Hot-Wall MOCVD for Highly Efficient and Uniform Growth of AlN A. Kakanakova-Georgieva,* R. R. Ciechonski, U. Forsberg, A. Lundskog, and E. Janze´n Department of Physics, Chemistry and Biology, Linko¨ping UniVersity, SE 581 83 Linko¨ping, Sweden ReceiVed May 30, 2008; ReVised Manuscript ReceiVed October 21, 2008
CRYSTAL GROWTH & DESIGN 2009 VOL. 9, NO. 2 880–884
ABSTRACT: We demonstrated successful growth of AlN at a temperature of 1200 °C in a set of hot-wall MOCVD systems with the possibility of straightforward scaling up the process on larger wafer areas to meet the demand of device technologies. We outlined several aspects of the carefully optimized design and process parameters with relevance to achievement of a high overall growth rate (1 and up to 2 µm/h), efficiency, and uniformity, which to a great extent depends on how consumption of growthlimiting species by gas-phase adduct formation can actively be prevented. Mixing of the precursors upstream from the deposition area facilitates uniform epitaxial growth, while the greater uniformity of substrate temperature inherent to the hot-wall reactor and rotation of the wafer are of fundamental importance for layer-growth uniformity. The AlN layer thickness can be controlled with an accuracy of (1.3% on 2 in. wafers. The low-temperature cathodoluminescence spectrum of the AlN epitaxial material is strongly dominated by the intense near band-gap deep UV emission at about 208 nm.
1. Introduction Development of epitaxial techniques for growth of III-V compound semiconductors over the past decades is closely related to successful development of III-V-based multiple layers heterostructure devices such as light-emitting and laser diodes operating from the ultraviolet (UV) to infrared regions of the spectrum. The semiconductor light-emitting technology is dominated by two material systems: the AlGaInN system, grown either on sapphire or SiC, for the range of near-UV over blue to green, and the AlGaInP/GaAs system for yellow to red.1 There has been a strong push toward developing high Al content (>40%) AlGaN-based deep-UV light-emitting devices2 intended for a broad area of applications covering fluorescence-based biological agent detection, non-line-of-sight communications, water purification, solid-state lighting via phosphor excitation, and extra-high-density optical data storage. High-quality AlGaN epitaxy suffers from the lack of large area and high-quality AlN substrates. SiC and sapphire are the only widely available substrates for AlGaN epitaxy. To improve device performance, use of dislocation reduction techniques such as epitaxial lateral overgrowth (ELO) is demanding, especially elaborative maskless ELO techniques3,4 to prevent interaction of Al with the SiO2 or SiNX mask materials and enhance the lateral growth rate of AlGaN.5 A major issue of the AlGaN technology is development of the vapor-phase growth process including both hydride vapor phase epitaxy, HVPE, for growth of thick layers and metal-organic vapor-phase epitaxy, MOVPE (or alternatively metal-organic chemical vapor deposition, MOCVD), for heterostructure device fabrication. Specific reaction kinetics and thermodynamics are involved in the deposition processes of III-V semiconductor materials by HVPE and MOVPE, implementing certain reactor concepts and precursors.6-9 In HVPE classical hot-wall reactors with resistively heated multiple zone furnace systems are used. The group III species is transported to the substrate as a volatile group III chloride (e.g., GaCl) formed in a gas-solid reaction between HCl and the group III metal. In conventional MOVPE, cold-wall reactors are used. The vapor of the metal-organics * To whom correspondence should be addressed. Phone: +46-13-282649. Fax: +46-13-142337. E-mail:
[email protected].
