Surface & Coatings Technology 237 (2013) 118–125
Contents lists available at ScienceDirect
Surface & Coatings Technology journal homepage: www.elsevier.com/locate/surfcoat
Epitaxial growth of AlN on c-plane sapphire by High Temperature Hydride Vapor Phase Epitaxy: Influence of the gas phase N/Al ratio and low temperature protective layer R. Boichot a,⁎, N. Coudurier a,b, F. Mercier a, S. Lay a, A. Crisci a, S. Coindeau c, A. Claudel b, E. Blanquet a, M. Pons a a b c
SIMaP CNRS, Grenoble INP, UJF, 1130 Rue de la Piscine, 38402 Saint Martin d'Hères, France ACERDE, 354 Voie Magellan-Alpespace, 73800 Ste Hélène du Lac, France CMTC, Grenoble INP-CNRS, 38402 Saint Martin d'Hères, France
a r t i c l e
i n f o
Available online 22 August 2013 Keywords: AlN High Temperature-HVPE III–V heteroepitaxy c-Plane sapphire
a b s t r a c t AlN is epitaxially grown on c-plane sapphire by High Temperature Hydride Vapor Phase Epitaxy (HT-HVPE) at constant growth rate and thickness, while varying the N/Al ratio in the gas phase at 1500 °C. The influence of an additional low temperature (1200 °C) protective layer on AlN crystal quality is also assessed. The experiments and thermodynamic calculations show that the sapphire substrate is unstable at high temperature under hydrogen and ammonia while it is stable at low temperature or under a few hundred nanometers of AlN protective layer even at high temperature. In terms of AlN crystal quality, the optimal process developed here consists in depositing a 170 nm low temperature protective AlN layer with N/Al = 3 followed by a high temperature thick AlN layer grown with N/Al = 1.5. In this case, the interface between AlN and sapphire remains continuous (no etching) and the stress in the grown layer at room temperature is minimized by a balance of the growing tensile stress with the cooling compressive stress. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Aluminum nitride is a promising substrate for AlGaN-based UV LED and piezoelectric applications (Micro Electro Mechanical Systems, MEMS and Surface Acoustic Waves, SAW devices). The UV LED industry requires high quality single crystals (i.e. deep UV transparency, low density of defect) [1]. The requirements are less restrictive for piezoelectric applications, for which highly oriented c-axis layers are needed [2,3]. Among the different available processes for AlN growth (Physical Vapor Transport, PVT [4–7]; Metal–Organic Chemical Vapor Deposition, MO-CVD [8] and High Temperature Hydride Vapor Phase Epitaxy, HTHVPE), High Temperature HVPE (N1200 °C) becomes the most prospective technique to produce the required quality for both piezoelectric and semiconductor industry [9–21]. A better understanding of the phenomena leading to high quality AlN layers grown on sapphire is the key point to allow HT-HVPE becoming a new industrial reference in thick AlN layers processing. Indeed, it is currently one of the cheapest way to produce industrial grade AlN single crystals. Currently, the main concern is the lack of cheap compatible seed substrates that could present a thermodynamic stability at the temperature and gas mixture used for AlN growth. In particular, the deep UV transparent sapphire undergoes severe etching under hydrogen atmosphere at
high temperature [21,22]. The main solution to overcome this issue is to protect the sapphire surface with a thin (b200 nm) epitaxial AlN layer grown at low temperature prior to the thick high temperature deposit (N1 μm). This buffer intermediate layer is called “protective layer”. Growth of AlN templates on sapphire by HT-HVPE was historically assessed since 2001, in majority on c-plane surfaces [9–16]. Some growths were also attempted on a-plane, tilted c-plane or semi-polar orientations [17]. The poor thermodynamic stability of sapphire in HTHVPE growth conditions was noticed and several authors worked on processes that are able to avoid sapphire decomposition by means of templates, protective or nucleation layers [18–20]. On the contrary, the thermodynamic instability of sapphire substrates in HVPE conditions was used to promote mechanical fragility of the sapphire/AlN interface to produce freestanding substrates [21]. Due to the high dislocation density obtained by direct growth on flat substrates, the current trend is to transpose the ELO (Epitaxial Layer Overgrowth) technique, mastered with GaN, to AlN layers [23–26]. The aim of this study is to assess the feasibility of a one step growth of AlN on c-plane sapphire by only varying the N/Al ratio in the gas phase and compare the crystalline quality obtained with layers grown on a thin low temperature protective layer (in situ grown template). Results are analyzed with the help of thermodynamic considerations. 2. Experiments
