As an important wide-bandgap semiconductor, gallium nitride (GaN) has attracted considerable attention. This paper describes the use of confocal Raman spectroscopy to characterize undoped GaN, n-type GaN, and p-type GaN through depth profiling using 405-, 532-, and 638-nm wavelength lasers. The Raman signal intensity of the sapphire substrate at different focal depths is studied to analyze the depth resolution. Based on the shift of the E2H mode of the GaN epitaxial layer, the interfacial stress for different types of GaN is characterized and calculated. The results show that the maximum interfacial stress appears approximately at the junction of the GaN and the sapphire substrate. Local interfacial stress analysis between the GaN epitaxial layer and the substrate will be very helpful in furthering the applications of GaN devices.

  • This paper characterizes the interfacial stress of n-type, undoped, and p-type GaN on a sapphire substrate using confocal Raman spectroscopy.

  • Based on the shift of the E2H mode of the GaN epitaxial layer, the interfacial stress for different types of GaN is characterized and calculated.

  • Depth profiling of n-type, undoped, and p-type GaN on a sapphire substrate is characterized by 405-, 532-, and 638-nm wavelength lasers.

Gallium nitride (GaN) is a third-generation semiconductor material that has a high breakdown electric field, large forbidden bandgap, high thermal conductivity, high electron saturation velocity, and strong radiation resistance. Its high electron mobility makes GaN highly suitable in the fabrication of high-frequency microwave devices. The suitability of high-frequency device materials is often evaluated using Johnson’s figure of merit. The Johnson's figure of merit is 1 for Si, 2.96 for GaAs, and 22 for SiC—for GaN, it can be up to 37.1 Both SiC and GaN can output high power. Moreover, GaN can output high power at high frequencies.1 Therefore, GaN has obvious advantages in the field of 5G radio frequency, power amplifier materials, power devices, and fast charging.

The GaN epilayer (EPI) exhibits two types of epitaxial growth: homoepitaxial and heteroepitaxial. To date, heteroepitaxial growth has been widely used for producing GaN epitaxial layers using metal organic chemical vapor deposition (MOCVD) or halide vapor phase epitaxy. The selection of heterogeneous substrates generally follows the principles of structure matching, lattice constant matching, and thermal expansion coefficient matching. The substrate material has a significant influence on the crystal quality of heteroepitaxial GaN. The main substrate materials used for GaN heteroepitaxial growth are sapphire, SiC, Si, diamond, and LiAlO2, among which sapphire, SiC, and Si are the most commonly used.2–4 

The main problems with GaN heteroepitaxial growth stem from lattice mismatch and thermal stress mismatch of heterogeneous substrates. The epitaxial growth of GaN on a substrate with a large lattice mismatch will cause a high density of dislocations (108–1010 cm-2) in the GaN epitaxial layer. High-density dislocations reduce the carrier mobility, lifetime, and material thermal conductivity, while forming nonradiative recombination centers and light-scattering centers, thus reducing the luminous efficiency of optoelectronic devices. In addition, electrode metal and impurity metal elements will diffuse into the dislocations to form a leakage current path, thus reducing the output power of the device and seriously affecting its stability.5 Growing GaN on a substrate with an excessively different coefficient of thermal expansion will produce a large biaxial stress during the cooling process and may cause microcracks, which will degrade the optoelectronic properties of the epitaxial layer.4 In addition, the existence of residual stress can lead to the separation of the GaN from the substrate, whereby thermal mismatch stress and shear stress become the driving force for interface separation.5,6 Adding a buffer layer or an insertion layer is a good way of controlling the interfacial stress. Common buffer layers for GaN include AlN and AlGaN graded buffer layers. Therefore, interfacial studies on defect characterization, stress analysis, and new buffer layer settings are the key to preparing high-quality GaN epilayers.

Confocal microscopes use a pinhole or slit placed on the back-image plane of the microscope objective to block light outside the confocal plane. Confocal Raman spectrometry offers greatly improved depth resolution and achieves rapid, nondestructive, and noncontact detection.7 Song et al. used confocal Raman spectrometry to distinguish the longitudinal optical phonon (LO) mode of the epitaxial layer and the longitudinal optical phonon–plasmon coupled (LOPC) mode of a 4H-SiC substrate layer,8 while Yamaguchi et al. applied this technique to characterize the mechanical stress of Si.9 Through confocal micro-Raman spectroscopy, Kladko et al. revealed the epitaxial structure of the nitride layer and the micrometer-scale depth distribution of the deformation gradient in a sapphire substrate interface region.10 Tada et al. studied the stress distribution in the shallow trench isolation structure of Si,11 Kudrin et al. identified the cleavage edge section of the heterostructure of a GaMnAs layer,12 and Holmi et al. characterized the volumetric stress distribution in α-GaN grown by the ammonothermal method, and further determined the types of linear dislocations.13 However, for the stress analysis of GaN on a sapphire substrate, previous analyses have focused on the sample surface rather than depth profiling.

In this paper, undoped, n-type, and p-type GaN materials on a sapphire substrate are characterized with depth profiling through confocal Raman spectroscopy. The interfacial stress distribution and resolution corresponding to different excitation light wavelengths and different depths are analyzed and discussed.

The experiments described in this paper were carried out at room temperature using two confocal Raman spectrometers. The first was a Horiba iHR550 spectrometer with a focal length of 550 mm. The spectral excitation light source was generated by a solid-state Nd:YAG SHG laser with wavelengths of 405 nm at 20 mW and 532 nm at 200 mW. The second was a Horiba XploRA PLUS spectrometer with a 250-mm focal length. The spectral excitation light source was generated by a solid-state Nd:YAG SHG laser with a wavelength of 638 nm at 9.53 mW. Different wavelength lasers have different excitation efficiencies, and so choosing different powers ensures that the Raman scattering intensity of the different wavelength lasers can obtain a clear spectrum without damaging the sample. To avoid photocarbonization during Raman characterization, the laser power attenuation ratio was selected to be 25%, 1%, and 100% for the 405-, 532-, and 638-nm wavelengths, respectively.

