Growth of GaN single crystals by Ca3N2 flux

Growth of GaN single crystals by Ca3N2 flux

Available online at www.sciencedirect.com Scripta Materialia 58 (2008) 319–322 www.elsevier.com/locate/scriptamat Growth of GaN single crystals by C...

337KB Sizes 0 Downloads 5 Views

Available online at www.sciencedirect.com

Scripta Materialia 58 (2008) 319–322 www.elsevier.com/locate/scriptamat

Growth of GaN single crystals by Ca3N2 flux G. Wang,a W.X. Yuan,b J.K. Jian,c H.Q. Bao,a J.F. Wang,b X.L. Chena,* and J.K. Lianga,d a

Beijing National Laboratory for Condensed Matter Physics, Institute of Physics, Chinese Academy of Sciences, P.O. Box 603, Beijing 100080, China b Department of Chemistry, School of Applied Science, University of Science and Technology Beijing, Beijing 100083, China c Department of Physics, Xinjiang University, Urumchi 830046, China d International Center for Materials Physics, Academia Sinica, Shenyang 110016, China Received 8 June 2007; revised 15 August 2007; accepted 22 September 2007

This paper reports recent progress on GaN single crystal growth by Ca3N2 flux. The isothermal phase diagrams of the Ca–Ga–N system were predicted from the corresponding binary systems by CALPHAD. Well-crystallized GaN crystals up to 1.5 mm were grown from the Ca–Ga–N system at 900 C under 0.2 MPa N2 pressure. It was found that the crystal size depended on the molar ratio of starting materials, the temperature and the duration of growth. A growth mechanism involving two-step reactions is proposed.  2007 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: CALPHAD; Single crystal growth; Nitride

GaN, a wide direct band gap semiconductor, is a promising material for fabrication of short-wavelength optical devices and high-power electronic devices [1,2]. To date, due to lack of commercially available GaN substrates, most GaN-based thin film devices are fabricated on sapphire, silicon and silicon carbide substrates by heteroepitaxy. However, high-density dislocations, introduced by large lattice constants and thermal expansion coefficient mismatches between the epitaxial layer and the substrate, seriously deteriorate the performance and lifetime of these devices. Therefore, high-quality GaN single crystals are strongly desired as substrates for homoepitaxy. Bulk GaN single crystal growth has been tried by various methods including hydride vapor-phase epitaxy (HVPE) [3,4], high-pressure solution [5,6], and flux methods using sodium [7–9] and Li3N [10–12]. In the HVPE method, free-standing GaN wafers 2 in. in size have been obtained at a growth rate greater than 100 lm h1. However, because of the heteroepitaxial growth nature of the method, dislocation densities of more than 104 cm2 are common in the grown GaN wafers. Better-quality GaN single crystals over 10 mm in size with a dislocation density of less than 102 cm2 * Corresponding author. Tel.: +86 10 8264 9039; fax: +86 10 8264 9646; e-mail: [email protected]

have been prepared by the high-pressure solution method [5,6]. However, this method requires high temperature and high nitrogen pressure because of very low solubility of nitrogen in metal Ga [13] and the extremely high decomposition pressure of GaN [14]. A sodium flux method reported by Yamane et al. [7–9] has been applied to prepare transparent GaN single crystals at 730–800 C under 5 MPa nitrogen pressure. Growth of bulk GaN single crystals at nitrogen pressures of 0.1– 0.2 MPa has been reported recently by a Li3N flux method at about 800 C [10–12]. These works suggest that the flux method is perhaps promising for low-cost and largescale growth of GaN crystals. More recently, we have demonstrated that a new flux, Ca3N2, can be employed to grow bulk GaN single crystals under moderate conditions [15]. In this paper, we report the recent progress on growing GaN single crystals by using Ca3N2 flux from the Ca–Ga–N system. The dependence of GaN crystal size on the growth conditions was investigated. A growth mechanism involving two reactions was proposed based on the obtained results. The phase relations in the ternary Ca–Ga–N system are completely unknown in the literature. Relevant binary subsystems Ca–Ga [16], Ca–N [17] and Ga–N [18] have been assessed. Ternary nitrides, such as CaGaN [19], Ca6GaN5 [20], Ca5Ga2N4 [21], Ca3Ga2N4 [22] and Ca7Ga1.33N4[23], have been reported with their structures. However, no thermodynamic properties or

