Journal Search Engine
Search Advanced Search Adode Reader(link)
Download PDF Export Citaion korean bibliography PMC previewer
ISSN : 1225-0562(Print)
ISSN : 2287-7258(Online)
Korean Journal of Materials Research Vol.27 No.12 pp.699-704

Surface Analysis of Plasma Pretreated Sapphire Substrate for Aluminum Nitride Buffer Layer

Woo Seop Jeong1, Dae-Sik Kim1, Seung Hee Cho1, Chul Kim1, Junggeun Jhin2, Dongjin Byun1
1Department of Materials Science & Engineering, Korea University, 145 Anam-ro, Seongbuk-gu, Seoul 02841, Republic of Korea
2LED Procurement Team, LG Innotek, 570 Hyuam-ro, Munsan-eup, Paju-si, Gyeonggi-do 10842, Republic of Korea
Corresponding author : (D. Byun, Korea Univ.)
20171102 20171102 20171128


Recently, the use of an aluminum nitride(AlN) buffer layer has been actively studied for fabricating a high quality gallium nitride(GaN) template for high efficiency Light Emitting Diode(LED) production. We confirmed that AlN deposition after N2 plasma treatment of the substrate has a positive influence on GaN epitaxial growth. In this study, N2 plasma treatment was performed on a commercial patterned sapphire substrate by RF magnetron sputtering equipment. GaN was grown by metal organic chemical vapor deposition(MOCVD). The surface treated with N2 plasma was analyzed by x-ray photoelectron spectroscopy(XPS) to determine the binding energy. The XPS results indicated the surface was changed from Al2O3 to AlN and AlON, and we confirmed that the thickness of the pretreated layer was about 1 nm using high resolution transmission electron microscopy(HR-TEM). The AlN buffer layer deposited on the grown pretreated layer had lower crystallinity than the as-treated PSS. Therefore, the surface N2 plasma treatment on PSS resulted in a reduction in the crystallinity of the AlN buffer layer, which can improve the epitaxial growth quality of the GaN template.


    Ministry of Trade, Industry and Energy

    © Materials Research Society of Korea. All rights reserved.

    This is an Open-Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License ( which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.


    GaN-based light emitting diodes have been studied to improve light efficiency. Methods for improving light efficiency include patterning the substrate, forming roughness on the chip surface, and improving the quality of the epitaxial layer.1-3) In the methods, quality improvement of epitaxial layer is an important factor in chip manufacturing. 4,5) Substrates are important for high-quality GaN epitaxial layer growth. Generally, the substrate for GaN growth uses sapphire or Si in consideration of cost. The substrates are difficult to grow a high-quality GaN epi layer due to a large difference in lattice parameter. Thus, a buffer layer is used to reduce lattice mis-match and helps to grow high-quality GaN epitaxial layer.6) Particularly, this buffer layer has a great importance in the use of the patterned substrate to improve the light efficiency. 2,7,8) In order to obtain high-quality GaN on the patterned substrate, it is grown by the epitaxial lateral over-growth(ELO) method, which requires selective growth on the substrate.9,10) However, the growth behavior of GaN is changed on the patterned substrate depending on the state of the buffer.11) The reason is that the patterned substrate has a surface exposed various crystal planes. By this surface, GaN has various growth behaviors. Therefore, the buffer layer needs optimization and characterization studies in order to obtain a high-quality GaN epitaxial layer on a patterned substrate. We have studied the improvement of GaN characteristics by using AlN buffer on the patterned sapphire substrate(PSS) and also we have studied the pretreatment of the substrate.

    In this study, we confirmed that different growth behaviors were observed when GaN was grown using AlN buffer layer by N2 plasma pretreatment of PSS. We in- vestigated the surface properties of plasma pretreated substrates. We demonstrate how the N2 plasma pretreatment process changes properties of the AlN buffer layer and how these changes affect the growth behavior of GaN.


