Open Access
Add to BibSonomy
Abstract
The effect of the carrier gas type on the crystal quality of GaN is investigated in detail. Compared with a single carrier gas (H2 or N2), the employment of N2 during the nucleation layer and H2 in the high temperature buffer layer growth process will lead to smoother surface, stronger photoluminescence spectral strength, and lower threading dislocation density. Furthermore, it is found that conical and snowflake-like protrusions appear on the surface for the sample under pure N2 atmosphere.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Gallium nitride (GaN) is a direct band gap semiconductor material with a band gap up to 3.42 eV [1]. It has high thermal conductivity, high saturated electron velocity, high electron mobility, and strong chemical stability [2]. Due to these material properties, GaN has shown great potential in the field of optoelectronic devices including light-emitting diodes (LEDs), lasers, detectors, as well as high-frequency microwave devices [3–6]. Since GaN based devices are mainly grown on foreign substrates, the large lattice and thermal coefficient mismatch generate high density of dislocations. Dislocation is considered as the nonradiative recombination center in GaN-based LEDs, which hinders the development of high efficiency LEDs. High quality epitaxial growth of GaN is an extremely critical step for these applications. In 1986, H. Amano et al successfully grew GaN film possessing smooth surface and high crystal quality with two-step epitaxy by metal organic chemical vapor deposition (MOCVD) [7], which laid a foundation for the subsequent growth of nitride epitaxial materials. Nowadays, the mainstream methods for GaN epitaxy are two-step epitaxy and other approaches based on it, including epitaxial lateral overgrowth [8], pendeo-epitaxial growth [9], facet-controlled epitaxial lateral overgrowth [10], etc. During the two-step epitaxial growth of GaN, low temperature GaN atomic groups are initially deposited on the surface of substrate to form nucleation islands. After that, the GaN nucleation islands will undergo recrystallization and lateral growth in the high temperature (HT) buffer layer growth step. The lateral growth process induces the nucleation islands to gradually merge and combine into a quasi-two-dimensional plane. At last, the GaN layer performs a layer-by-layer growth mode. The crystal quality of the GaN epilayer and corresponding device performance strongly depend on the growth parameters, such as V/III ratio, pressure, and temperature [11–13]. Although the carrier gas type for the GaN growth has attracted many interests, to our best knowledge, the specific physical mechanism is not yet clear. H. Amano et al. initially adopted pure hydrogen (H2) as the carrier gas in 1986 [7]. A. Ougazzaden et al. adjusted the carrier gas from H2 to pure nitrogen (N2) for GaN growth [14]. In the study of Y. S. Cho et al., the GaN nucleation layer was grown under H2 atmosphere, while different proportions of H2/N2 mixture as the carrier gas are investigated at the HT buffer layer growth stage [15,16]. X. Su et al. adopted different carrier gases including N2, H2 and different proportions of H2/N2 to grow the GaN nucleation layer and characterized the surface topography by Atomic Force Microscope (AFM) [17].
In this work, we use different carrier gas in the growth of nucleation layer and buffer layer, and experimental results are compared with using single carrier gas at the two stages. The sample with N2 in the nucleation layer and H2 in the HT buffer layer exhibits a smoother surface topography and better crystal quality.
