Mosaic Structure of GaN Film Grown on Sapphire Substrate by AP-MOCVD: Impact of Thermal Annealing on the Tilt and Twist Angles

Abstract

A GaN layer with a thickness of 2 µm was grown on a sapphire substrate using atmospheric pressure metal–organic chemical vapor deposition (AP-MOCVD). Subsequently, the layer was annealed under a nitrogen atmosphere at temperatures ranging from 1000 °C to 1120 °C. High-resolution X-ray diffraction (HRXRD) analysis reveals the impact of thermal annealing on the mosaic structure of the GaN, specifically the tilt and twist variations in four planes: (00.2), (10.3), (10.2), and (10.1). Interestingly, the observed trends suggest a differential effect of annealing on screw and edge dislocation densities. The annealing process reduces the edge and screw dislocation density. Lower values (D_screw = 1.2 × 10⁸ cm⁻²; D_edge = 1.6 × 10⁹ cm⁻²) were obtained for the sample annealed at 1050 °C. Notably, both tilt and twist angles exhibited a minimum at 1050 °C (tilt = 252 arcsecs, and twist = 558 arcsecs), indicating improved crystal quality at this specific temperature. Photoluminescence (PL) spectroscopy further complemented the structural analysis. The intensity and broadening of the yellow band (YL) in the PL spectra progressively increased with the increasing annealing temperature, suggesting the presence of additional defect states. The near band edge PL emission (3.35 and 3.41 eV) variation upon thermal annealing was correlated with the mosaic structure evolution.

1. Introduction

III-nitride semiconductors have been widely used in electronic applications due to their allowed direct optical bandgap. In recent decades, researchers have exhibited increasing interest in III-Nitride systems due to the availability of a diverse range of direct band-gap semiconductors. These materials are capable of spanning deep ultraviolet to near-infrared regions due to their good absorption coefficient, high thermal stability, and higher carrier mobility. III-nitride materials, such as InN, GaN, AlN, and their alloys, have been identified as particularly promising for electrical and optoelectronic applications [1]. GaN films have been used in a variety of applications, including photovoltaic (PV) applications [2,3], optoelectronics devices [4,5], power switch tools [6,7], and numerous optical devices that necessitate high-crystalline-quality GaN films. Different methods have been applied to enhance crystal quality, including those detailed in prestigious research articles. Lahrèche et al. [8] explored the complex connection between growth parameters and the development of crystal defects. The study clarified that accurate regulation of variables, like temperature gradients, growth rates, and V/III ratios, greatly affects the quantity and kinds of defects in the GaN crystal lattice. By fine-tuning these parameters, researchers can efficiently reduce the occurrence of harmful defects, like dislocations and point defects, that can significantly impair device performance. Additionally, epitaxial growth entails applying a thin film of GaN onto a substrate with a similar lattice structure, which can considerably diminish the density of dislocations, a primary contributor to defects in GaN crystals [9]. Through the precise choice of substrate material and the enhancement of growth conditions, scientists have successfully produced high-quality GaN films characterized by low dislocation densities. Tao et al. [10] indicated that Al-ion implantation pretreatment on sapphire substrates significantly facilitates the high-quality, orderly nucleation of GaN, resulting in better crystal quality and improved optical performance in GaN-based devices. This method, by promoting high-quality nucleation, greatly decreases the density of threading dislocations, which is a significant issue in hetero-epitaxial GaN growth.

Although III-nitride materials exhibit a wide range of applications due to their excellent thermal, chemical, and mechanical properties, they can also present fabrication challenges. Polishing, doping, and forming ohmic contacts often prove difficult to achieve. Post-growth heat treatment typically resolves these issues. However, it is crucial to determine the heat treatment conditions that ensure the annealed materials’ stability. Temperature influences the realization of devices and all their properties [11,12]. High-temperature regimes are typically employed for the activation of magnesium acceptors, enabling the successful p-type doping of GaN and fabrication of ohmic contacts. Furthermore, post-growth thermal annealing at elevated temperatures constitutes a vital tool for GaN-based materials. It induces a redistribution of defects, consequently altering its optical, electrical, and structural parameters [13]. Rapid thermal annealing has thus far been a widely applied thermal treatment method in numerous research investigations [14]. While widely utilized, it suffers from the disadvantage of uneven wafer heating despite providing rapid heat treatment. As an alternative, traditional furnace annealing can offer a more uniform temperature distribution across the wafer, affording atoms sufficient time to develop structural flaws, such as lattice strain and dislocations. Moreover, it serves as an effective technique for mitigating implantation-induced defects in semiconductors, activating implanted dopants [15], and enhancing crystalline quality in lattice-mismatched systems [16]. Several studies have investigated the thermal stability and ohmic contact of GaN films below 1100 °C [17]. These studies revealed that annealing GaN at temperatures below 900 °C exerts minimal influence on structural characteristics [18], whereas annealing at temperatures between 900 °C and 1100 °C degrades crystalline quality [19].

