- Email: [email protected]
Effect of dual buffer layer structure on the epitaxial growth of AlN on sapphire
Journal of Alloys and Compounds 544 (2012) 94–98
Contents lists available at SciVerse ScienceDirect
Journal of Alloys and Compounds journal homepage: www.elsevier.com/locate/jalcom
Effect of dual buffer layer structure on the epitaxial growth of AlN on sapphire D.G. Zhao a,⇑, D.S. Jiang a, L.L. Wu a, L.C. Le a, L. Li a, P. Chen a, Z.S. Liu a, J.J. Zhu b, H. Wang b, S.M. Zhang b, H. Yang a,b a b
State Key Laboratory on Integrated Optoelectronics, Institute of Semiconductors, Chinese Academy of Sciences, P.O. Box 912, Beijing 100083, China Suzhou Institute of Nano-tech and Nano-bionics, Chinese Academy of Sciences, Suzhou 215125, China
a r t i c l e
i n f o
Article history: Received 6 June 2012 Received in revised form 10 July 2012 Accepted 23 July 2012 Available online 1 August 2012 Keywords: Nitride materials Crystal growth X-ray diffraction
a b s t r a c t A dual AlN buffer layer structure, including an isolated layer and a nucleation layer, is proposed to improve the growth of AlN films on sapphire substrate by metal organic chemical vapor deposition. This method is aimed to weaken the negative nitridation effect and improve lateral growth condition in the initial growth stage. It is found that suitably increasing the thickness of the nucleation layer is in favor of a better structural quality of the AlN film. An examination of surface morphology by atomic force microscopy suggests that the thicker the dual AlN buffer layer, the rougher the surface, and a higher quality of AlN epilayer is resulted. Ó 2012 Elsevier B.V. All rights reserved.
1. Introduction Group-III nitrides have been investigated for many years due to their extensive applications in optoelectronics and microelectronics. For example, AlGaN solar-blind ultraviolet photodetectors are able to be used in missile detection, secure space-to-space communication, flame monitoring and so on. Great successes have been achieved in the fabrication of AlGaN solar-blind ultraviolet photodetectors, including the realization of focal plane arrays and avalanche photodiodes [1–4]. The quality of AlN films plays an important role in determining the performance of back-illuminated AlGaN ultraviolet photodetectors. However, it is very difficult to grow high quality AlN layers, mainly because the surface mobility of Al atoms is very low. Heteroepitaxial AlN films often exhibit poor crystalline quality or poor surface morphology due to the limited lateral growth. Usually, a special metalorganic chemical vapor deposition (MOCVD) system with an extremely high growth temperature around 1400 °C or with a migration-enhanced growth mode can be employed to promote two-dimensional (2D) grow mode and obtain high quality AlN films [5–7]. On the other hand, the two-step growth method had been widely used to grow nitride materials since it was developed for the growth of GaN epilayer on sapphire substrate by MOCVD [8,9]. The buffer layer is critical in the growth of nitride materials to overcome the problems caused by the large lattice and thermal mismatches with sapphire substrate, and the density of threading dislocations is largely reduced. In fact, both low- and high-temperature AlN buffer layers ⇑ Corresponding author. Tel.: +86 10 82304208; fax: +86 10 82304242. E-mail address: [email protected] (D.G. Zhao). 0925-8388/$ - see front matter Ó 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jallcom.2012.07.123
have been employed to AlN epitaxial growth [10–12], and the initial process is very important for the growth of high quality AlN epilayers [12,13]. It is known that the nitridation of sapphire substrate is a negative factor which hinders the high quality growth of AlN films since the Al and N faces have different polarities and different growth rates, leading to the formation of inversion domains and a rough surface morphology [13,14]. Therefore, suppressing the nitridation effect of sapphire substrate should also be useful to improve the quality of AlN [15]. In this paper, the dual AlN buffer layer, instead of commonly used single AlN buffer layer (or nucleation layer), is employed during the AlN growth by MOCVD, which is significantly helpful to grow high quality AlN epilayers even when a commercially-available conventional MOCVD system is used. The quality of AlN films could be improved further if the thickness of dual AlN buffer layer is made suitably larger. The additional in situ optical reflectance and AFM measurements of such a dual AlN buffer layer show that its surface roughness increases with increasing thickness, and the followed lateral growth of AlN epilayers is thus enhanced.
2. Experimental procedure All the AlN samples studied in this work are grown on c-plane sapphire substrate by MOCVD. The trimethylaluminum (TMAl) and ammonia (NH3) are used as Al and N sources, respectively. The H2 are used as carrier gas. Two kinds of AlN samples are grown by MOCVD in this work, one kind is with dual AlN buffer layer, the other is with single AlN buffer layer, as schematically shown in Fig. 1(a and b), respectively. For the growth of dual AlN buffer layer, a lower growth temperature Tg of 900 °C are used for its first sublayer, while a higher Tg of 1080 °C are used for its second sublayer. For the AlN samples with single AlN buffer layer whose growth condition is the same as that of the above-mentioned second AlN sublayer of the dual buffer layer, i.e., the growth temperature is 1080 °C. The growth
D.G. Zhao et al. / Journal of Alloys and Compounds 544 (2012) 94–98
95
Fig. 1. The schematic diagrams of AlN sample structures: (a) with dual AlN buffer layer, (b) with single AlN buffer layer. procedure of studied AlN samples is as follows, firstly a dual or single AlN buffer layer is deposited on the sapphire substrate, then the AlN epilayer is grown on the buffer layer at 1100 °C. All growth processes are monitored by the in situ optical reflectance, which is used as a powerful tool to study the GaN growth mechanism by MOCVD [16–18]. The quality of AlN films is characterized by the full-width-athalf-maximum (FWHM) of double crystal X-ray x-scan rocking curves (XRC), ex situ performed on a Rigaku SLX-1AL X-ray diffractometer. The surface morphology is examined by the atomic force microscopy (AFM) at room temperature.
