- Email: [email protected]
Crystal growth of gallium nitride and manganese nitride using an high pressure thermal gradient process
Solid-State Electronics 47 (2003) 1027–1030 www.elsevier.com/locate/sse
Crystal growth of gallium nitride and manganese nitride using an high pressure thermal gradient process F. Kelly, D.R. Gilbert, R. Chodelka, R.K. Singh *, S. Pearton Department of Materials Science and Engineering, University of Florida, P.O. Box 116400, Gainesville, FL 32611, USA Received 3 September 2002; received in revised form 30 October 2002; accepted 7 November 2002
Abstract Standard, semiconductor-industry bulk crystal growth processes are virtually impossible for the production of GaN as this is prohibited by both the high melt temperature of GaN and thermal decomposition of the compound into Ga metal and diatomic nitrogen gas. In this study, a novel hydrostatic pressure system was employed to grow GaN crystals in a very high pressure ambient. The ultra-high pressure, high temperature process uses a solid-phase nitrogen source to form GaN crystals in a metal alloy melt. Using a thermal gradient diffusion process, in which nitrogen dissolves in the high temperature region of the metal melt and diffuses to the lower temperature, lower solubility region, high quality crystals up to 0.5 mm in size were formed, as determined by scanning electron microscopy, X-ray diffraction, and micro-Raman analysis. Ó 2003 Published by Elsevier Science Ltd.
1. Introduction GaN is an important engineering material that has generated a great deal of industrial interest because of its optical and electronic properties. GaN has a wide, direct bandgap, and a high combined figure of merit for high temperature/high power/high frequency device applications [1]. Consistent growth of high quality, large area GaN substrates is a critical issue for the development of next-generation GaN-based optoelectronic and microwave devices. However, due to its extremely high melting temperature (2800 K) and the very high equilibrium nitrogen pressure (45 kbar) associated with GaN at its melting temperature [2], bulk growth of GaN singlecrystals is using the standard Czochralski or Bridgman growth methods used for most other semiconductors using stoichiometric melts is practically impossible. Therefore, GaN crystals must be grown by other methods that utilize lower temperatures.
*
Corresponding author. Tel.: +1-352-392-1032; fax: +1-352846-0326. E-mail address: [email protected]fl.edu (R.K. Singh).
Two techniques may be regarded as the most promising to produce GaN bulk crystals and substrates: high gas pressure crystal growth using nitrogen over high temperature Ga metal [3–8], and chloride–hydride vapor phase epitaxy (in combination with lateral regrowth) [9– 12]. The high pressure gas process has produced large, free-standing crystals (approximately 0:6 15 15 mm) with relatively low dislocation densities (as low as 101 – 102 cm2 [8]), but carries with it a need for highly specialized equipment for handling the dangerous, high gas pressure environment required for crystal growth. Vapor-phase methods usually result in relatively large defect densities due to the difference in the lattice parameters and thermal expansion coefficients between the GaN film and the heteroepitaxial substrates used in the process. Dislocation densities for the state of the art GaN crystals grown by vapor phase techniques are in the range of 108 –1010 cm2 , which is approximately 4–6 orders of magnitude higher than that obtained in other optoelectronic materials such as GaAs, InP, etc. In this manuscript, we present a novel process for the synthesis of GaN crystals using high mechanical pressures and high temperatures. A solid pellet of pressed GaN was used as a nitrogen source, while new GaN crystals were formed in a Ga0:7 Mn0:3 melt. This process differs
0038-1101/03/$ - see front matter Ó 2003 Published by Elsevier Science Ltd. doi:10.1016/S0038-1101(02)00470-7
1028
F. Kelly et al. / Solid-State Electronics 47 (2003) 1027–1030
significantly from other published work in that the system contains virtually no free volume for an ambient gas to occupy.
