Quality-enhanced GaN epitaxial films on Si(111) substrates by in situ deposition of SiN on a three-dimensional GaN template

Abstract

High-quality crack-free GaN epitaxial films were successfully grown on Si(111) substrates using metal–organic chemical vapor deposition by in situ depositing SiN on a 3-dimensional (3D) GaN template. The GaN epitaxial films with 0–600 s 3D GaN templates were grown. It was found that the crystalline quality of GaN epitaxial films could be optimized and the cracks could be suppressed within the window of the growth time of the 3D GaN template between 0 to 600 s. For the sample with 180 s 3D GaN template, X-ray rocking curve measurements revealed the minimum full-width at half-maximum values of 348 and 406 arcsec for GaN(0002) and GaN(10−12), respectively, indicating the best crystalline quality among all the samples. Furthermore, scanning electron microscopy and in situ reflectance curves suggested that the 3D GaN template growth time changed the size and density of the GaN islands on the 3D GaN template surface, thus affecting the subsequent coalescence process of GaN islands after SiN deposition, and consequently resulted in variation in the crystalline quality and the stress of GaN epitaxial films. This work broadens the approach to achieve high-quality crack-free GaN epitaxial films on Si substrates for applications in GaN-based devices.

---

1. Introduction

So far, enormous attention has been cast towards epitaxial growth of GaN films on Si substrates due to their numerous advantages,^1–3 such as being able to utilize a larger wafer size, much lower cost and the reasonably high thermal conductivity.^4,5 In addition, Si substrates have a potential for integrating nitride-on-Si structures for electronic applications.^6,7 All these advantages enable GaN-on-Si to be developed for applications in light-emitting diodes (LEDs), high electron mobility transistors (HEMTs), photo-detectors, and so on.^8–12

Despite the great potential of GaN-based devices on Si substrates, it is still a challenge to achieve high-performance GaN-based devices on Si substrates. On the one hand, the GaN epitaxial films usually contain high-density threading dislocations (TDs) as a result of the large 16.9% lattice mismatch of between GaN and Si. Having a high-density of TDs deteriorates the device performance severely.^2,13 On the other hand, the mismatch in the coefficient of thermal expansion (CTE) between GaN and Si is as high as 116%, which results in enormous tensile stress. The exorbitant tensile stress is manifested by generating cracks on the film surfaces during the cooling process,^14,15 thereby hindering the fabrication and use of the devices.^2,16

In situ SiN deposition is one of the most popular methods to suppress the TDs in GaN epitaxial films. It is commonly reported that the in situ SiN layer is deposited on the flat template, such as 2 dimensional (2D) GaN and flat AlGaN buffer layers. On this occasion, the SiN layer acts as an anti-surfactant to promote the transition from 2D to 3 dimensional (3D) mode for GaN growth,^17,18 thereby reducing the TD density in the GaN epitaxial films. Through this method, the quality of GaN epitaxial films can be significantly improved. However, both the deposited condition of SiN and the corresponding growth process of GaN have a significant influence on the GaN quality. Zhu et al. obtained high-quality GaN-on-Si epitaxial films by optimizing the deposition time of the SiN layer, the full-width at half-maximum (FWHM) values of X-ray rocking curves (XRCs) were as low as 373 and 501 arcsec for GaN(0002) and GaN(10−11), respectively.² Cheng et al. concluded that a high temperature process for GaN growth after the SiN deposition is beneficial to improve the quality of GaN epitaxial films.¹⁹ Except for the in situ SiN, the AlN interlayer was also reported to be able to stir up the 3D GaN growth to suppress the TDs.^20,21 The mechanism can be ascribed to the lattice mismatch between AlN (3.112 Å) and GaN (3.189 Å), which facilitates the island growth mode of GaN.²² Moreover, this lattice mismatch could induce compressive stress in the GaN layer to offset the tensile stress induced by the CTE mismatch between GaN and Si, thus suppressing the formation of cracks. Particularly, we inferred that if the in situ SiN is deposited onto the 3D GaN layer (3D GaN template), resulting from the AlN interlayer, the TD density could be further reduced. Meanwhile, the cracks can be effectively controlled.