⁎ Corresponding author at: SIMAP/Phelma Bâtiment Recherche, 1130 Rue de la Piscine, 38402 Saint Martin d'Hères, France. Tel.: +33 476826537; fax: +33 476826677. E-mail address:
[email protected] (R. Boichot). 0257-8972/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.surfcoat.2013.08.016
Experiments are conducted in a quartz, cold wall, and vertical CVD reactor. The experimental apparatus is depicted in [27]. Pressure is maintained at 1333 Pa during experiments. The reactor is
R. Boichot et al. / Surface & Coatings Technology 237 (2013) 118–125
fed with high purity gases (H2, NH3 and Cl2, 99.9999%). The nitrogen source is NH3, while the aluminum source is AlCl3 produced by the direct chlorination of high purity (99.9999%) aluminum pellets heated at 650 °C [16] by a lamp furnace surrounding the upper part of the reactor. Aluminum pellets are contained in a quartz tube. It is emphasized here that the chlorination reaction is complete (no remaining free chlorine at the chlorination tube outlet). In this case, the flow rate of AlCl3 entering the reactor is equal to 2/3 the flow rate of Cl2 entering the lamp furnace containing aluminum pellets. This assumption allows the calculation of the N/Al ratio in the gas phase. Prior to the beginning of the growth process, aluminum pellets are in situ cleaned using hydrogen at 650 °C under vacuum. Sapphire substrates are set down on a 55 mm diameter non-coated graphite susceptor heated by induction. The growth temperature is measured at the center of the substrate with a dual wavelength pyrometer. Quarters of 50.8 mm diameter “Epi-ready” c-plane (±0.1°) sapphire are used as substrates. The sapphire substrates are in-situ cleaned by an etching step under hydrogen at 1100 °C prior to the growth. Hydrogen is used as reactive carrier gas during growth and for heating and cooling down. The reaction by-products (mainly AlClx, HCl and NH4Cl) are condensed in a liquid nitrogen cold trap located prior to the pumping. The experiments are designed to assess:
119
Fig. 1. Schematic diagram of the one step growth sequence. sccm stands for “cubic centimeters per minute”, 6.929 × 10−7 mol/s.
- Firstly, the feasibility of a one step AlN growth on sapphire by varying the N/Al ratio in the gas phase from 0.75 to 30 at constant thickness (6.25 μm), constant growth rate (5 μm/h) and constant temperature (1500 °C). - Secondly the influence of a low temperature protective layer (1200 °C) prior to high temperature deposition and the characteristics of the protective layer itself.
protective AlN layer at 1200 °C. AlCl3 is injected 2 min prior to NH3 to favor the Al polar orientation [28]. A temperature of 1200 °C at this step is chosen because the sapphire etching rate in hydrogen is considered negligible below 1300 °C [21] and the higher the temperature is, the higher should be the crystalline quality of AlN due to surface adatom diffusion enhancement. As soon as the protective layer is deposited, the N/Al ratio in gas phase is changed and the ramp-up toward 1500 °C is made under a constant H2 + NH3 + AlCl3 mixture, so the growth rate is continuously increased between the two growth steps. The growth of the protective layer (1200 °C) is performed with a N/Al ratio of 3. Indeed, experiments carried out with N/Al ratios of 1.5 and 7.5 in the gas phase at 1200 °C led to polycrystalline deposits with rough surfaces. For the second step of the two step growth at 1500 °C, a N/Al = 1.5 is selected because it was the only ratio that leads to the growth of epitaxial AlN on AlN MO-CVD templates in our apparatus at 1400 °C [27]. The bare protective AlN layer is deposited by stopping the growth after the first step at 1200 °C and cooling down under H2 + NH3 mixture. Samples are characterized by Field Emission Gun Scanning Electron Microscopy (FEG-SEM), Raman spectroscopy in backscattering geometry using the 514 nm line of an Ar laser at about 7 mW, X-ray (XRD) rocking curves (ω-scans), using a 4-circles diffractometer and Cu Kα radiation (λ = 1.54 Å) of a rotating anode generator, Atomic Force Microscopy (AFM), Transmission Electron Microscopy (TEM) of the AlN/sapphire interface, Secondary Ions Mass Spectrometry (SIMS) and photoluminescence (PL) at 5 K. An excimer ArF laser at 193 nm (6.4 eV) was used as a deep-UV excitation source, above the AlN bandgap energy.