    The samples were prepared for c-plane commercial PSS and c-plane planar sapphire substrate for analysis. The pattern shape of the PSS is cone-shape type and height and diameter of the pattern is 1.65 μm and 2.45 μm, respectively. The size of the sample was prepared 1 × 1 cm2 by cutting a 2-inch wafer. N2 plasma treatment and the AlN buffer layer was deposited using RF sputter equipment in the prepared sample. The process of N2 plasma treatment on sapphire substrate surface is the same AlN sputtering process except for the deposition and process time. N2 plasma treatment was carried out for 10 minutes. The deposition conditions of AlN buffer layer are shown in Table 1. GaN was grown by metal organic chemical vapor deposition(MOCVD) on a sample deposited an AlN buffer layer. The growth conditions of GaN are shown in Table 1. Finally, the sample is prepared in two types of including N2 plasma treatment processing and not including.

    Field emission scanning electron microscope(Hitach, S4800) was carried out to confirm cross-section structure of GaN template with the samples. And XPS measured to obtain the binding energy data of pre-treatment PSS. We confirmed the atomic arrangement and selected area diffraction(SEAD) pattern through HR-TEM(FEI, TitanTM 80-300). X-ray diffraction was measured the crystal properties of AlN buffer layers. Atomic force microscopy (Park Scientific Instruments, Autoprobe CP) examination was performed to confirm surface of AlN. The extracted image was calculated the roughness of AlN surface by Nanoscope program(version 5.31).

    3.Result and Discussion

    In general, a buffer layer is used to improve the characteristics of GaN grown on a sapphire substrate. Companies and laboratories are using buffer layers of various materials, but we used the AlN buffer. Our research team found that deposited AlN buffer layer on special pretreated PSS surface affects growth behavior of GaN. Fig. 1(a) and (b) are SEM images of AlN buffer layer grown on PSS. In Fig. 1(a), the height and diameter of the pattern are known and are 1.65 μm and 2.45 μm, respectively. The thickness of the deposited AlN buffer layer is about 30 nm. In Fig. 2(b), the period and distance between the patterns were 3 μm and 0.55 μm, respectively. And the shape of the pattern is con-shape.

    Fig. 2(a) is a cross-sectional SEM image of a GaN template grown on AlN buffer layer using bare PSS. In the AlN buffer layer grown in bare PSS, many hillocks were found in the GaN template and many voids were observed at the top of the pattern. The reason for this phenomenon is that GaN growth occurs at the upper part of the pattern and selectively grown GaN at the bottom part acts as an obstacle in the ELOG process, resulting in voids and hillocks.12) Fig. 2(b) is a cross-sectional SEM image of the GaN template grown using the AlN buffer layer on N2 plasma pretreated PSS. Hillocks or voids were not found in the GaN template of this sample.Fig. 3

    XPS analysis was performed to confirm that the surface condition was changed by N2 plasma pretreatment. Depth profiling was performed to determine the depth of change from the surface. The gas used for sputtering was argon and the etching energy was 1 kV. The etching area is 1.5 * 1.5 mm and the etching rate is 0.13 nm / sec based on SiO2. The analyzing elements were N, O Al, C, and Ar. The difference between the two samples was found at N, O, and Al. Fig. 4 is a spectrums of N1s, O1s and Al2p showing a result obtained by measuring the depth profiling of a bare sapphire substrate. Bare sapphire was analyzed as a control group of N2 plasma pretreated sapphire. Depth profiling was performed with the cumulative etching time from 0 to 200 seconds. But the spectrum results were all the same after 20 seconds of etching time. So only the results of 0 and 20 seconds are shown in Fig. 4. Mixed peaks are represented by circles, the fitted spectrums are represented by red lines, and the divided subpeaks are represented by blue line, cyan line, and majenta line, respectively. The center of each subpeak is represented by a dash line and the atomic bonds are labeled. The binding energy of N1s was not confirmed on the bare sapphire. The results of fitting each data in O1s were confirmed the difference between 0 and 20 seconds. Two subpeaks were confirmed at 0 seconds and 529.8 and 530.6 eV, respectively. The 529.8 eV subpeak appeared only on the surface and was not observed at 20 seconds. In Weifang Lu group, the binding energy of near 529.8 eV was indicative of the C-O bond appearing on the surface.13) The 530.6 eV subpeak is considered to be an O-Al bond of A2O3 because it is the same position of binding energy as the 20 second spectrum. Similar phenomena were observed in Al2p at 0s. In the fitted data, two subpeaks were confirmed as 73.6 eV and 74.2 eV, respectively. The 73.6 eV subpeak is a peak shift phenomenon caused by carbon contamination on the surface.14) It was confirmed that the 74.2 eV subpeak occurred equally in 20 seconds due to the Al-O bond.