In this work, we report on the growth of crack-free high-quality GaN epitaxial films on Si substrates by in situ deposition of SiN on the 3D GaN template produced by an AlN interlayer. The influence and its mechanism of the growth time of 3D GaN template on the crystalline quality and the stress condition of GaN epitaxial films were systematically studied.

2. Experimental

All the samples in this work were grown on 4-inch Si(111) substrates by Veeco K465i metal–organic chemical vapor deposition (MOCVD). Trimethyl gallium (TMGa), trimethyl aluminum (TMAl), trimethyl indium (TMIn), silane (SiH₄) and ammonia were utilized as precursors for Ga, Al, In, Si and N sources, respectively. Before GaN growth, a 100 nm-thick AlN buffer layer was directly grown on the Si substrate, followed by a 550 nm-thick step-graded AlGaN buffer layer composed of 150 nm-thick Al_0.65Ga_0.35N, 200 nm-thick Al_0.40Ga_0.50N, 200 nm-thick Al_0.15Ga_0.85N, and a 30 nm-thick AlN interlayer. Afterwards, a certain growth time of 3D GaN template was grown on the AlN interlayer at 500 Torr, and the ultrathin SiN layer was then deposited by passing the SiH₄ and NH₃ in the absence of TMGa. The deposition time of the SiN layer was set at 30 s. Subsequently, the growth pressure was decreased to 200 Torr to promote the transition from 3D to 2D GaN growth. On this condition, a 1.5 μm thick n-doped GaN layer was grown. The identical structures with 0, 120, 180, 300 and 600 s 3D GaN templates were grown and labeled as samples A, B, C, D and E, respectively. The schematic structure of GaN epitaxial films are shown in Fig. 1a, and the typical X-ray diffraction (XRD) 2θ–ω scan of GaN(0002) confirmed the structures of AlN and step-graded Al_xGa_1−xN buffer layers, as shown in Fig. 1b. It should be noted that the SiN layer just had the thickness of several atomic layers,¹⁷ which is hard to be detected by XRD measurements. Also, another GaN epitaxial film with a 600 s grown 3D GaN layer but without in situ SiN deposition was grown as a reference, and is denoted as sample F. All the growth temperatures for the 3D GaN template (3D GaN layer), in situ SiN layer and n-doped GaN layer were set at 1050 °C.

Fig. 1 (a) The schematic of GaN epitaxial films grown on Si substrates, (b) typical XRD 2θ–ω scan of GaN(0002) for GaN epitaxial films on Si substrates.

The surface morphology was investigated by optical microscopy (Olympus, BX51M), and scanning electron microscopy (SEM: Nova Nano-SEM 430), atomic force microscopy (AFM, Bruker Dimension Edge), respectively. The crystalline quality of the as-grown films was characterized by high-resolution X-ray diffraction (HRXRD, Bruker D8 X-ray diffractometer with a Cu Kα1 X-ray source, λ = 1.5406 Å). The dislocation condition of as-grown films was studied by high-resolution transmission electron microscopy (HR-TEM, JEOL3000F). The residual stress was analyzed using micro-Raman spectroscopy (Renishaw inVia Raman spectrometer with a 532 nm laser as the excitation source). The in situ growth status was monitored using an in situ reflectance measurement.

