by the method reported in Ref. [16]. The AlN material quality is much better than a 1-μm AlN grown on FSS, which has (002) and (102) XRCs of 126 and 573 arcsec FWHM (data not shown here). Fig. 3(b) shows the Raman spectrum of the E2(high) phonon mode for AlN template layer grown on NPSS and FSS. The Raman shift peaks of E2(high) phonon mode for AlN template layer grown on NPSS and FSS are located at 658.7 and 660.2 cm 1, respectively. The vertical arrow indicates the stress-free frequency of 657.4 cm 1 [17]. The Raman shift peaks both appear on the higher frequency side, showing residual compressive stresses, while the peak of AlN on NPSS is more close to stress-free frequency, indicating strain relaxation. Fig. 1. (a) Schematic diagram of the nanosphere lithography (NSL) for fabricating nano-patterned sapphire substrate (NPSS). SEM images of the patterned photoresist (PR) (b) and wet-etched NPSS (c). Fig. 2. (a) A cross-sectional SEM image of AlN grown on NPSS. (b) An AFM image of the AlN grown on NPSS (5 5 μm2). P. Dong et al. / Journal of Crystal Growth 395 (2014) 9–13 The corresponding in-plane compressive stress (s) for AlN template layer on NPSS is therefore decreased according to the following expression [18]: ωE2 ðhighÞ ω0 ¼ C s where C is the biaxial strain coefficient, ωE2 ðhighÞ and ω0 are the Raman shift peaks for E2(high) mode of the AlN grown on NPSS and the stress-free AlN, respectively. The reduced stress has prevented the epilayers to suffer from cracking, which has been 11 indicated by our cross-sectional transmission electron microscope (TEM) measurements (not shown here), and can also improve the stress state of its upper layers. An AlGaN-based MQWs structure was regrown on the AlN/NPSS template (denoted as Structure A), consisting of 20 pairs of AlN/ AlGaN superlattices (SLs), a 3.5-μm-thick Si-doped Al0.55Ga0.45N layer, five 3-nm-thick un-doped Al0.4Ga0.6N quantum wells sandwiched by 12-nm-thick Si-doped Al0.5Ga0.5N barriers and a Mg-doped Al0.65Ga0.35N electron blocking layer. For comparison, the Fig. 3. (a) X-ray rocking curves (XRCs) of (002) and (102) diffractions for the AlN films grown on the NPSS and (b) Raman spectrum for the AlN films grown on the NPSS and flat sapphire substrate (FSS). Fig. 4. AFM images of the AlGaN based MQWs of Structure A (a) , Structure B (b). (c) Cross-sectional high-resolution TEM image of the AlGaN based MQWs of Structure A. 12 P. Dong et al. / Journal of Crystal Growth 395 (2014) 9–13 Fig. 5. Temperature-dependent PL spectra for Structure A (a), Structure B (b). (c) Arrhenius plot of the integrated PL spectra for Structure A and Structure B. same structure was also grown on AlN/FSS template with 1-μmthick AlN (denoted as Structure B). We also prepared samples with AlN/AlGaN SLs and 3.5-μm-thick Al0.55Ga0.45N layer grown on NPSS and FSS without MQWs for material quality comparison in XRD, the FWHM values of (002) and (102) reflections are 190 and 513 arcsec for AlGaN on NPSS, and those values for AlGaN on FSS are 228 and 631 arcsec. The surface morphology of Structures A and B was characterized by AFM measurements over an area of 1 1 μm2. The results are shown in Fig. 4(a) and (b), both demonstrating a step-flow growth mode, but structure A has a more