    Fig. 5 shows the XPS analysis of the N2 plasma pretreated sapphire substrate and fitting of each data. The spectrum results were all the same after 40 seconds of etching time. So only the results of 0, 20 and 40 seconds are shown in Fig. 5 Unlike the bare sapphire sample in N1s, the peak was observed at 0 and 20 seconds. There are three subpeaks at 0 s and divided into 394.9 eV, 396.0 eV, and 401.6 eV, respectively. The subspeak of 394.9 eV appeared only at 0 seconds with N-N bonds and could not be confirmed at other depths.15) The subpeak of 396.0 eV showed a relatively high peak intensity at 0 seconds with N-Al bond and decreased at 20 seconds. N-Al bond was not observed in 40 seconds. The subpeak of 401.6 eV binding energy is the N-Al-O bond observed in AlOxN1-x.16-18) N-Al-O was confirmed only at 0 seconds. In O1s, there was no difference from the mixed peak position of bare sapphire at 0s but the intensity was relatively decreased. In Al2p, three subpeaks were confirmed and 72.5 eV, 73.2 eV and 74.1 eV, respectively. The subpeak of 72.5 eV binding energy with Al-N bond was confirmed at 0, 20 seconds but the subpeak was not confirmed at 40 seconds. Consequently, the surfaces of the N2 plasma pretreated sapphire substrate was changed to a material composed of AlN or AlON.

    Fig. 5 shows the HR-TEM image and the SEAD pattern of the N2 plasma pretreated sapphire substrate. The surface of the sapphire was deposited on a metal(Au) surface to enhance the contrast and the specimen was fabricated by focused ion beam(FIB). Fig. 5(a) is a crosssection image of N2 plasma pretreated sapphire surface. The upper dark region of the image is a filler with Au deposited, and the center bright region is the pretreated layer. The region where the atomic arrangement is clearly visible is Al2O3. The intensity data of the dash line shown in the image are plotted to identify the atomic arrangement of the surface layer. The raw data image contains intensity information in pixels. Information about atomic arrangement can be obtained by line profiling from images. Fig. 5(b) Peak represents a specific atom. The Al2O3 region has a constant peak interval(atomic arrangement) and intensity. However, in the pretreated layer, irregular intervals and intensity changes were observed. In Fig. 5(a), the surface elements are different because of the difference in surface contrast. Therefore, it can be seen that N atoms are incorporated when compared to XPS data. The SEAD pattern analysis was carried out to confirm the crystal structure, and selected position was indicated by a yellow box, where (c) and (d) are number 1 and 2 respectively. In Fig. 5(c), the diffraction pattern image shows some diffraction spots, but the central region is hard to distinguish the spots. In Fig. 5(d) Diffraction pattern image shows the diffraction pattern of Al2O3. The direction of the Al2O3 crystal was confirmed by the reciprocal lattice calculation of the diffraction pattern, and the direction vector was shown in Fig. 5(a).

    Fig. 6 shows the HR-TEM image and SEAD pattern of the AlN deposited sample on the plasma pretreated sapphire substrate. It was confirmed that the image consists of metal(filer), AlN buffer layer and Al2O3 from top to bottom, as shown in Fig. 6(a). The AlN buffer layer was confirmed to be about 30 nm thickness. It is difficult to identify the pretreated layer between Al2O3 and AlN buffer layer. The pretreated layer is composed of AlN or AlON, and it is presumed that the contrast difference with the AlN buffer layer is not apparent in the image. At the bottom of the AlN buffer layer, the atomic arrangement is visible, but it has a relatively random arrangement at the top. Diffraction pattern analysis was performed for each position to confirm the correct crystallinity. Fig. 6(b), (c) and (d) show the SEAD patterns at the positions of the upper and lower AlN and Al2O3. In the SEAD pattern of Fig. 6(b) and (c), the crystal orientation of AlN was confirmed to be the (002) direction from the vertical direction of the surface. And the crystallinity of AlN decreased from the bottom to the top.