1359-6462/$ - see front matter  2007 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.scriptamat.2007.09.027

320

G. Wang et al. / Scripta Materialia 58 (2008) 319–322

decomposition temperatures are available for these compounds. In our growth of GaN single crystals using Ca3N2 flux, CaGaN was the only ternary nitride obtained in the products, and so here we assumed CaGaN to be the only stable ternary nitride. In the present work, the Gibbs energy functions of Ca, Ga and N were taken from the SGTE (Scientific Group Thermodata Europe) database for pure elements compiled by Dinsdale [24]. The gas phase was considered to be ideal and the corresponding expression for its Gibbs energy was: X ½xi  o Ggas þ xi RT lnðxi pÞ; ð1Þ Ggas i m ¼ i¼N;N2 ;N3

where p is the total pressure of the system. The Gibbs energy expressions for N, N2 and N3 in the gas phase were obtained from the database within the ThermoCalc program [25]. The Gibbs energy of the liquid phase could be expressed as: X X ¼ xi o Gliquid þ RT xi ln xi þ ex Gliquid ð2Þ Gliquid i m m i

i

ex liquid Gm

is the molar excess Gibbs energy of the where liquid, expressed in the Redlich–Kister polynomial as the following model: n X XX ex liquid k liquid Gm ¼ xi xj Li;j ðxi  xj Þk þ xi xj xk Lliquid i;j;k ; i

j>i

k¼0

ð3Þ where k Lliquid is the kth binary interaction parameter bei;j tween the components i and j. Lliquid i;j;k is the ternary interaction parameter to be evaluated. CaGaN was modeled as a stoichiometric compound. Its Gibbs energy in per mol-formula was given by the following expression: o

SER SER GðCam Gan N p Þ  mH SER Ca  nH Ga  pH N o orth o gas ¼ mo Gfcc Ca þ n GGa þ p GN þ A þ B  T ;

ð4Þ

where A and B are the parameters to be evaluated. o Gfcc Ca , o gas Gorth Ga and GN are the Gibbs energies of Ca, Ga and 1/2N2 with the crystal structures of face-centered cubic A1, orthorhombic and gas status, respectively. The results of the previously assessed Ca–Ga [16], Ca–N [17] and Ga–N [18] binary systems were accepted and extrapolated into the ternary system in this work. According to previous experimental work using Li3N flux [10,11], GaN single crystals can be grown from the liquid phase, implying that a ternary liquid phase must exist during the GaN crystal growth process, as shown in Ref. [11]. Since there is very little experimental information on the thermodynamics or phase diagram of the Ca–Ga–N system, we assumed here that one ternary liquid phase might form like that in the case of the Li–Ga–N system. We then attempted to make the liquid phase stable for calculation of isotherms at the temperature of GaN crystal growth. The negative starting values of interaction parameter were tried to perform the calculation. Finally we used an empirical value of of the Ca–Ga–N system to obtain 450,000 for Lliquid i;j;k o

