Optoelectronic Properties and Structural Characterization of Gan Thick Films on Different Substrates through Pulsed Laser Deposition

Approximately 4-µm-thick GaN epitaxial films were directly grown onto a GaN/sapphire template, sapphire, Si(111), and Si(100) substrates by high-temperature pulsed laser deposition (PLD). The influence of the substrate type on the crystalline quality, surface morphology, microstructure, and stress states was investigated by X-ray diffraction (XRD), photoluminescence (PL), atomic force microscopy (AFM), transmission electron microscopy (TEM), and Raman spectroscopy. Raman scattering spectral analysis showed a compressive film stress of −0.468 GPa for the GaN/sapphire template, whereas the GaN films on sapphire, Si(111), and Si(100) exhibited a tensile stress of 0.21, 0.177, and 0.081 GPa, respectively. Comparative analysis indicated the growth of very close to stress-free GaN on the Si(100) substrate due to the highly directional energetic precursor migration on the substrate's surface and the release of stress in the nucleation of GaN films during growth by the high-temperature (1000 • C) operation of PLD. Moreover, TEM images revealed that no significant GaN meltback (Ga–Si) etching process was found in the GaN/Si sample surface. These results indicate that PLD has great potential for developing stress-free GaN templates on different substrates and using them for further application in optoelectronic devices.


Introduction
Gallium nitride (GaN) and its related III-nitride materials are excellent wide direct band-gap (3.4 eV) semiconductors due to their potential properties of high saturation velocity in an electric field, high breakdown electric field, and electron mobility-all of which are necessary for the development of next-generation devices and applications that are high frequency, highly efficient, and can effectively power switching devices [1][2][3].However, due to the lack of suitable native or lattice-matched substrates, GaN epilyers are usually grown on sapphire, SiC, and Si substrates.This presents a serious problem, as a high defect density and a large biaxial stress in the heteroepitaxy of the GaN epilayers are generated by mismatches in the lattice structure and thermal expansion coefficients between the epilayers and the Si substrate.These growth-induced defects (such as threading dislocations, stacking faults, voids, and point defects) limit the performance and reliability of GaN-based devices [4][5][6].ZnO-related materials may be closely lattice-matched with GaN, but the drawback of the ZnO single crystalline wafer is that it is still expensive [7].Substrates that produce a low density of defects present the most effective possible approach for reducing defects in epitaxial films.The most widely used methods for growing GaN with low defect density are hydride vapor phase epitaxy (HVPE) and metalorganic chemical vapor deposition (MOCVD) [8,9].GaN thin films with high-quality and low-density of defects can also be grown by ion-beam assisted MBE [10,11].The reaction chamber in an HVPE system is often made of quartz, which is not operational under high temperature.An MOCVD system requires a high-temperature growth process, which consumes considerable electric power and thereby produces high running costs and the possibility of air pollution due to the toxicity of the metal-organic chemicals in the precursor gas.Pulsed laser deposition (PLD) is a promising technique that can address these problems [12][13][14].PLD is interesting, as it allows for in situ processing of the multilayer structure via multiple targets, stoichiometric transfer deposition from the target to the substrate, flexible doping options for complex compositions, and a highly directionally distributed energetic precursor produced by the laser ablation of a target.Most discussions on PLD focus on studying the influence of growth conditions on the properties of GaN films [15][16][17][18][19]. Several previous studies have reported how PLD enables the growth of high-quality III-nitrides on other substrates [20][21][22][23][24]. Since the considerable scale and production cost of native GaN substrates would be too much, GaN templates on foreign substrates are good choices for the heteroepitaxial deposition of GaN-based devices.In this study, the crystalline quality, surface morphology, optoelectronic and structural properties related to GaN thick film grown on different substrates as a GaN templates through high-temperature PLD are characterized and compared.