    XRD patterns show for AlN on bare sapphire and N2 plasma pretreated sapphire in Fig. 7. Two xrd peaks were observed and confirmed as peaks for AlN (002) and Al2O3 (006). The FWHM for the AlN (002) peak was calculated to confirm the crystallinity of the two samples. As a result, AlN/PSS and AlN/pretreated PSS was confirmed to be 1649 arcsec and 1742 arcsec, respectively.

    Fig. 8 AFM (a) and (b) show the AFM observation results for AlN / bare sapphire and AlN / pretreated sapphire. The AFM was measured in contact mode and the obtained image was expressed in 3D. Planar sapphire substrate was used to confirm the surface of micro scale area. The scan area was 5 × 5 um2 and the scan speed was 1 Hz. Topographical images obtained by AFM can be eliminated by artifact noise and tilt due to vertical error of surface and tip through flattening process.19) And a low-pass filter was used to enhance the contrast of the image.20) Z-scale range of 3 dimensional image was 4 nm. The root-mean-square(RMS) surface roughness was calculated before applying low-pass filter and was approximately 1.828 nm and 0.270 nm for Fig. 8(a) and (b), respectively. The roughness of the AlN surface deposited on the pretreated sapphire was improved. AlN growth on the surface of bare sapphire leads to epi-growth of AlN to maintain the lattice. However, when the internal stress reaches limit, AlN crystal divided into grains, which grow in the form of individual islands and islands expose on the surface.


    In this study, it was confirmed that the AlN buffer layer grown on the N2 plasma pretreated substrate affected the GaN growth. In order to confirm the effect of plasma pretreatment, we analyzed by XPS and confirmed the material change of the surface. Plasma pretreatment changed the surface of Al2O3 to AlN and AlON. The pretreated layer was confirmed to be about 1 nm thickness by HR-TEM. The crystal properties of the surface were changed by N2 plasma treatment. The crystallinity of AlN buffer layer decreases with distance from the substrate. In XRD, AlN buffer layer was grown as the (002) plane and the crystallinity was relatively low when compared with AlN-bare PSS. The surface state of the AlN was confirmed by AFM, and the surface roughness was determined to be 1.828 nm for AlN on bare sapphire and 0.270 nm for AlN on pretreated sapphire. In conclusion, AlN buffer changed by N2 plasma pretreatment has an advantageous impact on GaN growth on PSS.


    This work was supported by the Technology Innovation Program(10067492) funded by the Ministry of Trade, Industry & Energy(MOTIE, Korea).



    Cross-sectional (a) and plan view (b) SEM image of AlN buffer on PSS.


    Cross-sectional SEM image of GaN template on AlN/PSS (a) and AlN/pretreated PSS (b).


    Fitted X-ray photoelectron spectroscopy, N1s, O1s and Al2p spectra of bare sapphire substrate. The plots progress from top to bottom in cumulative sputter time. The dash line indicates the center position of the subpeaks.


    Fitted X-ray photoelectron spectroscopy, N1s, O1s and Al2p spectra of N2 plasma pretreated sapphire substrate. The plots progress from top to bottom in cumulative sputter time. The dash line indicates the center position of the subpeaks.


    (a) cross-sectional TEM image of pretreated sapphire without AlN. The metal(filler) was deposited for clear contrast. (b) Analysis of atom layers by scanning the intensity data of each pixel in the dot line. (c) Diffraction pattern of pretreated sapphire surface (1 square box). (d) Diffraction pattern of sapphire (2 square box).