The presence of twist and tilt angles within the crystal lattice constitutes a crucial parameter in assessing structural quality. The mismatch between the substrate’s and thin film’s crystallographic axes is quantified by the tilt and twist angles. These angles can significantly influence thin-film growth, deformation, and crystal grain alignment, ultimately impacting the films’ chemical and physical properties. According to Ikball et al. [20], successful reduction of the tilt and twist angles of CdS thin films produced by chemical bath deposition (CBD) can be achieved through annealing at temperatures exceeding 300 °C. This approach leads to improved optical properties and crystal quality. By meticulously controlling the annealing temperature and time, researchers can modulate the twist and tilt angles, thereby influencing the film’s properties. Furthermore, enhancement of the physical properties of SnS and Sn₂O₃ thin films produced by RF magnetron sputtering may be attainable by decreasing the tilt and twist angles through annealing at a temperature of 400 °C [21]. Conversely, annealing of ZnO thin films can result in a reduction of the tilt and twist angles, leading to improved optical transparency and electrical conductivity [22]. The effects of annealing on GaN dislocation density and mosaic structure have been examined in several studies [23,24]. After prolonged annealing at 800 °C under N₂, Chen et al. [23] observed a decrease in screw dislocations and a concomitant increase in edge dislocations. In contrast, Lui et al. [24] observed that rapid thermal annealing in the same temperature range diminished mosaic tilt. These contradictory findings suggest complex interactions between annealing conditions and dislocation types. Comprehending and refining this process are essential for realizing the full potential of GaN-based technology. Consequently, investigating the impact of thermal annealing on the tilt and twist angles of various GaN samples is crucial. To gain a comprehensive understanding of its effect on structural and optical properties, this study examines how thermal annealing influences the tilt and twist angles of GaN films. High-resolution X-ray diffraction is employed to quantify the tilt and twist angles of the samples both before and after thermal annealing. Additionally, the study investigates the impact of thermal annealing on the structural and optical characteristics of the GaN films.

Our objective is to provide a comprehensive understanding of the correlation between thermal annealing, tilt and twist angles, and the characteristics of thin films. The structural and optical properties of GaN/Al₂O₃ are investigated before and after annealing using laser reflectometry, HRXRD, and photoluminescence (PL).

2. Materials and Methods

A 2 μm thick GaN layer was grown on a c-axis sapphire substrate at 1120 °C using metal–organic chemical vapor deposition (MOCVD). Trimethylgallium (TMGa) and ammonia (NH₃) served as the Ga and N precursors, respectively, with a mixture of H₂ and N₂ as the carrier gas. More detailed information regarding the growth procedure can be found elsewhere [25]. Due to GaN’s complex decomposition behavior, thermal annealing presents challenges. Annealing was performed in a nitrogen atmosphere within the MOCVD vertical reactor. A pure N₂ ambient was maintained throughout the temperature ramp-up and sample cooling phases. To minimize intergrowth run variability, a single sample was grown. Four samples were cut from this layer and annealed for 25 min at temperatures ranging from 1000 °C to 1120 °C under a N₂ flow of 2 L/min.

The samples were labeled A00 (as-grown), A01 (1000 °C), A02 (1050 °C), A03 (1100 °C), and A04 (1120 °C) according to their annealing temperatures. The temperature was monitored using a thermocouple inserted into the graphite susceptor. The heating rate was 440 °C/min, and the cooling rate was 180 °C/min. In situ reflectometry measurements were conducted to investigate the thermal stability of the GaN layer. A 1 mW He-Ne laser (λ = 632.8 nm) was directed onto the stationary wafer through a quartz window in the reactor. The reflected signal was detected by a photodiode, demodulated by a lock-in amplifier, and fed to a computer via an IEEE 488 interface. Ex situ characterization techniques, including photoluminescence and high-resolution X-ray diffraction (HRXRD), were employed to assess the optical and structural properties of the annealed samples.