All tests were carried out using a 100× objective lens (NA=0.9) and 1800 lines/mm grating. The Horiba iHR550 spectrometer uses the slit confocal method with a slit width of 100 μm, blocking the light outside the confocal plane. According to the optical limit, the spatial resolution is better than 1 μm in the horizontal direction and 2 μm in the vertical direction (parallel to the optical axis). The integration time of spectrum acquisition was 10 s, and two accumulations were conducted. The LabSpec 6 software was used to fit the collected Raman spectrum according to the Gaussian–Lorentzian function to obtain the position, peak intensity, and half-width of the characteristic peak. The equipment was operated on a high-precision XYZ mechanical automatic platform, which could be adjusted through coarse and fine focusing, and a single-point depth test was performed in the Z direction with a minimum step size of 0.25 μm. To ensure the accuracy of the experiment, the single-window mode was used in the single-point depth test to avoid any possible errors from the movement of the grating during a multi-window scanning process. The center of the spectrum range was 568 cm−1.

During the experiment, the focus depth was changed by controlling the position of the XYZ-stage in the Z-axis direction. As shown in Fig. 1, a focus depth of 0 μm corresponds to the surface of the sample. Negative focus depths correspond to the inside of the sample, and positive focus depths correspond to the equipment being above the sample. In this way, the depth of the GaN layer was probed by changing the focus depth.

FIG. 1.

Confocal Raman spectroscopy focusing on different depths of GaN.

FIG. 1.

Confocal Raman spectroscopy focusing on different depths of GaN.

Close modal

All samples of GaN epitaxial wafers used in this paper were purchased from Hengchuan Electrical Co., Ltd. The samples were prepared using MOCVD. The GaN wafers were cut into pieces by laser for the Raman spectroscopy characterizations [see Fig. 2(a)].

FIG. 2.

Illustrations and details of GaN samples. (a) GaN wafer samples: (from left to right) n-type, undoped, and p-type. (b) Illustration of the internal structures of the GaN samples.

FIG. 2.

Illustrations and details of GaN samples. (a) GaN wafer samples: (from left to right) n-type, undoped, and p-type. (b) Illustration of the internal structures of the GaN samples.

Close modal

The test samples measured 5 mm × 5 mm, with a thickness of around 0.43 mm, roughness of approximately 0.5 nm, and dislocation density of 107 cm−2. The carrier concentration of the n-doped and p-doped layers was between 1017 and 1018 cm−3. The resistance was 0.005 Ω·cm. The detailed structures of the samples are shown in Fig. 2(b). There are AlN buffer layers of thickness 25 nm on the sapphire substrate to reduce the lattice mismatch, and 2000-nm-thick undoped GaN was grown on the buffer layer. The surface layer for the n-type GaN sample was a 2000-nm-thick n-type GaN thin layer, whereas the undoped sample had an additional 500-nm-thick undoped GaN thin layer and the p-type GaN sample had an additional 200-nm-thick p-type GaN layer.

For wurtzite GaN, the phonon normal mode at the Γ point, as predicted by group theory, has eight groups of the form 2A1+2B1+2E1+2E2. Among them, the A1 and E1 modes both have one set of acoustic modes, while the remaining six modes (A1+E1+2B1+2E2) are optical. The modes with Raman activity are 2A1+2E1+E2L+E2H, a total of six types.14 As shown in Fig. 3, the phonon dispersion of the hexagonal structure along [0001] (ΓA in the Brillouin zone) is approximated by folding the phonon dispersion of the cubic structure along [111] (ΓL). This folding reduces the transverse optical (TO) phonon mode at the L point of the Brillouin zone in the cubic structure to the E2 mode at the Γ point of the Brillouin zone in the hexagonal structure. This mode is written as E2H, where H denotes the higher-frequency branch of the E2 phonon. In the hexagonal structure, the polar phonons induce anisotropy in the macroscopic electric field. The atomic displacements of the E1 and E2 modes are perpendicular to the c-axis, while the others run along the c-axis. The atomic shift of the E2H mode is performed by the N atom, as shown in Fig. 3.14 

FIG. 3.

Optical phonon modes in the wurtzite structure.14 

FIG. 3.

Optical phonon modes in the wurtzite structure.14 

Close modal

Table I lists the typical phonon frequencies of GaN observed by Raman scattering.15 For the GaN single crystal with a wurtzite structure, there is an Al (TO) mode centered at 533 cm−1. The Alg (S) mode centered at 418.1 cm−1 is the characteristic peak of the sapphire substrate.10 In addition, the 578 cm−1 and 750 cm−1 peaks of the Eg(S) mode relate to the sapphire substrate.16 

TABLE I.

Typical phonon frequencies of wurtzite GaN (cm−1).