    (a) cross-sectional TEM image of P-AlN. The metal(filler) was deposited for clear contrast. (b) Diffraction pattern of upper PAlN (1 square box). (c) Diffraction pattern of P-AlN near sapphire (2 square box). (d) Diffraction pattern of sapphire (3 square box). The dash line is the boundary of each layer.


    X-ray diffraction pattern of AlN/PSS and AlN/pretreated PSS.


    Atomic force microscopy image of AlN/bare sapphire (a) and AlN/pretreated sapphire (b) surface with planar sapphire substrate. The scan size was 5 × 5 μm2. The z-scale range from 0 to 4 nm. Roughness was calculated 1.828 nm (a) and 0.270 nm (b).


    Growth condition of AlN and GaN.


    1. LeeY.J. HwangJ.M. HsuT.C. HsiehM.H. JouM.J. LeeB.J. LuT.C. KuoH.C. WangS.C. (2006) IEEE Photonics Technol. Lett., Vol.18 ; pp.1152
    2. AmanoH. SawakiN. AkasakiI. ToyodaY. (1986) Appl. Phys. Lett., Vol.48 ; pp.353
    3. FujiiT. GaoY. SharmaR. HuE.L. DenBaarsS.P. NakamuraS. (2004) Appl. Phys. Lett., Vol.84 ; pp.855
    4. KimS. KimB. JangS. KimA. ParkJ. ByunD. (2011) J. Cryst. Growth, Vol.326 ; pp.200
    5. KimD-S. JeongW.S. KoH. LeeJ-S. ByunD. (2017) Thin Solid Films, Vol.641 ; pp.2
    6. NingX.J. ChienF.R. PirouzP. (1996) J. Mater. Res., Vol.11 ; pp.580
    7. AmanoH. AkasakiI. HiramatsuK. KoideN. SawakiN. (1988) Thin Solid Films, Vol.163 ; pp.415
    8. NakamuraS. (1991) Jpn. J. Appl. Phys., Vol.30 ; pp.1705
    9. WuuD.S. WangW.K. WenK.S. HuangS.C. LinS.H. HuangS.Y. LinC.F. HorngR.H. (2006) Appl. Phys. Lett., Vol.89 ; pp.161105
    10. BeaumontB. VennA(c)guA"sPh. GibartP. (2001) Phys. Status Solidi, B Basic Res., Vol.227 ; pp.1
    11. ParkJ. KimD-S. JeongW.S. ChoS.H. KimC. KoH-A. LeeD. ByunD. (2017) Mater. Sci. Tech. Jpn, Vol.54 ; pp.112
    12. BockowskiM. GrzegoryI. (2005) J. Cryst. Growth, Vol.281 ; pp.11
    13. CelebiogluA. VempatiS. Ozgit-AkgunC. BiyikliabN. UyarT. (2014) RSC Adv, Vol.4 ; pp.61698
    14. LuW. IwasaY. OuY. JinnoD. KamiyamaS. PetersenP. M. OuH. (2017) RSC Adv, Vol.7 ; pp.8090
    15. RosenbergerL. BairdR. McCullenE. AunerG. ShreveG. (2008) Surf. Interface Anal., Vol.40 ; pp.1254
    16. ChoY. KimY. WeberE.R. RuvimovS. Liliental-WeberZ. (1999) J. Appl. Phys., Vol.85 ; pp.7909
    17. NamkoongG. DoolittleW.A. BrownA.S. LosurdoM. CapezzutoP. BrunoG. (2002) J. Vac. Sci. Technol. B, Vol.20 ; pp.1221
    18. SoaresG.V. BastosK.P. PezziR.P. MiottiL. DriemeierC. BaumvolI.J. (2004) Appl. Phys. Lett., Vol.84 ; pp.4992
    19. KienbergerF. PastushenkoV.P. KadaG. PuntheeranurakT. ChtcheglovaL. RiethmuellerC. RanklC. EbnerA. HinterdorferP. (2006) Ultramicroscopy, Vol.106 ; pp.822
    20. ShevchukA.I. HobsonP. LabM.J. KlenermanD. KrauzewiczN. KorchevY.E. (2008) Biophys. J., Vol.94 ; pp.4089