Seeded growth of AlN single crystals by physical vapor transport

Seeded growth of AlN single crystals by physical vapor transport

ARTICLE IN PRESS Journal of Crystal Growth 287 (2006) 372–375 Seeded growth of AlN single crystals by physical vapo...

424KB Sizes 0 Downloads 14 Views


Journal of Crystal Growth 287 (2006) 372–375

Seeded growth of AlN single crystals by physical vapor transport D. Zhuang, Z.G. Herro, R. Schlesser, Z. Sitar Department of Materials Science and Engineering, North Carolina State University, Campus box 7919, Raleigh, NC 27695-7907, USA Available online 4 January 2006

Abstract Seeded growth of AlN single crystals was achieved in an induction-heated, high-temperature reactor. The growth process was based on physical vapor transport (PVT), where presintered AlN powder was used as source material. AlN seeds were cut from a boule containing large single crystalline grains, which were grown by natural grain expansion of an initially polycrystalline, self-seeded deposit. Seeded growth was interrupted several times in order to refill the AlN powder source and a dedicated process scheme was used to ensure epitaxial re-growth on the seed surface after each exposure to air. The single crystalline seed expanded laterally at an angle of 451 resulting in an 18 mm large AlN single crystal. The crystal expansion rate, crystalline orientation, as well as growth morphology were characterized by optical microscopy and X-ray diffraction, respectively. r 2005 Elsevier B.V. All rights reserved. PACS: 61.10.Nz; 81.05.Ea; 81.10.Aj; 81.10.Bk Keywords: A1. Characterization; A2. Single-crystal growth; B1. Nitrides

1. Introduction Fabrication of single crystalline AlN substrates has become the subject of extensive research efforts in the past decade. In contrast to non-nitride substrates, e.g. sapphire and SiC, AlN has isomorphic crystal structure and closely matched thermal and lattice parameters with respect to AlGaN device layers. Therefore, employing AlN substrates in AlGaN epitaxial growth, especially for high Al concentration AlGaN, is anticipated to reduce thermal stress and defect density [1], leading to enhanced device performance and reliability [2,3]. In addition, the high electrical resistivity and excellent thermal conductivity, combined with its wide bandgap (6.2 eV) [4], makes AlN substrates particularly suitable for high-power microwave, as well as UV optoelectronic devices. Generally, there are three approaches to bulk AlN crystal growth: (1) seeded growth on non-nitride substrates, e.g. SiC or previously grown AlN single crystals, (2) growth of freestanding AlN crystals by self-seeding, and (3) crystal expansion using AlN seeds. Following the pioneerCorresponding author. Fax: 919 515 3419.

E-mail address: [email protected] (D. Zhuang). 0022-0248/$ - see front matter r 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.jcrysgro.2005.11.047

ing work of Slack and McNelly [5], several research groups reported progress on AlN bulk growth. Starting with SiC substrates, Edgar et al. [6–8] and Dalmau et al. [9] demonstrated growth of relatively large area, free-standing AlN crystals. Despite the advantage of availability of large SiC substrates, the growth temperature and resulting growth rate were limited in order to prevent substrate decomposition. Epelbaum and coworkers utilized spontaneously nucleated AlN crystals to study the growth habits as a function of growth temperature [10,11]. It was found that the morphology of freestanding crystals changed from hexagonal prisms to asymmetric platelets as growth temperatures increased. The largest freestanding crystals were up to 15 mm in their longest direction. Attempts to enlarge these crystals in subsequent growth runs, however, remained unsuccessful. Using a grain selection process, Schowalter et al. [12] reported AlN wafers larger than 10 mm in diameter. Most recently, our efforts at NCSU yielded up to 36 mm long AlN boules with single crystalline grains larger than 1 cm [13,14]. Theoretically, grain size expansion is proportional to the boule length, and eventually one single crystalline grain is likely to become dominant. In practice, however, the grain selection process is tedious and may be adversely affected by randomly

ARTICLE IN PRESS D. Zhuang et al. / Journal of Crystal Growth 287 (2006) 372–375

oriented re-nucleation, and by the introduction of significant stresses or even cracks in the grown boules. By contrast to the aforementioned crystal growth methods, seeded growth of AlN boules on single crystalline AlN seeds combines the advantages of high growth rates, large expansion angles, reduced stresses, and pre-defined crystal orientation. In this paper, we present recent progress on AlN seeded growth using seeds prepared from a boule grown using grain expansion of initially randomly nucleated material. 2. Experimental procedure AlN seeded growth was carried out in an inductively heated reactor, in which porous graphite foam was used as insulation, as shown in Fig. 1. A crucible containing AlN powder was placed on a rigid graphite pedestal. Two infrared pyrometers were employed to monitor the source (Tb) and seed (Tt) temperatures. The temperature gradient was established and altered by changing the position of the


