Manufacturing method for nitride semiconductor light-emitting devices
By employing in situ reflectance measurement and an absorption region to account for temperature-induced shifts, the method addresses the challenge of controlling resonator length in GaN-based lasers, achieving precise and reproducible resonator formation.
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- MEIJO UNIVERSITY
- Filing Date
- 2022-09-26
- Publication Date
- 2026-05-21
AI Technical Summary
The challenge in manufacturing GaN-based vertical-cavity surface-emitting lasers is the inability to accurately control the resonator length with high precision and reproducibility due to uncertainties in the refractive index shift and reflectance changes during epitaxial growth, which are not adequately addressed by existing methods.
A method involving in situ reflectance spectrum measurement during epitaxial growth, using a wavelength longer than the resonant wavelength at room temperature to account for temperature-induced shifts, and incorporating an absorption region within the resonator to enhance reflectance changes, allowing precise control of the resonator length.
Enables precise and reproducible control of the resonator length in GaN-based surface-emitting lasers, ensuring the desired resonator length is achieved with high accuracy and maintaining device performance.
Smart Images

Figure 0007863327000003 
Figure 0007863327000004 
Figure 0007863327000005
Abstract
Description
[Technical Field]
[0001] This invention relates to a method for manufacturing a nitride semiconductor light-emitting element. [Background technology]
[0002] In a vertical-cavity surface-emitting laser, the resonator is formed perpendicular to the wafer surface. Generally, the resonance condition is met and the device's inherent performance is realized when the laser's resonator length is an integer multiple of half the oscillation wavelength (1 / 2 wavelength). Here, the resonator length is synonymous with the thickness of the resonator. Therefore, in order to form a desired resonator length in a surface-emitting laser, it is necessary to control the resonator length during resonator formation (i.e., often during epitaxial growth). To perform such control, the growth rate of each layer constituting the resonator must be known, and the growth time must be calculated in conjunction with the design layer thickness, and epitaxial growth must be carried out based on this. The challenge here is that it is not possible to accurately determine the growth rate of all layers, and even if it is known, the value may change over time when actually forming the resonator, making it not always possible to form the desired resonator length with high precision and reproducibility.
[0003] To address the above challenges, in commercially available GaAs-based infrared surface-emitting lasers, the resonator length can be controlled with high precision by simultaneously measuring the reflectance spectrum (including wavelength dependence) of the growing wafer in situ during epitaxial growth, which forms the resonator. This method utilizes the change in reflectance at the resonant wavelength as epitaxial growth progresses (i.e., as the resonator length increases). Specifically, it utilizes the change in reflectance at the resonant wavelength under resonant and non-resonant conditions.
[0004] Generally, the refractive index of a semiconductor material at the growth temperature (hereinafter simply referred to as the growth temperature) during crystal growth changes to a value greater than that at room temperature (approximately 20 degrees Celsius). The growth temperature is higher than room temperature. Therefore, the resonance wavelength at the growth temperature becomes longer than the resonance wavelength at room temperature. Consequently, it is necessary to measure the reflectance value at a resonance wavelength different from that at room temperature, taking into account the wavelength shift of the resonance wavelength due to temperature changes. In GaAs-based materials, the temperature dependence of physical properties such as the refractive index has been studied in detail, and by using this, the resonance wavelength at the growth temperature can be calculated from the resonance wavelength at room temperature. Furthermore, instead of measuring the reflectance at a fixed wavelength, it is possible to correspond to the resonance wavelength for various growth temperatures by measuring the reflectance spectrum within a certain wavelength range (for example, a range of several hundred nanometers).
[0005] In contrast, to date, there have been no reports of measuring the reflectance spectrum during resonator formation (i.e., at the growth temperature) for GaN-based surface-emitting lasers. On the other hand, from the perspective of measuring reflectance at the growth temperature, Non-Patent Literature 1 has only observed the reflectance value at a fixed wavelength, the resonance wavelength at room temperature (see Figure 4 right in Non-Patent Literature 1). In other words, at the growth temperature during the manufacturing of GaN-based surface-emitting lasers, changes in reflectance are observed at wavelengths that are no longer the resonance wavelength. Therefore, even if oscillations in the reflectance value are observed, since it is not the resonance wavelength, the desired resonator length cannot be controlled with high precision by simply using the obtained oscillations in the reflectance value. [Prior art documents] [Patent Documents]
[0006] [Non-Patent Document 1] Tien-Chang Lu, Jun-Rong Chen, Shih-Wei Chen, Hao-Chung Kuo, Chien-Cheng Kuo, Cheng-Chung Lee, and Shing-Chung Wang, "Development of GaN-Based Vertical-Cavity Surface-Emitting Lasers", IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL.15, NO.3, 2009, p850. [Overview of the project] [Problems that the invention aims to solve]
[0007] The reason why the resonance wavelength at the growth temperature was not used in Non-Patent Literature 1 is that the refractive index value of GaN and other materials at the growth temperature, or the amount of wavelength shift of the resonance wavelength when changed from room temperature to the growth temperature, is not clear. Furthermore, even if the resonance wavelength at the growth temperature were known, it would be unclear whether the reflectivity at that wavelength would change significantly enough to be properly observed as the resonator length and thickness increase during resonator formation, or more specifically, whether it would oscillate in accordance with the resonance conditions.
[0008] The present invention has been made in view of the above-mentioned conventional circumstances, and provides a method for manufacturing a nitride semiconductor light-emitting element that can control the desired resonator length with high precision and reproducibility during the formation of a GaN-based surface-emitting laser resonator (using a reflectance spectrum measured in parallel in situ, etc.). [Means for solving the problem]
[0009] The present invention's method for manufacturing a nitride semiconductor light-emitting element is: A multilayer mirror lamination process for forming a multilayer mirror by stacking multiple layers, After the multilayer mirror lamination process is performed, a resonator lamination process is performed in which a resonator is laminated on the surface of the multilayer mirror, A reflectance measurement step is performed while carrying out the resonator stacking step, and the reflectance of the resonator is measured. Equipped with, The wavelength of light used in the reflectance measurement process is longer than the resonant wavelength of the multilayer mirror at room temperature.