2. Experiment
In this experiment, the GaN epitaxial growth was performed on c-plane sapphire substrate by using AIXTRON CRUIS I MOCVD system. The precursors for Ga and N were trimethyl-gallium (TMGa) and ammonia (NH3). The nucleation layer was grown at 520 °C for 145 s with an NH3 flow rate of 12500 ml/min and a TMGa flow rate of 75 ml/min. The pressure condition for growing the nucleation layer was 600 mbar. It took 478 s to increase the temperature from 520 °C to 974 °C.The flow rate of NH3 and TMGa were also increased to 24000 ml/min and 210 ml/min. It should be noted that the valves controlling TMGa and NH3 are not opened at this stage, and the sample did not actually grow. The total growth time of the GaN HT buffer layer was 2900 s, corresponding to the pressure condition of 400 mbar. The HT buffer layer growth can be divided into three stages. It took 900 s in the first growth stage, the flow rate of NH3 and TMGa were 24000 ml/min and 210 ml/min. Meanwhile, the temperature was also maintained at 974 °C. In the second 500 s growth stage, the NH3 and TMGa flow rates continue to increase to 28000 ml/min and 440 ml/min, and the temperature accordingly increased to 1058 °C. The third growth stage maintained the second final growth condition and continues to grow for 1500 s. Different stages of growth conditions are clearly shown in Fig. 1 and Table 1. In this study, pure H2was adopted in sample A as the carrier gas throughout the process, while pure N2was used in sample B. Sample C adopted N2 in the nucleation layer and H2 in the HT buffer layer. Meanwhile a 1×1 µm2 Ga2O3 substrate was also used for epitaxial growth.
Table 1. The experimental conditions of the sample A-C
After the growth process, the surface morphology was observed by AFM and scanning electron microscope (SEM). The Raman measurement instrument used in this experiment was a confocal Jobin Yvon LavRam HR800 micro-Raman spectrometer with a charge coupled device detector and an optical microscope system. The photoluminescence (PL) measurement was performed by a He-Cd laser excitation at a wavelength of 325 nm at room temperature (300 K). The crystal quality of GaN epilayer was investigated by Bruker D8 discovery high resolution X-ray diffraction (HRXRD).
3. Results and discussion
The surface topographies of all samples investigated by Bruker Icon AFM and MLA 650F SEM are shown in Fig. 2. Both samples A and C exhibit a smooth surface with distinct atomic step flow. Sample C shows a smoother surface with a lower root mean square roughness (RMS) (0.126 nm for sample C and 0.341 nm for sample A). However, Sample B exhibits a completely different surface topography. The surface topography of sample B can be divided into three types as tagged in Fig. 2 (e): (I) conical protrusions, (II) snowflake-like protrusions, and (III) relatively flat areas. The height of the protrusions is up to 108.7 nm. Consequently, the RMS of sample B (23.2 nm) is much higher than that of sample A and C.
Fig. 2. AFM and SEM images of GaN epilayers grow with different carrier gas. The AFM of (a) sample A (5×5 µm2), (b) sample B (50×50 µm2), and (c) sample C (5×5 µm2). The SEM of (d) sample A (e) Sample B and (f) sample C.
The polarity of sample B is further analyzed. As N-polar GaN is easily etched by alkali solution, while etching is negligible in Ga-polar GaN in comparison to regions of N-polar GaN [18]. Figure 3 shows the SEM images before and after etching with KOH solution.
It can be clearly seen that the surface morphologies before and after corrosion have not changed significantly. Conical and snowflake-like protrusions still maintain their original topography after corrosion. The conical and snowflake-like protrusions surfaces are determined to be Ga polarity. These morphologies are rarely observed in Ga-polar GaN by natural growth.
The residual strain is analyzed by Raman spectroscopy (Fig. 4). The frequency of E2 (high) peak in the Raman spectrum is sensitive to stress, and the frequency of E2 (high) peak of strain-free GaN is 567.6 cm-1 [19]. The E2 (high) mode phonon peaks of all samples have a redshift from the frequency of 567.6 cm-1, which indicates that all samples are subjected to compressive stress. Moreover, sample B is subject to minimum compressive stress (a slight redshift of 1.11 cm-1), which will be explained in more detail below. The intensity and full width at half maximum (FWHM) of E2 (high) peak can also reveal crystal quality to some extent [20]. According to the order of samples C, A, and B, the lower intensity of E2 (high) peak, the larger FWHM, and the worse crystal quality can be revealed.
Fig. 4. The Raman spectra of sample A, B, and C (a) ranging from 300 to 800 cm-1. (b) ranging from 560 to 576 cm-1.