3. Results and discussion

The effect of growth time of the 3D GaN template, on the crystalline quality of complete GaN epitaxial films was evaluated by the FWHM values of XRCs for GaN(0002) and GaN(10−12), which is shown in Fig. 2. One can easily find that the FWHM values for both GaN(0002) and GaN(10−12) change like V curves when the growth time of the 3D GaN template is increased from 0 to 600 s. First, the FWHM values of GaN(0002) and GaN(10−12) were reduced from 421 to 348 and 563 to 406 arcsec, respectively, as the growth time of the 3D GaN template was increased from 0 to 180 s. Subsequently, the FWHM values of GaN(0002) and GaN(10−12) increased from 348 to 414, and 406 to 547 arcsec, respectively, when the growth time of the 3D GaN template was further increased from 180 to 600 s. As we know, the FWHM of GaN is associated with the dislocation density in the GaN layer. The larger the FWHM is, the higher the dislocation density would be.²³ Therefore, the change in FWHMs of GaN(0002) and GaN(10−12) implies that dislocation density of the samples decreased at first and then increased, when the growth time of the 3D GaN template increased from 0 to 600 s. The lowest dislocation density was obtained when the growth time of the 3D GaN template was 180 s. Regarding sample F without SiN, the FWHM values of GaN(0002) and GaN(10−12) were as high as 423 and 582 arcsec, respectively, which were higher than those of all the samples with SiN. These results demonstrated that the decrease in the dislocation density for samples A–E could be ascribed to the in situ SiN deposition. However, the effect of dislocation suppression, induced by SiN, was strongly influenced by the growth time of the 3D GaN template grown before the SiN deposition. In addition, the XRC FWHMs for GaN(0002) and GaN(10−12) were 348, and 406 arcsec, respectively, which were slightly smaller than the results in most of the reports involving the growth of GaN epitaxial films on Si substrates.^2,24–27 This demonstrates that depositing SiN on the 3D GaN template is an effective method to suppress the dislocation in GaN epitaxial films on Si substrates.

Fig. 2 Growth time of 3D GaN template dependence of FWHM values for GaN(0002) and GaN(10−12) for samples A–F.

The crack condition was observed by optical microcopy, as shown in Fig. 3. It can be easily found that samples A, D and E showed clear cracks. It can be attributed to the AlN interlayer and AlGaN graded buffer, which erected compressive stress in the GaN layer during the growth period of GaN to offset the tensile stress formed during the cooling process. In contrast, samples B, C and F yielded crack-free surfaces, shown in Fig. 3b, c and f. The cracks were caused by the relatively high residual tensile stress. The tensile stress formed during the cooling process is found to exceed the limitation and has to be released by generating cracks. The crack-free surfaces of samples B and C suggest that the crack can be also controlled through introducing the in situ SiN. The difference in the crack condition for samples A–E with the in situ SiN deposition demonstrates that the growth time of the 3D GaN template played a significant role on the stress control of the GaN epitaxial films on Si substrates.

Fig. 3 Optical microscopy images of samples (a) A, (b) B, (c) C, (d) D, (e) E and (f) F.

Raman spectroscopy was utilized to further analyze the residual stress of samples. Fig. 4a shows the Raman spectra of samples A–F, where both the GaN E₂ peaks and the Si phonon peaks are indicated. Fig. 4b exhibits that the GaN E₂ peak positions of samples A–F are at 564.9, 565.5, 565.6, 565.3, 565.0 and 566.1 cm⁻¹, respectively. One can notice that all the GaN E₂ peaks deviated to the low-frequency side in contrast to the relaxed GaN E₂ peak located at 567.5 cm⁻¹.^23,24 This indicates that all the samples were under tensile stress. The precise residual tensile stress could be calculated as follows:^28,29

Δω = 4.3σ_χχ cm⁻¹ GPa⁻¹ (1)

where, Δω is the shift of GaN E₂, and σ_χχ represents the residual stress of GaN. Based on this equation, we calculated the residual tensile stress of all the samples, as shown in Fig. 4c. The residual tensile stress for samples A–F was 0.605, 0.465, 0.442, 0.512, 0.581 and 0.326 GPa, respectively. These data well support the results of optical microcopy, which states that higher residual tensile stress leads to the formation of cracks. Furthermore, it can be easily found that, sample F without SiN reveals the lowest residual tensile stress, which suggests that the in situ SiN layer increases the residual tensile stress of GaN on Si. According to T. Riemann et al.,³⁰ the in situ SiN is harmful for the stress control on the growth of GaN on Si. This is because the inserted SiN isolates the stress between upper and lower GaN layer, thereby neutralizing the compensative compressive stress caused by the AlN interlayer or AlGaN buffer layer during the growth period of the GaN layer, and eventually increases the residual tensile stress after the cooling process. Therefore, the increase in the residual tensile stress in the samples with in situ SiN can be ascribed to the identical mechanism. However, for samples A–E with in situ SiN, the residual tensile stress alleviates first, and then increases as the growth time of 3D GaN template increases from 0 to 600 s. The minimum residual tensile stress can be obtained for sample C with 180 s of 3D GaN template growth. This demonstrates that the increase in the residual tensile stress, induced by the SiN, can be impaired by optimizing the growth time of 3D GaN template. As a result, both the crack-free surface and a significant improvement in the crystalline quality can be achieved for the GaN epitaxial films on Si substrates through this optimization.