[0010] With this configuration, when growing nitride semiconductor light-emitting elements by stacking them, the resonant wavelength of the nitride semiconductor light-emitting element becomes longer at the growth temperature. Therefore, by making the wavelength of light used in the reflectance measurement process longer than the resonant wavelength (reflection center wavelength) of the multilayer mirror at room temperature, it is possible to measure the resonant wavelength at an appropriate level for controlling the resonator length with high precision and reproducibility while growing the nitride semiconductor light-emitting element. [Brief explanation of the drawing]
[0011] [Figure 1] (A) is a graph showing the temperature dependence of the reflectance spectrum of an AlInN / GaN multilayer mirror with a resonant wavelength of 417 nm at room temperature, and (B) is a graph showing the temperature dependence of the reflectance spectrum of an AlInN / GaN multilayer mirror with a resonant wavelength of 527 nm at room temperature. [Figure 2] Figure 2 is a graph showing the relationship between the reflection center wavelength, i.e., the wavelength shift amount at the substrate temperature, and the resonant wavelength of two different types of AlInN / GaN multilayer mirrors with different resonant wavelengths. [Figure 3] Figure 3 is a graph showing the results of theoretical calculations of the change in reflectivity when an AlInN / GaN multilayer mirror is epitaxially grown to increase crystal growth and thus increase the resonator length. [Figure 4] Figure 4 is a graph showing the measured reflectance spectrum of an AlInN / GaN multilayer mirror at the growth temperature. [Figure 5] With respect to the resonance wavelength at the growth temperature, (A) is a schematic diagram showing the structure of a sample in which the entire GaN resonator is the absorption region, and (B) is a schematic diagram showing the structure of a sample in which a portion of the stacking direction of the GaN resonator is the absorption region. [Figure 6]FIG. 6 is a graph showing the calculation results of the reflectance spectrum when the entire inside of a GaN resonator with a thickness of one wavelength is taken as an absorption region with respect to the resonance wavelength at the growth temperature, and when the absorption coefficient of the absorption region is changed from 0 to 100,000 cm-1. [Figure 7] (A) is the reflectance spectrum measured in parallel in situ for five different resonator lengths while growing the resonator, and (B) is a graph showing the dependence of the reflection center wavelength in the reflectance spectrum of the AlInN / GaN multilayer mirror in (A), that is, the reflectance intensity at the resonance wavelength of 439 nm, on the crystal growth time (i.e., the resonator length). [Figure 8] FIG. 8 is a graph showing the dependence of the reflectance intensity at the resonance wavelength at the growth temperature on the crystal growth time (i.e., the resonator length) during the crystal growth of a GaN resonator including an n-GaN layer, a GaInN light-emitting layer, and a p-GaN layer. [Figure 9] FIG. 9 is a schematic diagram showing the structure of the sample in Example 3. [Figure 10] FIG. 10 is a graph showing the results of calculating the reflectance spectrum by changing the thickness of the n-GaInN absorption layer in FIG. 9 in four steps. [Figure 11] FIG. 11 is a schematic diagram showing the structure of the sample in Example 4. [Figure 12] FIG. 12 is a graph showing the results of calculating the change in reflectance at the resonance wavelength at the growth temperature in the GaN resonator of Example 4.
Embodiments for Carrying out the Invention
[0012] Preferred embodiments of the present invention will be described.
[0013] The wavelength of light used in the reflectance measurement process can be 11.8 nm to 26.7 nm greater than the resonant wavelength (reflection center wavelength) of the multilayer mirror at room temperature. This configuration allows for almost complete coverage of the wavelength shift from room temperature to the growth temperature in the wavelength range (375 to 675 nm) that can be emitted from nitride semiconductor light-emitting elements. Therefore, it is possible to measure the reflectance at the resonant wavelength that takes into account the wavelength shift during crystal growth of a resonator having a desired resonant wavelength at room temperature.
[0014] The reflection center wavelength (resonance wavelength) of a multilayer mirror at its growth temperature can be determined by calculating a shift amount based on two parameters: the resonance wavelength at room temperature and the growth temperature. This shift amount is then added to the resonance wavelength of the multilayer mirror at room temperature. With this configuration, once the two parameters—the resonance wavelength at room temperature and the growth temperature—are determined, the reflection center wavelength (resonance wavelength) of the multilayer mirror at its growth temperature can be easily determined.
[0015] The system further includes an absorption region lamination process in which an absorption region is laminated within the resonator during the resonator lamination process. The absorption region laminated in the absorption region lamination process does not absorb light at the resonant wavelength of the multilayer mirror at room temperature, but at the growth temperature, it can absorb light with a longer wavelength than the resonant wavelength of the multilayer mirror at room temperature. With this configuration, the absorption region can absorb light with a longer wavelength than the resonant wavelength at room temperature at the growth temperature, thereby making it easier to observe the change in reflectivity at the resonant wavelength of the resonator during the resonator lamination process, i.e., at the growth temperature.
[0016] By performing an absorption region stacking process, the entire resonator can be set as an absorption region. With this configuration, light within the resonator can be reliably absorbed.
[0017] When the entire resonator is considered an absorption region, the absorption coefficient in the absorption region is 100 cm². -1 Above and 30,000 cm -1 The following is possible. With this configuration, the change in the reflectivity of the resonator can be made to be greater than 1%.
[0018] By performing an absorption region stacking process, a portion of the resonator's stacking direction can be designated as an absorption region. With this configuration, since only a portion of the resonator needs to be designated as an absorption region, the configuration of the resonator outside the absorption region does not need to be modified.