3. Results

3.1. In Situ Monitoring

Figure 1 illustrates the in situ laser reflectivity during thermal annealing. The reflectivity measured at an annealing temperature of 1000 °C (A01) for 25 min is nearly constant, indicating that the GaN surface exhibits thermal stability up to this temperature in a nitrogen atmosphere. Conversely, when the annealing temperature reaches 1050 °C (A02), the average reflectivity signal decreases, suggesting a partial degradation of the GaN layer surface. This decrease may be attributed to the partial desorption of Ga and/or N elements from the surface layer, potentially leading to increased surface roughness [26,27,28]. Furthermore, heat treatment can induce localized and non-uniform decomposition of the GaN material, particularly favored near regions where threading dislocations emerge at the surface [14]. At 1100 °C (A03), the reflectivity signal begins to oscillate, indicating GaN thermal decomposition. This observation aligns with previous findings and is dependent on the MOCVD reactor geometry. Fathallah et al. [29] reported that the stripping of GaN under H₂ begins at lower temperatures around 700 °C in a similar vertical reactor. The observed dramatic changes in reflectivity as a function of the annealing temperature in Figure 1 can be attributed to several factors. These include (i) phase transformations driven by even small temperature variations, altering the crystal structure and microstructure; (ii) grain growth, leading to reduced grain boundary scattering; (iii) changes in surface roughness, with smoother surfaces generally exhibiting higher reflectivity; and (iv) the influence of oxidation and impurity diffusion, which can modify the optical properties of the surface. In this work, we focus on the evolution of the optical and structural properties of GaN upon annealing, particularly around the critical temperature of thermal decomposition.

3.2. Surface Morphology

Figure 2 shows SEM pictures of the GaN layers before and after annealing. The untreated layer (A00) has a flat surface free of imperfections. Pits emerge on the layer’s surface following 25 min of annealing at 1050 °C (A02). These pits are most likely caused by selective and non-uniform etching near the points where dislocations appear on the surface. The pit density is around 10⁷ cm⁻². This implies a partial exposure of the dislocations. Figure 2 (c) and (d) illustrate the surface morphologies of samples that were annealed at 1100 °C (A03) and 1120 °C (A04). They are distinguished by the development of pores on the GaN surface. The size and distribution of these pores are not uniform. These observations show the onset of GaN sublimation, culminating in the appearance of the first whitish gallium droplets at an annealing temperature of 1120 °C.

The observed decrease in reflectivity at higher annealing temperatures (Figure 1) correlates well with the morphological changes observed in the SEM images (Figure 2). The increasing surface roughness and the formation of larger pits at higher temperatures likely contribute to the observed decrease in reflectivity.

3.3. Tilt and Twist Angles

For the GaN microstructure analysis, HRXRD was employed. GaN epitaxial films grown on sapphire substrate are often characterized as crystals with mosaic structures, which can be defined by their tilt and twist angles [30]. GaN typically exhibits a columnar structure. These columns are disoriented with respect to each other, with diameters ranging from a few hundred nanometers to a few micrometers. Tilt and twist are two parameters that define this mosaic structure. Tilt represents the disorientation of the GaN columns relative to the growth direction [001], while twist describes the rotation of the GaN columns in the growth plane (001). Figure 3 illustrates a schematic representation of the GaN mosaic structure. The tilt and twist angles of the grains can be determined by analyzing the broadening of the XRD peaks, which provides valuable information about the crystallographic texture and defects within the film.