source material (e.g., Ga(CH3)3, Al(CH3)3), as a precursor for group III element, is injected into a stream of carrier gas (e.g., H2). In both growth techniques hydrides (e.g., AsH3, PH3, or NH3) serve to deliver the group V element. As commonly considered, advantages of HVPE, compared to MOVPE, include greater uniformity of substrate temperature, higher growth rates, potential for higher throughput, and greatly reduced V/III ratios.10 It is also suited to the growth of lowimpurity III-V layers, but it is not often used for device fabrication as the HVPE technique is limited in providing the precise submicrometer thickness control and sharp interfaces required by multiple layer device heterostructures.11 This limitation stems from the difficulty in changing the flow rates of group III monochlorides which are formed in gas-solid reactions in separate zones upstream in the reactor.10 It is possible to extend the performance of hot-wall reactors to include metal-organics precursors as their use allows for improved control of the group III precursor mole fraction in the reactor12 and opens possibilities for growing heteroepitaxial structures with abrupt junctions, unlike conventional HVPE. For example, epitaxial growth of GaAs in a horizontal hot-wall resistively heated reactor has been investigated using the growth precursors diethylgallium chloride [(C2H5)2GaCl] and arsine, AsH3, instead of the typical HVPE inorganic precursor combination like Ga/HCl/AsH3.12 Epitaxial growth of InP by coinjection of trimethylindium, In(CH3)3, and HCl to form gaseous InCl in a horizontal hot-wall reactor with resistively heated furnaces has been demonstrated. Upon further reaction with PH3, InP can be deposited. The technique was introduced by the name of merged hydride-organometallic VPE.10 A similar technique by the name of hydride-metallorganic vapor-phase epitaxy for deposition of GaN has also been explored.13 While the merged HVPE-MOVPE approach in the area of conventional III-Vs (GaAs, InP, examples above) serves the purpose of using HVPE for heterostructure growth, the merged HVPE-MOVPE approach in the area of GaN serves the purpose of alternating between either MOVPE or HVPE in the same reactor. Recently, in a hybrid HVPE-MOVPE single reactor system independent heating of the substrate holder and reactor walls was demonstrated to provide suitable conditions for both types of growth processes applied at different stages of the GaN-onsapphire growth.11 The HVPE operation mode of the reactor
10.1021/cg8005663 CCC: $40.75 2009 American Chemical Society Published on Web 12/18/2008
Hot-Wall MOCVD
Figure 1. Schematic view of hot-wall MOCVD assembly in the deposition zone.
allows inexpensive high growth rate process for the low defect density thick GaN buffer layer. When in MOVPE operation mode a very thin initial starting layer as well as the final device layers can be grown. Very recently, an approach referred to as metal-organic hydride vapor-phase epitaxy and combining MOCVD and HVPE has been demonstrated for growth of high-quality thick (>20 µm) AlN buffer layers on specially prepared grooved templates followed by deposition of deep-UV LED structure in the same run in one continuous growth process.14 Growth of AlN by HVPE, which has been known for more than 30 years15 and has a principal similarity with growth of GaN by HVPE, is affected by the strong quartz etching effects at high growth temperatures by the aluminum chlorides. Elaborations on the design and materials of the parts inside the reactor to prevent the corrosive action of the aluminum chlorides15,16 and/or favor formation of AlCl3, which is less reactive with quartz than AlCl,17 are necessary. The MOVPE growth of high Al content AlGaN alloys and AlN, on the other side, is often complicated by parasitic gasphase chemical reactions between conventional precursors (Ga(CH3)3, Al(CH3)3, and NH3) that diminish the efficiency of Al incorporation and hamper control over the AlGaN alloy composition. Additional difficulties in the growth of high Al content AlGaN and AlN arise because of the high Al-N bond strength (2.88 eV) compared to that of Ga-N (1.93 eV).18 As a consequence, the probability of prereactions increases and adsorbed Al species possess a low surface mobility.19 To induce a step-flow 2D growth mode the surface diffusion length of the adsorbed Al species should be increased, as can be achieved by increasing the growth temperature. It is anticipated that elevated temperatures (>1100 °C) during AlN epitaxy promote better crystal quality of the grown material. In this paper, we present growth of AlN at a temperature of 1200 °C in a reactor based on the hot-wall MOCVD concept. In the context of the above introduction, exploration of the hotwall MOCVD concept for growth of AlN is inscribed in the general trends toward development of more sophisticated epitaxial growth technologies aiming to produce heterostructures of high complexity and better quality materials.