Fig. 4 (a) Raman spectrum, (b) Raman shifts and (c) residual stresses of samples A–F.

AFM was deployed to study the surface morphology of the as-grown GaN epitaxial films, shown in Fig. 5. It is clear that all the samples displayed step-and-terrace structured surfaces. When the growth time for 3D GaN template was 0 s, named sample A, the corresponding root-mean square (RMS) surface roughness was measured to be 0.46 nm. When the growth time of 3D GaN template increased from 0 to 180 s, the step-and-terrace structure unveiled a better arrangement. The surface morphology became smoother, which was confirmed by a 0.36 nm RMS roughness. This can be ascribed to the reduced TD density. The TDs would pin at the edge of the terrace, and thereby disorder the step-terrace structure. As the growth time of 3D GaN template was further increased, the TD density also increased, and thereby leading the step-and-terrace structure to disorder again. Accordingly, the RMS roughness increased from 0.36 to 0.44 nm as the growth time of 3D GaN template increased from 180 to 600 s. In addition, sample F revealed the disordered step-and-terrace surface with the largest RMS roughness of 0.48 nm. Evidently, the SiN layer is helpful to improve the surface quality of GaN epitaxial films.

Fig. 5 AFM images of samples (a) A, (b) B, (c) C, (d) D, (e) E and (f) F.

Fig. 6 shows the surface morphology of 0, 180 and 600 s 3D GaN templates, respectively. It is clear that the 0 s 3D GaN template, namely the AlN interlayer, had a flat surface, which is indicated in Fig. 6a. In the case of 180 s grown 3D GaN template, shown in Fig. 6b, it exhibited high-density independent GaN islands, which suggest an early stage of 3D GaN growth process. When the growth time of 3D GaN template reached 600 s, a nearly coalesced surface was formed, and only a few un-coalesced voids were left behind, Fig. 6c. It is clear that GaN islands tend to coalesce as the growth time increases. Therefore, the surface morphology of 120 s and 300 s 3D GaN templates can be deduced. In comparison to 180 s 3D GaN template, the former had a lower-density of GaN islands with a smaller size and larger area of exposed AlN surface, and the latter had relatively coalesced GaN islands and a smaller area of exposed AlN surface. These results clearly express that the density and the size of 3D GaN islands on the 3D GaN template surface were strongly influenced by its growth time.

Fig. 6 SEM images of (a) 0 s (AlN interlayer), (b) 180 s and (c) 600 s 3D GaN template.