Diffraction rocking curves (ω-scans) provide valuable insights into the layer’s structural quality. The mean tilt angle of the mosaic structure is directly correlated with the full width at half maximum (FWHM) for the (00.2) plane of the XRD rocking curves. Figure 4 presents the XRD rocking curves around the (00.2) diffraction peak for the prepared samples, both before and after the annealing process. A notable reduction in the (00.2) rocking curve scan broadening is observed after annealing within the 1000–1120 °C temperature range. The tilt angle was significantly reduced, reaching approximately half the value of the as-grown sample after annealing at 1050 °C. In this study, the mean twist angle was determined using the extrapolation method developed by Srikant et al. [31]. This method relies on the phenomenological observation that the broadening of the diffraction spectra exhibits a characteristic variation as a function of the inclination angle (ψ) of the (hkl) planes with respect to the sample surface (00.1). By extrapolating this variation toward the maximum inclination of ψ = 90°, the twist angle can be determined. Furthermore, this method allows for the extraction of a parameter (m) that describes the interdependence between tilt and twist. The planes investigated in this paper—(00.2), (10.3), (10.2), and (10.1)—are inclined at angles of 0°, 32°, 43°, and 61°, respectively, with respect to the growth plane (00.1). The rocking curves were measured for these four planes on each sample

The ω-scans were fitted into a curve determined by a Pseudo–Voigt function, which is given by the following equation:

𝑃(𝑥) = (1 − 𝑓) 𝐺(𝑥) + 𝑓𝐿(𝑥)

(1)

where G(x) and L(x) are Gaussian and Lorentzian functions, respectively, and f gives the fraction of the Lorentzian character of the distribution. The value of f can be obtained by fitting every rocking curve of every plane.

The composition of the measured FWHM as a function of the inclination angle ψ, W(ψ), can be expressed as [25,31]:

$$W\left(\psi \right)={\{{\left[W_{eff}^{twist}\left(\psi \right)\right]}^n+{\left[W_{eff}^{tilt}\left(\psi \right)\right]}^n\}}^{1/n}$$

(2)

where W(ψ) is the FWHM of the experimental result, and n is determined by the following formulation:

𝑛 = 1 + (1 − 𝑓)²

(3)

$W_{eff}^{twist}$ and $W_{eff}^{tilt}\left(\psi \right)$ are, respectively, the twist and the tilt components of the measured FWHM and can be calculated from the following equations:

$$W_0^{tilt}\left(\mathsf{\psi }\right)={cos}^{-1}[{cos}^2\left(\mathsf{\psi }\right)\cos \left(W_y)+{sin}^2\left(\mathsf{\psi }\right)\right]$$

(4)

$$W_0^{twist}\left(\mathsf{\psi }\right)={cos}^{-1}\left[{sin}^2\left(\mathsf{\psi }\right)\mathrm{c}\mathrm{o}\mathrm{s}\left(W_z\right)+{cos}^2\left(\mathsf{\psi }\right)\right]$$

(5)

Assuming this interaction to be exponential in nature, a new effective W is defined as follows [25,31]:

$$W_{eff}^{tilt}\left(\mathsf{\psi }\right)=W_0^{tilt}\left(\mathsf{\psi }\right)\mathrm{e}\mathrm{x}\mathrm{p}(-\mathrm{m}{\displaystyle \frac{W_0^{twist}\left(\mathsf{\psi }\right)}{W_0^{twist}\left(90\right)}}),$$

(6)

$$W_{eff}^{twist}\left(\mathsf{\psi }\right)=W_0^{twist}\left(\mathsf{\psi }\right)\mathrm{e}\mathrm{x}\mathrm{p}(-\mathrm{m}{\displaystyle \frac{W_0^{tilt}\left(\mathsf{\psi }\right)}{W_0^{tilt}\left(0\right)}}),$$

(7)

where Wy and Wz values represent the tilt value corresponding to the FWHM of the (00.2) reflection and the desired value of the twist angle, respectively. m is a parameter characterizing the interdependence between the tilt and twist distributions.

Examples of the twist determination for the two samples A00 and A03 are shown in Figure 5a and Figure 5b, respectively. Solid lines in Figure 4 show the best fitting curves according to Equation (2). We notice that the effect of the tilt $W_{eff}^{tilt}$ on the distortion of the planes (hkl) decreases with the inclination angle ψ. An opposite effect is observed for the twist $W_{eff}^{twist}$. The measured tilt angles and the extracted twist angles are summarized in

Table 1. Tilt, twist, screw, and edge dislocation densities before and after annealing.