We further analyze the stress and crystallization quality of different regions of the special morphology on sample B. Figure 5 illustrates the intensity, frequency and FWHM of the E2 (high) peak of the cones in sample B measured in a 25×25 µm2 zone. The individual E2 (high) peak is fitted through the Lorentzian-Guassian mixed function to extract the corresponding values. The mapping scan date shows that the E2 (high) phonon peak strength for the sidewall of the cones is weaker and has a larger FWHM in comparison with the top (Fig. 5 (a) and (c)), which suggests worse crystal quality of the sidewall. The threading dislocations will bend toward the sidewall of the cone in the growth process, resulting in higher threading dislocation density of the sidewall than the top. Thus, the worse crystal quality of the sidewall can be reasonably explained. Furthermore, it has been proven or reported in the previous reports that threading dislocations can effectively release stress [21]. As shown in Fig. 5 (b), the high threading dislocation density of the sidewall leads to more release of stress. Shin. H. Y. et al. used a conical shaped pattern substrate to grow GaN epilayer with conical surface and found that the cone top exhibits a lower threading dislocation density than the sidewall [22], which supports our conclusions. As shown in Fig. 6 (a) and (b), the near band edge emission (NBE) is observed in the PL spectrum of sample A (3.427 eV) and C (3.425 eV). The peak intensity of sample C is stronger than that of sample A, and the FWHM results (4.24 nm for sample A and 4.12 nm for sample C) suggest that sample C possesses better crystal quality. Meanwhile, the peak around 450 nm is observed in the PL spectrum of samples A and C (Fig. 6 (a)), which is induced by C impurities [23–24]. Relatively, no matter in the protrusions (cones and snowflakes) or flat region of sample B, the intensity of the peak near 365 nm is significantly lower than that of sample A and C, and the FWHM value is also the largest, indicating the worst crystal quality of sample B. These are consistent with the results of Raman. It should be noted that the peak intensity of the snowflake-like protrusions of sample B at 386 nm is much stronger than NBE intensity (Fig. 6 (c)). This peak is not found in the PL spectra of samples A and C, even in the flat and conical region of sample B. Light with a wavelength of 386 nm has a corresponding energy of 3.21 eV. The sidewalls of the snowflakes belong to semi-polar or a non-polar orientations. It has been confirmed that the non-polar or semi-polar surfaces can adhere to more oxygen impurities during growth [25]. And the oxygen impurity typically incorporates into N sites and forms a shallow donor level in GaN. Bermudez. V. M had reported O1s X-ray photoelectron spectroscopy data showing a peak with FWHM (3.2 eV) [26]. Meanwhile, G. A. Slacka et al. ’s work also proved that 3.2 eV absorption band is very likely to be produced by the oxygen impurities [27]. This reasonably explains the source of the 386 nm peak of the snowflakes in the PL spectrum. Based on above analysis, the formation of snowflakes is closely related to oxygen impurities.
Fig. 5. (a) Raman mapping of (a) E2 (high) mode intensity, (b) E2 (high) mode peak frequency, and (c) E2 (high) mode FWHM.
Fig. 6. The PL spectra of (a) samples A, B (flat, conical and snowflake-like regions) and Crangingfrom 300 nm to 600 nm, (b) samples A and C near 365 nm, and (c) flat, conical and snowflake-like regions of sample B ranging from 320 nm to 500 nm.
HRXRD is performed on all samples to further analyze the crystal quality of GaN. The screw and edge dislocation density are closely related to the FWHM of (002) and (102) X-ray rocking curves (XRCs) [28]. It can be clearly observed in Fig. 7 that the FWHM values show a decreasing trend for the sample B to A and C, revealing the decreasing trend in the threading dislocation density. This phenomenon agrees with Raman and PL results. Sample C has a FWHM of 239.08 arcsec for (002) reflection and 312.04 arcsec for (102) reflection (Fig. 7), which is the smallest of the three samples. Moreover, it has been confirmed that H2 can corrode GaN [29]. At the initial nucleation stage, the GaN in the N2 atmosphere is not corroded compared to the H2 atmosphere. Therefore, using N2 as the carrier gas at the nucleation stage will be more conducive to the subsequent HT buffer layer growth. The smaller FWHM of sample C compared to sample A (307.31 arcsec for (002) reflection and 388.02 arcsec for (102) reflection) also confirms this conclusion (Fig. 7).