E2LAl (TO)E1(TO)E2HA1(LO)E1(LO)Reference
144 531.8 558.8 567.6 734 741 17  
145 533 559 568 … … 18  
144 532 560 569 … … 19  
143 533 559 568 … 726 20  
… … … 570 738 … 21  
144 533 561 569 735 743 22  
E2LAl (TO)E1(TO)E2HA1(LO)E1(LO)Reference
144 531.8 558.8 567.6 734 741 17  
145 533 559 568 … … 18  
144 532 560 569 … … 19  
143 533 559 568 … 726 20  
… … … 570 738 … 21  
144 533 561 569 735 743 22  

The 532 nm laser was used to perform Raman characterizations of the three samples by focusing the laser beam on the samples’ surface. The Raman spectra are shown in Fig. 4. The Raman spectra have been fitted to a Gaussian–Lorentzian function using LabSpec 6. The main peak of the GaN Raman spectra is the E2H mode peak at about 568 cm−1.14 The Raman characteristic peak intensity of the sapphire at 418.1 cm−1 is 771, 687, and 330 counts for the n-type GaN, undoped GaN, and p-type GaN, respectively. As shown in Fig. 2, the n-type GaN has the largest GaN thickness, whereas the p-type GaN has the smallest thickness. The n-type GaN has the strongest sapphire substrate signal, while the p-type GaN has the weakest sapphire substrate signal, as shown in Fig. 4. The absorption coefficients of the samples for the 532-nm laser are in the order p-type GaN > undoped GaN > n-type GaN.

FIG. 4.

Raman spectra with 532-nm-wavelength laser using Horiba iHR550 spectrometer.

FIG. 4.

Raman spectra with 532-nm-wavelength laser using Horiba iHR550 spectrometer.

Close modal

From Fig. 4, it is clear that the p-type GaN has an extra peak on the left side of the E2H mode. The p-type GaN is doped with Mg ions. This peak may be due to the Mg ion doping in the p-type GaN, which increases the number of defects in the sample and enhances the dislocation.23 

Depth profiling of the GaN and sapphire signals was performed on the three samples using the 532 nm laser. Each sample was tested at three different points, as shown in Fig. 5. The depth analysis from 15 μm to −15 μm was achieved by moving the stage along the Z-axis, where negative depths correspond to focusing the laser beneath the surface and positive depths correspond to focusing the laser above the sample. Figure 5(a) shows the variation of the phonon line intensity of A1g(S) in the three samples with respect to the focus depth. The sapphire phonon line intensity gradually decreases as the focus depth moves from the inner part to the sample surface. Figure 5(b) shows the Raman spectra results for the surface and at a depth of −15 μm in the n-type GaN. When focusing at −15 μm, the Raman intensity of the sapphire phonon mode A1g(S) significantly exceeds the E2H mode of GaN. When focusing on the sample surface, the E2H mode is the main Raman signal, and the sapphire signal is weak.

FIG. 5.

Depth profiling of three samples by the excitation of 532-nm laser from the Horiba iHR550 spectrometer. (a) Curve of sapphire A1g(S) mode Raman peak intensity and focal depth for three samples with 100× objective lens with step size of 1 μm. (b) Raman spectra results when focusing on the surface and at −15 μm with n-type GaN. (c) Depth profiling of n-type GaN with a finer step size of 0.5 μm.

FIG. 5.

Depth profiling of three samples by the excitation of 532-nm laser from the Horiba iHR550 spectrometer. (a) Curve of sapphire A1g(S) mode Raman peak intensity and focal depth for three samples with 100× objective lens with step size of 1 μm. (b) Raman spectra results when focusing on the surface and at −15 μm with n-type GaN. (c) Depth profiling of n-type GaN with a finer step size of 0.5 μm.

Close modal

The depth test was repeated with the n-type GaN and a 532-nm laser with a finer step size of 0.5 μm. As shown in Fig. 5(c), the A1g(S) and Eg(S) signals gradually decrease as the focus depth moves from the inside of the sample to the surface, whereas the E2H mode signal for GaN gradually increases. The effective depth profiling of different layers indicates that the Horiba iHR550 spectrometer has a good resolution in the depth (Z) direction.

Different wavelength lasers have different penetration depths. The penetration depth can be calculated from the absorption coefficient as24 

dp=ln0.12α=2.32α,
(1)

where dp is the penetration depth and α is the absorption coefficient. The absorption coefficient can be determined using the relation25 

α=4πkλ,
(2)

where k is an extinction coefficient and λ is the wavelength of the laser. Therefore, the relationship between penetration depth and extinction coefficient can be obtained as

dp=2.3λ8πk.
(3)

The extinction coefficient of bulk GaN has been investigated in previous studies. The curve of extinction coefficient with respect to wavelength is shown in Fig. 6.26 The extinction coefficient of bulk GaN at 405, 532, and 638 nm can be obtained from the curve. The absorption coefficient and penetration depth can then be calculated using Eqs. (1)(3) (see Table II).

FIG. 6.

Extinction coefficient of bulk GaN.26 

FIG. 6.

Extinction coefficient of bulk GaN.26 

Close modal
TABLE II.

Extinction coefficient, absorption coefficient, and penetration depth of bulk GaN by different lasers.

Laser wavelength Extinction Absorption Penetration
λ (nm)coefficient kcoefficient αdepth (μm)
405 1.30×10−2 4.02×10−4 2.86 
532 3.25×10−3 7.68×10−5 14.98 
638 1.38×10−3 2.72×10−5 42.30 
Laser wavelength Extinction Absorption Penetration
λ (nm)coefficient kcoefficient αdepth (μm)
405 1.30×10−2 4.02×10−4 2.86 
532 3.25×10−3 7.68×10−5 14.98 
638 1.38×10−3 2.72×10−5 42.30 

The generation of the Raman scattering spectrum is related to the lattice vibration of the sample. Raman spectroscopy can accurately measure the lattice vibration energy of the material. When there is no stress in the crystal, the crystal lattice vibrates at an intrinsic frequency corresponding to the elastic constant. When the sample is subjected to tensile or compressive residual stress, its atomic bond lengths will extend or shorten accordingly, and the lattice vibration energy will change. Given the relationship between the force constant and the bond length, the force constant of the sample will increase or decrease, so the vibration frequency of the atom will change and the peak of the Raman spectrum will shift to a lower or higher frequency.27–29 

The frequency shift of the Raman spectrum is sensitive to the stress of the sample. Generally, when there is compressive stress inside the sample, the bond length of the molecule will decrease and the force constant will increase. This increases the vibration frequency and shifts the Raman spectrum to the right, with the wave number increasing (more inelastic energy lost during scattering). Conversely, when there is a tensile stress inside the sample, the Raman spectrum shifts to the left and the wave number decreases. Therefore, a frequency shift in the characteristic peak of the Raman spectrum can effectively characterize the local stress of the sample.