reasonable results. Similarly, 82,710 and 15 were assigned to A and B in Eq. (4), respectively. The calculated isothermal sections of the Ca–Ga–N system at 900 and 800 C under 0.2 MPa nitrogen pressure are shown in Figure 1. There were a liquid phase, 2 two-phase regions of (Liq + GaN), and one three-phase region of (Liq + GaN + CaGaN). Based on the Ca–Ga– N phase diagrams, GaN single crystals can be grown from the liquid phase, and the nitrogen content in the liquid phase even reaches about 2 at.% (e.g. point ‘‘a’’ in Fig. 1). In addition, the area of the (Liq + GaN) region becomes larger as the temperature increases, indicating that high temperature is more favorable for GaN crystal growth. The starting materials used for the growth of GaN single crystals were Ga (99.9999%) and Ca3N2. Ca3N2 was synthesized by the direct reaction of metal Ca grains (Alfa, 99.5%) and high-purity N2 (99.999%) at about 900 C. Powder X-ray diffraction (PXRD) analysis indicated that there was a little impurity of CaO and Ca(OH)2 in the as-prepared Ca3N2. The starting materials Ga and Ca3N2 were put in a tungsten crucible in predetermined molar ratios [10,11,15]. The growth apparatus used here was an induction furnace, as used previously for Li3N flux growth [26]. After evacuation to a vacuum of about 6.0 · 103 Pa, the system was filled with nitrogen gas up to 0.2 MPa at room temperature. The tungsten crucible filled with starting materials was heated to 850–950 C by an induction-heater, maintained at that temperature for about 12 h, and then slowly cooled at a rate of about 3 C day1. The whole growth process lasted 60–160 h. Finally, the sample in the crucible was cooled to room temperature by switching off the power. Some of the as-obtained products were taken and analyzed by PXRD. Then, most products were washed with HCl solution and distilled water, and some colorless transparent crystals were obtained. PXRD data were collected on a MAC-M18XHF diffractometer with Cu Ka radiation. An optical microscope and a scanning electron microscope (SEM, FEI, XL30) were used to characterize the morphology of the crystals. Raman scattering measurement of products was performed at room temperature by a Raman system (JY-HR800) using the 532 nm line of a solid-state laser as excitation source. Photoluminescence (PL) measurement was carried out at room temperature by means of a Hitachi F-4500 using 280 nm line of a xenon lamp.

Figure 1. Calculated isothermal section of the Ca–Ga–N ternary system at (a) 900 C and (b) 800 C under a nitrogen pressure of 0.2 MPa.

G. Wang et al. / Scripta Materialia 58 (2008) 319–322

A series of experiments under varying conditions was conducted. The main results are summarized in Table 1. It was found that the crystallization of GaN in the Ca– Ga–N ternary system was possible over a wide range of molar ratios of Ga-rich starting materials, temperatures and growth durations. The results clearly demonstrated that the size of GaN crystals strongly depended on the growth conditions. The size increased with increasing Ga/Ca3N2 molar ratio when the ratio was lower than 8:1. However, it saturated when the ratio was higher than 8:1. The largest GaN crystals were grown under the following conditions: Ga/Ca3N2 molar ratio 8:1, growth temperature 900 C, cooling rate 3 C day1, and nitrogen pressure 0.2 MPa. Under such conditions, GaN crystals with a length of up to 1.5 mm and a diameter of tens of microns have been grown. Figure 2 shows the typical prismatic GaN single crystals grown at such conditions. SEM observation (Fig. 3) further shows the morphology of the obtained GaN crystals. Most GaN crystals had hexagonal pyramidal caps. Figure 3b shows a single crystal having the hexagonal pyramidal cap with {10–11} side faces. Some prismatic crystals had smooth plane faces, while others had growth stripes on the side faces. A growth mechanism is suggested as follows based on the results presented above. When the starting materials are heated to a certain temperature, liquid Ga reacts with solid-state Ca3N2 according to reaction (1): 2Ga + Ca3N2 = 2CaGaN + Ca. Our experiments have confirmed the existence of CaGaN. Metal Ca obtained by reaction (1) will react with Ga and form a Ca–Ga melt with an appreciable concentration of Ca. Partial CaGaN may then react with metal Ga and form fine GaN particles by reaction (2): CaGaN + Ga = GaN + Ca–Ga alloy. Other CaGaN might dissolve in the Ca–Ga melt and form a Ca–Ga–N ternary liquid phase. The composition of this liquid phase will enter into the two-phase region (Liq + GaN) as the temperature decreases, as seen in Figure 1. Subsequently, GaN crystallizes and grows from the melt. According to this growth mechanism, the dependence of GaN crystal size on the growth conditions can be well understood. At a lower Ga/Ca3N2 molar ratio (lower than 4:1), no GaN is found; only CaGaN is found due to lack of Ga. As the Ga/Ca3N2 molar ratio increases from 4:1 to 8:1, more Ca–Ga melt forms and retards reaction (2) to some extent, resulting in much fewer GaN fine particles. Upon cooling, GaN crystallizes from the Ca–Ga–N melt and grows on those fewer GaN particles. Therefore, the size of GaN crystals increases with increasing Ga/Ca3N2 molar ratio. At a higher Ga/Ca3N2 molar ratio (higher

321

Figure 2. Optical image of GaN single crystals grown by Ca3N2 flux.