2. Experimental Crystals were synthesized in an ultra-high pressure (UHP) crystal growth system. The apparatus used in this work is based on a split-sphere hydraulic system, a cutaway diagram of the system is shown in Fig. 1. Hemispherical hydraulic membranes apply pressure to a set of eight large steel dies, which in turn compress a smaller set of six cemented tungsten carbide dies, finally applying an amplified pressure on a cubic reaction cell located at the center of the spherical region. This pressure amplification results from the successive decrease in contact surface area in transferring pressure from the larger set of dies to the smaller, then ultimately, to the core cell. At pressures exceeding 30 kbar, the tri-axial compressive stress applied to the cell by the tungsten carbide dies (labeled ‘‘small dies’’ in Fig. 1) effectively seals to core cell volume to the point where gasses cannot escape, and thus produces a hydrostatic pressure inside the cell volume under which crystal growth proceeds in the metal alloy melt. This system is capable of applying up to 65 kbar mechanical compression to a reaction cell while simultaneously heating the cell to greater than 1500 °C. For this work, the cell contained pressed GaN powder, which acted as a source of nitrogen upon heating, and a metal alloy consisting of Ga (70 at.%) and Mn (30 at.%) as reactants. The Mn was added in order to increase the solubility of nitrogen in the melt above that which can be achieved using pure Ga alone. Crystal growth proceeded by first applying compressive mechanical loading of the reactor core to the desired pressure level, after which the cell temperature was ramped up to the desired
Fig. 1. Schematic cutaway diagram of UHP furnace
value by applying an electrical current to the graphite heater inside the core cell. A calibration of the heater power vs. equilibrium temperature achieved was done on the system and, using this information, time-dependent heater power profiles were preprogrammed and controlled by computer to maintain the desired temperature. After processing, the reaction cell was extracted from the reactor, opened, and the remaining unreacted metal dissolved in an HNO3 /HCl acid solution, leaving the GaN crystals available for collection and analysis. Crystal morphology was evaluated using scanning electron microscopy (SEM), while structural characteristics were evaluated using X-ray diffraction (XRD) and micro-Raman spectroscopy. Compositional analysis was performed by energy dispersive X-ray spectroscopy (EDS), which was performed in situ with the SEM.
3. Results and discussion Multiple GaN crystals were formed during the course of each experiment as a result of spontaneous nucleation in the Ga metal melt under the conditions used. SEM micrographs of crystals synthesized in this UHP process are shown in Fig. 2. These crystals exhibit plate-like morphologies suggestive of a hexagonal structure and are 600–800 lm in size. Small (10 lm) triangular growth features visible on the faces of the crystals displayed preferential orientation with respect to each other, suggesting either single crystal or highly textured growth. Dendritic and needle-like crystals were also commonly observed after processing. EDS semi-quantitative analysis indicated that crystals formed in this process tended to be nitrogen deficient (N/Ga < 1) and also showed the presence of varying amounts of Mn from various selected crystals. AES performed on randomly selected crystals showed small concentrations of Mn and O on the surface that disappeared after sputtering for 180 s, suggesting that the small triangular surface features might be composed of a Mn-containing oxide phase which precipitated out of the melt after the GaN crystals were formed and grew to their observed size. Structural characterization of the crystals was accomplished using XRD and micro-Raman spectroscopy. XRD was performed using a Rigaku Cu rotating anode target, a singly bent (vertical focusing) LiF monochromator, and a four-circle Huber diffractometer (standard 2h, h, v, and u circles). The XRD scan (Fig. 3) shows the sample to be composed of single-crystal hexagonal GaN with the broad face oriented parallel to (0 0 0 1) and another phase which is most likely Mn3 N2 exhibiting the (1 0) and (2 0 0) reflections. The measured lattice pa and c ¼ 5:186 A , rameters for the GaN were: a ¼ 3:19 A
F. Kelly et al. / Solid-State Electronics 47 (2003) 1027–1030
1029
Fig. 2. SEM micrographs showing typical GaN crystals formed in UHP thermal gradient process and detail of the small triangular surface features.