[0019] The thickness in the stacking direction within the absorption region can be 0.3 nm or more and 10 nm or less. With this configuration, it is possible to maintain the quality of the resonator while reducing the change in the reflectivity of the resonator to a magnitude of 1% or more.
[0020] The absorption region can be the active layer. With this configuration, the active layer can be used as the absorption region, eliminating the need to create a separate absorption region.
[0021] <Example 1> This section describes how to measure the reflectance spectrum in situ while simultaneously growing crystals in a GaN surface-emitting laser, where an AlInN / GaN multilayer reflector is located at the bottom in the stacking direction, and a GaN resonator is stacked on its surface.
[0022] [Regarding the shift in the resonant wavelength from room temperature at the growth temperature] To clarify the shift in the resonant wavelength from room temperature at the growth temperature, the temperature dependence of two different resonant wavelengths of AlInN / GaN multilayer reflectors was investigated. Here, the resonant wavelength is the reflection center wavelength of the AlInN / GaN multilayer reflector. The reflection center wavelength of a multilayer reflector coincides with the design wavelength, allowing for the most effective utilization of the multilayer reflector's performance. For this reason, in typical device design, this reflection center wavelength is matched with the resonant wavelength. Regarding the resonant wavelength of AlInN / GaN multilayer reflectors at room temperature, one sample had a resonant wavelength of 417 nm, and the other had a resonant wavelength of 527 nm. These samples were placed in an organometallic compound vapor phase epitaxy (MOVPE) apparatus equipped with a measuring device (hereinafter simply referred to as the measuring device) that can measure reflectance spectra in parallel in situ while increasing the substrate temperature. Here, without crystal growth, the substrate temperature was changed to a temperature higher than room temperature under a nitrogen atmosphere, and the reflectance spectra were measured using the measuring device. The resulting reflectance spectra are shown in Figures 1(A) and (B). As expected, it became clear that the resonant wavelength became longer as the substrate temperature increased.
[0023] Figure 2 shows the wavelength shift of the resonant wavelength with respect to the substrate temperature. In most cases, the growth temperature for most GaN resonators is between 900°C and 1100°C. From this, as shown in Figure 2, it was found that for these two types of samples, the shift in the resonant wavelength between room temperature and the growth temperature is approximately 16 nm to 25 nm. For GaAs-based multilayer mirrors, based on the temperature dependence of the refractive index, the shift in the resonant wavelength between room temperature and the growth temperature is expected to be around 50 nm to 70 nm. In other words, the shift in the resonant wavelength of conventional GaAs-based materials and GaN-based materials are in different ranges, so the shift in the resonant wavelength of GaN-based materials cannot be inferred from the shift in the resonant wavelength of conventional GaAs-based materials.
[0024] From the experimental results described above, it was found that the reflection center wavelength (set as the resonance wavelength) of a GaN-based multilayer mirror at a growth temperature higher than room temperature can be determined by calculating the shift amount Δλ of the resonance wavelength based on a function using two parameters: the resonance wavelength λ at room temperature and the growth temperature T. This shift amount Δλ is then added to the resonance wavelength λ at room temperature. With this configuration, if the two parameters, the resonance wavelength λ at room temperature and the growth temperature T, are determined, the resonance wavelength (reflection center wavelength) of the multilayer mirror at the growth temperature can be easily determined. When the wavelength shift amount Δλ(λ, T) is approximated as a linear function using these two parameters (resonance wavelength λ at room temperature and growth temperature T), it was found in this study that it has the relationship shown in Equation 1.
[0025]
number
[0026] In the case of a GaN-based surface-emitting laser with a GaInN quantum well active layer, the resonant wavelength at room temperature is expected to be between 375 nm and 675 nm. Furthermore, in most cases, the growth temperature of the GaN resonator is between 900°C and 1100°C. From this, the shift amount Δλ in the range of resonant wavelength (375 nm to 675 nm) and growth temperature (900°C to 1100°C) was calculated using Equation 1 and, as shown in Table 1, is found to be within the range of 11.8 nm to 26.7 nm. From Table 1, it can be seen that the shift amount of the resonant wavelength decreases as the resonant wavelength at room temperature increases, and that the shift amount of the resonant wavelength increases as the growth temperature increases.
[0027] [Table 1]
[0028] [Regarding the change in reflectivity intensity at the resonant wavelength at growth temperature] Next, we investigated the change in reflectance intensity at the resonant wavelength at the growth temperature. Here, we selected 40 pairs of AlInN / GaN multilayer mirrors with a resonant wavelength of 417 nm at room temperature and considered the case where the GaN resonator is grown at 1000°C. In this case, Δλ = 22 is obtained from Equation 1. Therefore, it can be seen that the resonant wavelength at a growth temperature of 1000°C is 417 nm + 22 nm = 439 nm.
[0029] We theoretically calculated the change in reflectivity when an AlInN / GaN multilayer mirror meeting these specifications is epitaxially grown to increase crystal growth and thus increase the resonator length. The resonant wavelength of the light irradiated onto the AlInN / GaN multilayer mirror during reflectivity measurement was set to 439 nm. The results are shown in Figure 3. As shown in Figure 3, the reflectivity intensity at the resonant wavelength clearly oscillated every half wavelength, but the amplitude of this oscillation was only about 0.4%.
[0030] Here, Figure 4 shows the measured reflectance spectra of AlInN / GaN multilayer mirrors at their growth temperatures. As shown in Figure 4, when actually measuring the reflectance spectrum, various noises are included during the measurement, causing the spectral waveform to become distorted. Due to this distortion, it was found that accurately observing intensity oscillations of less than 1% is extremely difficult and impractical.