The theoretical AlN growth rate is calculated according to a numerical modeling [27], so that a constant growth rate can be maintained with varying N/Al ratio in gas phase. Table 1 summarizes the process parameters used for each growth. All the experimental conditions presented in Table 1 led to epitaxial growth of AlN on sapphire, i.e. the X-ray diffraction measurements of the samples in Bragg–Brentano configuration present only the (000n) diffraction peaks of AlN and sapphire (see end of the section for details). A slight discrepancy between computed and experimental growth rates is observed, essentially due to thickness measurement uncertainties and to simplifying assumptions in the numerical modeling [27]. Fig. 1 represents the schematic diagram for the one step growth process. The sapphire etching at 1100 °C is followed by a heating-up toward growth temperature (1500 °C) under a H2 + NH3 mixture. Ammonia is added to hydrogen as an attempt to protect the sapphire surface from any further reaction with H2 and to contribute to the formation of a native AlN nucleation layer. Prior to the growth, a two minute AlCl3 preflow is applied, but as the sapphire was heating up under NH3, the preflow may not avoid the formation of polarity inversion domains [28]. Fig. 2 represents the schematic diagram for the two step growth process. The sapphire etching at 1100 °C is followed by the growth of the
Table 1 Parameters used during AlN growth. sccm stands for “cubic centimeters per minute”, 6.929 × 10−7 mol/s. The two step growth is the combination between a protective layer deposition and a one step growth. N/Al ratio in gas phase
NH3 flow rate (sccm)
Cl2 flow rate (sccm)
H2 flow rate (sccm)
Estimated growth rate (μm/h)
Deposition duration (min)
Measured thickness (μm)
Measured growth rate (μm/h)
Comment
0.75 1.5 3 7.5 15 30 3 1.5
2.20 3.50 6.10 13.9 27.5 52.5 3.0 3.5
4.44 3.50 3.05 2.78 2.75 2.62 1.50 3.50
1000 1000 1000 1000 1000 1000 1000 1000
5 5 5 5 5 5 – 5
75 75 75 75 75 75 10 75
8.05 7.90 5.86 5.36 7.01 4.54 0.17 4.03
6.44 6.32 4.68 4.29 5.68 3.59 1.03 3.22
One step growth One step growth One step growth One step growth One step growth One step growth Protective layer Two step growth
120
R. Boichot et al. / Surface & Coatings Technology 237 (2013) 118–125
Fig. 2. Schematic diagram of the two step growth sequence. sccm stands for “cubic centimeters per minute”, 6.929 × 10−7 mol/s.