Experimental
All GaN film samples were deposited on different substrates by PLD at 1000 • C in a nitrogen plasma ambient atmosphere.The chamber was pumped down to 10 −6 Torr before the deposition process began, and N 2 gas (with a purity of 99.999%) was introduced.The working pressure once the N 2 plasma was injected was 1.13 × 10 −4 Torr.A KrF excimer laser (λ = 248 nm, Lambda Physik, Fort Lauderdale, FL, USA) was employed as the ablation source and operated with a repetition rate of 1 Hz and a pulse energy of 60 mJ.The average growth rate of the GaN film was approximately 1 µm/h.The laser beam was incident on a rotating target at an angle of 45 • .The GaN target was fabricated by HVPE and set at a fixed distance of 9 cm from the substrate before being rotated at 30 rpm during film deposition.In this case, ~4 µm-thick GaN films were grown on a GaN/sapphire template (sample A), sapphire (sample B), Si(111) (sample C), and Si(100) (sample D).For the GaN on sample A, a 2-µm GaN layer was firstly deposited on sapphire substrate by MOCVD.Scanning electron microscopy (SEM, S-3000H, Hitachi, Tokyo, Japan), transmission electron microcopy (TEM, H-600, Hitachi, Tokyo, Japan), atomic force microscopy (AFM, DI-3100, Veeco, New York, NY, USA), double-crystal X-ray diffraction (XRD, X'Pert PRO MRD, PANalytical, Almelo, The Netherlands), low-temperature photoluminescence (PL, Flouromax-3, Horiba, Tokyo, Japan), and Raman spectroscopy (Jobin Yvon, Horiba, Tokyo, Japan) were employed to explore the microstructure and optical properties of the GaN templates deposited on different substrates.The electrical properties of the GaN films were determined by Van der Pauw-Hall measurement under liquid nitrogen cooling at 77 K.