Fig. 3. XRD spectrum showing the GaN (0 0 0 2), (0 0 0 4) and (1 1 2 2) and the Mn3 N2 (1 0 3) and (2 0 6) reflections.
which are in good agreement with published values for GaN. The quality of the GaN crystal was such that the rocking curve full-width-at-half-maximum (FWHM) was no wider than the resolution limit of the spectrometer (10800 ), indicating a low defect density crystal. Fig. 4 shows a u-scan of the crystal which illustrates that the (1 0–1 0) planes are spaced at regular 60° increments, further verifying the hexagonal structure of the GaN. A micro-Raman spectrum taken from the same crystal is shown in Fig. 5. The Raman spectrum was taken in a backscattering mode using the 514 nm line of an Ar-ion laser at a spectral resolution of 1 cm1 . This spectrum shows the characteristic A1 (TO), A1 (LO), E1 (TO) and E2 modes of the hexagonal GaN structure at 528, 730, 554 and 565 cm1 , respectively. The A1 (LO) and E2 modes are the theoretically allowed modes in the backscattering geometry. Presence of the weak A1 (TO) mode in the spectrum may be indicative of an imperfect backscattering alignment during spectral acquisition.
Fig. 4. XRD u-scan of the GaN crystal showing the 60° spacing of the {1 0–1 0} reflections.
Fig. 5. Micro-Raman spectrum of UHP grown crystal showing characteristic GaN spectral features.
The observed crystal growth proceeded from spontaneously formed nuclei in the Ga:07 Mn0:3 alloy melt. Atomic nitrogen was provided in the melt through the
1030
F. Kelly et al. / Solid-State Electronics 47 (2003) 1027–1030
thermal dissolution/decomposition of the GaN source pellet at the operating pressure of the system, via the following process:
region, such as residual oxide powder crystals from the other parts of the core cell.
GaNðsÞ $ GaðmeltÞ þ NðmeltÞ
4. Conclusions
ð1Þ
where GaðmeltÞ and NðmeltÞ indicate gallium and nitrogen dissolved in the melt region. The reaction cell was designed to maintain a temperature gradient across the Ga melt region adjacent to the source pellet, keeping the GaN source in the highest temperature zone, thus allowing the dissolved nitrogen to diffuse to the lower temperature (lower solubility) region of the melt where it then reacted to form new GaN. The crystal growth rate is controlled by the flux of the dissolved nitrogen in the Ga melt in accordance with the expression: JN /
Crystalline material was grown under high pressure in a specially designed UHP, high temperature furnace. The material, grown from a solid-phase nitrogen source and a metal alloy melt was found by micro-Raman analysis and XRD to be composed primarily of good crystalline quality GaN and individual crystals produced ranged in size up to 800 lm. XRD and EDS analysis also showed a small quantities of a second phase which is most likely Mn3 N2 . Acknowledgements
oðcN ðx; T ÞÞ ; ox
ð2Þ
where JN is the flux of dissolved nitrogen, and the righthand term is the temperature dependent concentration gradient of the dissolved nitrogen. Nitrogen readily exceeded the supersaturation concentration necessary for spontaneous, homogeneous nucleation in these experiments. This is due to the very low solubility of nitrogen in Ga metal, which is only 0.1% at a temperature of 1600 K and 32 kbar pressure [1]. Increasing the pressure and the temperature of the melt results in increased nitrogen solubility for faster growth due to the increased flux. The addition of Mn to the melt also increases the overall solubility in the melt, as nitrogen dissolves more readily in Mn than in Ga and the overall solubility of the melt is then a linear admixture of the solubilities of the components, weighted by their individual concentrations in the melt as in the expression:
SðGa0:7Mn0:3Þ ¼ 0:7 SðGaÞ þ 0:3 SðMnÞ
ð3Þ
But as we have shown, Mn alloying can also lead to the formation of unwanted phases upon cooling, but fortunately the crystals recovered showed good GaN growth despite this. Under the experimental conditions used, we expect the otherwise thermodynamically favored reaction: NðmeltÞ þ NðmeltÞ $ N2ðgÞ
ð4Þ