[0031] To significantly change the reflectivity intensity at the resonant wavelength according to the resonator length, it is necessary to significantly change the proportion of reflected light at the resonant wavelength, i.e., light returning to the substrate surface, between the resonant and non-resonant states. Here, the amount of light at the resonant wavelength accumulated in the resonator during crystal growth differs between the resonant and non-resonant states. Specifically, a relatively large amount of light accumulates in the resonator in the resonant state, while a relatively small amount accumulates in the non-resonant state. On the other hand, as shown in Figure 3, the difference between the resonant state (e.g., 0.5λ and 1.0λ) and the non-resonant state (e.g., 0.75λ and 1.25λ) regarding the amount of light that is ultimately reflected back to the incident side (i.e., reflected) is less than 1%, and does not change significantly. This is because, in a resonator where a highly reflective mirror exists at the bottom of the wafer, the light accumulated in the resonator is ultimately reflected by the lower mirror and returns to the incident side, i.e., the wafer surface.
[0032] Therefore, it would be desirable to significantly change the amount of light returning to the incident side in the resonant and non-resonant states. One way to achieve this is to create an absorption region within the resonator that actively absorbs light. In the resonant state, a large amount of light accumulates in the resonator, resulting in a large amount of light absorption and ultimately a small amount of light being reflected back to the incident side. In the non-resonant state, the amount of light absorbed within the resonator is small, so the amount of light ultimately reflected back to the incident side does not change much. On the other hand, if the amount of light absorbed within the resonator is too large, light will be absorbed even in the non-resonant state, and although the amount of reflected light returning to the incident side will decrease, the difference in reflectivity between the resonant and non-resonant states will disappear.
[0033] Therefore, in a nitride semiconductor light-emitting element equipped with this AlInN / GaN multilayer mirror and GaN resonator, we considered intentionally creating an absorption region within the GaN resonator to absorb light. Here, we considered two cases: one in which the entire GaN resonator 12 is set as the absorption region Ab1 to absorb light, as shown in Figure 5(A), and another in which a portion of the GaN resonator 112 in the stacking direction (up and down direction in Figure 5) is set as the absorption region Ab2 to absorb light, as shown in Figure 5(B).
[0034] <Example 2> [When light is absorbed throughout the entire resonator] We consider the case where the entire GaN resonator 12 absorbs light (see Figure 5(A)). Specifically, a multilayer mirror stacking process is performed to stack 40 pairs of AlInN / GaN multilayer mirrors 11. After the multilayer mirror stacking process, a resonator stacking process is performed to stack the GaN resonator 12, which is the resonator, on the surface of the AlInN / GaN multilayer mirrors 11. The thickness of the GaN resonator 12 is 1 wavelength (corresponding to the resonant state). Figure 6 shows the calculation results of the reflectance spectrum when the entire GaN resonator 12 is set as the absorption region Ab1. The resonant wavelength at the growth temperature was set to 439 nm (417 nm at room temperature), the same as in Example 1. Since it is in a resonant state, 439 nm light is accumulated inside the GaN resonator 12, but the absorption coefficient α = 0 cm -1 In this case, the reflectance at the resonant wavelength remains high (corresponding to the position of 1λ in Figure 3, where the reflectance is approximately 99.6%). On the other hand, the absorption coefficient α = 100 cm -1 At this point, the reflectance at the resonant wavelength of 439 nm becomes equal to the absorption coefficient α = 0 cm. -1 It can be seen that this is more than 1% lower than in the previous case.
[0035] Furthermore, increasing the amount of light absorbed in the GaN resonator 12 (i.e., increasing the absorption coefficient α) results in an absorption coefficient α = 30000 cm². -1 Up to this point, the reflectivity clearly decreased at the resonant wavelength of 439 nm. On the other hand, the absorption coefficient α was 100,000 cm -1 It was also found that if the value is increased to this extent, the phenomenon of selectively decreasing reflectivity at the resonant wavelength of 439 nm no longer appears, and information for determining the resonator length cannot be obtained. Absorption coefficient α = 100,000 cm -1 This value is equivalent to the interband absorption value of nitride semiconductors, including GaN. Therefore, it was found that if enough light is absorbed throughout the entire resonator to cause interband absorption, the objective of obtaining information for determining the resonator length cannot be achieved.
[0036] As a method of changing the absorption coefficient α of the semiconductor crystal constituting the resonator, adjusting the bandgap or the doping amount is mentioned on the premise of using interband absorption or absorption between impurity levels. However, it is essential that it absorbs at the resonance wavelength of the growth temperature but does not absorb at the resonance wavelength at room temperature. This is because if it absorbs at the resonance wavelength at room temperature, the device characteristics will be significantly deteriorated. On the other hand, the resonance wavelength at the growth temperature (for example, 439 nm) is 11.8 nm to 26.7 nm larger than the coexistence wavelength at room temperature (for example, 417 nm). Generally, in interband transitions, all are absorbed at wavelengths shorter than the wavelength corresponding to the bandgap (this is called the absorption edge), the absorption amount decreases rapidly at wavelengths longer than that, and no absorption occurs at wavelengths sufficiently separated from the absorption edge. Therefore, it seems that the above requirements cannot be satisfied. In fact, the bandgap and impurity levels also have temperature dependence. From room temperature to the growth temperature, the wavelength becomes longer by about 60 nm, and the amount of wavelength shift is larger than the wavelength shift amount of the above resonance wavelength (less than 30 nm). Therefore, as the semiconductor material constituting the resonator in this embodiment, for example, if GaN (bandgap: 3.4 eV) is selected, the absorption edge at room temperature is 370 nm, and it does not absorb at the resonance wavelength of 417 nm at room temperature. On the other hand, at the growth temperature, the absorption edge is shifted to a longer wavelength up to around 430 nm, and at the resonance wavelength of 439 nm at the growth temperature, by moderately receiving the influence of the above interband absorption, the reflectivity of the resonance wavelength vibrates according to the resonator length. Based on this idea, when the resonance wavelength is set to a shorter wavelength (close to 375 nm), instead of GaN, AlGaN with a larger bandgap or the like may be used. Also, when the resonance wavelength is set to a longer wavelength (close to 675 nm), GaInN with a smaller bandgap or the like may be used.