than 8:1), reactions (1) and (2) proceed rapidly to completion under the existence of much excess of Ga. Under these conditions, the crystal size does not increase further because small GaN crystals dominate. It is well known that high temperature will enhance the solubility of N in the Ca–Ga melt and promote the critical supersaturation of GaN in the ternary melt. The isothermal phase diagrams predicted by CALPHAD for the Ca–Ga–N system (Fig. 1) confirmed this conclusion. But much higher temperatures will induce the decomposition of GaN. Based on the experimental results, 900 C is more appropriate for the growth. Raman scattering measurement was employed to further characterize the crystals. According to Ref. [27], hexagonal GaN has six Raman active modes: E2 (low), E2 (high), A1 (TO), A1 (LO), E1 (TO) and E1 (LO). Figure 4 shows the Raman spectrum of the obtained GaN single crystal recorded at room temperature by the back-scattering X ðY ; Y ÞX configuration. Four phonon modes were observed at 147, 534, 571 and 740 cm1, which were in good agreement with the reported E2 (low), A1 (TO), E2 (high) and E1 (LO) [27]. The very sharp peak corresponding to E2 (high) mode indicated that the crystal was well-crystallized. In the spectrum, the E1 (LO) mode was weakly observed although it was forbidden in this configuration, which may be attributed to the so-called leakage effect [28]. Figure 5 shows the room temperature PL spectrum of the GaN crystals. Besides a broad luminescence band centered at 2.65 eV, a weak characteristic band-edge emission at 3.40 eV was observed, while the well known yellow band [29] was not detected in the spectrum. The broad band has been observed several times previously [29–32] and was thought to be related to the defect level of nitrogen vacancy (VN) or oxygen impurity (ON) [29,31], which can be present in our case due to the deficiency of nitrogen and oxygen impurity in the GaN crystals grown here. The isothermal phase diagrams of the ternary Ca–Ga–N system were predicted by CALPHAD based

Table 1. Experimental conditions of crystalline GaN obtained by Ca3N2 flux (P N2 ¼ 0:2 MPa) Cases

Molar ratios (Ga/Ca3N2)

Temperatures (C)

Duration (h)

Main products

Maximum size (mm)

1 2 3 4 5 6 7 8 9

4:1 6:1 8:1 8:1 8:1 8:1 10:1 10:1 12:1

900 900 900 900 950 850 900 950 900

160 160 60 160 160 160 160 160 160

CaGaN CaGaN, Ca–Ga alloy, GaN Ca–Ga alloy, CaGaN, GaN Ca–Ga alloy, GaN, CaGaN Ca–Ga alloy, Ga Ca–Ga alloy, CaGaN, GaN Ca–Ga alloy, CaGaN, GaN Ca–Ga alloy, Ga Ca–Ga alloy, CaGaN, GaN

– <1 <0.1 1.5 – <1 <1 – <1

322

G. Wang et al. / Scripta Materialia 58 (2008) 319–322

Figure 3. SEM images of GaN crystals. (a) Overall observation of crystals. (b) The hexagonal pyramidal cap of a single crystal.

Figure 4. Raman spectrum of GaN single crystal.

Figure 5. Room temperature PL spectrum of GaN crystals.

on the extrapolation of three corresponding binary systems, Ca–Ga, Ca–N and Ga–N. Well-crystallized hexagonal GaN prism-like crystals up to 1.5 mm have been grown from the Ca–Ga–N system at 900 C under a nitrogen pressure of 0.2 MPa. The size of GaN crystals depended on the molar ratio of the starting materials, the growth temperature and the duration of growth. A growth mechanism based on reactions (1) and (2) was proposed based on the obtained results. The formation of numerous fine GaN particles by reaction (2) confined the enlargement of the crystal size. These results suggested that further efforts are still needed to obtain larger GaN crystals. The authors gratefully acknowledge Pandat for supplying the phase diagram calculation software. This work was financially supported by the National Natural Science Foundation of China (Grant Nos. 50472075 and 50304002). [1] S.J. Pearton, J.C. Zolper, R.J. Shul, F. Ren, J. Appl. Phys. 86 (1999) 1. [2] B. Monemar, J. Mater. Sci. Mater. Electron. 10 (1999) 227.