3. Results Fig. 3 presents the FEG-SEM images of surface morphology for layers obtained with N/Al ratio of 0.75, 1.5, 7.5 and 30 during the one step growth process. The surface is generally flat with the presence of particularities depending on the N/Al ratio. The sample grown with N/Al = 0.75 presents open cracks and severe layer disorientation (strong bending of the peeled AlN layer). For the N/Al ratio of 1.5, the surface presents only oriented cracks (stripes) that tend to coalesce during growth. These stripes are the consequence of an early directional AlN layer peeling that is clearly visible in Fig. 4. Fig. 4 presents the FEG-SEM images of cross-sections for layers obtained with N/Al ratio of 0.75, 1.5, 7.5 and 30 during the one step
growth process. The interface presents voids that indicate a poor AlN adherence coupled with a sufficient tensile stress to separate and bend AlN flakes. On the contrary, for N/Al ratio above 7.5, the interface is nearly continuous and the surface stripes vanished. The lower the N/Al ratio is, the more voids at the interface are observed. In these samples, cracks do not follow any obvious orientation regarding the crystal structure of substrate. Fig. 5 presents the cross section and surface morphology of the protective layer grown with a N/Al of 3 and 1200 °C and the two step growth with additional “thick” layer grown at 1500 °C and N/Al = 1.5. The interface between AlN and sapphire is continuous. The surfaces of the protective layer and the layer processed in two steps are flat. The surface of the protective layer presents small disoriented AlN inclusions (white dots in Fig. 5, see also Fig. 8), not found after the second growth step. Fig. 6 presents the Raman E2(h) shifts of all the grown AlN layers measured at room temperature. The E2(h) phonon mode is commonly used as a strain or stress indicator due to its high intensity in backscattering configuration. The E2(high) peak frequency of unstressed AlN is 657.4 ± 0.2 cm−1 [29]. Raman E2(h) peak frequency shift indicates that only the protective layer presents tensile stress after cooling down. All the other AlN samples are in a compressive state. The compressive stress is maximal for the one step growth at N/Al = 3 and minimal for N/Al = 15 in gas phase. The influence of N/Al ratio on the AlN E2(high) shift will be discussed in the next section. Fig. 7 presents the FWHM (Full Width at Half Maximum) values of the AlN (0002) diffraction peaks from ω rocking curves. The FWHM of symmetric diffraction is correlated to the screw dislocation density in the layer [30]. The FWHM values globally follow the trend of the AlN E2(high) shift: the more the layer is relaxed at room temperature, the better is the crystalline quality, except for the lowest N/Al ratios (N/Al = 0.75 and 1.5). For N/Al = 0.75, a FWHM of 5011 arcsec is measured with a layer undergoing a nearly complete lack of adherence. At N/Al = 1.5 a local FWHM minimum is measured, in spite of a very poor adherence of the AlN layer on sapphire. This trend (good intrinsic crystalline quality at N/Al = 1.5) is consistent with another study on this experimental apparatus [27]. The lowest
Fig. 3. Surface morphology of various samples grown with the one step process.
R. Boichot et al. / Surface & Coatings Technology 237 (2013) 118–125
121
Fig. 4. Cross section of various samples grown with the one step process.
FWHM for the one step growth is found for N/Al = 15 (950 arcsec). The FWHMs of the protective layer and the two step grown layer are 327 and 403 arcsec, respectively, corresponding to screw dislocation densities of 1.4 × 108 cm−2 and 2.2 × 108 cm−2 [30]. The low temperature protective layer effectively increases strongly the crystal quality of AlN grown on flat sapphire. Fig. 8 presents AFM pictures and RMS (Root Mean Squared) roughness of the protective layer and the two step grown layer. The RMS roughnesses of surfaces are 9.4 nm and 14.9 nm for the protective
layer and two step grown, respectively. The surfaces present at least hexagonal features (spiral growth) for the low temperature protective layer and hexagonal grain boundaries for the two step growth. TEM imaging of the AlN/sapphire interface for the two step grown layer is presented in Fig. 9. The electron diffraction patterns indicate a single crystal growth of AlN near the interface and an epitaxial relationship (0001)AlN//(0001)Al2O3, with b10–10NAlN//b11–20NAl2O3. The bright field image highlights the very high density of threading dislocations emerging from the interface AlN/sapphire. Dark field analysis (not
Fig. 5. Surface morphology and cross section of the low temperature protective layer and the layer grown with the two step process.
122
R. Boichot et al. / Surface & Coatings Technology 237 (2013) 118–125
Fig. 6. Raman E2(h) shift of AlN with varying gas phase N/Al ratio and deposition process.
represented here) indicates an absence of inversion domains on the analyzed area. Wet KOH etching (5 min, 80 °C, and 5% KOH in deionized water) revealed the characteristic hexagonal pits appearing during etching of Al polar (0001) AlN facets [31,32], confirming the influence of the AlCl3 preflow on polarity [28]. The common admitted theoretical bandgap for AlN at 300 K is 6.2 eV [33,34]. Perry and Rutz [34] measured a bandgap of 6.28 eV at 5 K with PVT AlN single crystal. Fig. 10 shows the photoluminescence (PL) spectra at 5 K for the two step grown layer. PL spectrum shows two weak Near Band Edge emission (NBE) peaks at wavelength around 212 nm (5.85 eV) and 225 nm (5.51 eV). The AlN layer also shows broad defect band emissions of high intensity, emitting around 285 nm and 400 nm. These defect band emissions are related to Alvacancy (VAl) and contamination by impurities such as oxygen and the formation of VAl–O complexes. The high concentration of oxygen contaminants and Al vacancies [35] inside these layers is certainly highly detrimental for UV transparency and further efforts are needed to reduce their incorporation.