The frequency shift has a linear relationship with the stress in the material. The stress at different depths can be calculated according to the frequency shift as

Δω=Kσ,
(4)

where Δω is the frequency shift, σ is the stress, and K is the stress coefficient.30 

According to previous studies, the position of the E2H mode of a standard GaN single crystal is 568 cm−1, and the stress coefficient of GaN grown on c-direction sapphire is 2.56.30 In previous research on different types of GaN substrates, Jiang31 found the compressive stress to be 1.277 GPa when the position of the E2H mode of GaN on sapphire substrate was 571.27 cm−1.

σ=ΔωK=shiftE2H5682.56.
(5)

Raman spectrometry has a certain spectral resolution, but data fitting can effectively increase the sensitivity and elucidate the Raman spectrum shift. For a grating of 1800 lines/mm, the spectral resolution is 1.17 cm−1, and a nominal sub-pixel resolution of 0.02 cm−1 can be obtained after fitting the Raman peak with a Gaussian–Lorentz function, greatly improving the measurement accuracy.32 

The three kinds of GaN samples were characterized by Raman spectroscopy depth profiling from −15 μm to 15 μm with 405 and 532 nm lasers. The Raman spectra of the E2H mode under different focal depths were fitted and further analyzed according to the above calculation method. The frequency shift over all depths was found to be greater than 0.02 cm−1, indicating that the measurement accuracy was sufficient. The interface stress value was then calculated according to the frequency shift. The peak fitting results, frequency shifts, and corresponding stress values of the E2H modes are presented in Tables III and IV. All samples exhibit a frequency shift toward the red end of the spectrum. Therefore, the interfacial state is compressive stress, which is consistent with the compressive stress of the sapphire substrate sample.15,31

TABLE III.

Depth profiling of GaN E2H mode Raman spectra with 532 nm laser (frequency shift and stress of n-type GaN, undoped GaN, and p-type GaN, 100× objective lens, 1800 lines/mm grating, Horiba iHR550 spectrometer).

n-type GaNUndoped GaNp-type GaN
DepthE2H modeShiftStressE2H modeShiftStressE2H modeShiftStress
(μm)(cm−1)(cm−1)(GPa)(cm−1)(cm−1)(GPa)(cm−1)(cm−1)(GPa)
−15 569.33 1.33 0.5195 569.62 1.62 0.6328 568.98 0.98 0.3828 
−12 569.33 1.33 0.5195 569.54 1.54 0.6016 568.97 0.97 0.3789 
−9 569.32 1.32 0.5156 569.57 1.57 0.6133 569.02 1.02 0.3984 
−6 569.37 1.37 0.5352 569.59 1.59 0.6211 569.01 1.01 0.3945 
−3 569.4 1.40 0.5469 569.58 1.58 0.6172 569.09 1.09 0.4258 
−2 569.4 1.40 0.5469 569.56 1.56 0.6094 569.12 1.12 0.4375 
−1 569.44 1.44 0.5625 569.56 1.56 0.6094 569.18 1.18 0.4609 
569.52 1.52 0.5937 569.61 1.61 0.6289 569.28 1.28 0.5000 
569.47 1.47 0.5742 569.60 1.6 0.6250 569.40 1.4 0.5469 
569.38 1.38 0.5391 569.54 1.54 0.6016 569.27 1.27 0.4961 
569.34 1.34 0.5234 569.5 1.5 0.5859 569.16 1.16 0.4531 
569.21 1.21 0.4727 569.42 1.42 0.5547 569.00 1.00 0.3906 
569.25 1.25 0.4883 569.44 1.44 0.5625 568.98 0.98 0.3828 
12 569.23 1.23 0.4805 569.43 1.43 0.5586 568.94 0.94 0.3672 
15 569.29 1.29 0.5039 569.52 1.52 0.5937 568.91 0.91 0.3555 
n-type GaNUndoped GaNp-type GaN
DepthE2H modeShiftStressE2H modeShiftStressE2H modeShiftStress
(μm)(cm−1)(cm−1)(GPa)(cm−1)(cm−1)(GPa)(cm−1)(cm−1)(GPa)
−15 569.33 1.33 0.5195 569.62 1.62 0.6328 568.98 0.98 0.3828 
−12 569.33 1.33 0.5195 569.54 1.54 0.6016 568.97 0.97 0.3789 
−9 569.32 1.32 0.5156 569.57 1.57 0.6133 569.02 1.02 0.3984 
−6 569.37 1.37 0.5352 569.59 1.59 0.6211 569.01 1.01 0.3945 
−3 569.4 1.40 0.5469 569.58 1.58 0.6172 569.09 1.09 0.4258 
−2 569.4 1.40 0.5469 569.56 1.56 0.6094 569.12 1.12 0.4375 
−1 569.44 1.44 0.5625 569.56 1.56 0.6094 569.18 1.18 0.4609 
569.52 1.52 0.5937 569.61 1.61 0.6289 569.28 1.28 0.5000 
569.47 1.47 0.5742 569.60 1.6 0.6250 569.40 1.4 0.5469 
569.38 1.38 0.5391 569.54 1.54 0.6016 569.27 1.27 0.4961 
569.34 1.34 0.5234 569.5 1.5 0.5859 569.16 1.16 0.4531 
569.21 1.21 0.4727 569.42 1.42 0.5547 569.00 1.00 0.3906 
569.25 1.25 0.4883 569.44 1.44 0.5625 568.98 0.98 0.3828 
12 569.23 1.23 0.4805 569.43 1.43 0.5586 568.94 0.94 0.3672 
15 569.29 1.29 0.5039 569.52 1.52 0.5937 568.91 0.91 0.3555 
TABLE IV.