The in situ monitoring reflectance curves were adopted to investigate the GaN growth process evolution. Fig. 7 represents the in situ reflectance curves of samples A–F. Both the positions of SiN deposition and the recovery duration of GaN were identified by the oscillation amplitude of the reflectance signals reaching a maximum,¹⁹ and are respectively marked. In the case of sample F, a gradually broadening oscillation amplitude was observed, which is consistent with the results of Fig. 6, and indicates a gradual coalescence of GaN islands without SiN deposition.^31,32 In large contrast to sample F, the oscillation amplitudes of the reflectance for samples A–E were dramatically reduced after the SiN deposition, leading to a significant extension in the recovery duration. This extension in the recovery duration suggests that the coalescence process of GaN islands was dramatically prolonged because of the SiN deposition. Moreover, it is worth noting that the recovery durations for samples A–E were also different. It is clearly shown that the recovery duration was extended at first when the growth time of 3D GaN template increased from 0 to 180 s. Then, the recovery duration was shortened, as the growth time of 3D GaN template further increased from 180 to 600 s. The longest recovery duration was found with sample C, with 180 s of 3D GaN template growth. As mentioned above, the density and the size of GaN islands on the 3D GaN template surface were strongly affected by its growth time, as is shown in Fig. 6. Therefore, when compared with the 3D GaN templates with a lower-density of GaN islands or those with relatively coalesced GaN islands, the SiN deposited on the 3D GaN template with a high-density of independent GaN islands contributed most to delaying the coalescence process of GaN islands. It is well known that the coalescence process of GaN islands plays a significant role on the dislocation behavior. In the early stage of 3D GaN template growth, a high density of dislocations will propagate from the underlayer into the GaN islands. In the following process, the GaN islands begin to laterally grow and gradually coalesce. During this process, the dislocations will be bent to the slant side facets of GaN islands, i.e., either {1−101} or {1−102}, eventually being terminated on the coalesced island boundary.^20,33 If the coalescence process of GaN islands is suitably prolonged, more dislocations can be bent and terminated.^34,35 Based on this mechanism, the improvement in the crystalline quality of samples B and C can be attributed to the extension in the coalescence process of GaN islands.

Fig. 7 In situ monitoring reflectance curves of samples A–F collected during GaN growth.

In addition, the change in the coalescence process of GaN islands also accounts for the variation in residual stress of samples with SiN deposition. It is well known that the coalescence of GaN islands will produce tensile stress,^14,36 and the magnitude of this tensile stress is inversely proportional to the coalescence time of GaN islands.³⁴ In this regard, the prolongation in the coalescence process of GaN islands resulted in alleviating the tensile stress induced by the coalescence of GaN islands. Therefore, for sample C with the longest recovery duration, the lower residual tensile stress can be achieved in comparison with the rest of samples with SiN deposition.

The cross-sectional TEM was employed to investigate the TD condition in sample C. The dislocations were first suppressed by the step-graded AlGaN buffer layer. When the dislocations penetrated into the GaN layer, the dislocations were found to bend, as shown in Fig. 8a. This result supports the above-mentioned viewpoint about the dislocation behavior during the 3D GaN growth. The bending leads to the annihilation of dislocations, and thereby reduces the TD density. However, it is worth noting that SiN layer is hard to be distinguished from the detailed TEM image, as shown in Fig. 8b. On the one hand, the SiN layer had the thickness of just several atomic layers, which is too thin to be clearly observed. On the other hand, SiN was not deposited in a regular shape, which makes it hard to distinguish the SiN layer from the bending dislocations. Furthermore, we deduced that SiN may not be deposited uniformly. The difference in surface morphologies of 3D GaN template may lead to the various distribution conditions of the segregation and aggregation for SiN layer, thereby inducing the different effects on the quality of GaN epitaxial films.^{17,26,37–39} In the case of 3D GaN templates grown for times shorter or longer than 180 s, the distribution condition of SiN layer had a relatively weak influence on the 3D GaN coalescence. In the case of 180 s grown 3D GaN template, the distribution of SiN was more beneficial to hindering the 3D GaN coalescence. As a result, it shows a trade-off effect on the quality of GaN epitaxial films with the growth time of 3D GaN template increasing.

Fig. 8 (a) Cross-sectional TEM image of sample C and (b) the detailed cross-sectional TEM image.

4. Conclusion

In conclusion, deposition of in situ SiN on the 3D GaN template was demonstrated to be able to significantly improve the crystalline quality of as-grown GaN epitaxial films on Si substrates. The in situ monitoring reflectance curves suggest that the coalescence process of GaN islands can be prolonged by the SiN deposition, but nevertheless, is strongly affected by the density and the size of GaN islands on the 3D GaN template surface. The longest coalescence process of GaN islands was obtained for the sample with 180 s grown 3D GaN template that was covered by a high-density of independent GaN islands. The delayed coalescence of GaN islands facilitates the bending and annihilation of dislocations, thus suppressing the propagation of dislocations. The excellent crystalline quality for the optimized sample with the 180 s grown 3D GaN template was confirmed by high resolution X-ray diffraction, revealing a minimum full width at half maximum of 348 and 406 arcsec for GaN(0002) and GaN(10−12) rocking curves, respectively. In addition, the introduced SiN increased the residual tensile stress and, nevertheless, the delayed coalescence of GaN islands induced by optimizing the 3D GaN template impaired this kind of increase in the residual tensile stress, thereby guaranteeing crack-free surfaces.