[0037] Subsequently, actually at a growth temperature of 1000 °C, the resonance wavelength is 439 nm, and the absorption coefficient α = 11000 cm for the entire GaN resonator 12 with a thickness of one wavelength -1We considered the case where this is introduced. Here, as mentioned above, we utilized the property that the interband absorption of GaN (more precisely, the wavelength slightly longer than at which interband absorption occurs completely, and the absorption coefficient α is smaller than that of interband absorption) becomes longer in wavelength as the temperature increases. Specifically, at 1000°C (growth temperature), the absorption coefficient α = 11000 cm⁻¹ at 439 nm. -1 (Interband absorption (absorption coefficient α = 100,000 cm) -1 An absorption region Ab1 (less than ) is made to exist throughout the GaN resonator 12. In other words, during the execution of the resonator stacking process, an absorption region stacking process is performed in which the absorption region Ab1 is stacked within the GaN resonator 12. At the growth temperature, the absorption region Ab1 absorbs not only 417 nm light but also 439 nm light with an absorption coefficient α = 11000 cm². -1 It absorbs light at 439 nm, but at room temperature, it does not absorb light at 417 nm, nor does it absorb light at 439 nm. In other words, at room temperature, absorption region Ab1 does not absorb light at the resonant wavelength of the multilayer mirror at room temperature (417 nm), but at the growth temperature, it absorbs light with a longer wavelength (439 nm) than the resonant wavelength of the multilayer mirror at room temperature (417 nm).
[0038] In any case, if the entire GaN resonator 12 is considered the absorption region Ab1, then the absorption coefficient α in the absorption region Ab1 is 100 cm². -1 Above and 30,000 cm -1 The settings should be as follows. More preferably, the absorption coefficient α in the absorption region Ab1 is 1000 cm -1 From 30,000 cm -1 Set it so that the absorption coefficient α is 0 cm. -1 Compared to the previous case, it is possible to change the reflectivity at the growth temperature by more than 10%. Furthermore, if light is absorbed at room temperature, the nitride semiconductor light-emitting element will not be able to efficiently oscillate as a laser, so it is also necessary that there is no absorption of light at the resonant wavelength at room temperature (417 nm in this case). In other words, the absorption region Ab1 does not absorb light at the resonant wavelength (417 nm in this case) at room temperature, but at the growth temperature it has the function of absorbing light at the resonant wavelength (439 nm in this case).
[0039] The following experiment was conducted to measure the reflectance spectrum. A multilayer mirror stacking process was performed to form 40 pairs of AlInN / GaN multilayer mirrors 11, and then a resonator stacking process was performed to stack GaN resonators 12 with a thickness of 1 / 4 wavelength on the surface of the AlInN / GaN multilayer mirrors 11. Therefore, at this point, the resonator length of the GaN resonator 12 is 1 / 4 wavelength. This wafer was placed in the reaction furnace of a MOVPE apparatus equipped with a measuring device that can measure the reflectance spectrum in parallel in situ, and the resonator stacking process was continued at a growth temperature of 1000°C under a hydrogen atmosphere to further crystallize the GaN resonators 12. Trimethylgallium was supplied as the Ga raw material and ammonia as the N raw material. If necessary, Si which provides n-type electrical conductivity or Mg which provides p-type electrical conductivity may be added. Furthermore, an emissive layer may be inserted. Then, while the resonator stacking process was being performed, a reflectance measurement process was carried out to measure the reflectance of the GaN resonator 12. Figure 7 shows the reflectance spectra measured in parallel in situ while crystal growth was performed on samples with five different resonator lengths, as well as the dependence of the reflectance intensity at the resonator thickness on the resonator thickness. The wavelength of light used in the reflectance measurement process (439 nm) is 22 nm longer than the resonant wavelength (417 nm) of the AlInN / GaN multilayer mirror 11 at room temperature.
[0040] Here, the spectrum of d in Figure 7(A) corresponds to the spectrum of α = 10000 cm in Figure 6. -1 This is roughly equivalent to the absorption coefficient α = 10000 cm in Figure 6. -1It can be seen that a spectrum almost equivalent to the above can be observed. As shown in Figure 7(A), at odd multiples of 1 / 4 wavelength (a, c, e), the conditions for a non-resonant state are met, and the reflectance at the resonant wavelength is high. On the other hand, at integer multiples of 1 / 2 wavelength (b, d), the conditions for a resonant state are met, and the reflectance at the resonant wavelength is significantly reduced. Figure 7(B) shows the change in the value of the reflectance at the resonant wavelength with respect to the growth time (i.e., the thickness of the resonator) of the GaN resonator 12. As shown in Figure 7(B), the reflectance changes so that it oscillates greatly every 1 / 2 wavelength (0.5λ) of the resonator length. In this way, by performing an absorption region stacking process during the execution of the resonator stacking process and introducing an absorption region Ab1 set to an appropriate absorption coefficient α to the entire GaN resonator 12, the change in reflectance at the growth temperature can be clearly observed, and the thickness of the GaN resonator 12 during crystal growth can be accurately grasped in parallel in place, and the thickness of the GaN resonator 12 can be controlled with high precision while forming the GaN resonator 12.
[0041] In reality, the inside of the GaN resonator 12 contains an n-GaN layer 12A, a GaInN light-emitting layer 12B, and a p-GaN layer 12C, as shown in Figure 8. Figure 8 shows the dependence of the reflectance at the resonant wavelength (431 nm in this case) on the thickness of the resonator at the growth temperature. Here, the thickness of the p-GaN layer 12C is made thicker than the actual design value in order to observe the change in reflectance. To obtain a nitride semiconductor light-emitting element that functions well at room temperature, we would like to set the thickness of the GaN resonator 12 to 3.7 wavelengths as an example of the design value. Therefore, if the growth of the GaN resonator 12 is terminated at 3.7 wavelengths (dotted line time) in Figure 8, a GaN resonator 12 that matches the design value can be formed. In this way, the epitaxial growth of the nitride semiconductor layer in the GaN surface-emitting laser is completed.