3. Results and discussion Fig. 2(a) shows the in situ measured trace of optical reflectance from the surface of sample A which is grown with the dual AlN buffer layer structure by MOCVD. As shown in Fig. 2(a), the trace is divided into three parts as follows, corresponding to the three growth stages of AlN film deposited on the dual AlN buffer layer [16–18]: (I) the growth of AlN ‘‘isolated’’ layer (AlN sublayer 1) at 900 °C which is employed to separate sapphire substrate from nucleation layer and suppress the nitridation effect. The thickness of sublayer 1 is about 75 nm, the corresponding growth rate is about 0.04 nm/s; (II) the growth of AlN ‘‘nucleation’’ layer (AlN sublayer 2) at 1080 °C. The thickness of sublayer 2 is about 375 nm, the corresponding growth rate is about 0.32 nm/s; (III) the growth of thick AlN epilayer at 1100 °C, the corresponding growth rate is about 0.2 nm/s. In order to suppress the nitridation, the conditions of growth temperature, growth rate, and reactor pressure during the growth of isolated AlN layer are carefully adjusted. In Fig. 2(a) the trace in stages I and II corresponds to the grow process of dual AlN buffer layer. During the growth stage III, the reflectance increases gradually, and the growth rate is intentionally reduced to allow a better surface migration of Al atoms. It is noted that during the growth of AlN nucleation layer (in stage II), the intensity of in situ optical reflectance decreases gradually, i.e., the surface roughness of material increases. A rough AlN layer may contain a lot of islands which may provide nucleation centers. However, afterwards in stage III the surface of the AlN epilayer gradually turns to be optically smoother, which means that the lateral growth and coalescence of AlN islands emerge. Finally, the quasi 2D growth of the AlN layer occurs. An oscillation of the reflectivity intensity with large and equal amplitude is well observed. In fact, the practical conditions used for the AlN buffer layer growth are very different from those used for the GaN buffer layer growth. For example, the growth temperature of AlN buffer layer must be much higher (up to 900 and 1080 °C as in our case), otherwise the crystalline quality will be very poor
Fig. 2. The three stages of in situ optical reflectance traces in the whole growth process of AlN: (a) sample A grown with dual AlN buffer layer including an isolated layer and a nucleation layer. (b) sample B grown with single AlN buffer layer (an AlN nucleation layer).
due to the limited lateral migration of Al atoms, and it will be difficult to form necessary islands and develop enough amount of nucleation centers for further epitaxial growth. However, there is still a basic rule valid for both AlN and GaN growth by MOCVD, i.e., it is very important to better control the processes of nucleation and transition from three-dimensional (3D) to quasi-2D growth mode, as what has been widely reported for the initial growth stage of GaN [16–18]. As a comparison, we have also investigated the growth process of AlN film sample B which employs a single AlN buffer layer (the structure is shown in Fig. 1(b)). As shown by the optical reflectance trace of sample B in Fig. 2(b), in the beginning stage of the growth process, i.e., the temperature ramping process, the reflectance intensity is very low, indicating that the stage of ‘‘isolated’’ AlN layer grown at 900 °C is omitted for this sample, only the growth of the single AlN buffer layer (also named as AlN nucleation layer) is remained and is marked as growth stages II in Fig. 2(b). There is actually a great difference between the in situ optical reflectance traces of samples A and B. The average intensity of optical reflectance of sample B in stages II and III almost keeps constant, i.e., there is nearly no any obvious growth mode transformation from 3D to quasi 2D. However, for sample A with the dual AlN buffer layer, there is an extended transformation process from 3D to quasi 2D growth mode, which is very beneficial to the improvement of structural quality of epilayers. The structural quality of AlN samples A and B is checked by the x-scan XRC. As shown in Table 1, the (0 0 2) and (1 0 2) plane FWHMs of XRC of sample A are narrow, i.e., 472 and 796 arcsec, respectively, while they are much broader for sample B, i.e., 810 and 966 arcsec, respectively. It is known that the screw and edge dislocation densities could be indirectly represented by the (0 0 2)
96
D.G. Zhao et al. / Journal of Alloys and Compounds 544 (2012) 94–98
Table 1 The growth conditions of AlN dual or single buffer layers and the FWHM results of X-ray diffraction rocking curves for grown AlN epilayers. Samples
Thickness of buffer layer (nm) Isolated layer
Nucleation layer
A, D B C E F
75 0 75 75 75
375 450 225 675 1275
and (1 0 2) FWHMs of XRC, respectively [19,20]. That is to say, both screw and edge dislocation densities of sample A are much lower than those of sample B. Correspondingly, the structural quality of sample A is better than sample B. It is noted that the clear difference between two samples is not only in the structural quality, but also in surface morphology. Fig. 3(a and b) show the surface morphology of AlN samples A and B (5 5 lm2) obtained by AFM, respectively. Their surface roughness is 0.13 nm and 40.65 nm, respectively. The surface of sample A is much smoother than sample B. Furthermore, a step-flow growth mode could be detected from the surface morphology of sample A. This result indicates that the dual AlN buffer layer structure can play an important role in improving the structural quality of AlN epilayers. It is well known that the epitaxial growth of GaN and AlN has remarkable differences. Especially, the surface mobility of Al atoms is much lower than that of Ga atoms under similar growth conditions [21,22], making it much more difficult for AlN to achieve the 2D growth mode. In addition, the possible nitridation of sapphire substrate is another reason to deteriorate the quality of AlN layers because the different growth velocities on the Al and N polarity surfaces will lead to an unfavorable deposition condition, introducing inversion domain defects in the epilayers, and the N polaritydominant layer growth always results in a relatively rough surface morphology [13]. Some practical methods had been suggested to avoid nitridation and thus had improved the quality of AlN films. For example, the technology