Depth profiling of GaN E2H mode Raman spectra with 405 nm laser (frequency shift and stress of n-type GaN, undoped GaN, and p-type GaN, 100× objective lens, 1800 lines/mm grating, Horiba iHR550 spectrometer).

n-type GaNUndoped GaNp-type GaN
DepthE2H modeShiftStressE2H modeShiftStressE2H modeShiftStress
(μm)(cm−1)(cm−1)(GPa)(cm−1)(cm−1)(GPa)(cm−1)(cm−1)(GPa)
−15 568.43 0.43 0.1680 569.52 1.52 0.5937 569.03 1.03 0.4023 
−12 568.43 0.43 0.1680 569.47 1.47 0.5742 569.09 1.09 0.4258 
−9 568.42 0.42 0.1641 569.52 1.52 0.5937 569.10 1.10 0.4297 
−6 568.43 0.43 0.1680 569.56 1.56 0.6094 569.11 1.11 0.4336 
−3 568.55 0.55 0.2148 569.66 1.66 0.6484 569.19 1.19 0.4648 
−2 568.55 0.55 0.2148 569.69 1.69 0.6602 569.16 1.16 0.4531 
−1 568.62 0.62 0.2422 569.81 1.81 0.7070 569.19 1.19 0.4648 
568.61 0.61 0.2383 569.84 1.84 0.7188 569.19 1.19 0.4648 
568.45 0.45 0.1758 569.86 1.86 0.7266 569.20 1.20 0.4688 
568.44 0.44 0.1719 569.93 1.93 0.7539 569.21 1.21 0.4727 
568.79 0.79 0.3086 569.8 1.8 0.7031 569.34 1.34 0.5234 
568.45 0.45 0.1758 569.21 1.21 0.4727 568.98 0.98 0.3828 
568.41 0.41 0.1602 569.31 1.31 0.5117 569.08 1.08 0.4219 
12 568.40 0.40 0.1562 569.35 1.35 0.5273 569.06 1.06 0.4141 
15 568.42 0.42 0.1641 569.35 1.35 0.5273 569.05 1.05 0.4102 
n-type GaNUndoped GaNp-type GaN
DepthE2H modeShiftStressE2H modeShiftStressE2H modeShiftStress
(μm)(cm−1)(cm−1)(GPa)(cm−1)(cm−1)(GPa)(cm−1)(cm−1)(GPa)
−15 568.43 0.43 0.1680 569.52 1.52 0.5937 569.03 1.03 0.4023 
−12 568.43 0.43 0.1680 569.47 1.47 0.5742 569.09 1.09 0.4258 
−9 568.42 0.42 0.1641 569.52 1.52 0.5937 569.10 1.10 0.4297 
−6 568.43 0.43 0.1680 569.56 1.56 0.6094 569.11 1.11 0.4336 
−3 568.55 0.55 0.2148 569.66 1.66 0.6484 569.19 1.19 0.4648 
−2 568.55 0.55 0.2148 569.69 1.69 0.6602 569.16 1.16 0.4531 
−1 568.62 0.62 0.2422 569.81 1.81 0.7070 569.19 1.19 0.4648 
568.61 0.61 0.2383 569.84 1.84 0.7188 569.19 1.19 0.4648 
568.45 0.45 0.1758 569.86 1.86 0.7266 569.20 1.20 0.4688 
568.44 0.44 0.1719 569.93 1.93 0.7539 569.21 1.21 0.4727 
568.79 0.79 0.3086 569.8 1.8 0.7031 569.34 1.34 0.5234 
568.45 0.45 0.1758 569.21 1.21 0.4727 568.98 0.98 0.3828 
568.41 0.41 0.1602 569.31 1.31 0.5117 569.08 1.08 0.4219 
12 568.40 0.40 0.1562 569.35 1.35 0.5273 569.06 1.06 0.4141 
15 568.42 0.42 0.1641 569.35 1.35 0.5273 569.05 1.05 0.4102 

According to the Raman spectrum characterization results, the depth profiling of the calculated interfacial stress values is shown in Fig. 7. The intensity of the main characteristic peak of the epitaxial layer should be the strongest when the laser focuses on the sample’s surface.7,33 In this paper, the red triangles mark the depth at which the peak intensity of the E2H mode is the strongest. The E2H mode shift and the calculated local stress display variations around the interfacial layers, where the maximum interfacial stress value appears a few micrometers below the sample surface for both the 405 and 532 nm lasers. The thermal conductivity coefficient of GaN is four times greater than that of sapphire below 100°C. There is a large lattice mismatch and thermal mismatch between the sapphire and GaN layers, although the 25-nm AlN thin layer should lessen the mismatch to some extent. Interfacial residual stress will be generated during the growth process of the GaN epitaxial layer. Large residual stress appears at the interface between the sapphire and GaN and extends to the surface of the sample, as shown in Fig. 7. The residual stress gradually drops as the focus point moves towards the sample surface. This is mainly because of the high lattice quality of the EPI layer of the GaN samples.

FIG. 7.

(a) Stress depth profiling of GaN samples with step length of 1 μm under the parameters in Table III. (b) Stress depth profiling of GaN samples with step length of 1 μm under the parameters in Table IV. Red triangles represent the depth at which the E2H mode peak intensity of the GaN is the strongest.

FIG. 7.

(a) Stress depth profiling of GaN samples with step length of 1 μm under the parameters in Table III. (b) Stress depth profiling of GaN samples with step length of 1 μm under the parameters in Table IV. Red triangles represent the depth at which the E2H mode peak intensity of the GaN is the strongest.