RF coil relative to the crucible. Under typical growth conditions, the source temperature was maintained at 2200–2250 1C with an axial temperature gradient in the range of 5–10 1C/cm; the reactor pressure used in these experiments was in a range of 400–900 Torr of UHP-grade nitrogen. To ensure epitaxial re-growth, the seeds underwent wet chemical etching before being loaded into the system. The seeds were first cleaned in acetone, isopropanol and methanol in an ultrasonic bath for 5 min to remove any organic contaminants and particles on the surface, then they were dipped in a phosphoric and sulfuric acid mixture (1:3) for 5 min, and finally underwent etching in diluted hydrofluoric acid (25%) for 5 min. X-ray diffraction (XRD) and optical microscopy were employed to characterize the crystalline orientation and surface morphology of the grown crystals.

3. Results and discussions 3.1. Seed preparation and impurity reduction The seeds used in this work were prepared from a selfseeded boule [14] by cutting off a top layer parallel to the growth surface. As seen in Fig. 2, the harvested crystal contained several cracks due to significant stress present in the boule. To avoid adverse effects in subsequent growth, only the crack-free center region that was approximately 10 mm in diameter was used as a seed. The y=2y scan of this seed crystal performed in a general area detection diffraction system (GADDS) showed two major peaks corresponding to the ð1 0 1¯ 0Þ and ð1 1 2¯ 0Þ planes, respectively, as shown in Fig. 3. High-resolution XRD studies confirmed that the seed was single crystalline and its orientation was

Fig. 1. Schematic of the RF-heated AlN reactor.

Fig. 2. AlN seed prepared from a self-seeded boule.

ARTICLE IN PRESS D. Zhuang et al. / Journal of Crystal Growth 287 (2006) 372–375


ð4 1 5¯ 0Þ, which is 10.91 off ð1 0 1¯ 0Þ and 19.11 off ð1 1 2¯ 0Þ planes, respectively. It is known that the presence of oxygen may result in secondary nucleation in AlN sublimation growth. At low temperatures, oxygen in the source material assists in mass transport of Al species leading to randomly nucleated aluminum oxynitride [13]. In addition, a naturally formed, thin oxide layer on AlN seeds may disturb or inhibit epitaxial growth by affecting the ordering of adatoms on the seed surface. To reduce oxygen concentration in our system and thus avoid secondary nucleation, we generally (1) pre-sinter AlN powder at temperatures above 2200 1C, (2) remove the oxide layer on the seed surface by wet chemical etching [15], as described above, and (3) apply an inverted temperature gradient during the ramp-up stage to inhibit undesired, low-temperature deposition on the seed [13,16]. Chemical analysis by glow discharge mass spectrometry (GDMS) (Table 1) confirmed that the crystals grown in our system have low oxygen impurity levels (100 ppm), similar to those reported in the literature [11], carbon contamination levels below 100 ppm wt, and transition metals present only in the sub-ppm range. 3.2. Crystal expansion and growth morphology In order to evaluate the single crystal expansion rate in the lateral direction, we prepared a longitudinal cut of a boule after several growth runs, as shown in Fig. 4. Following a mechanical polish on both sides, the grown single crystal was highly transparent, even though the cut was relatively thick (2 mm). The most striking feature was that the seed expanded from 10 to 18 mm in diameter after

only 4 mm of growth, i.e., the crystal laterally expanded under an angle of approximately 451. The optical images of as-grown surface morphology after the first two growth runs are shown in Fig. 5. The seed expanded from 10 mm diameter to 11  14 mm2 after the first growth run and grew further to 13  15 mm2 during the second growth run. The round seed was elongated along the direction perpendicular to the c-axis, suggesting that the growth rate along the c-axis was lower than that along the a-axis. After the first growth run, the crystal featured smooth surface morphology and retained the ð4 1 5¯ 0Þ facet of the seed. After the second growth run, however, the asgrown surface comprised three facets, as marked in Fig. 5b. The orientation of these facets was determined by XRD as ð2 5 7¯ 0Þ, ð1 1 2¯ 0Þ, and ð4 1 5¯ 0Þ, respectively. In addition to their different crystallographic orientation, these three facets also featured different microscopic surface morphology. Optical microscopy revealed that the ð1 1 2¯ 0Þ facet was smooth and completely feature-less while both the ð2 5 7¯ 0Þ and ð4 1 5¯ 0Þ facets showed a distinctly serrated structure, as shown in Fig. 6. 4. Summary We have, for the first time, demonstrated seeded growth of AlN single crystals on AlN seeds previously prepared using a natural grain expansion process. A single

2500 (1120)

(1010) Intensity (A.U.)