Acknowledgements

This work is supported by National Science Fund for Excellent Young Scholars of China (No. 51422203), National Natural Science Foundation of China (No. 51572091 and 51372001), Distinguished Young Scientist Foundation of Guangdong Scientific Committee (No. S2013050013882), Key Project in Science and Technology of Guangdong Province (No. 2014B010119001, 2014B010121004 and 2011A080801018), and Strategic Special Funds for LEDs of Guangdong Province (No. 2011A081301010, 2011A081301012 and 2012A080302002). Yunhao Lin and Meijuan Yang contributed equally to this work.

References

1. L. Liu and J. H. Edgar, Mater. Sci. Eng., R, 2002, 37, 61–127.
2. D. Zhu, D. J. Wallis and C. J. Humphreys, Rep. Prog. Phys., 2013, 76, 106501.
3. C. Florian, T. Cappello, R. P. Paganelli, D. Niessen and F. Filicori, IEEE Trans. Microwave Theory Tech., 2015, 63, 2589–2602.
4. Y. Honda, Y. Kuroiwa, M. Yamaguchi and N. Sawaki, Appl. Phys. Lett., 2002, 80, 222–224.
5. A. Dadgar, A. Strittmatter, J. Blasing, M. Poschenrieder, O. Contreras, P. Veit, T. Riemann, F. Bertram, A. Reiher, A. Krtschil, A. Diez, T. Hempel, T. Finger, A. Kasic, M. Schubert, D. Bimberg, F. A. Ponce, J. Christen and A. Krost, Phys. Status Solidi C, 2003, 1583–1606.
6. S. Raghavan and J. Redwing, J. Appl. Phys., 2005, 98, 23515.
7. J. Y. Clames, C. Giesen, T. Meyer and M. Heuken, Phys. Status Solidi C, 2006, 3, 2191–2194.
8. G. Li, W. Wang, W. Yang and H. Wang, Surf. Sci. Rep., 2015, 70, 380–423.
9. M. Wosko, B. Paszkiewicz, T. Szymanski and R. Paszkiewicz, J. Cryst. Growth, 2015, 414, 248–253.
10. A. Szyszka, L. Lupina, G. Lupina, M. Mazur, M. A. Schubert, P. Storck, S. B. Thapa and T. Schroeder, Appl. Phys. Lett., 2014, 104, 11106.
11. W. Wang, W. Yang, H. Wang and G. Li, J. Mater. Chem. C, 2014, 2, 9342–9358.
12. L. S. Wang, S. Tripathy, B. Z. Wang and S. J. Chua, Appl. Surf. Sci., 2006, 253, 214–218.
13. J. Ma, X. Lu, X. Zhu, T. Huang, H. Jiang, P. Xu and K. M. Lau, J. Cryst. Growth, 2015, 414, 237–242.
14. A. Krost, A. Dadgar, G. Strassburger and R. Clos, Phys. Status Solidi A, 2003, 200, 26–35.
15. A. Dadgar, F. Schulze, T. Zettler, K. Haberland, R. Clos, G. Strassburger, J. Blasing, A. Diez and A. Krost, J. Cryst. Growth, 2004, 272, 72–75.
16. R. F. Xiang, Y. Fang, J. N. Dai, L. Zhang, C. Y. Su, Z. H. Wu, C. H. Yu, H. Xiong, C. Q. Chen and Y. Hao, J. Alloys Compd., 2011, 509, 2227–2231.