The reduced FWHM of sample C compared to sample B (596.62 arcsec for (002) reflection and 1534.97 arcsec for (102) reflection) shows that H2 is an optimal choice in the HT buffer layer (Fig. 7). At the HT buffer layer stage, the corresponding islands of sample C experience the coarsening process, and the surface becomes relatively rough and the reflectance decreases drastically. This process can be reflected from the growth reflectance curve of Fig. 9. When the islands contact with each other during the coarsening process, the threading dislocations may merge and quench. In contrast, the islands of sample B under the atmosphere of N2 are directly polymerized at this stage, and the reflectance curve oscillates quickly. Therefore, the threading dislocations hardly merge and quench, resulting in higher dislocation density of sample B. Obviously, the largest FWHM of sample B confirms this conclusion. As discussed above, threading dislocations can effectively release stress. The minimum compressive stress of sample B (Fig. 4) can be reasonably explained. The reaction process of samples A, B and C described above can be shown in Fig. 8.
As for conical and snowflake-like protrusions, it can be clearly seen from Fig. 1 (a) that the TMGa flow rate in the HT buffer layer is increased compared to the nucleation layer, which may result in excess Ga source in some regions. The etching of H2 in sample C will cause the nucleation islands to not directly polymerize and undergo the coarsening process. This induces the secondary distribution of Ga source and the accumulation of Ga source can be well eliminated. While the nucleation islands of sample B under N2 atmosphere are directly polymerized, and the Ga source is not redistributed. The growth rate generally depends on the Ga source under high V/III ratio, and the initial Ga source aggregation region of sample B will grow more quickly to form protrusions. It has been confirmed that there is an excessive Ga layer on the surface of Ga polar GaN [30], resulting in a faster growth rate along the [001] direction. These protrusions will gradually grow into cones or snowflakes in the subsequent growth.
4. Conclusion
In summary, the GaN epilayer is grown by MOCVD under different carrier gas conditions, and the quality of the epilayer is investigated by AFM, SEM, PL, Raman, HRXRD. It is confirmed that the GaN epilayer with a better crystal quality can be obtained by using N2 as the carrier gas in the nucleation layer and H2 in the HT buffer layer. There is almost no coarsening process after the nucleation stage under the atmosphere of pure N2, which leads to the appearance of conical and snowflake-like protrusions.
Funding
National Key Research and Development Program of China (2016YFB0400801); National Natural Science Foundation of China (NSFC) (61574108); Key Research and Development program in Shaanxi Province (2018ZDCXL-GY-01-02-02).
References
1. S. Nakamura, M. Senoh, S. Nagahama, N. Iwasa, T. Yamada, T. Matsushita, H. Kiyoku, and Y. Sugimoto, “InGaN-Based Multi-Quantum-Well-Structure Laser Diodes,” Jpn. J. Appl. Phys. 35(Part 2), L74–L76 (1996). [CrossRef]