Crystal Growth & Design, Vol. 9, No. 2, 2009 881
Figure 2. Cross-sectional sketch of the hot-wall MOCVD assembly together with the upstream gas delivery quartz liners. The whole package is pushed inside a quartz tube. The process gases are forced through the opening of the susceptor. Energy is fed into the system from the RF inductive coil outside the quartz tube. The susceptor, made from high-purity graphite, is coated with either SiC or TaC to encapsulate the impurities in the graphite and avoid undesirable impurity doping during epitaxial growth. The substrate is heated by physical contact with the RF-heated susceptor and additional radiative heating from the susceptor surfaces. The susceptor design gives efficient heating and good temperature homogeneity. The temperature gradients in the growth zone, in both horizontal and vertical directions, are reduced, which provides an intrinsic advantage for uniformity of the material properties of the grown material. Growth Details. Initially, basic process investigations of the hotwall MOCVD growth of AlN were performed in a small-scale (wafer diameter of less than 30 mm) reactor. Currently, we are equipped with a horizontal large-scale hot-wall CVD reactor where the gas delivery system is modified to suit growth of AlN, GaN, and their derivatives. Ammonia (NH3), trimethylaluminum (TMAl or Al(CH3)3), and trimethylgallium (TMGa or Ga(CH3)3) are used as precursors for N, Al, and Ga, respectively. The reactor features flow separation of metal-organics and ammonia at the inlet flange and continuing by double tube liners inside the reactor. The main carrier flow and NH3 flow enter the growth chamber through the outer quartz liner, which is connected to the susceptor inlet by a graphite liner (Figure 2). A second carrier flow and the metal-organics enter through the inner quartz liner, which ends at a certain distance upstream from the susceptor inlet for reasons to be explained further in the text. Either the main or the second carrier flow can be either H2/N2 or a mixture of both (H2 + N2). The single-wafer load capacity of the reactor can be up to 4 in. The wafer is placed on a satellite whose rotation speed can be controlled separately by a distinct H2/N2 flow using the gas foil rotation principle22 in order to allow maximum uniformity of growth. In the following we will present results related to the performance of both reactors. Growth of AlN is optimized at a total reactor pressure of 100 mbar in H2 or N2 in the small-scale reactor and 50 mbar in a mixture of H2 and N2 in the large-scale reactor. SiC wafers, of either the 4H or the 6H polytype, 2 in. size and on-axis cut, are employed as substrates for growth. The substrates are heat treated at a temperature of 1250 °C or higher in H2 for 10 min prior to epitaxy. The temperature is then lowered to the AlN growth temperature of 1200 °C. The temperature is measured by a pyrometer pointing in a hole drilled in the roof of the susceptor. Characterization Details. The luminescence properties of the AlN epitaxial layers are examined by cathodoluminescence (CL) in situ in a Leo 1550 field-emission gun scanning electron microscope equipped with a MonoCL2 system (Oxford Res. Instr.). The CL measurements are done by employing an electron beam energy of 10 keV. The experiments are carried out at a temperature of 4.6 K.
2. Experimental Section General Concept. The hot-wall (MO)CVD technique has been proved as a worldwide dominating technique for SiC epitaxy. It was previously been introduced in the 1990s for growth of thick device quality SiC homoepitaxial layers at deposition temperatures between 1400 and 1650 °C.20,21 At these growth temperatures much of the thermal losses are in the form of thermal radiation. Graphite insulation is wrapped around the susceptor (Figure 1).
3. Results and Discussion Overall Performance of the Hot-Wall MOCVD System. In the growth of an exemplary AlxGa1-xN/GaN/AlN/SiC device heterostructure (Table 1) in the small-scale reactor, the TMAl flow during deposition of the AlxGa1-xN barrier is adjusted so that the Al molar fraction, x ) 0.20, in the gas phase, is
882 Crystal Growth & Design, Vol. 9, No. 2, 2009
Kakanakova-Georgieva et al.
Table 1. IRSE Best-Fit Values for the Thickness of the Individual Layers in the Al0.2Ga0.8N/GaN/AlN Heterostructurea thickness, nm
Al0.2Ga0.8Nb
GaNc
AlNd
24
3500
160
a
NH3 is delivered at 1 L/min with a N/Al ratio of 970 (AlN deposition) and N/Ga ratio of 1670 (GaN deposition). b The growth is done at 1150 °C for 1.5 min. c The growth is done at 1150 °C for 60 min. d The growth is done at 1200 °C for 10 min.