Samples	Annealing Temperature	FWHM (tilt) ×10−3 (rad)	FHWM (twist) × 10−3 (rad)	Dscrew (tilt) ×108 (cm−2)	Dedge ×109 (cm−2)	Ddislocations ×109 (cm−2)
A00	As grown	2.47	5.24	5.2	6.2	6.7
A01	1000 °C	1.85	3.98	2.9	3.6	3.9
A02	1050 °C	1.22	2.70	1.2	1.6	1.7
A03	1100 °C	2.09	4.19	3.7	4.1	4.4
A04	1120 °C	2.69	4.36	6.5	4.3	5.1

. Namely, m is the parameter which describes the interaction between the edge and screw-type TDs. Interaction between tilt and twist in the mosaic could lead to its eventual annihilation.

Figure 6 illustrates the variation in the extracted interdependence parameter (m) as a function of annealing temperature. The parameter ‘m’ quantifies the correlation between the tilt and twist distributions of dislocations within the GaN films. Higher ‘m’ values indicate a stronger correlation, while lower values suggest a weaker one. As observed in the figure, the ‘m’ values exhibit a decrease with increasing the annealing temperature from 1000 °C to 1100 °C, followed by a plateau at higher temperatures. This indicates a diminishing interdependence between tilt and twist distributions as the thermal treatment intensifies. This phenomenon can be linked to various mechanisms activated by the annealing process, including dislocation annihilation and rearrangement. At higher temperatures, dislocation annihilation results in a decreased overall density of dislocations, which disrupts the established relationships between tilt and twist distributions. Additionally, the annealing process may facilitate the rearrangement of dislocations into configurations that possess lower energy, thereby modifying the interactions between tilt and twist components within the dislocation network. The observed decrease in ‘m’ aligns well with the expected reduction in tilt and twist after annealing, suggesting that the annealing process not only reduces the overall dislocation density but also alters the distribution and interaction patterns. The positive ‘m’ values, all smaller than 0.3 for our samples, are consistent with the findings reported by Heinke et al. [32], who observed a similar range of positive values for a parameter reflecting the correlation between tilt and twist distributions in epitaxial GaN films using X-ray diffraction analysis.

The variations in the tilt and twist angles with the annealing temperature are shown in Figure 7. The values for as-grown mosaic inclination, tilt, and twist were inserted into the same figure. As we can see, the tilt values for all samples are less than the twist angles. This is consistent with previous studies [13]. It can be observed that the tilt and twist have the same tendency as a function of annealing temperature. Both tilt and twist angles have lower values (tilt = 252 arcsecs, and twist = 558 arcsecs) for the sample annealed at 1050 °C. The twist reduction is twice as large as the tilt reduction. The twist angle is approximately 900 arcsecs when the sample is annealed at 1120 °C, whereas the tilt angle is approximately 555 arcsecs. These detected discrepancies in the variation in tilt and twist suggest that annealing affects the screw and edge dislocation densities differently. This phenomenon could be linked to the defect thickness profiling in hetero-epitaxial films, where a high density is typically detected near the substrate–epilayer interface [33]. Screw and edge dislocation generation and/or annihilation may occur in varying amounts depending on the annealing temperature. Twist and tilt inclination angles were reduced after annealing when compared to the as-grown sample, indicating an improvement in crystal quality. We noted that this improvement is enhanced until an annealing temperature of 1050 °C. However, when the annealing temperature is greater than 1050 °C, the tilt and twist drops are reduced by increasing the temperature.

The decrease in tilt and twist angles in thin films during thermal annealing is caused by a number of significant causes. Grain development, stress relaxation, defect annihilation, and surface diffusion are some of the mechanisms [34,35,36,37]. The activation barrier can be overcome more easily by thermal energy, which enables atoms to rearrange themselves into more energetically favorable shapes. The smaller grains clump together and the larger grains grow to produce a larger average grain size. The total angles of tilt and twist in the film tend to decrease as the size of the grains increases because their borders become straighter and more parallel [34]. Stress can also be created in the film as a result of factors such as lattice mismatch with the substrate or thermal expansion and contraction. Annealing relaxes tensions in the film by promoting atomic diffusion and mobility along grain boundaries. The film’s stress reduction produces a more stable and structured crystal structure, as seen by reduced tilt and twist angles [35]. On the other hand, annealing promotes surface diffusion, a process in which atoms migrate along a surface and tend to occupy step edges and other imperfections. Smoothing the surface can also help decrease tilt and twist angles by decreasing the total surface area. During deposition, various defects, like dislocations, vacancies, and interstitials, can be incorporated into the crystal lattice of the film. Annealing provides the thermal energy needed for these defects to move and interact with each other. This can lead to their annihilation or rearrangement, reducing the overall defect density in the film. Lower defect density translates to a more ordered crystal structure, contributing to lower tilt and twist angles [36,37].