of initially alternating supply of ammonia may eliminate the negative effect of nitridation [15]. In our work, it is proposed to add a low temperature-grown isolated AlN layer to weaken the nitridation effect. It is assumed to be helpful to improve the structural quality and surface morphology of the AlN film. It is found that a relatively larger thickness of nucleation layer in the dual AlN buffer layer structure has a positive effect on the AlN growth by MOCVD. Fig. 4 shows the traces of in situ optical reflectance during the growth of four AlN samples, i.e., samples C, D, E and F, with varying thickness of AlN nucleation layer, where Fig. 4(a and d) correspond to the thickness of AlN nucleation layer of 225, 375, 675, and 1275 nm, respectively. For these samples the thickness of isolated layer is kept the same as 75 nm. The dashed lines in Fig. 4 denote the start of AlN epilayer’s growth after the completion of the buffer layer. It is interesting to note that there is a large difference in the four optical reflectance curves. For example, for sample C which has the thinnest nucleation layer thickness, the optical reflectance of AlN epilayer quickly increases to a stable value after 3280 s after the growth of dual AlN buffer layer. The transformation process from 3D to quasi-2D occurs shortly. However, for sample D (here sample D is the same one as sample A, but is called a different name here in this series), the case is different. After the growth of 375 nm AlN nucleation layer, as shown in Fig. 4(b), the surface optical reflectance is lower than that of sample C, implying that the surface is rougher. In the followed growth of thick AlN epilayer, the average intensity of in situ optical reflectance increases gradually. After about 4502 s, the average intensity of optical reflectance reaches a stable value. Clearly, it takes a longer time to complete the growth transformation from 3D to 2D. Such
Thickness of AlN epilayer (lm)
XRC FWHM of AlN (arcsec) (0 0 2)
(1 0 2)
1 0.75 0.75 1.65 2.4
472 810 778 435 360
796 966 932 558 480
Fig. 3. The surface morphology of AlN epilayers A (a) and B (b) (5 5 lm2). The surface roughness is 0.13 nm and 40.65 nm, respectively.
transformation process is even longer for samples E and F. After growth of AlN nucleation layer, as shown in Fig. 4(c and d), the intensity of in situ optical reflectance of samples E and F decreases further, and it takes much longer time, i.e., about 7867 and 11222 s, respectively, for the growth mode transformation from 3D to 2D. From the four traces of in situ optical reflectance of growth processes it is found that when the thickness of nucleation layer is larger, it will take longer time for the optical reflectance to reach a
D.G. Zhao et al. / Journal of Alloys and Compounds 544 (2012) 94–98
97
Fig. 4. The in situ optical reflectance traces of 4 AlN samples with dual AlN buffer layer. The thickness of AlN nucleation sublayer is: (a) 225 nm, (b) 375 nm, (c) 675 nm, (d) 1275 nm. The red dashed lines denote the start of AlN epilayer’s growth. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
stable intensity. Correspondingly, the structural quality of the four AlN samples is different. As shown in Table 1, sample C has the widest FWHMs of XRC, which are 778 and 932 arcsec for (002) and (102) plane, respectively. However, sample F has the narrowest FWHMs of XRC, which are 360 and 480 arcsec for (0 0 2) and (1 0 2) plane, respectively. Clearly, sample F has the best quality, indicating that the quality of AlN sample can be significantly improved when the thickness of nucleation layer increases. In addition, the thickness of those five AlN epilayers are also shown in Table 1. It is noted that the epilayer thickness of AlN samples A–F is different. Normally, for the growth of GaN films, if the film is not extremely thin and when the growth process of 3D to quasi-2D has been completed, the structural quality will not change too much when only the film thickness is increased [23]. Therefore, it is believed that the improvement of quality of the AlN samples (i.e., the narrowing of XRC HWFM) is not resulted from the thickness increase itself, but mainly due to the longer dual buffer growth process. In order to gain a further insight into the effect of dual AlN buffer layer on the quality of AlN epilayers, the surface morphology of two different dual AlN buffer layers is carefully examined. Two AlN dual buffer layer samples ‘‘a’’ and ‘‘b’’ are prepared and examined
by AFM. Actually, the growth of these two buffer layers was finished just before the further growth of thick AlN epilayers (as has been indicated by the arrows in Fig. 4(a) and (d), respectively). Therefore, ‘‘a’’ and ‘‘b’’ are grown under the same conditions as the AlN buffer layers of samples C and F, i.e., sample ‘‘a’’ has a 75 nm isolated layer plus a 225 nm nucleation layer, and ‘‘b’’ has a 75 nm isolated layer plus a 1275 nm nucleation layer, respectively. The obtained surface morphology of two AlN buffer layers is shown in Fig. 5(a and b) (10 10 lm2), respectively. The two surface images are quite different. As shown in Fig. 4, sample ‘‘a’’ has a relatively flat surface with the roughness value of 2.71 nm, while sample ‘‘b’’ has a much rougher surface with the roughness of 12.15 nm. The difference in surface morphology of AlN buffer layers leads to a difference in the further grown of AlN epilayer. The result is consistent very well with the afore-mentioned in situ optical reflectance measurement, i.e., if the dual AlN buffer layer has a flatter surface, the AlN islands in the initial growth stage will coalescence quicker. A lot of dislocations which are formed at the interface with sapphire will go through into the AlN epilayer, leading to a deteriorated structural quality [16,18]. However, a rougher surface of the buffer layer will be much better as the lateral growth of AlN will be promoted. It will take more time for the islands to
98
D.G. Zhao et al. / Journal of Alloys and Compounds 544 (2012) 94–98
is found that suitably increasing the thickness of nucleation layer is helpful to make the quality of AlN film better. The examination of surface morphology of dual AlN buffer layers reveals that the thicker the dual AlN buffer layer grown in the initial stage, the rougher the surface, and then the lateral growth of followed AlN epilayer is significantly promoted, leading to a better structural quality of the epilayers.