SIMS measurements indicate the following contaminants for this layer, in mean value: Si: 2 × 1018 at. cm−3, C: 8 × 1017 at. cm−3, O: 3 × 1017 at. cm−3, H: 4 × 1017 at. cm−3, and Cl: 2 × 1016 at. cm−3. The contaminant profiles present a decreasing slope from the interface AlN/sapphire to the near surface, and a large increase at the free surface. We attribute the three main contaminants (Si, C and O) to the incorporation of susceptor and reactor wall etched products during growth. Cl and H levels reflect the gas mixture contribution to the layer contamination and may be present as dissolved species in the crystal structure. 4. Discussion 4.1. Void formation at the interface AlN/sapphire AlN deposits made without protective layer (one step growth process) undergo severe detachment and bending in the case of low N/Al, probably during the very early steps of growth (see Fig. 4). The instability of the AlN/alumina interface in the one step growth process will be
Fig. 7. FWHM of the (0002) diffraction peak of AlN in XRD with varying gas phase N/Al ratio and deposition process.
R. Boichot et al. / Surface & Coatings Technology 237 (2013) 118–125
123
Fig. 8. Phase mode AFM pictures and RMS roughness of the protective layer (left) and the two step grown layer (right). The images cover a 20 μm × 20 μm area.
now discussed from a thermodynamic point of view, by simulating the thermodynamic equilibrium between alumina and a H2 + NH3 mixture in a large temperature range. Fig. 11 represents the thermodynamic equilibrium of 1000 mol H2, 20 mol NH3 and 1 mol Al2O3 at 1333 Pa in the range [800–1500 °C] in a closed volume, calculated with the Factsage 6.0 software, SGTE database (http://www.sgte.org/). Solid alumina bAl2O3N is stable in these conditions only in equilibrium with gaseous N2, H, H2O and Al (mainly) and a small quantity of solid bAlNN (even if N atoms are in large excess compared to Al atoms). Then, above 1240 °C solid AlN is not stable and should vaporize. In our heating-up conditions between etching (1100 °C) and growing (1500 °C) temperature for the one step growth process, the vacuum is dynamically maintained with 1000 sccm H2 + 20 sccm NH3, depleting the partial pressure of gases containing oxygen and aluminum; the equilibrium should be globally displaced in favor of Al2O3 vaporization. It should be noticed that according
to thermodynamic calculations, the bAlNN formed is in equilibrium with bAl2O3N and gaseous compounds, which means that nothing indicates that the bAlNN forms a continuous tight protective layer on sapphire. It would be the case only if the equilibrium was strongly in favor of bAlNN compared to bAl2O3N. Kinetically, it has be proven that the c-sapphire etching under hydrogen, negligible at 1100 °C [36] (at atomic step level), is considered to arise at 1300 °C [21] to reach 6 μm/h at 1500 °C [36,37]. Kinetic of AlN etching under hydrogen is considered to be substantially lower. The etching rate of Al polar AlN at 1500 °C under hydrogen reaches 0.5 μm/h only [38]. Additionally, H2 can diffuse through a few nanometers of bAlNN layer and create voids at the interface AlN/Al2O3 at 1500 °C [21]. In consequence, the thin native bAlNN eventually formed under H2 + NH3 mixture during heating up may not protect the sapphire from being etched by the carrier gas at high temperature.
Fig. 9. Annotated TEM diffraction patterns (left) and interface bright field image (right) of the AlN two step grown layer.
124
R. Boichot et al. / Surface & Coatings Technology 237 (2013) 118–125
Fig. 10. Photoluminescence spectrum at 5 K of the two step AlN grown layer with, in inset, a focus near the theoretical bandgap of AlN.