The screw and edge dislocation densities in GaN films are related to the mosaic tilt and twist values, respectively [38]. The screw and edge dislocations densities are given by the following equations:

$$D_{screw}={\displaystyle \frac{{FWHM}_{\left(ooo2\right)}^2}{4.35b_{screw}^2}}{\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}D}_{edge}={\displaystyle \frac{{FWHM}_{twist}^2}{\begin{array}{c}4.35b_{edge}^2\end{array}}}$$

(8)

where b is the length of the Burger vector (b_screw = 0.5185 nm, and b_edge = 0.3189 nm).

The overall dislocation (D_dis) of the structures can be estimated by summation of the D_edge and D_screw equation:

$$D_{dislocation}=D_{screw}+D_{edge}$$

(9)

The calculated dislocation densities before and after annealing are shown in Table 1. GaN grown directly on smooth sapphire substrates often has a threading dislocation density (TD) of more than 10⁹ cm⁻² [39]. It was also reported that the TD density slowly decreases with greater film thickness when it reaches 10⁷ cm⁻² by growing GaN to a thickness of about 300 μm [40]. From Table 1, the edge dislocation density is greater than the screw dislocation density. Edge dislocations are the dominant dislocation type in hexagonal GaN grown by MOCVD on (00.1) sapphire [41]. We can clearly observe that the annealing process reduces the edge and screw dislocation density. The lower values (D_screw = 1.2 × 10⁸ cm⁻²; D_edg = 1.6 × 10⁹ cm⁻²) were obtained for the sample A02 annealed at 1050 °C. When the annealing temperature exceeds 1050 °C, we note the increase in the dislocation densities, which is attributed to the decrease in layer thickness by the decomposition of the GaN layer [25]. Indeed, the decomposition thickness deduced from the reflectivity oscillation is around 0.4 µm and 0.78 µm for annealing temperatures of 1100 °C and 1120 °C, respectively. This could reveal deeper dislocation in the heterostructure GaN/Al₂O₃. According to Chen et al. [23], annealing of GaN under N₂ at 800 °C for a long annealing time (up to 48 h) leads to an increase in the edge dislocation density and a reduction in the screw type one. On the other hand, Lui et al. [24] reported that upon rapid thermal annealing of GaN under nitrogen in the temperature range of 800 °C to 1000 °C, the tilting of the mosaic structure is reduced. The lower dislocation density translates to a more ordered crystal structure, contributing to the lower tilt and twist angles observed in this study. In the study of Q. Shen et al. [42], a significant reduction in both the tilting and twisting of grain features was observed in GaN films grown on vicinal substrates with angles exceeding 0.5 degrees. This improvement in tilt and twist, in turn, led to a decrease in the density of threading dislocations (TDs) within the GaN films. The above-reported discrepancy in the annealing effect on the GaN structural quality is related to the different starting GaN properties and growing process. In this work, GaN films were grown by the sapphire SiN treatment method. While a drop in edge dislocation density after annealing is to be expected due to defect annihilation and rearrangement, our finding of a rise in screw dislocation density for the high thermal annealed sample (A04) compared to the as-grown sample (A00) is noteworthy. The dislocation may multiply or annihilate further, depending on the interaction type between dislocations. The key dislocation generation mechanisms in Ill-nitride epitaxial layers are dislocations at the interface and their propagation into the GaN layer. The dislocation generation is also related to the mechanical properties of the layer and substrate materials, as well as the stress level in the epitaxial system that formed during growth and post-growth annealing. The threading dislocations are typically arranged in small angle boundaries, dividing crystalline layers into domains. We believe that edge and screw dislocations will be influenced differently due to their distinct properties, specifically for sample A04, which is annealed at a critical temperature, announcing the start of GaN thermal decomposition.