Acknowledgments The authors acknowledge the support from the National Science Fund for Distinguished Young Scholars (Grant No. 60925017), the National Natural Science Foundation of China (Grant Nos. 10990100, 60836003 and 60976045), and Tsinghua National Laboratory for Information Science and Technology (TNList) Cross-discipline Foundation.
References
Fig. 5. AFM images of the surface morphology of two AlN dual buffer layer samples ‘‘a’’ and ‘‘b’’ (10 10 lm2), which are used for the further growth of samples C and F, respectively. The roughness of ‘‘a’’ and ‘‘b’’ is 2.71 nm and 12.15 nm, which are shown in (a) and (b), respectively.
coalescence. In this process, a lot of dislocations will be able to bend or to eliminate each other, leading to an increased volume of defect-free columnar domains and improves the structural quality of AlN epilayer [16,18]. From the above analysis it is understood why the quality of AlN epilayers could be improved by employing relatively thick dual AlN buffer layer. 4. Summary In summary, a dual AlN buffer layer structure is proposed for the growth of high quality AlN films by MOCVD, which includes an isolated layer and a nucleation layer. The quality of AlN films is improved since the negative nitridation effect is weakened. It
[1] R. McClintock, K. Mayes, A. Yasan, D. Shiell, P. Kung, M. Razeghi, Appl. Phys. Lett. 86 (2005) 011117. [2] T. Tut, M. Gokkavas, B. Butun, S. Butun, E. Ulker, E. Ozbay, Appl. Phys. Lett. 89 (2006) 183524. [3] R. McClintock, A. Yasan, K. Minder, P. Kung, M. Razeghi, Appl. Phys. Lett. 87 (2005) 241123. [4] L. Sun, J.L. Chen, J.F. Li, H. Jiang, Appl. Phys. Lett. 97 (2010) 191103. [5] M. Imura, K. Nakano, N. Fujimoto, N. Okada, K. Balakrishnan, M. Iwaya, S. Kamiyama, H. Amano, I. Akasaki, T. Noro, T. Takagi, A. Bandoh, Jpn. J. Appl. Phys. 46 (2007) 1458–1462. [6] M.A. Khan, J.N. Kuzina, R.A. Skogman, D.T. Olson, M. MacMillan, W.J. Choyke, Appl. Phys. Lett. 61 (1992) 2539–2541. [7] A. Chitnis, J.P. Zhang, V. Adivarahan, M. Shatalov, S. Wu, R. Pachipulusu, V. Mandavilli, M.A. Khan, Appl. Phys. Lett. 82 (2003) 2565–2567. [8] H. Amano, N. Sawaki, I. Akasaki, Y. Toyoda, Appl. Phys. Lett. 48 (1986) 353–355. [9] S. Nakamura, Jpn. J. Appl. Phys. 30 (1991) L1705–L1707. [10] M. Takeuchi, H. Shimizu, R. Kajitani, K. Kawasaki, Y. Kumagai, A. Koukitu, Y. Aoyagi, J. Cryst. Growth 298 (2007) 336–340. [11] M. Takeuchi, S. Ooishi, T. Ohtsuka, T. Maegawa, T. Koyama, S.F. Chichibu, Y. Aogi, Appl. Phys.Express 1 (2008) 021102. [12] R.G. Banal, M. Funato, Y. Kawakami, Appl. Phys. Lett. 92 (2008) 241905. [13] Q.S. Paduano, D.W. Weyburne, J. Jasinski, Z. Liliental-Weber, J. Cryst. Growth 261 (2004) 259–265. [14] J. Jasinski, Z. Liliental-Weber, Q.S. Paduano, D.W. Weyburne, Appl. Phys. Lett. 83 (2003) 2811–2813. [15] F. Yan, M. Tsukihara, A. Nakamura, T. Yadani, T. Fukumoto, Y. Naoi, S. Sakai, Jpn. J. Appl. Phys. 43 (2004) L1057–L1059. [16] J. Han, T.B. Ng, R.M. Biefeld, M.H. Crawford, D.M. Follstaedt, Appl. Phys. Lett. 71 (1997) 3114–3116. [17] D.G. Zhao, J.J. Zhu, Z.S. Liu, S.M. Zhang, H. Yang, D.S. Jiang, Appl. Phys. Lett. 85 (2004) 1499–1501. [18] D.G. Zhao, D.S. Jiang, J.J. Zhu, Z.S. Liu, S.M. Zhang, H. Yang, J.W. Liang, J. Cryst. Growth 303 (2007) 414–418. [19] B. Heying, X.H. Wu, S. Keller, Y. Li, D. Kapolnek, B.P. Keller, S.P. DenBaars, J.S. Speck, Appl. Phys. Lett. 68 (1996) 643–645. [20] H. Heinke, V. Kirchner, S. Einfeldt, D. Hommel, Appl. Phys. Lett. 77 (2000) 2145–2147. [21] T.F. Kuech, D.J. Wolford, E. Veuhoff, V. Deline, P.M. Mooney, R. Potemski, J. Bradley, J. Appl. Phys. 62 (1987) 632–643. [22] J.P. Zhang, H.M. Wang, W.H. Sun, V. Adivarahan, S. Wu, A. Chitnis, C.Q. Chen, M. Shatalov, E. Kuokstis, J.W. Yang, M.A. Khan, J. Electron. Mater. 32 (2003) 364– 370. [23] D.G. Zhao, D.S. Jiang, J.J. Zhu, X. Guo, Z.S. Liu, S.M. Zhang, Y.T. Wang, Hui Yang, J. Alloys Compd. 487 (2009) 400–403.