to be suppressed inside the reaction zone because of the large activation barrier associated with the high applied mechanical pressure and lack of free volume. By carefully controlling the nitrogen solubility via the temperature gradient in and composition of the melt, as well as providing a pre-existing seed crystal in the lowest temperature region to suppress homogeneous nucleation, we expect that the only obstacle in the way of realizing seed growth of GaN is the homogeneous nucleation of crystals in the melt on solid-phase impurities in the melt
We would like to acknowledge the valuable assistance of S. MacDonald of the University of Florida for the acquisition of micro-Raman spectral data and J. Budai at Oak Ridge National Laboratory, Oak Ridge, TN for acquisition of XRD data. We would also like to thank Dr. R. Abbaschian of the University of Florida, and Drs. A. Novikov and N. Patrin of the Gemesis Corporation for all of their technical expertise in high pressure systems. Working in collaboration with the University of Florida, The Gemesis Corporation conducted its work under SBIR grant no. N00014-99-M0151, which was monitored by Colin Wood of the Office of Naval Research. References [1] Kucheyev SO, Williams JS, Pearton SJ. Mater Sci Eng 2001;33:51–107. [2] Porowski S, Grzegory I. In: Edgar J, editor. Properties of group III nitrides. INSPEC, the Institution of Electrical Engineers; 1994. p. 80. [3] Porowski S, Jun J, Perliu P, Gregory I, Teissere H, Suski T. Inst Phys Conf, Ser 1994;137:369. [4] Grzegory I et al. Physica B 1993;185:99. [5] Porowski S. J Cryst Growth 1996;166:583. [6] Grzegory I et al. MRS Internet J Nitride Semicond Res 1996;1:20. [7] Porowski S, Grzegory I. J Cryst Growth 1997;178:174. [8] Porowski S. MRS Internet J Nitride Semicond Res 1999;4S1:G1.3. [9] Usui A, Sunakawa H, Sakia A, Yamaguchi A. Jpn J Appl Phys 1997;36:1899. [10] Melnik Y, Vassilevski K, Nikitina I, Babanin A, Davydov V, Dmitriev V. MRS Internet J Nitride Semicond Res 1997;2:39. [11] Shibata T, Sone H, Yahashi K, Yamaguchi M, Hiramatsu K, Sawaki N, et al. J Cryst Growth 1998;189/190:67. [12] Zhang R, Zhang L, Hansen D, Boleslawski M, Chen K, Lu D, et al. MRS Internet J Nitride Semicond Res 1999; 4S1:G4.7.
Crystal growth of gallium nitride and manganese nitride using an high pressure thermal gradient process F. Kelly, D.R. Gilbert, R. Chodelka, R.K. Singh *, S. Pearton Department of Materials Science and Engineering, University of Florida, P.O. Box 116400, Gainesville, FL 32611, USA Received 3 September 2002; received in revised form 30 October 2002; accepted 7 November 2002
Abstract Standard, semiconductor-industry bulk crystal growth processes are virtually impossible for the production of GaN as this is prohibited by both the high melt temperature of GaN and thermal decomposition of the compound into Ga metal and diatomic nitrogen gas. In this study, a novel hydrostatic pressure system was employed to grow GaN crystals in a very high pressure ambient. The ultra-high pressure, high temperature process uses a solid-phase nitrogen source to form GaN crystals in a metal alloy melt. Using a thermal gradient diffusion process, in which nitrogen dissolves in the high temperature region of the metal melt and diffuses to the lower temperature, lower solubility region, high quality crystals up to 0.5 mm in size were formed, as determined by scanning electron microscopy, X-ray diffraction, and micro-Raman analysis. Ó 2003 Published by Elsevier Science Ltd.
1. Introduction GaN is an important engineering material that has generated a great deal of industrial interest because of its optical and electronic properties. GaN has a wide, direct bandgap, and a high combined figure of merit for high temperature/high power/high frequency device applications [1]. Consistent growth of high quality, large area GaN substrates is a critical issue for the development of next-generation GaN-based optoelectronic and microwave devices. However, due to its extremely high melting temperature (2800 K) and the very high equilibrium nitrogen pressure (45 kbar) associated with GaN at its melting temperature [2], bulk growth of GaN singlecrystals is using the standard Czochralski or Bridgman growth methods used for most other semiconductors using stoichiometric melts is practically impossible. Therefore, GaN crystals must be grown by other methods that utilize lower temperatures.