found to correspond to an Al content of 0.20 in the solid phase of AlxGa1-xN as derived from data analysis of the infrared spectroscopic ellipsometry (IRSE) response of the heterostructure. IRSE is a nondestructive characterization method used to determine the thickness and composition of individual epitaxial layers. Details of the ellipsometric data analysis can be found elsewhere.23,24 Under the applied growth conditions the AlN growth rate is estimated to be about 1 µm/h. The typical growth rate of GaN is 3-4 µm/h. Hall measurements at room temperature yielded an electron mobility of 1380 cm2/V s and a sheet carrier density of 6 × 1012 cm-2 in the two-dimensional electron gas formed at the Al0.20Ga0.80N/GaN interface (the thickness of the Al0.20Ga0.80N barrier being 24 nm), which is consistent with that theoretically calculated.25 The hot-wall MOCVD concept has also been applied to growth of AlGaN high electron mobility transistor heterostructures (HEMTs) on 2 in. SiC wafers in the large-scale reactor, yielding good properties of the two-dimensional electron gas including a sheet carrier density of 1.05 × 1013 cm-2 ( 4%, sheet resistance of 449 Ω ( 1%, and drift mobility of 1330 cm2/V s, which meet the requirements for fabrication of AlGaN HEMTs.26 The overall performance of the hot-wall MOCVD system demonstrates a versatile approach that can successfully support growth of device heterostructures with thin layers and controlled composition as well as transfer of the growth process on a large scale with excellent uniformity of the material properties. Highly Efficient AlN Epitaxy. In the following we will consider extension of the hot-wall MOCVD to growth of AlN, which is a demanding material, and important features of the growth, e.g., growth rate, and crystal quality are determined by the yet not very well understood chemistry of the processes that govern AlN deposition. In the Al(CH3)3-NH3 system development of parasitic gas-phase chemistry, connected to the initial (CH3)3Al:NH3 adduct formation upon mixing of the precursors and further propagation of less volatile adduct-derived species, may have a severe impact on epitaxial growth. It can be displayed in the typically low growth rate due to an uncontrolled loss of the Al species from the vapor. Previously, the ratio of the growth rate to the input group III molar flow rate has been introduced as a measure of growth efficiency expressed in units of µm/mol.27 For a system with no parasitic reactions leading to gas-phase depletion upstream of the substrate such growth efficiency should be in the vicinity of 104 µm/mol.27,28 As defined, this quantity suffers from not being dimensionless and ignoring the area over which epitaxial growth takes place. However, it is still a relevant measure if the major part of the gas flow passes over the wafer. Growth efficiency values on the order of several thousands of micrometers per mole have been obtained for the GaN and AlN layers grown in the small-scale hot-wall MOCVD reactor (Table 2). The values presented in Table 2 are taken with reference to Table 1 and ref 29. With N2 as a carrier gas and an Al/N ratio of ∼50 and 10, respectively, the thickness of the high crystal quality AlN layers grown in 15 min, as determined from the
Table 2. Growth Efficiency [µm/mol] As Defined in Ref 27 for Diversity of Epitaxy Conditions in the Small-Scale Hot-Wall MOCVD Reactor at a Pressure of 100 mbar and TMGa and TMAl Flow Rates of 17.5 and 15 µmol/min, Respectively type of growth carrier gas, material temperature, °C 3 L/min GaN AlN AlN AlN
1150 1200 1200 1200
H2 H2 N2 N2
V/III ratio 1670a 970a 48b 10c
growth growth efficiency, rate, µm/h µm/mol 3.5 1 1.3 1.9
3300 1100 1444 2166
a NH3 flow of 1 L/min. b NH3 flow of 50 mL/min.29 c NH3 flow of 10 mL/min.29
Figure 3. Deposition patterns on the susceptor cassette as affected by certain process parameters: (a) H2, 19 L/min, N2, no; T, 1200 °C; (b) H2, 25 L/min, N2, no; T, 1200 °C; (c) H2, 25 L/min; N2, 6 L/min; T, 1200 °C; (d) H2, 25 L/min; N2, 6 L/min; T, 1100 °C. Precursors are delivered at an Al/N ratio of 1000 with a NH3 flow rate of 2 L/min.
line shape analysis of the respective IRSE spectra, is 328 and 488 nm, yielding a growth rate of about 1.3 and up to 2 µm/ h.29 Obviously, the interaction of several growth parameters in the performance of the small-scale hot-wall MOCVD system has favored a less intensive parasitic gas-phase chemistry and smaller material losses: (i) operating at a low NH3 flow rate
Hot-Wall MOCVD
Figure 4. Thickness map of the AlN layer grown in 1 h on a stationary 2 in. SiC wafer as done by white-light reflectance measurements.26 Thickness mapping is performed on a 1 × 1 mm grid covering the whole wafer, which gives 1868 measurement points for 2 in. wafers; 69 of the total 1868 points are labeled on the map.