Contents lists available at SciVerse ScienceDirect
Journal of Alloys and Compounds journal homepage: www.elsevier.com/locate/jalcom
Effect of dual buffer layer structure on the epitaxial growth of AlN on sapphire D.G. Zhao a,⇑, D.S. Jiang a, L.L. Wu a, L.C. Le a, L. Li a, P. Chen a, Z.S. Liu a, J.J. Zhu b, H. Wang b, S.M. Zhang b, H. Yang a,b a b
State Key Laboratory on Integrated Optoelectronics, Institute of Semiconductors, Chinese Academy of Sciences, P.O. Box 912, Beijing 100083, China Suzhou Institute of Nano-tech and Nano-bionics, Chinese Academy of Sciences, Suzhou 215125, China
a r t i c l e
i n f o
Article history: Received 6 June 2012 Received in revised form 10 July 2012 Accepted 23 July 2012 Available online 1 August 2012 Keywords: Nitride materials Crystal growth X-ray diffraction
a b s t r a c t A dual AlN buffer layer structure, including an isolated layer and a nucleation layer, is proposed to improve the growth of AlN films on sapphire substrate by metal organic chemical vapor deposition. This method is aimed to weaken the negative nitridation effect and improve lateral growth condition in the initial growth stage. It is found that suitably increasing the thickness of the nucleation layer is in favor of a better structural quality of the AlN film. An examination of surface morphology by atomic force microscopy suggests that the thicker the dual AlN buffer layer, the rougher the surface, and a higher quality of AlN epilayer is resulted. Ó 2012 Elsevier B.V. All rights reserved.
1. Introduction Group-III nitrides have been investigated for many years due to their extensive applications in optoelectronics and microelectronics. For example, AlGaN solar-blind ultraviolet photodetectors are able to be used in missile detection, secure space-to-space communication, flame monitoring and so on. Great successes have been achieved in the fabrication of AlGaN solar-blind ultraviolet photodetectors, including the realization of focal plane arrays and avalanche photodiodes [1–4]. The quality of AlN films plays an important role in determining the performance of back-illuminated AlGaN ultraviolet photodetectors. However, it is very difficult to grow high quality AlN layers, mainly because the surface mobility of Al atoms is very low. Heteroepitaxial AlN films often exhibit poor crystalline quality or poor surface morphology due to the limited lateral growth. Usually, a special metalorganic chemical vapor deposition (MOCVD) system with an extremely high growth temperature around 1400 °C or with a migration-enhanced growth mode can be employed to promote two-dimensional (2D) grow mode and obtain high quality AlN films [5–7]. On the other hand, the two-step growth method had been widely used to grow nitride materials since it was developed for the growth of GaN epilayer on sapphire substrate by MOCVD [8,9]. The buffer layer is critical in the growth of nitride materials to overcome the problems caused by the large lattice and thermal mismatches with sapphire substrate, and the density of threading dislocations is largely reduced. In fact, both low- and high-temperature AlN buffer layers ⇑ Corresponding author. Tel.: +86 10 82304208; fax: +86 10 82304242. E-mail address: [email protected] (D.G. Zhao). 0925-8388/$ - see front matter Ó 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jallcom.2012.07.123
have been employed to AlN epitaxial growth [10–12], and the initial process is very important for the growth of high quality AlN epilayers [12,13]. It is known that the nitridation of sapphire substrate is a negative factor which hinders the high quality growth of AlN films since the Al and N faces have different polarities and different growth rates, leading to the formation of inversion domains and a rough surface morphology [13,14]. Therefore, suppressing the nitridation effect of sapphire substrate should also be useful to improve the quality of AlN [15]. In this paper, the dual AlN buffer layer, instead of commonly used single AlN buffer layer (or nucleation layer), is employed during the AlN growth by MOCVD, which is significantly helpful to grow high quality AlN epilayers even when a commercially-available conventional MOCVD system is used. The quality of AlN films could be improved further if the thickness of dual AlN buffer layer is made suitably larger. The additional in situ optical reflectance and AFM measurements of such a dual AlN buffer layer show that its surface roughness increases with increasing thickness, and the followed lateral growth of AlN epilayers is thus enhanced.
2. Experimental procedure All the AlN samples studied in this work are grown on c-plane sapphire substrate by MOCVD. The trimethylaluminum (TMAl) and ammonia (NH3) are used as Al and N sources, respectively. The H2 are used as carrier gas. Two kinds of AlN samples are grown by MOCVD in this work, one kind is with dual AlN buffer layer, the other is with single AlN buffer layer, as schematically shown in Fig. 1(a and b), respectively. For the growth of dual AlN buffer layer, a lower growth temperature Tg of 900 °C are used for its first sublayer, while a higher Tg of 1080 °C are used for its second sublayer. For the AlN samples with single AlN buffer layer whose growth condition is the same as that of the above-mentioned second AlN sublayer of the dual buffer layer, i.e., the growth temperature is 1080 °C. The growth
D.G. Zhao et al. / Journal of Alloys and Compounds 544 (2012) 94–98
95
Fig. 1. The schematic diagrams of AlN sample structures: (a) with dual AlN buffer layer, (b) with single AlN buffer layer. procedure of studied AlN samples is as follows, firstly a dual or single AlN buffer layer is deposited on the sapphire substrate, then the AlN epilayer is grown on the buffer layer at 1100 °C. All growth processes are monitored by the in situ optical reflectance, which is used as a powerful tool to study the GaN growth mechanism by MOCVD [16–18]. The quality of AlN films is characterized by the full-width-athalf-maximum (FWHM) of double crystal X-ray x-scan rocking curves (XRC), ex situ performed on a Rigaku SLX-1AL X-ray diffractometer. The surface morphology is examined by the atomic force microscopy (AFM) at room temperature.