[0042] Next, a device capable of receiving current is formed using the completed wafer. First, a circular mesa structure with a diameter of 40 μm is formed. Specifically, the mesa structure is formed using photolithography and dry etching techniques. At this time, the mesa structure covers the surface of the p-GaN layer 12C with a circular photoresist, and etching is performed around the circular photoresist to a depth of approximately 100 nm so that the surface of the n-GaN layer 12A inside the GaN resonator 12 is exposed.
[0043] Next, an 8 μm diameter resist is formed in the center of the mesa structure's surface, followed by the deposition of a 10 nm thick SiO2 film and then lift-off. This creates an 8 μm diameter opening in the center of the mesa structure's surface and forms a 10 nm thick SiO2 film that covers the sides of the mesa structure.
[0044] Next, a 20 nm ITO layer and a 32 nm Nb2O5 layer, which will serve as p-contact electrodes, are deposited on the surface of the SiO2 film and the surface of the mesa structure where the SiO2 film openings are formed, by repeatedly patterning, depositing, and lift-off with photoresist. Through these processes, the ITO layer and Nb2O5 layer are deposited on the GaN surface-emitting laser having the aforementioned 3.7 wavelength resonator, completing a resonator with exactly 4 wavelengths. Since the GaN resonator 12 formed by epitaxial growth is formed with high precision control of its layer thickness by performing reflectance spectrum measurements in parallel in situ, the ITO layer and Nb2O5 layer only need to be deposited with the thickness specified in the design, and no adjustments considering the misalignment of the GaN resonator 12 are necessary.
[0045] Next, the p-pad electrode and n-contact electrode are formed. The p-pad electrode is formed on the outer edge of the surface of the ITO layer. The n-contact electrode is formed on the surface of the n-GaN layer 12A exposed by etching. The p-pad electrode and n-contact electrode are composed of Cr / Ni / Au (10 / 20 / 370 nm). Finally, by stacking eight pairs of Nb2O5 / SiO2 multilayer reflectors, which have a high reflectivity of 99.9% or more, on the surface of the Nb2O5 layer, the GaN surface-emitting laser as a nitride semiconductor light-emitting element is completed. Because the thickness of the resonator length is controlled with high precision, the gain wavelength of the light-emitting layer and the resonance wavelength of the resonator can be formed in an appropriate relationship as designed, making it possible to reproducibly extract the original performance of the element.
[0046] The method for manufacturing a nitride semiconductor light-emitting element comprises a multilayer mirror stacking step of stacking AlInN / GaN multilayer mirrors 11 to form a reflector; a resonator stacking step of stacking a GaN resonator 12 on the surface of the AlInN / GaN multilayer mirror 11 after the multilayer mirror stacking step; and a reflectance measurement step of measuring the reflectance of the GaN resonator 12 while performing the resonator stacking step. The wavelength of light used in the reflectance measurement step is longer than the reflection center wavelength of the multilayer mirror at room temperature, i.e., the resonance wavelength. With this configuration, at the growth temperature when stacking and growing the nitride semiconductor light-emitting element, the resonance wavelength of the nitride semiconductor light-emitting element becomes longer. Therefore, by making the wavelength of light used in the reflectance measurement step longer than the resonance wavelength (reflection center wavelength) of the AlInN / GaN multilayer mirror 11 at room temperature, the reflectance of the GaN resonator 12 during crystal growth can be accurately measured while growing the nitride semiconductor light-emitting element.
[0047] The system further includes an absorption region stacking step in which an absorption region Ab1 is stacked within the GaN resonator 12 during the resonator stacking process. The absorption region Ab1 stacked in the absorption region stacking step does not absorb light at the resonant wavelength of the AlInN / GaN multilayer mirror 11 at room temperature, but at the growth temperature, it absorbs light with a longer wavelength than the resonant wavelength of the AlInN / GaN multilayer mirror 11 at room temperature. With this configuration, the absorption region Ab1 can absorb light with a longer wavelength than the resonant wavelength of the AlInN / GaN multilayer mirror 11 at room temperature at the growth temperature. This allows for a large change in the reflectivity of the GaN resonator 12 during the resonator stacking process, i.e., at the growth temperature, and makes it easier to observe the change in reflectivity of the GaN resonator 12 at the resonant wavelength during the resonator stacking process.
[0048] By performing an absorption region stacking process, the entire GaN resonator 12 is set as an absorption region Ab1. With this configuration, light within the GaN resonator 12 can be reliably absorbed.
[0049] If the entire GaN resonator 12 is considered as the absorption region Ab1, the absorption coefficient α in the absorption region Ab1 is 100 cm². -1 Above and 30,000 cm -1 The following is the result. With this configuration, the change in reflectivity of the GaN resonator 12 can be made to be greater than 1%.
[0050] <Example 3> [When absorption occurs in a portion of the resonator] As an example of light absorption in a portion of the GaN resonator 112 (see Figure 5(B)), we consider the case where the resonant wavelength at room temperature is 520 nm. Specifically, as shown in Figure 9, a GaN resonator 212 is laminated on the surface of 46 pairs of AlInN / GaN multilayer mirrors 111 that have been laminated by a multilayer mirror lamination process. The GaN resonator 212 has a thickness of 4 wavelengths (corresponding to the resonant state) that does not absorb light of the resonant wavelength. Then, during the execution of the resonator lamination process, an absorption region lamination process is performed to provide an n-GaInN absorption layer 212A as an absorption region Ab2 that absorbs light of the resonant wavelength in a portion of the GaN resonator 212 in the lamination direction.