The chemical stability of the AlN/sapphire interface in H2/NH3 at high temperature cannot be maintained without a continuous (tight) AlN protective layer sufficiently thick to suppress H2 diffusion at the time and temperature scale of heating up of the sample. In that sense, the peeling of AlN at the AlN/sapphire interface during the one step growth of our samples is easily explained by the interface instability during the early growth time. The fact that the interface remains continuous for N/Al N 7.5 during one step growth could indicate that the increasing partial pressure of ammonia with increasing N/Al ratio slows the formation of voids at the AlN/sapphire interface, perhaps by stabilizing the growing AlN layer (increasing precursor supersaturation). During the two step growth, a continuous 170 nm AlN layer is deposited at 1200 °C on sapphire prior to the growth at high temperature (1500 °C). Even if the 170 nm AlN layer becomes thermodynamically unstable during the heating up toward growth temperature (see Fig. 11), the etching rate of AlN under hydrogen is sufficiently low and the remaining thickness is sufficiently high at 1500 °C to ensure that AlN is still an efficient diffusion barrier to hydrogen when the high
Fig. 11. Calculated molar fraction at equilibrium of a mixture of 1000 mol H2, 20 mol NH3 and 1 mol Al2O3 at 1333 Pa. bAlNN molar fraction refers to the ratio moles bAlNN formed/moles bAl2O3N remaining.
temperature growth begins. This explains why voids are not observed at the AlN/sapphire interface with the two step growth process and the subsequent increase in crystal quality. 4.2. Relationship between process parameters and stress state Analysis of results from the observation of Raman E2(high) shifts highlights the effect of interfacial etching of sapphire on the stress state of the grown AlN at room temperature. It was shown in Fig. 6 that the temperature stress in AlN layers grown in one step is always compressive. The compressive state for high temperature grown AlN layers on sapphire is mainly due to the thermal expansion coefficient difference between AlN and sapphire [39]. Lattice mismatch stress is only responsible for a negligible part of the overall stress in AlN [40]. On the other hand, at growth temperature the layer is in a tensile state due to the coalescence of primary AlN islands formed during the first stages of growth [39,40]. The growth temperature is constant in our experiments, so the only parameters that explain differences in stresses during the one step growth are the gaseous N/Al ratio and the interface stability (the two parameters are however correlated). Voids and compressive strain are maximal at N/Al = 3, this corresponds to the case where tensile stresses are relaxed in the early steps of growth by cracking and AlN peeling, leading to an unstressed AlN layer at high temperature that undergoes a compressive cooling stress only. On the other hand, when N/Al ratio is high in one step growth (N/Al N 7.5), the interface is stable and the layer undergoes first a tensile growing stress at high temperature that is balanced with a compressive stress during coolingdown. Surprisingly, the residual stresses are similar and low for N/Al ratios = 0.75 and 30. For N/Al = 30 the cooling stress is partially balanced by the coalescence stress, while for N/Al = 0.75 the layer is so cracked and poorly adhesive that the compression from substrate is only partly transmitted to the AlN layer and locally released by open cracks. The thin low temperature protective layer is in a strongly tensile state even after cooling. It results from a lower temperature difference with room temperature compared to high temperature grown samples that lower the cooling compressive part of the stress. Consequently, the two step growth leads to a layer in a slight compressive state by a growing and cooling stress balance nearly in equilibrium at room temperature. Influence of chlorides is not discussed here but the produced HCl should also be an efficient etching gas for sapphire and AlN playing its