3. Results and discussion Fig. 2(a) shows the in situ measured trace of optical reflectance from the surface of sample A which is grown with the dual AlN buffer layer structure by MOCVD. As shown in Fig. 2(a), the trace is divided into three parts as follows, corresponding to the three growth stages of AlN film deposited on the dual AlN buffer layer [16–18]: (I) the growth of AlN ‘‘isolated’’ layer (AlN sublayer 1) at 900 °C which is employed to separate sapphire substrate from nucleation layer and suppress the nitridation effect. The thickness of sublayer 1 is about 75 nm, the corresponding growth rate is about 0.04 nm/s; (II) the growth of AlN ‘‘nucleation’’ layer (AlN sublayer 2) at 1080 °C. The thickness of sublayer 2 is about 375 nm, the corresponding growth rate is about 0.32 nm/s; (III) the growth of thick AlN epilayer at 1100 °C, the corresponding growth rate is about 0.2 nm/s. In order to suppress the nitridation, the conditions of growth temperature, growth rate, and reactor pressure during the growth of isolated AlN layer are carefully adjusted. In Fig. 2(a) the trace in stages I and II corresponds to the grow process of dual AlN buffer layer. During the growth stage III, the reflectance increases gradually, and the growth rate is intentionally reduced to allow a better surface migration of Al atoms. It is noted that during the growth of AlN nucleation layer (in stage II), the intensity of in situ optical reflectance decreases gradually, i.e., the surface roughness of material increases. A rough AlN layer may contain a lot of islands which may provide nucleation centers. However, afterwards in stage III the surface of the AlN epilayer gradually turns to be optically smoother, which means that the lateral growth and coalescence of AlN islands emerge. Finally, the quasi 2D growth of the AlN layer occurs. An oscillation of the reflectivity intensity with large and equal amplitude is well observed. In fact, the practical conditions used for the AlN buffer layer growth are very different from those used for the GaN buffer layer growth. For example, the growth temperature of AlN buffer layer must be much higher (up to 900 and 1080 °C as in our case), otherwise the crystalline quality will be very poor
Fig. 2. The three stages of in situ optical reflectance traces in the whole growth process of AlN: (a) sample A grown with dual AlN buffer layer including an isolated layer and a nucleation layer. (b) sample B grown with single AlN buffer layer (an AlN nucleation layer).
due to the limited lateral migration of Al atoms, and it will be difficult to form necessary islands and develop enough amount of nucleation centers for further epitaxial growth. However, there is still a basic rule valid for both AlN and GaN growth by MOCVD, i.e., it is very important to better control the processes of nucleation and transition from three-dimensional (3D) to quasi-2D growth mode, as what has been widely reported for the initial growth stage of GaN [16–18]. As a comparison, we have also investigated the growth process of AlN film sample B which employs a single AlN buffer layer (the structure is shown in Fig. 1(b)). As shown by the optical reflectance trace of sample B in Fig. 2(b), in the beginning stage of the growth process, i.e., the temperature ramping process, the reflectance intensity is very low, indicating that the stage of ‘‘isolated’’ AlN layer grown at 900 °C is omitted for this sample, only the growth of the single AlN buffer layer (also named as AlN nucleation layer) is remained and is marked as growth stages II in Fig. 2(b). There is actually a great difference between the in situ optical reflectance traces of samples A and B. The average intensity of optical reflectance of sample B in stages II and III almost keeps constant, i.e., there is nearly no any obvious growth mode transformation from 3D to quasi 2D. However, for sample A with the dual AlN buffer layer, there is an extended transformation process from 3D to quasi 2D growth mode, which is very beneficial to the improvement of structural quality of epilayers. The structural quality of AlN samples A and B is checked by the x-scan XRC. As shown in Table 1, the (0 0 2) and (1 0 2) plane FWHMs of XRC of sample A are narrow, i.e., 472 and 796 arcsec, respectively, while they are much broader for sample B, i.e., 810 and 966 arcsec, respectively. It is known that the screw and edge dislocation densities could be indirectly represented by the (0 0 2)
96
D.G. Zhao et al. / Journal of Alloys and Compounds 544 (2012) 94–98
Table 1 The growth conditions of AlN dual or single buffer layers and the FWHM results of X-ray diffraction rocking curves for grown AlN epilayers. Samples
Thickness of buffer layer (nm) Isolated layer
Nucleation layer
A, D B C E F
75 0 75 75 75
375 450 225 675 1275
and (1 0 2) FWHMs of XRC, respectively [19,20]. That is to say, both screw and edge dislocation densities of sample A are much lower than those of sample B. Correspondingly, the structural quality of sample A is better than sample B. It is noted that the clear difference between two samples is not only in the structural quality, but also in surface morphology. Fig. 3(a and b) show the surface morphology of AlN samples A and B (5 5 lm2) obtained by AFM, respectively. Their surface roughness is 0.13 nm and 40.65 nm, respectively. The surface of sample A is much smoother than sample B. Furthermore, a step-flow growth mode could be detected from the surface morphology of sample A. This result indicates that the dual AlN buffer layer structure can play an important role in improving the structural quality of AlN epilayers. It is well known that the epitaxial growth of GaN and AlN has remarkable differences. Especially, the surface mobility of Al atoms is much lower than that of Ga atoms under similar growth conditions [21,22], making it much more difficult for AlN to achieve the 2D growth mode. In addition, the possible nitridation of sapphire substrate is another reason to deteriorate the quality of AlN layers because the different growth velocities on the Al and N polarity surfaces will lead to an unfavorable deposition condition, introducing inversion domain defects in the epilayers, and the N polaritydominant layer growth always results in a relatively rough surface morphology [13]. Some practical methods had been suggested to avoid nitridation and thus had improved the quality of AlN films. For example, the technology