[0051] As shown in Figure 9, the GaN resonator 212 has a thickness of 4 wavelengths, and the n-GaInN absorption layer 212A is positioned such that the center in the thickness direction coincides with a position 1 wavelength upward from the bottom end. The n-GaInN absorption layer 212A is sandwiched vertically by two n-GaN layers 212B. A GaInN light-emitting layer 212C, formed by a GaInN quantum well or the like, is laminated on the surface of the upper n-GaN layer 212B. Furthermore, a p-GaN layer 212D is laminated on the surface of the GaInN light-emitting layer 212C. Because the GaInN light-emitting layer 212C is sandwiched vertically by the n-GaN layer 212B and the p-GaN layer 212D, current injection is possible. In addition, the n-GaInN absorption layer 212A is n-type due to the addition of Si, etc., and does not hinder the above current injection.
[0052] The n-GaInN absorption layer 212A is required to absorb light at the resonant wavelength at the growth temperature but not at the resonant wavelength at room temperature. Assuming the resonant wavelength at room temperature is 520 nm and the growth temperature of the GaN resonator 212 is 1000°C, Equation 1 shows that the resonant wavelength at the growth temperature is 538 nm, which is 18 nm longer than 520 nm. Therefore, the n-GaInN absorption layer 212A needs to sufficiently absorb light at least at a resonant wavelength of approximately 540 nm at the growth temperature. Here, the absorption edge of GaN at room temperature is 370 nm, and as described in Example 2, the wavelength is lengthened to nearly 430 nm (about 60 nm) at the growth temperature.
[0053] Here, if we set the n-GaInN absorption layer 212A to absorb light with a resonant wavelength of 550 nm at the growth temperature, the absorption edge at room temperature will be 490 nm, which is 30 nm shorter than the device operating wavelength of 520 nm at room temperature, and therefore does not significantly affect the device characteristics. Thus, it is sufficient to select an appropriate absorption wavelength in the n-GaInN absorption layer 212A. Here, the n-GaInN absorption layer 212A was used as the absorption region Ab2. Because interband absorption occurs, the absorption coefficient α in the n-GaInN absorption layer 212A is 100,000 cm². -1 That is the case.
[0054] Here, it is important to determine the appropriate thickness of the n-GaInN absorption layer 212A in order to absorb a sufficient amount of light. Figure 10 shows the theoretical calculation results of the reflectance spectra when the thickness of the n-GaInN absorption layer 212A is varied in four steps: 0.01 nm, 0.1 nm, 1 nm, and 10 nm. As shown in Figure 10, it was found that the reflectance at the resonance wavelength decreases by more than 1% when the thickness is 0.1 nm or more. At 10 nm, the reflectance becomes 35%, which is less than 50%, and it can be seen that the change in reflectance can be clearly observed. However, the thickness of one atomic layer of nitride semiconductors such as GaInN, i.e., the thinnest layer, is 0.3 nm. Layers thinner than this will not be uniform films. Also, if the thickness of the n-GaInN absorption layer 212A is made thicker than 10 nm, there is a concern that the crystallinity will deteriorate due to lattice mismatch of the n-GaInN absorption layer 212A. Therefore, when introducing an n-GaInN absorption layer 212A, which is an absorption region Ab2, to only a portion of the thickness direction of the GaN resonator 212 in order to control the thickness of the GaN resonator 212, the thickness in the stacking direction should be 0.3 nm or more and 10 nm or less. A more preferable thickness for the n-GaInN absorption layer 212A is 1 nm or more and 10 nm or less, which results in a change in reflectivity of 10% or more.
[0055] The reflectance measurement process is performed to measure the change in reflectance, while the resonator stacking process is performed to grow the GaN resonator 212 crystal. When the GaN resonator 212 reaches the desired thickness, the resonator stacking process is terminated and the crystal growth is stopped. In this way, the epitaxial growth for the nitride semiconductor layer formation of the GaN surface-emitting laser is completed. Then, the GaN surface-emitting laser element as a nitride semiconductor light-emitting element is formed in the same manner as in Example 2. The method of element formation is the same as in Example 2, so the explanation is omitted.
[0056] By performing an absorption region stacking process, a portion of the stacking direction of the GaN resonator 212 is set as the absorption region Ab2. With this configuration, since only a portion of the GaN resonator 212 needs to be set as the absorption region Ab2, the configuration of the GaN resonator 212 excluding the absorption region Ab2 does not need to be modified.
[0057] The thickness in the stacking direction in the absorption region Ab2 is 0.3 nm or more and 10 nm or less. With this configuration, the change in reflectivity of the GaN resonator 212 can be made to be 1% or more while maintaining the quality of the GaN resonator 212.
[0058] <Example 4> [When the absorption region also functions as the light-emitting layer] As another example of light absorption in a portion of the GaN resonator 112 (see Figure 5(B)), we present an example using quantum wells in the light-emitting layer of a nitride semiconductor light-emitting device. Specifically, as shown in Figure 11, we considered the case where the GaInN absorption layer 312A, which is the absorption region Ab2, also serves as the light-emitting layer. When the resonance wavelength at room temperature is set to 520 nm, the emission wavelength of the GaInN absorption layer 312A (light-emitting layer) is set to approximately 510 nm. In this case, at the growth temperature of the GaN resonator 312, 1000°C, the absorption edge is extended to a longer wavelength of 570 nm. Therefore, even when the resonance wavelength at the growth temperature estimated from Equation 1 is 538 nm, the absorption region Ab2 absorbs light. Generally, the active layer (light-emitting layer) has 1 to 5 quantum wells with a thickness of 2 nm to 3 nm. Thus, the active layer has quantum wells with a thickness of approximately 2 nm to 15 nm. The thickness of this quantum well is sufficient for absorbing light, as can be seen from the results of the analysis of the absorption layer thickness in Example 3. Therefore, if the resonance wavelength at the growth temperature can be estimated using Equation 1, it is possible to use the GaInN absorption layer 312A as an emissive layer as well.