Figure 5. CL spectrum at 4.6 K taken on a spot of a AlN layer showing intense near band-gap UV emission at ∼208 nm.
known to shift the equilibrium Al(CH3)3 + NH3 T (CH3)3Al: NH3 toward adduct dissociation,30 (ii) operating at a low growth pressure known to reduce the concentration of precursors along with a reduction in residence time,28 (iii) early mixing of the precursors in a cool upstream region. While the first two points represent common measures to overcome the problem of the parasitic gas-phase reactions, the last point may seem to be in contradiction to the general practice of operating an III-nitride reactor where the flows of ammonia and metal-organics are kept separately and brought in contact only shortly before the deposition zone. We will elaborate more on this point further in the text. We also identified the direct impact of nitrogen on prevention of the premature gas-phase depletion in AlN MOCVD growth. Applying nitrogen instead of hydrogen as the carrier gas results in better growth efficiency.31 AlN Thickness Homogeneity Capacity in a Large Scale Hot-Wall MOCVD System. The same feature of mixing of the precursors in a cool upstream region has also been preserved when scaling up the growth process in the large-scale reactor. The inner quartz liner delivering the metal-organics to the deposition zone (Figure 2), which in the original assembly of the reactor ended inside the graphite liner, was replaced by a shorter version giving an opportunity to optimize the location
Crystal Growth & Design, Vol. 9, No. 2, 2009 883
Figure 6. Thickness map of a AlN layer grown on a rotating 2 in. SiC wafer as done by white-light reflectance measurements.26 Thickness mapping is performed on a 1 × 1 mm grid covering the whole wafer, which gives 1868 measurement points for 2 in. wafers, of which 69 are labeled on the map. Mean thickness: 0.52 µm. Max thickness: 0.56 µm. Min thickness: 0.50 µm. Standard deviation (σ): 0.007 µm. (σ/ mean) value: 1.3%. Precursors are delivered at an Al/N ratio of 1045 with a NH3 flow rate of 2 L/min.
of mixing of ammonia and metal-organics upstream from the susceptor inlet. We note that RF power inevitably dissipates not only at the susceptor but also at the graphite liner, producing a certain heat distribution upstream from the susceptor inlet where the temperature at the graphite liner can be in excess of 300 °C. This was taken into consideration together with two more facts: (i) TMAl decomposes upon heating in the presence of H2 at about 300 °C to form solid Al4C3,32 which obviously is to be avoided, and (ii) theoretical and experimental evidence has previously pointed out that mixing of the ammonia and metal-organics precursors in a cool upstream region contributes to suppress formation of excited adducts that can easily climb up the energy barrier to the oligomer state33 and initiate further high-order oligomers as unwanted nucleation centers,34 making them inaccessible to growth. Thus, after the initial evaluation of the performance of the system, further development of the growth process in the hot-wall MOCVD system has led to flows of NH3 and Al(CH3)3 being mixed over a certain distance (tens of centimeters) upstream of the susceptor inlet and their joint delivery to the deposition area. The arrangement of the turns of the RF induction coil was done to address the issue of reduction of heat load into the susceptor inlet together with the graphite liner in order to minimize undesirable deposition upstream of the substrate. Reduction of heat load into the susceptor inlet