2. E. A. Jones, F. F. Wang, and D. Costinett, “Review of Commercial GaN Power Devices and GaN-Based Converter Design Challenges,” IEEE J. Emerg. Sel. Topics Power Electron. 4(3), 707–719 (2016). [CrossRef]
3. B. Mitchell, V. Dierolf, T. Gregorkiewicz, and Y. Fujiwara, “Perspective: Toward efficient GaN-based red light emitting diodes using europium doping,” J. Appl. Phys. 123(16), 160901 (2018). [CrossRef]
4. M. J. Uren, M. Casar, M. A. Gajda, and M. Kuball, “Buffer transport mechanisms in intentionally carbon doped GaN heterojunction field effect transistors,” Appl. Phys. Lett. 104(26), 263505 (2014). [CrossRef]
5. X. C. Xia, R. S. Shen, Y. D. Liu, D. C. Yang, S. W. Song, L. Zhao, Z. F. Shi, X. P. Li, H. W. Liang, B. L. Zhang, and G. T. Du, “Dominant UV emission from p-MgZnO/n-GaN light emitting diodes,” Opt. Mater. Express 2(1), 38–44 (2012). [CrossRef]
6. S. Nakamura, “III-V nitride based light-emitting devices,” Solid State Commun. 102(2-3), 237–248 (1997). [CrossRef]
7. H. Amano, N. Sawaki, I. Akasaki, and Y. Toyoda, “Metalorganic vapor phase epitaxial growth of a high quality GaN film using an AlN buffer layer,” Appl. Phys. Lett. 48(5), 353–355 (1986). [CrossRef]
8. M. A. L. Johnson, Z. Yu, J. D. Brown, N. A. EI-Masry, J. W. Cook JR, and J. F. Schetzina, “Scanning electron microscopy and cathodoluminescence study of the epitaxial lateral overgrowth (ELO) process for gallium nitride,” J. Electron. Mater. 28(3), 295–300 (1999). [CrossRef]
9. A. M. Roskowski., P. Q. Miraglia., E. A. Preble, S. Einfeldt, and R. F. Davis, “Surface instability and associated roughness during conventional and pendeo-epitaxial growth of GaN (0001) films via MOVPE,” J. Cryst. Growth 241(1-2), 141–150 (2002). [CrossRef]
10. K. Hiramatsu, K. Nishiyama, M. Onishi, H. Mizutani, M. Narukawa, A. Motogaito, H. Miyake, Y. Iyechika, and T. Maeda, “Fabrication and characterization of low defect density GaN using facet-controlled epitaxial lateral overgrowth (FACELO),” J. Cryst. Growth 221(1-4), 316–326 (2000). [CrossRef]
11. S. W. Kima, J. T. Oha, J. S. Kanga, D. J. Kima, J. H. Wona, J. W. Kima, and H. K. Chob, “Two-step growth of high quality GaN using V/III ratio variation in the initial growth stage,” J. Cryst. Growth 262(1-4), 7–13 (2004). [CrossRef]
12. I. W. Feng, J. Li, J. Y. Lin, H. X. Jiang, and J. Zavada, “Effects of growth pressure on erbium doped GaN infrared emitters synthesized by metal organic chemical vapor deposition,” Opt. Mater. Express 2(8), 1095–1100 (2012). [CrossRef]
13. M. S. Yi, H. H. Lee, D. J. Kim, S. J. Park, and D. Y. Noh, “Effects of growth temperature on GaN nucleation layers,” Appl. Phys. Lett. 75(15), 2187–2189 (1999). [CrossRef]
14. A. Ougazzaden, S. Gautier, C. Sartel, N. Maloufi, J. Martin, and F. Jomard, “BGaN materials on GaN/sapphire substrate by MOVPE using N2 carrier gas,” J. Cryst. Growth 298, 316–319 (2007). [CrossRef]
15. Y. S. Cho, H. Hardtdegen, N. Kaluza, N. Thillosen, R. Steins, Z. Sofer, and H. Lüth, “Effect of carrier gas on GaN epilayer characteristics,” Phys. Status Solidi C 3(6), 1408–1411 (2006). [CrossRef]
16. Y. S. Cho, H. Hardtdegen, N. Kaluza, R. Steins, G. Heidelberger, and H. Lüth, “The growth mechanism of GaN with different H2/N2 carrier gas ratios,” J. Cryst. Growth 307(1), 6–13 (2007). [CrossRef]