*
Corresponding author. Tel.: +1-352-392-1032; fax: +1-352846-0326. E-mail address: [email protected]fl.edu (R.K. Singh).
Two techniques may be regarded as the most promising to produce GaN bulk crystals and substrates: high gas pressure crystal growth using nitrogen over high temperature Ga metal [3–8], and chloride–hydride vapor phase epitaxy (in combination with lateral regrowth) [9– 12]. The high pressure gas process has produced large, free-standing crystals (approximately 0:6 15 15 mm) with relatively low dislocation densities (as low as 101 – 102 cm2 [8]), but carries with it a need for highly specialized equipment for handling the dangerous, high gas pressure environment required for crystal growth. Vapor-phase methods usually result in relatively large defect densities due to the difference in the lattice parameters and thermal expansion coefficients between the GaN film and the heteroepitaxial substrates used in the process. Dislocation densities for the state of the art GaN crystals grown by vapor phase techniques are in the range of 108 –1010 cm2 , which is approximately 4–6 orders of magnitude higher than that obtained in other optoelectronic materials such as GaAs, InP, etc. In this manuscript, we present a novel process for the synthesis of GaN crystals using high mechanical pressures and high temperatures. A solid pellet of pressed GaN was used as a nitrogen source, while new GaN crystals were formed in a Ga0:7 Mn0:3 melt. This process differs
0038-1101/03/$ - see front matter Ó 2003 Published by Elsevier Science Ltd. doi:10.1016/S0038-1101(02)00470-7
1028
F. Kelly et al. / Solid-State Electronics 47 (2003) 1027–1030
significantly from other published work in that the system contains virtually no free volume for an ambient gas to occupy.
2. Experimental Crystals were synthesized in an ultra-high pressure (UHP) crystal growth system. The apparatus used in this work is based on a split-sphere hydraulic system, a cutaway diagram of the system is shown in Fig. 1. Hemispherical hydraulic membranes apply pressure to a set of eight large steel dies, which in turn compress a smaller set of six cemented tungsten carbide dies, finally applying an amplified pressure on a cubic reaction cell located at the center of the spherical region. This pressure amplification results from the successive decrease in contact surface area in transferring pressure from the larger set of dies to the smaller, then ultimately, to the core cell. At pressures exceeding 30 kbar, the tri-axial compressive stress applied to the cell by the tungsten carbide dies (labeled ‘‘small dies’’ in Fig. 1) effectively seals to core cell volume to the point where gasses cannot escape, and thus produces a hydrostatic pressure inside the cell volume under which crystal growth proceeds in the metal alloy melt. This system is capable of applying up to 65 kbar mechanical compression to a reaction cell while simultaneously heating the cell to greater than 1500 °C. For this work, the cell contained pressed GaN powder, which acted as a source of nitrogen upon heating, and a metal alloy consisting of Ga (70 at.%) and Mn (30 at.%) as reactants. The Mn was added in order to increase the solubility of nitrogen in the melt above that which can be achieved using pure Ga alone. Crystal growth proceeded by first applying compressive mechanical loading of the reactor core to the desired pressure level, after which the cell temperature was ramped up to the desired
Fig. 1. Schematic cutaway diagram of UHP furnace
value by applying an electrical current to the graphite heater inside the core cell. A calibration of the heater power vs. equilibrium temperature achieved was done on the system and, using this information, time-dependent heater power profiles were preprogrammed and controlled by computer to maintain the desired temperature. After processing, the reaction cell was extracted from the reactor, opened, and the remaining unreacted metal dissolved in an HNO3 /HCl acid solution, leaving the GaN crystals available for collection and analysis. Crystal morphology was evaluated using scanning electron microscopy (SEM), while structural characteristics were evaluated using X-ray diffraction (XRD) and micro-Raman spectroscopy. Compositional analysis was performed by energy dispersive X-ray spectroscopy (EDS), which was performed in situ with the SEM.