and, consequently, controlling the depletion of the precursors inside the reactor is additionally helped by tuning the total flow of the H2 carrier gas and adding N2 to the carrier flow. The impact of these growth parameters on the deposition profile can be followed in Figure 3 through pictures a-c. Delay in the longitudinal depletion due to consumption of limiting species is achieved by increasing of the total H2 flow and adding N2 to the carrier flow, which is demonstrated by the expansion of the deposition patterns on the susceptor further downstream. To support the understanding that reduction of heat load into the susceptor inlet can essentially improve penetration of reaction species along the susceptor length, a growth run has been performed at lower temperature, 1100 instead of 1200 °C, and the resulting deposition pattern is presented in Figure
884 Crystal Growth & Design, Vol. 9, No. 2, 2009
3d. Except for the lower temperature, all other growth parameters, including the flows of H2 and N2, are the same as for the deposition pattern in Figure 3c. The AlN deposition pattern on a stationary 2 in. SiC wafer is shown in Figure 4. The typical low-temperature CL spectrum of the AlN epitaxial material, as measured in several spots over the wafer, is strongly dominated by the intense near-band-gap deep UV emission at about 208 nm. The intensity of other often observed defect-related (and most probably carbon- and oxygenassisted35) emission bands, especially the one at ∼280 nm, is considerably quenched (Figure 5). Mixing of the precursors upstream from the deposition area facilitates uniform epitaxial growth, while the greater uniformity of substrate temperature inherent to the hot-wall reactor and rotation of the wafer are of fundamental importance for layer growth uniformity (Figure 6). It is confirmed that the AlN layer thickness can be controlled with an accuracy of (1.3%. In this particular case an AlN layer of sufficient thickness (0.5 µm) to be applicable to fabricating multiple thin layer AlN-based device heterostructures was grown.
5. Summary We demonstrated successful growth of AlN at a temperature of 1200 °C in a set of hot-wall MOCVD systems with the possibility of straightforward scaling up the process on larger wafer areas to meet the demand of device technologies. We outlined several aspects of the carefully optimized design and process parameters with relevance to achievement of high overall growth rate, efficiency, and uniformity, which to a great extent depend on how consumption of growth-limiting species by gas-phase adduct formation can actively be prevented. Acknowledgment. The Swedish Research Council (VR) and Knut and Alice Wallenberg Foundation (KAWS) are gratefully acknowledged. We appreciate the collaboration with V. Desmaris, N. Rorsman, and H. Zirath at Chalmers University of Technology, Gothenburg, Sweden, on Hall measurements and device processing and with A. Kasic on the IRSE measurements.
References (1) Duboz, J.-Y. Phys. Stat. Sol. (a) 1999, 176, 5–14. (2) Khan, M. A.; Shatalov, M.; Maruska, H. P.; Wang, H. M.; Kuoktis, E. Jpn. J. Appl. Phys. 2005, 44, 7191–7206. (3) Chen, Z.; Qhalid Fareed, R. S.; Gaevski, M.; Adivarahan, V.; Yang, J. W.; Khan, M. A.; Mei, J.; Ponce, F. A. Appl. Phys. Lett. 2006, 89, 081905-1081905-3. (4) Mei, J.; Ponce, F. A.; Qhalid Fareed, R. S.; Yang, J. W.; Khan, M. A. Appl. Phys. Lett. 2007, 90, 221909-1221909-3. (5) Heikman, S.; Keller, S.; Newman, S.; Wu, Y.; Moe, C.; Moran, B.; Schmidt, M.; Mishra, U. K.; Speck, J. S.; Denbaars, S. P. Jpn. J. Appl. Phys. 2005, 44, L405–L407.