R. Boichot et al. / Surface & Coatings Technology 237 (2013) 118–125
own role in interface stability. Cl and H are found in noticeable quantity by SIMS measurements, confirming the potential mobility (diffusion) of these species into AlN. 5. Conclusion The growth of AlN on c-plane sapphire is particularly influenced by the kinetic and thermodynamic stability of the sapphire at the interface with AlN. If no direct chemical reactions can occur between AlN and sapphire at such temperatures, H2 etches sapphire even through a few nanometers of AlN layer, leading to gaseous oxygen and aluminum hydrides and to the detachment of AlN layer from the substrate. In consequence, the conditions to succeed in growing epitaxial AlN on sapphire are the following (using hydrogen as carrier gas): - To grow a low temperature AlN protective layer at a temperature where sapphire is kinetically stable, but at the higher temperature possible to increase the quality of this first layer: 1200 °C seems a good compromise. The protective layer should be sufficiently thick to slow hydrogen diffusion during the ramp-up to the second layer growth temperature. A N/Al = 3 ratio in the gas phase during growth is found to reproducibly allow an epitaxial relationship between AlN and sapphire at 1200 °C in our experimental conditions. - To grow a second layer at high temperature (1500 °C, to increase crystalline quality and growth velocity) with low N/Al ratio (typically N/Al = 1.5 [27] to 4 [20]) on the previously grown layer. The thickness and growth rate of this second layer seem to have a low influence on the final quality of the AlN layer [20], only the temperature plays a key role. The question of the influence of the process parameters on the dislocation density is still open. References [1] T. Kinoshita, K. Hironaka, T. Obata, T. Nagashima, R. Dalmau, R. Schlesser, B. Moody, J.Q. Xie, S. Inoue, Y. Kumagai, A. Koukitu, Z. Sitar, Appl. Phys. Express 5 (2012) 122101. [2] S. Ballandras, A. Reinhardt, V. Laude, A. Soufyane, S. Camou, W. Daniau, T. Pastureaud, W. Steichen, R. Lardat, M. Solal, P. Ventura, J. Appl. Phys. 96 (2004) 7731–7741. [3] G. Piazza, V. Felmetsger, P. Muralt, R.H. Olsson, R. Ruby, MRS Bull. 37 (2012) 1051–1061. [4] G.A. Slack, T.F. McNelly, J. Cryst. Growth 42 (1977) 560–563. [5] M. Bickermann, O. Filip, B.M. Epelbaum, P. Heimann, M. Feneberg, B. Neuschl, K. Thonke, E. Wedler, A. Winnacker. J. Cryst. Growth 339 (1) (2012) 13–21. [6] B. Gao, S. Nakano, K. Kakimoto. J. Cryst. Growth 338 (1) (2012) 69–74. [7] S. Zuo, X. Chen, L. Jiang, H. Bao, J. Wang, L. Guo, W. Wang. Mater. Sci. Semicond. Process. 15 (4) (2012) 401–405. [8] W. Tian, W.Y. Yan, J.N. Dai, S.L. Li, Y. Tian, X. Hui, J.B. Zhang, Y.Y. Fang, Z.H. Wu, C.Q. Chen, J. Phys. D Appl. Phys. 46 (2013) 065303. [9] Yu. Melnik, D. Tsvetkov, A. Pechnikov, I. Nikitina, N. Kuznetsov, V. Dmitriev, Phys. Status Solidi A 188 (1) (2001) 463–466.
125