3. Results and discussion Multiple GaN crystals were formed during the course of each experiment as a result of spontaneous nucleation in the Ga metal melt under the conditions used. SEM micrographs of crystals synthesized in this UHP process are shown in Fig. 2. These crystals exhibit plate-like morphologies suggestive of a hexagonal structure and are 600–800 lm in size. Small (10 lm) triangular growth features visible on the faces of the crystals displayed preferential orientation with respect to each other, suggesting either single crystal or highly textured growth. Dendritic and needle-like crystals were also commonly observed after processing. EDS semi-quantitative analysis indicated that crystals formed in this process tended to be nitrogen deficient (N/Ga < 1) and also showed the presence of varying amounts of Mn from various selected crystals. AES performed on randomly selected crystals showed small concentrations of Mn and O on the surface that disappeared after sputtering for 180 s, suggesting that the small triangular surface features might be composed of a Mn-containing oxide phase which precipitated out of the melt after the GaN crystals were formed and grew to their observed size. Structural characterization of the crystals was accomplished using XRD and micro-Raman spectroscopy. XRD was performed using a Rigaku Cu rotating anode target, a singly bent (vertical focusing) LiF monochromator, and a four-circle Huber diffractometer (standard 2h, h, v, and u circles). The XRD scan (Fig. 3) shows the sample to be composed of single-crystal hexagonal GaN with the broad face oriented parallel to (0 0 0 1) and another phase which is most likely Mn3 N2 exhibiting the (1 0) and (2 0 0) reflections. The measured lattice pa and c ¼ 5:186 A , rameters for the GaN were: a ¼ 3:19 A
F. Kelly et al. / Solid-State Electronics 47 (2003) 1027–1030
1029
Fig. 2. SEM micrographs showing typical GaN crystals formed in UHP thermal gradient process and detail of the small triangular surface features.
Fig. 3. XRD spectrum showing the GaN (0 0 0 2), (0 0 0 4) and (1 1 2 2) and the Mn3 N2 (1 0 3) and (2 0 6) reflections.
which are in good agreement with published values for GaN. The quality of the GaN crystal was such that the rocking curve full-width-at-half-maximum (FWHM) was no wider than the resolution limit of the spectrometer (10800 ), indicating a low defect density crystal. Fig. 4 shows a u-scan of the crystal which illustrates that the (1 0–1 0) planes are spaced at regular 60° increments, further verifying the hexagonal structure of the GaN. A micro-Raman spectrum taken from the same crystal is shown in Fig. 5. The Raman spectrum was taken in a backscattering mode using the 514 nm line of an Ar-ion laser at a spectral resolution of 1 cm1 . This spectrum shows the characteristic A1 (TO), A1 (LO), E1 (TO) and E2 modes of the hexagonal GaN structure at 528, 730, 554 and 565 cm1 , respectively. The A1 (LO) and E2 modes are the theoretically allowed modes in the backscattering geometry. Presence of the weak A1 (TO) mode in the spectrum may be indicative of an imperfect backscattering alignment during spectral acquisition.
Fig. 4. XRD u-scan of the GaN crystal showing the 60° spacing of the {1 0–1 0} reflections.
Fig. 5. Micro-Raman spectrum of UHP grown crystal showing characteristic GaN spectral features.
The observed crystal growth proceeded from spontaneously formed nuclei in the Ga:07 Mn0:3 alloy melt. Atomic nitrogen was provided in the melt through the
1030
F. Kelly et al. / Solid-State Electronics 47 (2003) 1027–1030
thermal dissolution/decomposition of the GaN source pellet at the operating pressure of the system, via the following process:
region, such as residual oxide powder crystals from the other parts of the core cell.