17. T. Markurt, L. Lymperakis, J. Neugebauer, P. Drechsel, P. Stauss, T. Schulz, T. Remmele, V. Grillo, E. Rotunno and M. Albrecht, Phys. Rev. Lett., 2013, 110, 36103.
18. K. Y. Zang, Y. D. Wang, L. S. Wang, S. Y. Chow and S. J. Chua, J. Appl. Phys., 2007, 101, 93502.
19. K. Cheng, M. Leys, S. Degroote, M. Germain and G. Borghs, Appl. Phys. Lett., 2008, 92, 192111.
20. Y. Lin, S. Zhou, W. Wang, W. Yang, H. Qian, H. Wang, Z. Lin, Z. Liu, Y. Zhu and G. Li, J. Mater. Chem. C, 2015, 3, 1484–1490.
21. J. Liu, J. Zhang, Q. Mao, X. Wu and F. Jiang, CrystEngComm, 2013, 15, 3372.
22. X. Liu, D. Li, X. Sun, Z. Li, H. Song, H. Jiang and Y. Chen, CrystEngComm, 2014, 16, 8058–8063.
23. M. A. Moram and M. E. Vickers, Rep. Prog. Phys., 2009, 72, 36502.
24. J. Ma, X. Zhu, K. M. Wong, X. Zou and K. M. Lau, J. Cryst. Growth, 2013, 370, 265–268.
25. S. Zheng, L. Xin, Z. Gangyi, W. Zhenhai, P. Grunberg, Z. Hongbo and W. Yongjin, Appl. Phys. Express, 2014, 7, 82102–82104.
26. T. Y. Wang, S. L. Ou, R. H. Horng and D. S. Wuu, CrystEngComm, 2014, 16, 5724–5731.
27. Y. Feng, H. Wei, S. Yang, H. Zhang, S. Kong, G. Zhao and X. Liu, CrystEngComm, 2014, 16, 7525.
28. D. Wang, S. Jia, K. J. Chen, K. M. Lau, Y. Dikme, P. van Gemmern, Y. C. Lin, H. Kalisch, R. H. Jansen and M. Heuken, J. Appl. Phys., 2005, 97, 56103.
29. S. Tripathy, S. J. Chua, P. Chen and Z. L. Miao, J. Appl. Phys., 2003, 93, 789.
30. T. Riemann, T. Hempel, J. Christen, P. Veit, R. Clos, A. Dadgar, A. Krost, U. Haboeck and A. Hoffmann, J. Appl. Phys., 2006, 99, 123518.
31. S. Figge, T. Bottcher, S. Einfeldt and D. Hommel, J. Cryst. Growth, 2000, 221, 262–266.
32. T. Yang, K. Uchida, T. Mishima, J. Kasai and J. Gotoh, Phys. Status Solidi A, 2000, 180, 45–50.
33. L. Weike, L. Liang, L. Zhonghui, D. Xun, P. Daqing, Z. Dongguo and X. Xiaojun, J. Alloys Compd., 2015, 633, 494–498.
34. T. Bottcher, S. Einfeldt, S. Figge, R. Chierchia, H. Heinke, D. Hommel and J. S. Speck, Appl. Phys. Lett., 2001, 78, 1976.
35. D. G. Zhao, D. S. Jiang, J. J. Zhu, X. Guo, Z. S. Liu, S. M. Zhang, Y. T. Wang and H. Yang, J. Alloys Compd., 2009, 487, 400–403.
36. B. W. Sheldon, A. Rajamani, A. Bhandari, E. Chason, S. K. Hong and R. Beresford, J. Appl. Phys., 2005, 98, 043509.
37. R. A. Oliver, M. J. Kappers, J. Sumner, R. Datta and C. J. Humphreys, J. Cryst. Growth, 2006, 289, 506–514.
38. M. J. Kappers, R. Datta, R. A. Oliver, F. Rayment, M. E. Vickers and C. J. Humphreys, J. Cryst. Growth, 2007, 300, 70–74.
39. S. Tanaka, M. Takeuchi and Y. Aoyagi, Jpn. J. Appl. Phys., 2000, 39, L831–L834.

---