atomically flat surface than structure B. The RMS roughness of the two samples is found to be 0.14 nm and 0.23 nm, respectively. By counting pits in the AFM figures [19], the TDDs are estimated to be about 2.0 109 cm 2 and 3.0 109 cm 2, respectively, for structures A and B. These results qualitatively demonstrate improvement in material quality for AlGaN-based MQWs on NPSS compared to that on FSS. Additionally, that is very close to the TD densities estimated from cross-sectional TEM measurement reported previously [20]. Fig. 5(c) shows a cross-sectional high-resolution TEM image of Al0.5Ga0.5N/ Al0.4Ga0.6N MQWs of structure A. The image was taken with diffraction vector g¼[0002] near the [112̄ 0] zone. The thicknesses of the quantum wells and the quantum barriers are indicated to be 3 nm and 12 nm, respectively. Moreover, MQWs in the image display decent periodicity and uniformity, as well as atomically abrupt and smooth interfaces between barriers and wells. Temperature-dependent photoluminescence (PL) measurements were performed on structures A and B to compare their IQE. The samples were placed inside a closed-cycle refrigerator, and the temperature ranged from 10 K to 300 K. A 4th harmonic of Q-switched YAG:Nd laser (λ¼ 266 nm, pulse width ¼ 7 ns) was used for excitation, and an Ocean Optics USB2000þ VIS-NIR fiber optic spectrometer recorded the PL spectra. Fig. 5(a) and (b) shows the normalized temperature-dependent PL spectra of structures A and B under identical excitation condition, respectively, both demonstrating a single peak around 283 nm at 300 K. The peak red-shift is low for structure A than that of structure B with temperature increasing from 10 K to 300 K. Fig. 5(c) shows the normalized integrated PL intensities as a function of inverse temperature in an Arrhenius plot. Assuming that IQE equals to 100% for both AlGaN based MQWs at 10 K, their IQE at 300 K (¼ PL intensity at 300 K/PL intensity at 10 K) are 43% and 30%, respectively, for structures A and B, demonstrating about 43% enhancement for structure A. We believe that the significant improvement in IQE can be partly attributed to the decrease in TDs, which act as non-radiative centers. Nanoscale ELO on NPSS decreases TD densities in AlN template and upper epilayers, such as n-AlGaN and MQWs. In addition, the less in-plane compressive stress defined by Raman analysis will decrease the strain-induced piezoelectric polarization in AlGaN-based MQWs, which can be indicated by the decreased red-shift. Thus, the recombination efficiency will also be increased. 4. Conclusions We have demonstrated AlGaN-based DUV LEDs fabricated on an AlN/NPSS template with significant IQE improvement. The NPSS is prepared by NSL and wet etching. The AlN layer has shown strain relaxation and a coalescence thickness of only 2.5 μm by nanoscale ELO on NPSS. Narrow XRC FWHMs and smooth surface both demonstrated high material quality of AlN on NPSS. When grown on the AlN/NPSS template, the Al0.5Ga0.5N/Al0.4Ga0.6N MQWs showed better surface morphology, decreased straininduced piezoelectric polarization and decreased TD density, contributing to the significant improvement in IQE from 30% to 43% observed in temperature-dependent PL measurement. Acknowledgments This work was