3.4. Optical Properties

Figure 8 illustrates the photoluminescence (PL) at room temperature of GaN samples labeled as A00, A02, and A03. The photoluminescence (PL) spectra exhibit two primary transitions: a wide (YL) luminescence band and the ultraviolet (UVL) band, with centers at 2.3 eV and 3.4 eV, respectively. It is commonly acknowledged that the YL is linked to faults related to gallium vacancies [43]. We can see the YL band broadening and the intensity increase upon thermal annealing, indicating a rise in the density of the defect. Thermal annealing of the V and/or III elements causes the desorption of these elements, resulting in the formation of a nonstoichiometric epitaxially defective thin layer of GaN.

The ultraviolet (UVL) peaks, located at 3.4 eV in the various samples, exhibit an overlap of wide peaks. The deconvolution of the near band edge emission, as shown in Figure 9a–c, reveals two distinct peaks at 3.35 eV and 3.41 eV, suggesting the presence of multiple recombination mechanisms within the GaN bandgap. The peak at 3.41 eV appears to have a somewhat higher intensity compared to the peak at 3.35 eV. The presence of two broad and closely spaced peaks indicates the potential occurrence of overlapping contributions from various defect states or transitions. A shallow donor-deep acceptor (SDA-DA) transition with defects may be the cause of the 3.35 eV peak [43,44]. A deeper donor-deep acceptor (DDA-DA) transition involving acceptor-related defects may be responsible for the presence of the 3.41 eV peak [44]. The peaks may also be associated with transitions close to the GaN band edges, though this is less likely. Recombination close to the band edge, specifically between shallow donors and the valence band, may be indicated by the 3.35 eV peak. The 3.41 eV peak might be related to deeper defect states that are situated within the bandgap. Conversely, the observed peaks might be the result of both band edge transitions and DAP transitions. After thermal annealing, we can see the change in peak intensities and FWHM (full width at half maximum) of the two obtained PL peaks, but the peak positions remain constant. Peak 1 was centered at 3.35 eV but slightly varied in position, with 3.359 eV for A00 and 3.352 eV for A02 and A03, with an FWHM of 0.082 eV for sample A00 and 0.062 eV for samples A02 and A03. P2 centered around 3.41 eV, with a slight variation in samples of 3.411 eV for A00 and 3.401 eV for A02 and A03, which was much narrower, and with an FWHM of 0.044 eV for A00 and 0.051 eV for A02 and A03. In the analysis, Peak 1 for sample A00 was of higher intensity, with a value of 0.075, compared to samples A02 and A03, with values of 0.061.

From the XRD analysis, variations in the observed PL spectra can be combined with the detected structural changes. The reduction in the tilt and twist angles presented in Figure 7 indicates the reduced density of the threading dislocations. This should have an impact on the 3.41 eV peak, which is possibly linked to DDA-DA transitions involving such dislocations-related defects. Furthermore, the reduction in the tilt angles, which is related to edge-type dislocations, may also influence the intensity of the 3.35 eV peak due to changes in the density and distribution of the point defects associated with these dislocations.

4. Conclusions

Thermal annealing is a powerful technique for improving crystalline quality and reducing the mosaic structure of GaN thin films. The epitaxial GaN layers deposited on the sapphire substrate were grown using the metal–organic chemical vapor deposition technique, with thicknesses of 2 µm. The prepared sample was exposed to annealing temperatures from 1000 °C to 1120 °C for 25 min under N₂ (2 L/min). The studied mosaic nature for epitaxial layers and the two parameters of tilt and twist angles were carried out. The values of the tilt and twist angles for GaN/Al₂O₃ after various annealing temperatures were calculated. Both the tilt and twist angles have lower values (tilt = 252 arcsecs, and twist = 558 arcsecs) for the sample annealed at 1050 °C. For all the samples, the tilt values are lower than the twist angles. These discrepancies in the variation in tilt and twist suggest that annealing affects the screw and edge dislocation densities differently. The value of the interdependence parameter indicates that in GaN films annealed at 1100 °C and 1120 °C, the tilt and the twist distributions are independent, while those annealed at 1000 °C and 1050 °C are strongly inter-dependent. The change in the tilt and twist angles of GaN samples as a function of thermal annealing potentially affects the intensity of the two peaks observed at 3.35 eV and 3.41 eV in the UVL- PL analysis obtained after deconvolution.