[0059] However, the optimal structure for the light-emitting layer and the optimal structure for achieving the objectives of the present invention do not always coincide, so a configuration in which the absorption region Ab2 also serves as the light-emitting layer is not always possible. In this case, as shown in Figure 11, the light-emitting layer (GaInN absorption layer 312A) is often positioned towards the top of the GaN resonator 312 in the stacking direction (here, the center in the thickness direction is located at a position 3 wavelengths from the bottom and 1 wavelength from the top). More specifically, the light-emitting layer (GaInN absorption layer 312A) is stacked on the surface of the n-GaN layer 312B, and the p-GaN layer 312C is stacked on the surface of the light-emitting layer. This is because the resistance of the p-layer is about 100 times higher than that of the n-layer, so the thickness of the p-layer is kept to a minimum to reduce its resistance.
[0060] Figure 12 shows the change in reflectance at the resonant wavelength at the growth temperature in this case. As shown in Figure 12, no clear vertical oscillation of the reflectance is observed until the light-emitting layer (GaInN absorption layer 312A) is formed (i.e., until the thickness of the n-GaN layer 312B is stacked by three wavelengths), and the intermediate progress of the resonator length of the GaN resonator 312 cannot be determined. Once the formation of the GaInN absorption layer 312A (light-emitting layer) is complete, the reflectance begins to oscillate significantly, and this oscillation becomes detectable by the measuring device, making it possible to complete the formation of the GaN resonator 312 at the desired thickness.
[0061] Then, when the resonator length of the GaN resonator 312 reaches the desired thickness, crystal growth is terminated. Then, a GaN surface-emitting laser element is formed as a nitride semiconductor light-emitting element, similar to Example 2. The method of element formation is the same as in Example 2, so the explanation is omitted.
[0062] The absorption region Ab2 is the active layer. With this configuration, the active layer can be used as the absorption region Ab2, eliminating the need to separately create the absorption region Ab2.
[0063] The embodiments disclosed herein should be considered in all respects to be illustrative and not restrictive. The scope of the present invention is not limited to the embodiments disclosed herein, but is indicated by the claims, and all modifications within the meaning and scope equivalent to the claims are intended to be included. (1) Unlike in Example 3 above, the number of absorption regions within the resonator may be two or more. (2) Unlike Example 2 above, the wavelength of light used in the reflectance measurement process can be set to a value corresponding to the desired resonance wavelength and its growth temperature, depending on the reflection center wavelength of the multilayer mirror, and should be set within the range of the shift amount shown in Table 1. Specifically, it should be 11.8 nm to 26.7 nm greater than the reflection center wavelength of the multilayer mirror at room temperature, i.e., the resonance wavelength. In other words, the wavelength of light used in the reflectance measurement process is 11.8 nm to 26.7 nm greater than the resonance wavelength (reflection center wavelength) of the AlInN / GaN multilayer mirror at room temperature. With this configuration, the range of wavelength shift from room temperature to the growth temperature in the wavelengths that can be emitted from nitride semiconductor light-emitting elements (375 nm to 675 nm) can be almost covered. Therefore, it is possible to measure the reflectance at the resonance wavelength that takes into account the wavelength shift when growing a resonator with a desired resonance wavelength at room temperature. (3) The material of the multilayer mirror may be different from the material disclosed in Examples 1 to 4. (4) In the embodiment, the resonator length was one wavelength or four wavelengths, but the amount of absorption in the absorption region can be adjusted while selecting the optimal resonator length according to the required element characteristics. [Explanation of Symbols]
[0064] 11,111...AlInN / GaN multilayer reflector (multilayer reflector) 12,112,212,312…GaN resonator (resonator) 312A…GaInN absorption layer (active layer) Ab1, Ab2… Absorption regions α... Absorption coefficient
Claims
1. A multilayer mirror lamination process for forming a multilayer mirror by stacking multiple layers, After the multilayer mirror lamination process is performed, a resonator lamination process is performed in which a resonator is laminated on the surface of the multilayer mirror, A reflectance measurement step is performed while carrying out the resonator stacking step, and the reflectance of the resonator is measured. Equipped with, The wavelength of light used in the reflectance measurement process is longer than the resonant wavelength of the multilayer mirror at room temperature. The process further comprises an absorption region stacking step in which an absorption region is stacked within the resonator during the execution of the resonator stacking step, The absorption regions to be stacked in the aforementioned absorption region stacking process are At room temperature, the multilayer mirror does not absorb light of the resonant wavelength at room temperature. A method for manufacturing a nitride semiconductor light-emitting element, wherein, at the growth temperature, it absorbs light with a wavelength longer than the resonant wavelength of the multilayer reflecting mirror at room temperature.
2. The method for manufacturing a nitride semiconductor light-emitting element according to claim 1, wherein the wavelength of light used in the reflectance measurement step is 11.8 nm to 26.7 nm greater than the resonance wavelength of the multilayer mirror at room temperature.
3. The method for manufacturing a nitride semiconductor light-emitting element according to claim 1 or 2, wherein the reflection center wavelength of the multilayer reflector at the growth temperature is determined by calculating a shift amount based on two parameters: the resonance wavelength of the multilayer reflector at room temperature and the growth temperature, and adding the shift amount to the resonance wavelength.
4. A method for manufacturing a nitride semiconductor light-emitting element according to claim 1 or claim 2, wherein the entire resonator is set as the absorption region by performing the absorption region stacking step.
5. The method for manufacturing a nitride semiconductor light-emitting element according to claim 4, wherein, when the entire resonator is defined as the absorption region, the absorption coefficient in the absorption region is 100 cm⁻¹ or more and 30,000 cm⁻¹ or less.
6. A method for manufacturing a nitride semiconductor light-emitting element according to claim 1 or 2, wherein by performing the absorption region stacking step, a portion of the stacking direction of the resonator is set as the absorption region.
7. The method for manufacturing a nitride semiconductor light-emitting element according to claim 6, wherein the thickness in the stacking direction in the absorption region is 0.3 nm or more and 10 nm or less.
8. The method for manufacturing a nitride semiconductor light-emitting element according to claim 6, wherein the absorption region is an active layer.