[3] X.P. Xu, R.P. Vaudo, C. Loria, A. Salant, G.R. Brandes, J. Chaudhuri, J. Cryst. Growth 246 (2002) 223. [4] Y. Oshima, T. Eri, M. Shibata, H. Sunakawa, K. Kobayashi, T. Ichihashi, A. Usui, Jpn. J. Appl. Phys. 42 (2003) L1. [5] S. Porowski, J. Cryst. Growth 189/190 (1998) 153. [6] I. Grzegory, J. Phys.: Condens. Matter. 13 (2001) 6875. [7] H. Yamane, M. Shimada, S.J. Clarke, F.J. DiSalvo, Chem. Mater. 9 (1997) 413. [8] M. Aoki, H. Yamane, M. Shimada, S. Sarayama, F.J. DiSalvo, J. Cryst. Growth 242 (2002) 70. [9] M. Aoki, H. Yamane, M. Shimada, S. Sarayama, H. Iwata, F.J. DiSalvo, J. Cryst. Growth 266 (2004) 461. [10] Y.T. Song, W.J. Wang, W.X. Yuan, X. Wu, X.L. Chen, J. Cryst. Growth 247 (2003) 275. [11] X.L. Chen, Sci. Technol. Adv. Mater. 6 (2005) 766. [12] G. Wang, J.K. Jian, W.X. Yuan, X.L. Chen, Cryst. Growth Des. 6 (2006) 1157. [13] S. Porowski, I. Grzegory, J. Cryst. Growth 178 (1997) 174. [14] J. Karpinski, J. Jun, S. Porowski, J. Cryst. Growth 66 (1984) 11. [15] J.K. Jian, G. Wang, C. Wang, W.X. Yuan, X.L. Chen, J. Cryst. Growth 291 (2006) 72. [16] J.F. Wang, W.X. Yuan, X. Zhao, W.L. Qian, Z.F. Cai, Calphad (31) (2007) 120. [17] G. Wang, W.X. Yuan, J.F. Wang, X. Zhao, J.K. Jian, X.L. Chen, Mater. Lett. 61 (2007) 2266. [18] A.V. Davydov, W.J. Boettinger, U.R. Katiner, T.J. Anderson, Phys. Stat. Sol. (a) 188 (2001) 407. [19] P.P. Verdier, P. L’Haridon, M. Maunaye, R. Marchand, Acta. Cryst. B 30 (1974) 226. [20] G. Cordier, R. Kniep, P. Hoehn, A. Rabenau, Z. Anorg. Allg. Chem. 591 (1990) 58. [21] G. Cordier, Z. Naturforsch. B 43 (1988) 1253. [22] S.J. Clarke, F.J. DiSalvo, Inorg. Chem. 36 (1997) 1143. [23] P. Hoehn, G. Auffermann, R. Ramlau, H. Rosner, W. Schnelle, R. Kniep, Angew. Chem. Int. Ed. 45 (2006) 6681. [24] A.T. Dinsdale, Calphad 15 (1991) 317. [25] B. Sundman, B. Jansson, J.O. Andersson, Calphad 9 (1985) 153. [26] G. Wang, J.K. Jian, B. Song, X.L. Chen, W.J. Wang, Y.T. Song, Appl. Phys. A 85 (2006) 169. [27] J.W. Orton, C.T. Foxon, Rep. Prog. Phys. 61 (1998) 1. [28] H. Harima, J. Phys.: Condens. Matter 14 (2002) R967. [29] T. Ogino, M. Aoki, Jpn. J. Appl. Phys. 19 (1980) 2395. [30] X.L. Chen, J.Y. Li, Y.G. Cao, Y.C. Lan, H. Li, M. He, C.Y. Wang, Z. Zhang, Z.Y. Qiao, Adv. Mater. 12 (2000) 1432. [31] Y.G. Cao, X.L. Chen, Y.C. Lan, J.Y. Li, Y. Zhang, Y.P. Xu, T. Xu, J.K. Liang, Modern Phys. Lett. B 14 (2000) 583. [32] J.K. Jian, X.L. Chen, Q.Y. Tu, Y.P. Xu, L. Dai, M. Zhao, J. Phys. Chem. B 108 (2004) 12024.