of initially alternating supply of ammonia may eliminate the negative effect of nitridation [15]. In our work, it is proposed to add a low temperature-grown isolated AlN layer to weaken the nitridation effect. It is assumed to be helpful to improve the structural quality and surface morphology of the AlN film. It is found that a relatively larger thickness of nucleation layer in the dual AlN buffer layer structure has a positive effect on the AlN growth by MOCVD. Fig. 4 shows the traces of in situ optical reflectance during the growth of four AlN samples, i.e., samples C, D, E and F, with varying thickness of AlN nucleation layer, where Fig. 4(a and d) correspond to the thickness of AlN nucleation layer of 225, 375, 675, and 1275 nm, respectively. For these samples the thickness of isolated layer is kept the same as 75 nm. The dashed lines in Fig. 4 denote the start of AlN epilayer’s growth after the completion of the buffer layer. It is interesting to note that there is a large difference in the four optical reflectance curves. For example, for sample C which has the thinnest nucleation layer thickness, the optical reflectance of AlN epilayer quickly increases to a stable value after 3280 s after the growth of dual AlN buffer layer. The transformation process from 3D to quasi-2D occurs shortly. However, for sample D (here sample D is the same one as sample A, but is called a different name here in this series), the case is different. After the growth of 375 nm AlN nucleation layer, as shown in Fig. 4(b), the surface optical reflectance is lower than that of sample C, implying that the surface is rougher. In the followed growth of thick AlN epilayer, the average intensity of in situ optical reflectance increases gradually. After about 4502 s, the average intensity of optical reflectance reaches a stable value. Clearly, it takes a longer time to complete the growth transformation from 3D to 2D. Such
Thickness of AlN epilayer (lm)
XRC FWHM of AlN (arcsec) (0 0 2)
(1 0 2)
1 0.75 0.75 1.65 2.4
472 810 778 435 360
796 966 932 558 480
Fig. 3. The surface morphology of AlN epilayers A (a) and B (b) (5 5 lm2). The surface roughness is 0.13 nm and 40.65 nm, respectively.
transformation process is even longer for samples E and F. After growth of AlN nucleation layer, as shown in Fig. 4(c and d), the intensity of in situ optical reflectance of samples E and F decreases further, and it takes much longer time, i.e., about 7867 and 11222 s, respectively, for the growth mode transformation from 3D to 2D. From the four traces of in situ optical reflectance of growth processes it is found that when the thickness of nucleation layer is larger, it will take longer time for the optical reflectance to reach a
D.G. Zhao et al. / Journal of Alloys and Compounds 544 (2012) 94–98
97
Fig. 4. The in situ optical reflectance traces of 4 AlN samples with dual AlN buffer layer. The thickness of AlN nucleation sublayer is: (a) 225 nm, (b) 375 nm, (c) 675 nm, (d) 1275 nm. The red dashed lines denote the start of AlN epilayer’s growth. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
stable intensity. Correspondingly, the structural quality of the four AlN samples is different. As shown in Table 1, sample C has the widest FWHMs of XRC, which are 778 and 932 arcsec for (002) and (102) plane, respectively. However, sample F has the narrowest FWHMs of XRC, which are 360 and 480 arcsec for (0 0 2) and (1 0 2) plane, respectively. Clearly, sample F has the best quality, indicating that the quality of AlN sample can be significantly improved when the thickness of nucleation layer increases. In addition, the thickness of those five AlN epilayers are also shown in Table 1. It is noted that the epilayer thickness of AlN samples A–F is different. Normally, for the growth of GaN films, if the film is not extremely thin and when the growth process of 3D to quasi-2D has been completed, the structural quality will not change too much when only the film thickness is increased [23]. Therefore, it is believed that the improvement of quality of the AlN samples (i.e., the narrowing of XRC HWFM) is not resulted from the thickness increase itself, but mainly due to the longer dual buffer growth process. In order to gain a further insight into the effect of dual AlN buffer layer on the quality of AlN epilayers, the surface morphology of two different dual AlN buffer layers is carefully examined. Two AlN dual buffer layer samples ‘‘a’’ and ‘‘b’’ are prepared and examined
by AFM. Actually, the growth of these two buffer layers was finished just before the further growth of thick AlN epilayers (as has been indicated by the arrows in Fig. 4(a) and (d), respectively). Therefore, ‘‘a’’ and ‘‘b’’ are grown under the same conditions as the AlN buffer layers of samples C and F, i.e., sample ‘‘a’’ has a 75 nm isolated layer plus a 225 nm nucleation layer, and ‘‘b’’ has a 75 nm isolated layer plus a 1275 nm nucleation layer, respectively. The obtained surface morphology of two AlN buffer layers is shown in Fig. 5(a and b) (10 10 lm2), respectively. The two surface images are quite different. As shown in Fig. 4, sample ‘‘a’’ has a relatively flat surface with the roughness value of 2.71 nm, while sample ‘‘b’’ has a much rougher surface with the roughness of 12.15 nm. The difference in surface morphology of AlN buffer layers leads to a difference in the further grown of AlN epilayer. The result is consistent very well with the afore-mentioned in situ optical reflectance measurement, i.e., if the dual AlN buffer layer has a flatter surface, the AlN islands in the initial growth stage will coalescence quicker. A lot of dislocations which are formed at the interface with sapphire will go through into the AlN epilayer, leading to a deteriorated structural quality [16,18]. However, a rougher surface of the buffer layer will be much better as the lateral growth of AlN will be promoted. It will take more time for the islands to
98
D.G. Zhao et al. / Journal of Alloys and Compounds 544 (2012) 94–98
is found that suitably increasing the thickness of nucleation layer is helpful to make the quality of AlN film better. The examination of surface morphology of dual AlN buffer layers reveals that the thicker the dual AlN buffer layer grown in the initial stage, the rougher the surface, and then the lateral growth of followed AlN epilayer is significantly promoted, leading to a better structural quality of the epilayers.