Results and Discussion
Figure 1 shows a low-temperature PL spectra (at 77 K) of GaN films grown on different substrates.PL spectra of GaN grown on different substrates are dominated by the near-band-edge emission at around 360 nm.The full width at half maximum (FWHM) of the GaN films produced on samples A (4 nm) and B (8 nm) are narrower than that of the films grown on samples C (10 nm) and D (13 nm), indicating the low defect density and high crystalline quality of the GaN films due to their lower lattice mismatch, which is consistent with the XRD results.Similar trends of the yellow band-emission peak on these samples were also observed (data not shown here).The yellow luminescence is related to deep level defects in GaN [25].Figure 2 shows a comparison of the typical XRD patterns of GaN (0002) films grown on different substrates.It can be seen that there is a variation in the FWHM value of the (0002) diffraction peak, and intensities of the GaN diffraction peak on the different substrates were Appl.Sci.2017, 7, 87 3 of 9 obtained at around 34.5 degrees.The intensity of GaN (0002) in sample A is the strongest among all samples, which indicates that the GaN films on the GaN/sapphire template are highly c-oriented and have better crystalline quality.The FWHM of GaN (0002) values for samples A, B, C, and D were measured at 0.19 • , 0.51 • , 0.79 • , and 1.09 • , respectively.However, the XRD peak intensity increases as FWHM decreases; this is attributed to the increase in the crystallite size due to either the aggregation of small grains or grain boundary movement during the growth process.Since the FWHM of the XRD diffraction peak is relative to the average crystallite grain size in the film [26], the grain size of GaN grown on the different substrates is calculated using the Debye-Scherer equation [27]: where D is the crystallite size, λ is the X-ray wavelength, and θ is the diffraction angle.The crystallite sizes of samples A, B, C, and D are estimated to be 57, 20, 13, and 9 nm, respectively.These results indicate that the crystalline quality of GaN films grown on samples A and B is better than that of the films grown on samples C and D.
Appl.Sci.2017, 7, 87 3 of 9 GaN/sapphire template are highly c-oriented and have better crystalline quality.The FWHM of GaN (0002) values for samples A, B, C, and D were measured at 0.19°, 0.51°, 0.79°, and 1.09°, respectively.However, the XRD peak intensity increases as FWHM decreases; this is attributed to the increase in the crystallite size due to either the aggregation of small grains or grain boundary movement during the growth process.Since the FWHM of the XRD diffraction peak is relative to the average crystallite grain size in the film [26], the grain size of GaN grown on the different substrates is calculated using the Debye-Scherer equation [27]: where D is the crystallite size, λ is the X-ray wavelength, and θ is the diffraction angle.The crystallite sizes of samples A, B, C, and D are estimated to be 57, 20, 13, and 9 nm, respectively.These results indicate that the crystalline quality of GaN films grown on samples A and B is better than that of the films grown on samples C and D.  Figure 3 shows plane-view SEM pictures of GaN films grown on various substrates.The surface morphologies show different features, as they are strongly dependent on the types of substrates used.The surface of GaN films in samples A and B was mirror-like, indicating less of a lattice mismatch between GaN and sapphire (Figure 3a,b).The smooth surface might be due to the high kinetic energy needed by PLD for GaN precursor migration and diffusion on the substrates' surface [28].A rough GaN film surface, meanwhile, was observed in sample C (Figure 3c).Sample D presented an incomplete island coalescence process with a hexagonal structure, as shown in Figure 3d.GaN/sapphire template are highly c-oriented and have better crystalline quality.The FWHM of GaN (0002) values for samples A, B, C, and D were measured at 0.19°, 0.51°, 0.79°, and 1.09°, respectively.However, the XRD peak intensity increases as FWHM decreases; this is attributed to the increase in the crystallite size due to either the aggregation of small grains or grain boundary movement during the growth process.Since the FWHM of the XRD diffraction peak is relative to the average crystallite grain size in the film [26], the grain size of GaN grown on the different substrates is calculated using the Debye-Scherer equation [27]: where D is the crystallite size, λ is the X-ray wavelength, and θ is the diffraction angle.The crystallite sizes of samples A, B, C, and D are estimated to be 57, 20, 13, and 9 nm, respectively.These results indicate that the crystalline quality of GaN films grown on samples A and B is better than that of the films grown on samples C and D.  Figure 3 shows plane-view SEM pictures of GaN films grown on various substrates.The surface morphologies show different features, as they are strongly dependent on the types of substrates used.The surface of GaN films in samples A and B was mirror-like, indicating less of a lattice mismatch between GaN and sapphire (Figure 3a,b).The smooth surface might be due to the high kinetic energy needed by PLD for GaN precursor migration and diffusion on the substrates' surface [28].A rough GaN film surface, meanwhile, was observed in sample C (Figure 3c).Sample D presented an incomplete island coalescence process with a hexagonal structure, as shown in Figure 3d. Figure 3 shows plane-view SEM pictures of GaN films grown on various substrates.The surface morphologies show different features, as they are strongly dependent on the types of substrates used.The surface of GaN films in samples A and B was mirror-like, indicating less of a lattice mismatch between GaN and sapphire (Figure 3a,b).The smooth surface might be due to the high kinetic energy needed by PLD for GaN precursor migration and diffusion on the substrates' surface [28].A rough GaN film surface, meanwhile, was observed in sample C (Figure 3c).Sample D