17. X. Su, T. Ye, S. Wang, Y. Shi, L. Fan, L. Liu, G. Zhang, X. Shi, M. Wei, H. Zhou, and H. Jiao, “Surface morphology of GaN nucleation layer grown by MOCVD with different carrier gas,” AIP Adv. 8(7), 075301 (2018). [CrossRef]
18. D. Li, M. Sumiya, S. Fuke, D. Yang, D. Que, Y. Suzuki, and Y. Fukuda, “Selective etching of GaN polar surface in potassium hydroxide solution studied by Xray photoelectron spectroscopy,” J. Appl. Phys. 90(8), 4219–4223 (2001). [CrossRef]
19. K. Hiramatsu, T. Detchprohm, and I. Akasaki, “Relaxation Mechanism of Thermal Stresses in the Heterostructure of GaN Grown on Sapphire by Vapor Phase Epitaxy,” Jpn. J. Appl. Phys. 32(Part 1), 1528–1533 (1993). [CrossRef]
20. T. Jiang, S. R. Xu, J. C. Zhang, Y. Xie, and Y. Hao, “Spatially resolved and orientation dependent Raman mapping of epitaxial lateral overgrowth nonpolar a-plane GaN on r-plane sapphire,” Sci. Rep. 6(1), 19955 (2016). [CrossRef]
21. P. Cantu, F. Wu, P. Waltereit, S. Keller, A. E. Romanov, S. P. DenBaars, and J. S. Specka, “Role of inclined threading dislocations in stress relaxation in mismatched layers,” J. Appl. Phys. 97(10), 103534 (2005). [CrossRef]
22. H. Y. Shin, Y. I. Chang, S. K. Kwon, K. T. Lee, M. J. Cho, and K. H. Park, “The growth characteristics of a GaN layer on a cone-shaped patterned sapphire substrate by TEM observation,” J. Korean Phys. Soc. 50(4), 1147–1151 (2007). [CrossRef]
23. F. Liang, D. Zhao, D. Jiang, Z. Liu, J. Zhu, P. Chen, J. Yang, S. Liu, Y. Xing, L. Zhang, M. Li, Y. Zhang, and G. Du, “Carbon-Related Defects as a Source for theEnhancement of Yellow Luminescence ofUnintentionally Doped GaN,” Nanomaterials 8(9), 744 (2018). [CrossRef]
24. F. Liang, D. Zhao, D. Jiang, Z. Liu, J. Zhu, P. Chen, J. Yang, S. Liu, Y. Xing, and L. Zhang, “Role of Si and C Impurities in Yellow and Blue Luminescence of Unintentionally and Si-Doped GaN,” Nanomaterials 8(12), 1026 (2018). [CrossRef]
25. S. C. Cruz, S. Keller, T. E. Mates, U. K. Mishra, and S. P. DenBaars, “Crystallographic orientation dependence of dopant and impurity incorporation in GaN films grown by metalorganic chemical vapor deposition,” J. Cryst. Growth 311(15), 3817–3823 (2009). [CrossRef]
26. V. M. Bermudez, “Study of oxygen chemisorption on the GaN (0001) - (1×1) surface,” J. Appl. Phys. 80(2), 1190–1200 (1996). [CrossRef]
27. G. A. Slacka, Leo. J. Schowaltera, D. Morellic, and J. A. Freitas Jr, “Some effects of oxygen impurities on AlN and GaN,” J. Cryst. Growth 246(3-4), 287–298 (2002). [CrossRef]
28. T. Metzger, R. Höpler, E. Born, O. Ambacher, M. Stutzmann, R. Stömmer, M. Schuster, H. Göbel, S. Christiansen, M. Albrecht, and H. P. Strunk, “Defect structure of epitaxial GaN films determined by transmission electron microscopy and triple-axis X-ray diffractometry,” Philos. Mag. A 77(4), 1013–1025 (1998). [CrossRef]
29. Y. T. Moon, Y. Fu, F. Yun, S. Dogan, M. Mikkelson, D. Johnstone, and H. Morkoç, “A study of GaN regrowth on the micro-facetted GaN template formed by in-situ thermal etching,” Phys. Status Solidi A 202(5), 718–721 (2005). [CrossRef]
30. T. Zywietz, J. Neugebauer, and M. Scheffler, “Adatom diffusion at GaN (0001) and (000(1) over-bar) surfaces,” Appl. Phys. Lett. 73(4), 487–489 (1998). [CrossRef]