[10] O. Kovalenkov, V. Soukhoveev, V. Ivantsov, A. Usikov, V. Dmitriev, J. Cryst. Growth 281 (2005) 87–92. [11] Y. Kumagai, T. Yamane, T. Miyaji, H. Murakami, Y. Kangawa, A. Koukitu, Phys. Status Solidi C (7) (2003) 2498(0). [12] Y. Kumagai, T. Yamane, A. Koukitu, J. Cryst. Growth 281 (2005) 62. [13] D.S. Kamber, Y. Wu, B.A. Haskell, S. Newman, S.P. DenBaars, J.S. Speck, S. Nakamura, J. Cryst. Growth 297 (2006) 321–325. [14] T. Nagashima, M. Harada, H. Yanagi, Y. Kumagai, A. Koukitu, K. Takada, J. Cryst. Growth 300 (2007) 42. [15] K. Eriguchi, H. Murakami, U. Panyukova, Y. Kumagai, S. Ohira, A. Koukitu, J. Cryst. Growth 298 (2007) 332. [16] A. Claudel, E. Blanquet, D. Chaussende, M. Audier, D. Pique, M. Pons, J. Cryst. Growth 311 (13) (2009) 3371–3379. [17] J.J. Wu, Y. Katagiri, K. Okuura, D.B. Li, H. Miyake, K. Hiramatsu, J. Cryst. Growth 311 (2009) 3801–3805. [18] J. Tajima, H. Murakami, Y. Kumagai, K. Takada, A. Koukitu, J. Cryst. Growth 311 (2009) 2837. [19] M. Balaji, A. Claudel, V. Fellmann, I. Gelard, E. Blanquet, R. Boichot, A. Pierret, B. Attal-Trétout, A. Crisci, S. Coindeau, H. Roussel, D. Pique, K. Baskar, M. Pons, J. Alloys Compd. 526 (2012) 103–109. [20] T. Nagashima, M. Harada, H. Yanagi, H. Fukuyama, Y. Kumagai, A. Koukitu, K. Takada, J. Cryst. Growth 305 (2007) 355–359. [21] Y. Kumagai, Y. Enatsu, M. Ishizuki, Y. Kubota, J. Tajima, T. Nagashima, H. Murakami, K. Takada, A. Koukitu, J. Cryst. Growth 312 (2010) 2530–2536. [22] K. Akiyama, T. Araki, H. Murakami, Y. Kumagai, A. Koukitu, Phys. Status Solidi C 4 (7) (2007) 2297–2300. [23] 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. [24] Y. Katagiri, S. Kishino, K. Okuura, H. Miyake, K. Hiramatu, J. Cryst. Growth 311 (2009) 2831–2833. [25] V. Kueller, A. Knauer, F. Brunner, U. Zeimer, H. Rodriguez, M. Kneissl, M. Weyers, J. Cryst. Growth 315 (2011) 200–203. [26] N. Okada, N. Kato, S. Sato, T. Sumii, N. Fujimoto, M. Imura, K. Balakrishnan, M. Iwaya, S. Kamiyama, H. Amano, I. Akasaki, T. Takagi, T. Noro, A. Bandoh, J. Cryst. Growth 300 (2007) 141–144. [27] R. Boichot, A. Claudel, N. Baccar, A. Milet, E. Blanquet, M. Pons, Surf. Coat. Technol. 205 (2010) 1294. [28] R. Togashi, T. Nagashima, M. Harada, H. Murakami, Y. Kumagai, H. Yanagi, A. Koukitu, J. Cryst. Growth 360 (2012) 197–200. [29] V.Y.u. Davydov, Y.u.E. Kitaev, I.N. Goncharuk, A.N. Smirnov, J. Graul, O. Semchinova, D. Uffmann, M.B. Smirnov, A.P. Mirgorodsky, R.A. Evarestov, Phys. Rev. B 58 (19) (1998) 12899–12907. [30] P. Gay, P.B. Hirsch, A. Kelly, Acta Metall. 1 (1953) 315. [31] M. Bickermann, S. Schmidt, B.M. Epelbaum, P. Heimann, S. Nagata, A. Winnacker, J. Cryst. Growth 300 (2007) 299–307. [32] D. Zhuang, J.H. Edgar, B. Strojek, J. Chaudhuri, Z. Rek, J. Cryst. Growth 262 (2004) 89–94. [33] H. Yamashita, K. Fukui, S. Misawa, S. Yoshida, J. Appl. Phys. 50 (1979) 896. [34] P.B. Perry, R.F. Rutz, Appl. Phys. Lett. 33 (1978) 319. [35] Y. Kumagai, Y. Kubota, T. Nagashima, T. Kinoshita, R. Dalmau, R. Schlesser, B. Moody, J. Xie, Hisashi Murakami, A. Koukitu, Z. Sitar, Appl. Phys. Express 5 (2012) 055504. [36] F. Dwikusuma, D. Saulys, T.F. Kuech, J. Electrochem. Soc. 149 (11) (2002) G603–G608. [37] K. Akiyama, H. Murakami, Y. Kumagai, A. Koukitu, Jpn. J. Appl. Phys. 47 (2008) 3434–3437. [38] Y. Kumagai, K. Akiyama, R. Togashi, H. Murakami, M. Takeuchi, T. Kinoshita, K. Takada, Y. Aoyagi, A. Koukitu, J. Cryst. Growth 305 (2007) 366–371. [39] B. Wu, J. Bai, V.L. Tassev, M.L. Nakarmi, W. Sun, X. Huang, M. Dudley, H. Zhang, D.F. Bliss, J. Lin, H. Jiang, J. Yang, M. Asif Khan, Mater. Res. Soc. Symp. Proc. 892 (2006) 653–658. [40] S. Raghavan, J.M. Redwing, J. Appl. Phys. 96 (2004) 2995.