Acknowledgments The authors acknowledge the support from the National Science Fund for Distinguished Young Scholars (Grant No. 60925017), the National Natural Science Foundation of China (Grant Nos. 10990100, 60836003 and 60976045), and Tsinghua National Laboratory for Information Science and Technology (TNList) Cross-discipline Foundation.
References
Fig. 5. AFM images of the surface morphology of two AlN dual buffer layer samples ‘‘a’’ and ‘‘b’’ (10 10 lm2), which are used for the further growth of samples C and F, respectively. The roughness of ‘‘a’’ and ‘‘b’’ is 2.71 nm and 12.15 nm, which are shown in (a) and (b), respectively.
coalescence. In this process, a lot of dislocations will be able to bend or to eliminate each other, leading to an increased volume of defect-free columnar domains and improves the structural quality of AlN epilayer [16,18]. From the above analysis it is understood why the quality of AlN epilayers could be improved by employing relatively thick dual AlN buffer layer. 4. Summary In summary, a dual AlN buffer layer structure is proposed for the growth of high quality AlN films by MOCVD, which includes an isolated layer and a nucleation layer. The quality of AlN films is improved since the negative nitridation effect is weakened. It
[1] R. McClintock, K. Mayes, A. Yasan, D. Shiell, P. Kung, M. Razeghi, Appl. Phys. Lett. 86 (2005) 011117. [2] T. Tut, M. Gokkavas, B. Butun, S. Butun, E. Ulker, E. Ozbay, Appl. Phys. Lett. 89 (2006) 183524. [3] R. McClintock, A. Yasan, K. Minder, P. Kung, M. Razeghi, Appl. Phys. Lett. 87 (2005) 241123. [4] L. Sun, J.L. Chen, J.F. Li, H. Jiang, Appl. Phys. Lett. 97 (2010) 191103. [5] M. Imura, K. Nakano, N. Fujimoto, N. Okada, K. Balakrishnan, M. Iwaya, S. Kamiyama, H. Amano, I. Akasaki, T. Noro, T. Takagi, A. Bandoh, Jpn. J. Appl. Phys. 46 (2007) 1458–1462. [6] M.A. Khan, J.N. Kuzina, R.A. Skogman, D.T. Olson, M. MacMillan, W.J. Choyke, Appl. Phys. Lett. 61 (1992) 2539–2541. [7] A. Chitnis, J.P. Zhang, V. Adivarahan, M. Shatalov, S. Wu, R. Pachipulusu, V. Mandavilli, M.A. Khan, Appl. Phys. Lett. 82 (2003) 2565–2567. [8] H. Amano, N. Sawaki, I. Akasaki, Y. Toyoda, Appl. Phys. Lett. 48 (1986) 353–355. [9] S. Nakamura, Jpn. J. Appl. Phys. 30 (1991) L1705–L1707. [10] M. Takeuchi, H. Shimizu, R. Kajitani, K. Kawasaki, Y. Kumagai, A. Koukitu, Y. Aoyagi, J. Cryst. Growth 298 (2007) 336–340. [11] M. Takeuchi, S. Ooishi, T. Ohtsuka, T. Maegawa, T. Koyama, S.F. Chichibu, Y. Aogi, Appl. Phys.Express 1 (2008) 021102. [12] R.G. Banal, M. Funato, Y. Kawakami, Appl. Phys. Lett. 92 (2008) 241905. [13] Q.S. Paduano, D.W. Weyburne, J. Jasinski, Z. Liliental-Weber, J. Cryst. Growth 261 (2004) 259–265. [14] J. Jasinski, Z. Liliental-Weber, Q.S. Paduano, D.W. Weyburne, Appl. Phys. Lett. 83 (2003) 2811–2813. [15] F. Yan, M. Tsukihara, A. Nakamura, T. Yadani, T. Fukumoto, Y. Naoi, S. Sakai, Jpn. J. Appl. Phys. 43 (2004) L1057–L1059. [16] J. Han, T.B. Ng, R.M. Biefeld, M.H. Crawford, D.M. Follstaedt, Appl. Phys. Lett. 71 (1997) 3114–3116. [17] D.G. Zhao, J.J. Zhu, Z.S. Liu, S.M. Zhang, H. Yang, D.S. Jiang, Appl. Phys. Lett. 85 (2004) 1499–1501. [18] D.G. Zhao, D.S. Jiang, J.J. Zhu, Z.S. Liu, S.M. Zhang, H. Yang, J.W. Liang, J. Cryst. Growth 303 (2007) 414–418. [19] B. Heying, X.H. Wu, S. Keller, Y. Li, D. Kapolnek, B.P. Keller, S.P. DenBaars, J.S. Speck, Appl. Phys. Lett. 68 (1996) 643–645. [20] H. Heinke, V. Kirchner, S. Einfeldt, D. Hommel, Appl. Phys. Lett. 77 (2000) 2145–2147. [21] T.F. Kuech, D.J. Wolford, E. Veuhoff, V. Deline, P.M. Mooney, R. Potemski, J. Bradley, J. Appl. Phys. 62 (1987) 632–643. [22] J.P. Zhang, H.M. Wang, W.H. Sun, V. Adivarahan, S. Wu, A. Chitnis, C.Q. Chen, M. Shatalov, E. Kuokstis, J.W. Yang, M.A. Khan, J. Electron. Mater. 32 (2003) 364– 370. [23] D.G. Zhao, D.S. Jiang, J.J. Zhu, X. Guo, Z.S. Liu, S.M. Zhang, Y.T. Wang, Hui Yang, J. Alloys Compd. 487 (2009) 400–403.