supported by the National Natural Sciences Foundation of China under Grant nos. 61376090, 61376047, 61006038, 61204053 and 51102226, by the National High Technology Program of China under Grant nos. 2014AA032608 and 2011AA03A111, by the Program of Science and Technology of Beijing under Grant no. D12110300140000 and the National 1000 Young Talents Program. References [1] Asif Khan, Krishnan Balakrishnan, Tom Katona, Nat. Photonics 2 (2008) 77. [2] M.S. Shur, R. Gaska, IEEE Trans. Electron Devices 57 (2010) 12. [3] M. Imura, K. Nakano, N. Fujimoto, N. Okada, K. Balakrishnan, M. Iwaya, S. Kamiyama, H. Amano, I. Akasaki, T. Noro, T. Takagi, A. Bandoh, Jpn. J. Appl. Phys. 46 (2007) 1458. [4] M. Kneissl, T. Kolbe, C. Chua, V. Kueller, N. Lobo, J. Stellmach, A. Knauer, H. Rodriguez, S. Einfeldt, Z. Yang, N.M. Johnson, M. Weyers, Semicond. Sci. Technol. 26 (2011) 014036. [5] K. Ban, J. Yamamoto, K. Takeda, K. Ide, M. Iwaya, T. Takeuchi, S. Kamiyama, I. Akasaki, H. Amano, Appl. Phys. Express 4 (2011) 052101. [6] R. Jain, W. Sun, J. Yang, M. Shatalov, X. Hu, A. Sattu, A. Lunev, J. Deng, I. Shturm, Y. Bilenko, R. Gaska, M.S. Shur, Appl. Phys. Lett. 93 (2008) 051113. [7] J.P. Zhang, H.M. Wang, M.E. Gaevski, C.Q. Chen, Q. Fareed, J.W. Yang, G. Simin, M.A. Khan, Appl. Phys. Lett. 80 (2002) 3542. [8] H. Hirayama, T. Yatabe, N. Noguchi, T. Ohashi, N. Kamata, Appl. Phys. Lett. 91 (2007) 071901. [9] V. Adivarahan, Q. Fareed, M. Islam, T. Katona, B. Krishnan, A. Khan, Jpn. J. Appl. Phys. 46 (2007) L877. P. Dong et al. / Journal of Crystal Growth 395 (2014) 9–13 [10] H. Hirayama, J. Norimatsu, N. Noguchi, S. Fujikawa, T. Takano, K. Tsubaki, N. Kamata, Phys. Status Solidi C 6 (2009) S474. [11] M. Kim, T. Fujita, S. Fukahori, T. Inazu, C. Pernot, Y. Nagasawa, A. Hirano, M. Ippommatsu, M. Iwaya, T. Takeuchi, S. Kamiyama, M. Yamaguchi, Y. Honda, H. Amano, I. Akasaki, Appl. Phys. Express 4 (2011) 092102. [12] M. Shatalov, W. Sun, A. Lunev, X. Hu, A. Dobrinsky, Y. Bilenko, J. Yang, M. Shur, R. Gaska, C. Moe, G. Garrett, M. Wraback, Appl. Phys. Express 5 (2012) 082101. [13] C. Li, G. Hong, P. Wang, D. Yu, L. Qi, Chem. Mater. 21 (2009) 891. [14] S. Hwang, D. Morgan, A. Kesler, M. Lachab, B. Zhang, A. Heidari, H. Nazir, I. Ahmad, J. Dion, Q. Fareed, V. Adivarahan, M. Islam, A. Khan, Appl. Phys. Express 4 (2011) 032102. [15] M. Imura, K. Nakano, G. Narita, N. Fujimoto, N. Okada, K. Balakrishnan, M. Iwaya, S. Kamiyama, H. Amano, I. Akasaki, T. Noro, T. Takagi, A. Bandoh, J. Cryst. Growth 298 (2007) 257–260. 13 [16] B.N. Pantha, R. Dahal, M.L. Nakarmi, N. Nepal, J. Li, J.Y. Lin, H.X. Jiang, Q.S. Paduano, D. Weyburne, Appl. Phys. Lett. 90 (2007) 241101. [17] T. Prokofyeva, M. Seon, J. Vanbuskirk, M. Holtz, S. Nikishin, N. Faleev, H. Temkin, S. Zollner, Phys. Rev. B 63 (2001) 125313. [18] P. Puech, F. Demangeot, J. Frandon, C. Pinquier, M. Kuball, V. Domnich, Y. Gogotsi, J. Appl. Phys. 96 (2004) 2853. [19] P. Hansen, Y. Strausser, A. Erickson, E. Tarsa, P. Kozodoy, E. Brazel, J. Ibbetson, U. Mishra, V. Narayanamurti, S. DenBaars, Appl. Phys. Lett. 72 (1998) 2247. [20] P. Dong, J. Yan, J. Wang, Y. Zhang, C. Geng, T. Wei, P. Cong, Y. Zhang, J. Zeng, Y. Tian, L. Sun, Q. Yan, J. Li, S. Fan, Z. Qin, Appl. Phys. Lett. 102 (2013) 241113.
© Copyright 2024