presented an incomplete island coalescence process with a hexagonal structure, as shown in Figure 3d.This result indicates that GaN films on Si(100) have a hexagonal phase.The different GaN film structure of the grains can be attributed to the different lattice structure of the Si substrate [29].The surface morphology and roughness of the GaN films grown on different substrates were carried out by AFM measurements with the scanning area of 10 × 10 µm 2 , as shown in Figure 4.In Figure 4, the root-mean-square RMS values for samples A, B, C, and D are 2.1, 3.4, 14.3, and 17.7 nm, respectively.The film grown in samples A and B exhibited quite a smooth surface, with the RMS roughness being 3.4 and 2.1 nm, respectively, and the RMS surface roughness of samples C and D was estimated as 14.3 and 17.7 nm, respectively.The large values for the surface roughness of the GaN films in samples C and D might be due to the large lattice mismatch between the film and the substrates.A decrease in surface roughness occurs with an increase in grain size, as mentioned in the XRD results.The electrical resistivity of the GaN films grown on different substrates is shown in Figure 5a.The electrical resistivity of the four samples was found to be in the range 16.2-32.8Ω•cm.The electrical resistivity of sample D was the largest, while that of sample A was the smallest.The electrical resistivity correlates with defect density, and the high defect density in the films may cause a decrease in the electrical resistivity [30].The values of electrical resistivity of samples C and D were very close, which is consistent with the structural features of the films grown on these substrates, as discussed above.As electrical resistivity is inversely proportional to the carrier concentration and carrier mobility, the electrical resistivity of the films grown on the different substrates can be determined from their measurements.Low-temperature Hall measurement data from GaN films    The electrical resistivity of the GaN films grown on different substrates is shown in Figure 5a.The electrical resistivity of the four samples was found to be in the range 16.2-32.8Ω•cm.The electrical resistivity of sample D was the largest, while that of sample A was the smallest.The electrical resistivity correlates with defect density, and the high defect density in the films may cause a decrease in the electrical resistivity [30].The values of electrical resistivity of samples C and D were very close, which is consistent with the structural features of the films grown on these substrates, as discussed above.As electrical resistivity is inversely proportional to the carrier concentration and carrier mobility, the electrical resistivity of the films grown on the different substrates can be determined from their measurements.Low-temperature Hall measurement data from GaN films The electrical resistivity of the GaN films grown on different substrates is shown in Figure 5a.The electrical resistivity of the four samples was found to be in the range 16.2-32.8Ω•cm.The electrical resistivity of sample D was the largest, while that of sample A was the smallest.The electrical resistivity correlates with defect density, and the high defect density in the films may cause a decrease in the electrical resistivity [30].The values of electrical resistivity of samples C and D were very close, which is consistent with the structural features of the films grown on these substrates, as discussed above.As electrical resistivity is inversely proportional to the carrier concentration and carrier mobility, the electrical resistivity of the films grown on the different substrates can be determined from their measurements.Low-temperature Hall measurement data from GaN films grown on the different substrates are shown in Figure 5b,c.Sample A showed the lowest carrier concentration and highest carrier mobility, thereby resulting in an increased number of conductive paths.The carrier concentration in sample D was higher than that in the others, whereas its carrier mobility was the lowest.This can be attributed to the existence of a high intrinsic defect and several grain boundaries in the film.These defects trap and scatter moving electrons, thus decreasing their mobility in the GaN films [31,32].
Appl.Sci.2017, 7, 87 5 of 9 grown on the different substrates are shown in Figure 5b,c.Sample A showed the lowest carrier concentration and highest carrier mobility, thereby resulting in an increased number of conductive paths.The carrier concentration in sample D was higher than that in the others, whereas its carrier mobility was the lowest.This can be attributed to the existence of a high intrinsic defect and several grain boundaries in the film.These defects trap and scatter moving electrons, thus decreasing their mobility in the GaN films [31,32].To further clarify the stress behaviors among the four samples, Raman scattering spectroscopy was performed, and the results are shown in Figure 6.The E2-high phonon mode is very sensitive to biaxial strain, and is extensively used to characterize the in-plane stress state of the GaN epilayer [33].The relationship between biaxial stress and Raman shift can be shown by the formula: where σ is the biaxial stress, Δω is the Raman shift, and k is the Raman stress coefficient of 6.2 cm −1 •GPa −1 for GaN [34].Generally, a blue shift in an E2-high phonon peak indicates compressive To further clarify the stress behaviors among the four samples, Raman scattering spectroscopy was performed, and the results are shown in Figure 6.The E 2 -high phonon mode is very sensitive to biaxial strain, and is extensively used to characterize the in-plane stress state of the GaN epilayer [33].