Kakanakova-Georgieva et al. (6) Molnar, R. J.; Go¨tz, W.; Romano, L. T.; Johnson, N. M. J. Cryst. Growth 1997, 178, 147–156. (7) Dupuis, R. D. J. Cryst. Growth 1997, 178, 56–73. (8) Lourdudoss, S.; Kjebon, O. IEEE J. Selected Top. Quantum Electron. 1997, 3, 749–767. (9) Ambacher, O. J. Phys. D: Appl. Phys. 1998, 31, 2653–2710. (10) Ban, V. S.; Rodefeld, D.; Flemish, J. R.; Jones, K. A. Appl. Phys. Lett. 1993, 62, 160–162. (11) Solomon, G. Compd. Semiconductors 2006, 12, 23–25. (12) Buchan, N. I.; Kuech, T. F.; Tischler, M. A.; Scilla, G.; Cardone, F.; Potemski, R. J. Electrochem. Soc. 1991, 138, 2789–2794. (13) Kryliouk, O.; Reed, M.; Dann, T.; Anderson, T.; Chai, B. Mater. Sci. Eng., B 1999, 59, 6–11. (14) Fareed, Q.; Adivarahan, V.; Gaevski, M.; Katona, T.; Mei, J.; Ponce, F. A.; Khan, M. A. Jpn. J. Appl. Phys. 2007, 46, L752–L754. (15) Yim, W. M.; Stofko, E. J.; Zanzucchi, P. J.; Pankove, J. I.; Ettenberg, M.; Gilbert, S. L. J. Appl. Phys. 1973, 44, 292–296. (16) Kovalenkov, O.; Soukhoveev, V.; Ivantsov, V.; Usikov, A.; Dmitriev, V. J. Cryst. Growth 2005, 281, 87–92. (17) Kumagai, Y.; Yamane, T.; Miyaji, T.; Murakami, H.; Kangawa, Y.; Koukitu, A. Phys. Stat. Sol. (c) 2003, 0, 2498–2501. (18) Keller, S.; Parish, G.; Fini, P. T.; Heikman, S.; Chen, C.-H.; Zhang, N.; DenBaars, S. P.; Mishra, U. K.; Wu, Y.-F. J. Appl. Phys. 1999, 86, 5850–5857. (19) Keller, S.; DenBaars, S. P. J. Cryst. Growth 2003, 248, 479–486. (20) Kordina, O.; Hallin, C.; Glass, R. C.; Henry, A.; Janze´n, E. Inst. Phys. Conf. Ser. 1994, 137, 41. (21) Kordina, O.; Hallin, C.; Henry, A.; Bergman, J. P.; Ivanov, I.; Ellison, A.; Son, N. T.; Janze´n, E. Phys. Stat. Sol. (b) 1997, 202, 321–334. (22) Frijlink, P. M. J. Cryst. Growth 1988, 93, 207–215. (23) Kasic, A.; Schubert, M.; Einfeldt, S.; Hommel, D.; Tiwald, T. E. Phys. ReV. B 2000, 62, 7365–7377. (24) Kasic, A.; Schubert, M.; Off, J.; Scholz, F. Appl. Phys. Lett. 2001, 78, 1526–1528. (25) Ambacher, O.; Smart, J.; Shealy, J. R.; Weimann, N. G.; Chu, K.; Murphy, M.; Schaff, W. J.; Eastman, L. F.; Dimitrov, R.; Wittmer, L.; Stutzmann, M.; Rieger, W.; Hilsenbeck, J. J. Appl. Phys. 1999, 85, 3222–3233. (26) Kakanakova-Georgieva, A.; Forsberg, U.; Ivanov, I. G.; Janze´n, E. J. Cryst. Growth 2007, 300, 100–103. (27) Stringfellow, G. B. Organometallic Vapor-Phase Epitaxy: Theory and Practice; Academic Press, Inc.: New York, 1989; p 10. (28) Chen, C. H.; Liu, H.; Steigerwald, D.; Imler, W.; Kuo, C. P.; Craford, M. G.; Ludowise, M.; Lester, S.; Amano, J. J. Electron. Mater. 1996, 25, 1004–1008. (29) Kakanakova-Georgieva, A.; Kasic, A.; Hallin, C.; Monemar, B.; Janze´n, E. Phys. Stat. Sol. (c) 2005, 2, 960–963. (30) Creighton, J. R.; Wang, G. T. J. Phys. Chem. A 2005, 109, 133–137. (31) Kakanakova-Georgieva, A.; Gueorguiev, G. K.; Stafstro¨m, S.; Hultman, L.; Janze´n, E. Chem. Phys. Lett. 2006, 431, 346–351. (32) Morita, M.; Uesugi, N.; Isogai, S.; Tsubouchi, K.; Mikoshiba, N. Jpn. J. Appl. Phys. 1981, 20, 17–23. (33) Matsumoto, K.; Tachibana, A. J. Cryst. Growth 2004, 272, 360–369. (34) Creighton, J. R.; Breiland, W. G.; Coltrin, M. E.; Pawlowski, R. P. Appl. Phys. Lett. 2002, 81, 2626–2628. (35) Freitas, J. A., Jr. J. Cryst. Growth 2005, 281, 168–182.
CG8005663