Close modal

Compared with the results from using the 532-nm wavelength laser, characterization by the 405 nm laser produces a larger difference between the maximum and minimum values of the interfacial stress. The stress results characterized by the 532 nm laser exhibit smoother transitions during the depth profiling. The difference between the two results is mainly due to the different penetration depths of the lasers.

The residual stress characterized by 532 nm laser Raman spectroscopy shows that the stress can be ordered as undoped GaN > n-type GaN > p-type GaN. Among the three samples, the p-type GaN has the thinnest GaN layer. However, the n-type GaN was found to have a smaller stress value when characterized by the 405 nm laser. Because the laser wavelength influences the attenuation (absorption) coefficient and Raman scattering efficiency,34 there may be a significant difference between the n-type GaN and the other two samples when using the 405 nm laser. Moreover, the penetration depth of GaN with the 405-nm wavelength laser is 2.86 μm (Table II), which is less than the thickness of the n-type GaN. This may affect the accuracy of the characterization, although the actual penetration depth can be more than 2.86 μm due to the influence of the refractive index.8 

To determine the cause of the difference in the residual stress of n-type GaN, confocal Raman depth characterization was performed using the 638-nm wavelength laser of the Horiba XploRA PLUS spectrometer. The peak fitting results, frequency shift, and corresponding stress values of the E2H modes are presented in Table V, and the depth profiling of the calculated interfacial stress values is shown in Fig. 8. The residual stress characterized by 638 nm laser Raman spectroscopy also runs in the order undoped GaN > n-type GaN > p-type GaN. Although the characterization with the 638-nm wavelength laser was not performed using the same spectrometer as for the 532- and 405-nm wavelength lasers, the magnitude relationship and trends for the three samples are the same. This proves that penetration depths of less than the sample thickness may degrade the accuracy of interfacial stress measurements for n-type GaN when using a 405 nm laser.

TABLE V.

Depth profiling of GaN E2H mode Raman spectra with 638 nm laser (frequency shift and stress of n-type GaN, undoped GaN, and p-type GaN, 100× objective lens, 1800 lines/mm grating, Horiba XploRA PLUS spectrometer).

n-type GaNUndoped GaNp-type GaN
DepthE2H modeShiftStressE2H modeShiftStressE2H modeShiftStress
(μm)(cm−1)(cm−1)(GPa)(cm−1)(cm−1)(GPa)(cm−1)(cm−1)(GPa)
−6 569.74 1.74 0.6797 569.94 1.94 0.7578 569.37 1.37 0.5352 
−5 569.62 1.62 0.6328 569.88 1.88 0.7344 569.45 1.45 0.5664 
−4 569.66 1.66 0.6484 569.88 1.88 0.7344 569.40 1.40 0.5469 
−3 569.69 1.69 0.6602 569.96 1.96 0.7656 569.53 1.53 0.5977 
−2 569.75 1.75 0.6836 570.07 2.07 0.8086 569.60 1.60 0.6250 
−1 569.92 1.92 0.7500 570.13 2.13 0.8320 569.62 1.62 0.6328 
570.05 2.05 0.8008 570.02 2.02 0.7891 569.47 1.47 0.5742 
569.95 1.95 0.7617 569.79 1.79 0.6992 569.26 1.26 0.4922 
569.79 1.79 0.6992 569.75 1.75 0.6836 569.35 1.35 0.5273 
569.76 1.76 0.6875 569.78 1.78 0.6953 569.31 1.31 0.5117 
569.77 1.77 0.6914 569.75 1.75 0.6836 569.34 1.34 0.5234 
569.69 1.69 0.6602 569.81 1.81 0.7070 569.47 1.47 0.5742 
569.73 1.73 0.6758 569.72 1.72 0.6719 569.39 1.39 0.5430 
n-type GaNUndoped GaNp-type GaN
DepthE2H modeShiftStressE2H modeShiftStressE2H modeShiftStress
(μm)(cm−1)(cm−1)(GPa)(cm−1)(cm−1)(GPa)(cm−1)(cm−1)(GPa)
−6 569.74 1.74 0.6797 569.94 1.94 0.7578 569.37 1.37 0.5352 
−5 569.62 1.62 0.6328 569.88 1.88 0.7344 569.45 1.45 0.5664 
−4 569.66 1.66 0.6484 569.88 1.88 0.7344 569.40 1.40 0.5469 
−3 569.69 1.69 0.6602 569.96 1.96 0.7656 569.53 1.53 0.5977 
−2 569.75 1.75 0.6836 570.07 2.07 0.8086 569.60 1.60 0.6250 
−1 569.92 1.92 0.7500 570.13 2.13 0.8320 569.62 1.62 0.6328 
570.05 2.05 0.8008 570.02 2.02 0.7891 569.47 1.47 0.5742 
569.95 1.95 0.7617 569.79 1.79 0.6992 569.26 1.26 0.4922 
569.79 1.79 0.6992 569.75 1.75 0.6836 569.35 1.35 0.5273 
569.76 1.76 0.6875 569.78 1.78 0.6953 569.31 1.31 0.5117 
569.77 1.77 0.6914 569.75 1.75 0.6836 569.34 1.34 0.5234 
569.69 1.69 0.6602 569.81 1.81 0.7070 569.47 1.47 0.5742 
569.73 1.73 0.6758 569.72 1.72 0.6719 569.39 1.39 0.5430 
FIG. 8.

Stress depth profiling of GaN samples with step length of 1 μm under the parameters in Table V.

FIG. 8.

Stress depth profiling of GaN samples with step length of 1 μm under the parameters in Table V.