500 (2020) 0 30


50 2-theta



Fig. 4. Longitudinal cut of a crystal boule after several growth runs showing the crystal expansion angle. 1 mm grid.

Fig. 3. y/2y scan of a seed prepared from a self-seeded boule.

Table 1 Impurity concentration in the sintered source and starting seeds, all in ppm wt

Sintered source Seeds










o100 o100

o100 o100

o10 o10

1.5 o0.5

100 80

o0.5 o0.5

5.5 0.35

10 o0.5

o0.5 o0.5

ARTICLE IN PRESS D. Zhuang et al. / Journal of Crystal Growth 287 (2006) 372–375


Fig. 5. Seeded growth morphology after the first (a) and second (b) growth runs, the edges of the ð1 1 2¯ 0Þ facet are highlighted by two white lines.

N00014-01-1-0716; Dr. C. Wood, project monitor. The authors thank Dr. B. Raghothamachar and Dr. M. Dudley for the XRD studies.


Fig. 6. Optical micrograph of ð2 5 7¯ 0Þ and ð4 1 5¯ 0Þ facets shown in Fig. 5b.

crystalline seed expanded from 10 to 18 mm over several growth runs, with an expansion angle of approximately 451. The as-grown surface featured three facets with the following crystallographic orientations: ð2 5 7¯ 0Þ, ð1 1 2¯ 0Þ, and ð4 1 5¯ 0Þ. The as-grown surface of the a-facet ð1 1 2¯ 0Þ was optically smooth and revealed no growth-related structures. On the other two facets, however, serrated morphology was observed. Acknowledgments This work was supported by the Office of Naval Research under the auspices of the MURI program; grant

[1] R. Schlesser, R. Dalmau, Z. Sitar, J. Crystal Growth 241 (2002) 416. [2] R. Gaska, C. Chen, J. Yang, E. Kuokstis, A. Khan, G. Tamulaitis, et al., Appl. Phys. Lett. 81 (2002) 4658. [3] X. Hu, J. Deng, N. Pala, R. Gaska, M.S. Shur, C.Q. Chen, et al., Appl. Phys. Lett. 82 (2003) 1299. [4] J.H. Edgar (Ed.), Properties of Group III Nitrides, Electronic Materials Information Service (EMIS), London, 1994. [5] G.A. Slack, T.F. McNelly, J. Crystal Growth 34 (1976) 263. [6] J.H. Edgar, L. Liu, B. Liu, D. Zhuang, J. Chaudhuri, M. Kuball, et al., J. Crystal Growth 246 (2002) 187. [7] L. Liu, B. Liu, Y. Shi, J.H. Edgar, MRS Internet J. Nitride Semicond. Res. 6 (2001) 7. [8] Y. Shi, Z.Y. Xie, L. Liu, J.H. Edgar, M. Kuball, J. Crystal Growth 233 (2001) 177. [9] R. Dalmau, R. Schlesser, B.J. Rodriguez, R.J. Nemanich, Z. Sitar, J. Crystal Growth 281 (2005) 68. [10] B.M. Epelbaum, C. Seitz, A. Magerl, M. Bickermann, A. Winnacker, J. Crystal Growth 265 (2004) 577. [11] M. Bickermann, B.M. Epelbaum, A. Winnacker, J. Crystal Growth 269 (2004) 432. [12] L.J. Schowalter, G.A. Slack, J.B. Whitlock, K. Morgan, S.B. Schujman, B. Raghothamachar, et al., Phys. Stat. Sol. (C) 0 (2003) 1997. [13] V. Noveski, R. Schlesser, B. Raghothamachar, M. Dudley, S. Mahajan, S. Beaudoin, et al., J. Crystal Growth 279 (2005) 13. [14] D. Zhuang, R. Schlesser, Z. Sitar, Mater. Res. Soc. Symp. Proc. 831 (2005) E11.1. [15] S.W. King, J.P. Barnak, M.D. Bremser, K.M. Tracy, C. Ronning, R.F. Davis, et al., J. Appl. Phys. 84 (1998) 5248. [16] V. Noveski, R. Schlesser, J. Freitas Jr., S. Mahajan, S. Beaudoin, Z. Sitar, Mater. Res. Soc. Symp. Proc. 798 (2004) Y2.8.