Appl.Sci.2017, 7, 87 5 of 9 grown on the different substrates are shown in Figure 5b,c.Sample A showed the lowest carrier concentration and highest carrier mobility, thereby resulting in an increased number of conductive paths.The carrier concentration in sample D was higher than that in the others, whereas its carrier mobility was the lowest.This can be attributed to the existence of a high intrinsic defect and several grain boundaries in the film.These defects trap and scatter moving electrons, thus decreasing their mobility in the GaN films [31,32].To further clarify the stress behaviors among the four samples, Raman scattering spectroscopy was performed, and the results are shown in Figure 6.The E2-high phonon mode is very sensitive to biaxial strain, and is extensively used to characterize the in-plane stress state of the GaN epilayer [33].The relationship between biaxial stress and Raman shift can be shown by the formula: where σ is the biaxial stress, Δω is the Raman shift, and k is the Raman stress coefficient of 6.2 cm −1 •GPa −1 for GaN [34].Generally, a blue shift in an E2-high phonon peak indicates compressive The relationship between biaxial stress and Raman shift can be shown by the formula: where σ is the biaxial stress, ∆ω is the Raman shift, and k is the Raman stress coefficient of 6.2 cm −1 •GPa −1 for GaN [34].Generally, a blue shift in an E 2 -high phonon peak indicates compressive stress, while a red shift indicates tensile stress [35].It has been found that an E 2 -high peak position is substrate dependent, which implies that there are different stress states in those samples.In the present case, the GaN E 2 -high peaks of samples MGS (MOCVD-grown GaN on sapphire), A, B, C, and D were evaluated as 520.7, 569.7, 565.5, 565.7, and 566.3 cm −1 , respectively.Compared to the intrinsic value of 566.8 cm −1 for the stress-free GaN, samples B, C, and D were under tensile stress, while sample A was under compressive stress [36].This can be due to the rapid release of stress in the nucleation of GaN films during the initial growth by high-temperature (1000 • C) PLD.This observed result is also consistent with those reported by Wang et al. [37].Sample D had minimum stress, likely caused by the growth of polygonal island structures and defects generated in the films, which is consistent with the SEM results [38].There is a large difference in the lattice mismatch and thermal expansion between GaN and Si when compared to the GaN/sapphire template and sapphire.The calculated values of stress for GaN grown on different substrates are shown in Figure 7.The Raman spectra of the MGS sample is displayed for comparison, as shown in Figure 7.The GaN E 2 peak of MGS was evaluated at 570.2 cm −1 with a compressive stress value of −0.548 GPa, which is larger than the compressive stress value of −0.468 for sample B. It can be concluded that the PLD growth method is beneficial for the release of stress in the films.
Appl.Sci.2017, 7, 87 6 of 9 stress, while a red shift indicates tensile stress [35].It has been found that an E2-high peak position is substrate dependent, which implies that there are different stress states in those samples.In the present case, the GaN E2-high peaks of samples MGS (MOCVD-grown GaN on sapphire), A, B, C, and D were evaluated as 520.7, 569.7, 565.5, 565.7, and 566.3 cm −1 , respectively.Compared to the intrinsic value of 566.8 cm −1 for the stress-free GaN, samples B, C, and D were under tensile stress, while sample A was under compressive stress [36].This can be due to the rapid release of stress in the nucleation of GaN films during the initial growth by high-temperature (1000 °C) PLD.This observed result is also consistent with those reported by Wang et al. [37].Sample D had minimum stress, likely caused by the growth of polygonal island structures and defects generated in the films, which is consistent with the SEM results [38].There is a large difference in the lattice mismatch and thermal expansion between GaN and Si when compared to the GaN/sapphire template and sapphire.The calculated values of stress for GaN grown on different substrates are shown in Figure 7.
The Raman spectra of the MGS sample is displayed for comparison, as shown in Figure 7.The GaN E2 peak of MGS was evaluated at 570.2 cm −1 with a compressive stress value of −0.548 GPa, which is larger than the compressive stress value of −0.468 for sample B. It can be concluded that the PLD growth method is beneficial for the release of stress in the films.Cross-sectional TEM images were used to investigate the GaN-on-Si meltback-etching reaction with PLD operating at a high temperature of 1000 °C.Previously, it was reported that the meltback-etching process caused by alloying reaction Ga with Si leads to a rough GaN surface and deep hollows in the Si substrate [39,40].Figure 8a,b shows the TEM images of the GaN films grown on Si(111) and Si(100), respectively.From Figure 8a,b, it can clearly be observed that no significant Ga-Si meltback occurred at the GaN/Si surface; this is likely because of the suppressed interaction between the GaN epitaxy films and the Si substrates developed through PLD.Cross-sectional TEM images were used to investigate the GaN-on-Si meltback-etching reaction with PLD operating at a high temperature of 1000 • C. Previously, it was reported that the meltback-etching process caused by alloying reaction Ga with Si leads to a rough GaN surface and deep hollows in the Si substrate [39,40].Figure 8a,b shows the TEM images of the GaN films grown on Si(111) and Si(100), respectively.From Figure 8a,b, it can clearly be observed that no significant Ga-Si meltback occurred at the GaN/Si surface; this is likely because of the suppressed interaction between the GaN epitaxy films and the Si substrates developed through PLD.
with PLD operating at a high temperature of 1000 °C.Previously, it was reported that the meltback-etching process caused by alloying reaction Ga with Si leads to a rough GaN surface and deep hollows in the Si substrate [39,40].Figure 8a,b shows the TEM images of the GaN films grown on Si(111) and Si(100), respectively.From Figure 8a,b, it can clearly be observed that no significant Ga-Si meltback occurred at the GaN/Si surface; this is likely because of the suppressed interaction between the GaN epitaxy films and the Si substrates developed through PLD.