Close modal

Confocal Raman spectroscopy has been used to nondestructively characterize and analyze the interfacial stress of GaN on a sapphire substrate by depth profiling. The distribution of interfacial stress for n-type, undoped, and p-type GaN was characterized and calculated using 405-, 532-, and 638-nm wavelength lasers. By calculating the E2H mode Raman shift, the interfacial compressive stress between the GaN and sapphire substrate was found to run in the order undoped-GaN > n-type GaN > p-type GaN. The stress results characterized by the 532 nm laser showed smoother transitions during the depth profiling. The depth profiling of interfacial stress is useful for solving key problems in the preparation of new semiconductor materials with high-quality GaN substrates and functional devices.

This work was supported by the National Natural Science Foundation of China (Grant Nos. 51575389 and 51761135106), the National Key Research and Development Program of China (Grant No. 2016YFB1102203), the State Key Laboratory of Precision Measuring Technology and Instruments (Pilt1705), and the ‘111’ Project of the State Administration of Foreign Experts Affairs and the Ministry of Education of China (Grant No. B07014).

1.
Zheng
Y
,
Wu
L
,
Shen
B
.
New materials for China’s strategic emerging industries-third generation semiconductor materials
.
Beijing, China
:
China Railway Press
.
2017
(in Chinese).
2.
Engl
K
,
Beer
M
,
Gmeinwieser
N
, et al 
Influence of an in situ-deposited SiNx intermediate layer inside GaN and AlGaN layers on SiC substrates
.
J Cryst Growth
2006
;
289
(
1
):
6
13
.
3.
Seredin
PV
,
Leiste
H
,
Lenshin
AS
, et al 
Effect of the transition porous silicon layer on the properties of hybrid GaN/SiC/por-Si/Si(111) heterostructures
.
Appl Surf Sci
2020
;
508
:
145267
.
4.
Liu
BT
,
Guo
SK
,
Ma
P
, et al 
High-quality and strain-relaxation GaN epilayer grown on SiC substrates using AlN buffer and AlGaN interlayer
.
Chin Phys Lett
2017
;
34
(
4
):
048101
.
5.
Sugahara
T
,
Sato
H
,
Hao
MS
, et al 
Direct evidence that dislocations are non-radiative recombination centers in GaN
.
Jpn J Appl Phys, Part 2
1998
;
37
(
4A
):
L398
L400
.
6.
Li
XB
,
Wu
JJ
,
Liu
NL
, et al 
Self-separation of two-inch-diameter freestanding GaN by hydride vapor phase epitaxy and heat treatment of sapphire
.
Mater Lett
2014
;
132
:
94
97
.
7.
Tabaksblat
R
,
Meier
RJ
,
Kip
BJ
.
Confocal Raman microspectroscopy: Theory and application to thin polymer samples
.
Appl Spectrosc
1992
;
46
(
1
):
60
68
.
8.
Song
Y
,
Xu
ZW
,
Liu
T
, et al 
Depth profiling of ion-implanted 4H–SiC using confocal Raman spectroscopy
.
Crystals
2020
;
10
(
2
):
131
.
9.
Yamaguchi
M
,
Ueno
S
,
Miura
I
, et al 
Experimental mechanical stress characterization of micro-electro-mechanical-systems device using confocal laser scanning microscope combined with Raman spectrometer system
.
Jpn J Appl Phys, Part 1
2007
;
46
(
10A
):
6860
6862
.
10.
Kladko
VP
,
Kolomys
AF
,
Slobodian
MV
, et al 
Internal strains and crystal structure of the layers in AlGaN/GaN heterostructures grown on a sapphire substrate
.
J Appl Phys
2009
;
105
(
6
):
063515
.
11.
Tada
T
,
Poborchii
V
,
Kanayama
T
.
Study of stress distribution in a cleaved Si shallow trench isolation structure using confocal micro-Raman system
.
J Appl Phys
2010
;
107
(
11
):
113539
.
12.
Kudrin
AV
,
Plankina
SM
,
Vikhrova
OV
, et al 
Characterization of the cleaved edge cross section of the heterostructures with GaMnAs layer by the confocal micro-Raman spectroscopy
.
Micron
2017
;
93
:
38
42
.
13.
Holmi
JT
,
Bairamov
BH
,
Suihkonen
S
, et al 
Identifying threading dislocation types in ammonothermally grown bulk α-GaN by confocal Raman 3-D imaging of volumetric stress distribution
,
J Cryst Growth
2018
;
499
:
47
54
.
14.
Harima
H
.
Properties of GaN and related compounds studied by means of Raman scattering
.
J Phys: Condens Matter
2002
;
14
(
38
):
R967
R993
.
15.
Hushur
A
,
Manghnani
MH
,
Narayan
J
.
Raman studies of GaN/sapphire thin film heterostructures
.
J Appl Phys
2009
;
106
(
5
):
054317
.
16.
Li
GH
,
Zhang
W
,
Han
HX
, et al 
Raman scattering in hexagonal GaN epitaxial layer grown on MgAl2O4 substrate
.
J Appl Phys
1999
;
86
(
4
):
2051
2054
.
17.
Siegle
H
,
Eckey
L
,
Hoffmann
A
, et al 
Quantitative determination of hexagonal minority phase in cubic GaN using Raman spectroscopy
.
Solid State Commun
1995
;
96
(
12
):
943
949
.
18.
Manchon
DD
, Jr.
,
Barker
AS
, Jr.
,
Dean
PJ
, Jr.
, et al 
Optical studies of the phonons and electrons in gallium nitride
.
Solid State Commun
1970
;
8
(
15
):
1227
1231
.
19.
Burns
G
,
Dacol
F
,
Marinace
JC
, et al 
Raman scattering in thin-film waveguides
.
Appl Phys Lett
1973
;
22
(
8
):
356
357
.
20.
Cingolani
A
,
Ferrara
M
,
Lugará
M
, et al 
First order Raman scattering in GaN
.
Solid State Commun
1986
;
58
(
11
):
823
824
.
21.
Murugkar
S
,
Merlin
R
,
Botchkarev
A
, et al 
Second order Raman spectroscopy of the wurtzite form of GaN
.
J Appl Phys
1995
;
77
(
11
):
6042
6043
.
22.
Azuhata
T
,
Sota
T
,
Suzuki
K
, et al 
Polarized Raman spectra in GaN
.
J Phys: Condens Matter
1995
;
7
(
10
):
L129
L133
.
23.
Xu
B
,
Yu
Q
,
Wu
Q
, et al 
Effects of strain and Mg-dopant on the photoluminescence spectra in p-type GaN
.
Chin J Phys
2004
;
53
(
1
):
204
209
(in Chinese).
24.
Xu
ZW
,
He
ZD
,
Song
Y
, et al 
Topic review: Application of Raman spectroscopy characterization in micro/nano-machining
.
Micromachines
2018
;
9
(
7
):
361
.
25.
Fouad
SS
,
Amin
GAM
,
El-Bana
MS
.
Physical and optical characterizations of Ge10Se90−xTex thin films in view of their spectroscopic ellipsometry data
.
J Non-Cryst Solids
2018
;
481
:
314
320
.
26.
Lee
JH
,
Lee
B
,
Kang
JH
, et al 
Optical characterization of nanoporous GaN by spectroscopic ellipsometry
.
Thin Solid Films
2012
;
525
:
84
87
.
27.
Iwasaki
T
,
Sasaki
N
,
Moriya
H
, et al 
Molecular dynamics study of stress effects on Raman frequencies of crystalline silicon
.
Trans Jpn Soc Mech Eng Ser A
1997
;
63
(
611
):
1511
1517
.
28.
Li
M
,
Xin
L
,
Yu
J
,
Li
T
.
Measurements of residual stress in oxide scales by Raman spectroscopy
.
J Chin Soc Corro Prot
1999
;
19
(
3
):
185
188
(in Chinese). https://www.jcscp.org/CN/Y1999/V19/I3/185.
29.
Wu
H
,
Meng
Y
,
Su
C
, et al 
Experimental micro-Raman analysis of residual stress in polysilicon films
.
J Mech Strength
2007
;
29
(
2
):
233
236
(in Chinese).
30.
Zhao
DG
,
Xu
SJ
,
Xie
MH
, et al 
Stress and its effect on optical properties of GaN epilayers grown on Si(111), 6H-SiC(0001), and c-plane sapphire
.
Appl Phys Lett
2003
;
83
(
4
):
677
679
.
31.
Jiang
T.
Applied research on Raman spectroscopy characteristics of group III nitride semiconductor materials
.
Xian, China
:
Xidian University
,
2016
(Doctoral dissertation, in Chinese).
32.
Bauer
M
,
Gigler
AM
,
Huber
AJ
, et al 
Temperature-depending Raman line-shift of silicon carbide
.
J Raman Spectrosc
2009
;
40
(
12
):
1867
1874
.
33.
De Wolf
I
,
Pozzat
G
,
Pinardi
K
, et al 
Experimental validation of mechanical stress models by micro-Raman spectroscopy
.
Microelectron Reliab
1996
;
36
(
11-12
):
1751
1754
.
34.
Ren
XY
,
Tian
ZS
,
Sun
LJ
, et al 
Effects of laser wavelength on both water temperature measurement precision and detection depth of Raman scattering lidar system
.
Acta Phys Sin
2014
;
63
(
16
):
164209
.