GaNðsÞ $ GaðmeltÞ þ NðmeltÞ
4. Conclusions
ð1Þ
where GaðmeltÞ and NðmeltÞ indicate gallium and nitrogen dissolved in the melt region. The reaction cell was designed to maintain a temperature gradient across the Ga melt region adjacent to the source pellet, keeping the GaN source in the highest temperature zone, thus allowing the dissolved nitrogen to diffuse to the lower temperature (lower solubility) region of the melt where it then reacted to form new GaN. The crystal growth rate is controlled by the flux of the dissolved nitrogen in the Ga melt in accordance with the expression: JN /
Crystalline material was grown under high pressure in a specially designed UHP, high temperature furnace. The material, grown from a solid-phase nitrogen source and a metal alloy melt was found by micro-Raman analysis and XRD to be composed primarily of good crystalline quality GaN and individual crystals produced ranged in size up to 800 lm. XRD and EDS analysis also showed a small quantities of a second phase which is most likely Mn3 N2 . Acknowledgements
oðcN ðx; T ÞÞ ; ox
ð2Þ
where JN is the flux of dissolved nitrogen, and the righthand term is the temperature dependent concentration gradient of the dissolved nitrogen. Nitrogen readily exceeded the supersaturation concentration necessary for spontaneous, homogeneous nucleation in these experiments. This is due to the very low solubility of nitrogen in Ga metal, which is only 0.1% at a temperature of 1600 K and 32 kbar pressure [1]. Increasing the pressure and the temperature of the melt results in increased nitrogen solubility for faster growth due to the increased flux. The addition of Mn to the melt also increases the overall solubility in the melt, as nitrogen dissolves more readily in Mn than in Ga and the overall solubility of the melt is then a linear admixture of the solubilities of the components, weighted by their individual concentrations in the melt as in the expression:
SðGa0:7Mn0:3Þ ¼ 0:7 SðGaÞ þ 0:3 SðMnÞ
ð3Þ
But as we have shown, Mn alloying can also lead to the formation of unwanted phases upon cooling, but fortunately the crystals recovered showed good GaN growth despite this. Under the experimental conditions used, we expect the otherwise thermodynamically favored reaction: NðmeltÞ þ NðmeltÞ $ N2ðgÞ
ð4Þ
to be suppressed inside the reaction zone because of the large activation barrier associated with the high applied mechanical pressure and lack of free volume. By carefully controlling the nitrogen solubility via the temperature gradient in and composition of the melt, as well as providing a pre-existing seed crystal in the lowest temperature region to suppress homogeneous nucleation, we expect that the only obstacle in the way of realizing seed growth of GaN is the homogeneous nucleation of crystals in the melt on solid-phase impurities in the melt
We would like to acknowledge the valuable assistance of S. MacDonald of the University of Florida for the acquisition of micro-Raman spectral data and J. Budai at Oak Ridge National Laboratory, Oak Ridge, TN for acquisition of XRD data. We would also like to thank Dr. R. Abbaschian of the University of Florida, and Drs. A. Novikov and N. Patrin of the Gemesis Corporation for all of their technical expertise in high pressure systems. Working in collaboration with the University of Florida, The Gemesis Corporation conducted its work under SBIR grant no. N00014-99-M0151, which was monitored by Colin Wood of the Office of Naval Research. References [1] Kucheyev SO, Williams JS, Pearton SJ. Mater Sci Eng 2001;33:51–107. [2] Porowski S, Grzegory I. In: Edgar J, editor. Properties of group III nitrides. INSPEC, the Institution of Electrical Engineers; 1994. p. 80. [3] Porowski S, Jun J, Perliu P, Gregory I, Teissere H, Suski T. Inst Phys Conf, Ser 1994;137:369. [4] Grzegory I et al. Physica B 1993;185:99. [5] Porowski S. J Cryst Growth 1996;166:583. [6] Grzegory I et al. MRS Internet J Nitride Semicond Res 1996;1:20. [7] Porowski S, Grzegory I. J Cryst Growth 1997;178:174. [8] Porowski S. MRS Internet J Nitride Semicond Res 1999;4S1:G1.3. [9] Usui A, Sunakawa H, Sakia A, Yamaguchi A. Jpn J Appl Phys 1997;36:1899. [10] Melnik Y, Vassilevski K, Nikitina I, Babanin A, Davydov V, Dmitriev V. MRS Internet J Nitride Semicond Res 1997;2:39. [11] Shibata T, Sone H, Yahashi K, Yamaguchi M, Hiramatsu K, Sawaki N, et al. J Cryst Growth 1998;189/190:67. [12] Zhang R, Zhang L, Hansen D, Boleslawski M, Chen K, Lu D, et al. MRS Internet J Nitride Semicond Res 1999; 4S1:G4.7.