Conclusions
We investigated the GaN thick films grown on a GaN/sapphire template, sapphire, Si(111), and Si(100) by high-temperature PLD.The substrate effect on GaN crystalline growth quality, surface morphology, stress behavior, and interface property were studied.This paper demonstrates the potential of using high-temperature PLD as a growth method for preparing GaN templates that exhibit improved device performance.

Figure 1 .
Figure 1.Low-temperature photoluminescence (PL) spectra (at 77 K) of GaN films grown on different substrates.FWHM: full width at half maximum.

Figure 2 .
Figure 2. X-ray diffraction (XRD) measurements results of GaN films grown on different substrates.

Figure 1 .
Figure 1.Low-temperature photoluminescence (PL) spectra (at 77 K) of GaN films grown on different substrates.FWHM: full width at half maximum.

Figure 1 .
Figure 1.Low-temperature photoluminescence (PL) spectra (at 77 K) of GaN films grown on different substrates.FWHM: full width at half maximum.

Figure 2 .
Figure 2. X-ray diffraction (XRD) measurements results of GaN films grown on different substrates.

Figure 2 .
Figure 2. X-ray diffraction (XRD) measurements results of GaN films grown on different substrates.
that GaN films on Si(100) have a hexagonal phase.The different GaN film structure of the grains can be attributed to the different lattice structure of the Si substrate[29].The surface morphology and roughness of the GaN films grown on different substrates were carried out by AFM measurements with the scanning area of 10 × 10 μm 2 , as shown in Figure4.In Figure4, the root-mean-square RMS values for samples A, B, C, and D are 2.1, 3.4, 14.3, and 17.7 nm, respectively.The film grown in samples A and B exhibited quite a smooth surface, with the RMS roughness being 3.4 and 2.1 nm, respectively, and the RMS surface roughness of samples C and D was estimated as 14.3 and 17.7 nm, respectively.The large values for the surface roughness of the GaN films in samples C and D might be due to the large lattice mismatch between the film and the substrates.A decrease in surface roughness occurs with an increase in grain size, as mentioned in the XRD results.
that GaN films on Si(100) have a hexagonal phase.The different GaN film structure of the grains can be attributed to the different lattice structure of the Si substrate[29].The surface morphology and roughness of the GaN films grown on different substrates were carried out by AFM measurements with the scanning area of 10 × 10 μm 2 , as shown in Figure4.In Figure4, the root-mean-square RMS values for samples A, B, C, and D are 2.1, 3.4, 14.3, and 17.7 nm, respectively.The film grown in samples A and B exhibited quite a smooth surface, with the RMS roughness being 3.4 and 2.1 nm, respectively, and the RMS surface roughness of samples C and D was estimated as 14.3 and 17.7 nm, respectively.The large values for the surface roughness of the GaN films in samples C and D might be due to the large lattice mismatch between the film and the substrates.A decrease in surface roughness occurs with an increase in grain size, as mentioned in the XRD results.

Figure 5 .
Figure 5. Variation in (a) resistivity; (b) carrier concentration; and (c) mobility of GaN films with different substrates.

Figure 6 .
Figure 6.Raman spectra of GaN films for samples MGS (metalorganic chemical vapor deposition (MOCVD)-grown GaN on sapphire), A, B, C, and D.

Figure 5 .
Figure 5. Variation in (a) resistivity; (b) carrier concentration; and (c) mobility of GaN films with different substrates.

Figure 5 .
Figure 5. Variation in (a) resistivity; (b) carrier concentration; and (c) mobility of GaN films with different substrates.

Figure 6 .
Figure 6.Raman spectra of GaN films for samples MGS (metalorganic chemical vapor deposition (MOCVD)-grown GaN on sapphire), A, B, C, and D.

Figure 6 .
Figure 6.Raman spectra of GaN films for samples MGS (metalorganic chemical vapor deposition (MOCVD)-grown GaN on sapphire), A, B, C, and D.

Figure 7 .
Figure 7. Residual stress and its corresponding E2 Raman shift for samples MGS, A, B, C, and D.

Figure 7 .
Figure 7. Residual stress and its corresponding E 2 Raman shift for samples MGS, A, B, C, and D.

Figure 8 .
Figure 8. Cross-sectional TEM pictures of GaN films on samples (a) C and (b) D.Figure 8. Cross-sectional TEM pictures of GaN films on samples (a) C and (b) D.

Figure 8 .
Figure 8. Cross-sectional TEM pictures of GaN films on samples (a) C and (b) D.Figure 8. Cross-sectional TEM pictures of GaN films on samples (a) C and (b) D.