Zengqi Zhang, State Key Laboratory of Precision Measuring Technology and Instruments, School of Precision Instruments and Opto-Electronics Engineering, Tianjin University, China. Mr. Zhang is currently studying for a Master’s degree. His research interests include the interfacial stress characterization of GaN epitaxial layers by confocal Raman spectroscopy.

Zongwei Xu, Associate Professor, State Key Laboratory of Precision Measuring Technology and Instruments, School of Precision Instrument and Opto-Electronics Engineering, Tianjin University, China. His research interests include defect engineering in wide-bandgap semiconductors, micro/nanofabrication using focused ion beams, and Raman and fluorescence spectroscopy.

Ying Song, State Key Laboratory of Precision Measuring Technology and Instruments, School of Precision Instruments and Opto-Electronics Engineering, Tianjin University, China. Ms. Song is currently studying for a PhD. Her research interests include the manufacture and spectral characterization of silicon carbide color centers, including the preparation of silicon carbide color centers by ion-implantation, three-dimensional Raman and photoluminescence spectral characterization, and models of spectral depth profiling.

Tao Liu, State Key Laboratory of Precision Measuring Technology and Instruments, School of Precision Instruments and Opto-Electronics Engineering, Tianjin University, China. Mr. Liu is currently studying for a Master’s degree. His research interests include Raman and fluorescence spectrum characterization.

Bing Dong, State Key Laboratory of Precision Measuring Technology and Instruments, School of Precision Instrument and Opto-Electronics Engineering, Tianjin University, China. Mr. Dong is currently studying for a Master’s degree.

Jiayu Liu, State Key Laboratory of Precision Measuring Technology and Instruments, School of Precision Instruments and Opto-Electronics Engineering, Tianjin University, China. Mr. Liu is currently studying for a Master’s degree. His research interests include the preparation of silicon carbide color centers by femtosecond laser, and Raman and fluorescence spectrum characterization.

Hong Wang, Associate Professor, School of Materials Science and Engineering, Tianjin Polytechnic University. Ms. Wang is mainly engaged in inorganic catalytic membrane materials, organic electrochemical synthesis, and electrochemical research and development of electro-catalytic membrane reactors and sewage treatments.