A GaN-based micro-LED epitaxial structure, device and preparation method

Abstract

This application discloses a GaN-based micro-LED epitaxial structure, device, and fabrication method. The GaN-based micro-LED epitaxial structure includes, along the growth direction, a substrate, a buffer layer, an N-type doped GaN electron injection layer, an active region, and an Al2O3 layer. y Ga 1‑y The substrate consists of an N-type electron blocking layer and a P-type doped GaN hole injection layer; the substrate is selected from a (111) crystal plane single-crystal silicon substrate, a c-axis single-crystal sapphire substrate, a c-axis single-crystal silicon carbide substrate, or a c-axis self-supporting single-crystal gallium nitride substrate; the active region includes an InGaN-based light-emitting structure, which is at least one of an InGaN / GaN alternating multi-quantum-well structure, a GaN / InGaN / GaN single-quantum-well structure, and an InGaN quantum dot structure. This application shortens the carrier radiative recombination lifetime, weakens the quantum-confined Stark effect, and improves the radiative recombination rate by employing a thin quantum well or quantum dot active region structure, thereby enhancing the device's modulation bandwidth and photoelectric efficiency, meeting the performance requirements of light sources for applications such as data center optical interconnects, co-packaged optics, and high-speed visible light communication.

Description

Technical Field

[0001] This application relates to the field of semiconductor light-emitting technology, specifically to a GaN-based micro-LED epitaxial structure, device, and fabrication method. Background Technology

[0002] With the rapid development of artificial intelligence, cloud computing, and big data technologies, the data transmission demand of data centers is experiencing explosive growth. It is predicted that by 2025, global data center IP traffic will exceed 20 ZB per year, posing unprecedented challenges to optical interconnect technologies within data centers. Traditional data center optical interconnect solutions mainly rely on optical modules based on III-V semiconductor lasers (such as VCSELs and DFB lasers). However, as data rates evolve towards 400G, 800G, and even 1.6T, traditional laser solutions face an increasingly severe "power wall" problem—the power consumption and thermal management costs of the laser itself account for a significant proportion of the total power consumption of the optical module. Furthermore, lasers are extremely sensitive to temperature, typically requiring thermoelectric coolers (TECs) for precise temperature control, which further increases system power consumption and cost. Co-Packaged Optics (CPO) technology is considered a key path to solving the power consumption bottleneck of data center optical interconnects. Its core idea is to tightly integrate the optical engine with the switching chip (ASIC), shortening the electrical interconnect distance and reducing system power consumption. However, the CPO architecture places more stringent requirements on the light source: it needs to operate stably in high-temperature environments (temperatures near the switching chip can reach 85°C~105°C), possess high modulation bandwidth to support high-speed direct modulation, and simultaneously reduce power consumption and cost significantly. While existing InP-based lasers perform excellently in long-distance communication, their high manufacturing cost, strong temperature sensitivity, and poor compatibility with silicon-based CMOS processes limit their large-scale application in CPO scenarios.

[0003] Against this backdrop, GaN-based micro-LEDs, as a novel semiconductor light source, exhibit unique advantages. The GaN material system possesses a wide bandgap (3.4 eV), high thermal conductivity, excellent chemical stability, and high-temperature resistance. GaN-based devices can operate stably in high-temperature environments without thermoelectric cooling, which perfectly aligns with the high-temperature operating requirements of CPO architectures. Furthermore, GaN-based micro-LEDs employ direct modulation, eliminating the need for external modulators, resulting in a simple system structure, low power consumption, and controllable cost.

[0004] Visible light communication (VLC) is another important application area for GaN-based micro-LEDs. It utilizes visible light signals for data transmission and, with its abundant spectrum resources, lack of electromagnetic interference, and high security, is considered an important supplement to next-generation wireless communication technologies. Whether in data center optical interconnects or visible light communication, the modulation bandwidth of the light source is one of the key factors determining the communication rate.

[0005] In the field of GaN-based light-emitting devices, traditional large-area GaN-based LEDs, due to their structural characteristics, typically exhibit large junction capacitance and long carrier lifetimes, resulting in modulation bandwidths usually only on the order of tens of MHz, far from meeting the demands of high-speed communication. To address the high-speed modulation problem, researchers have introduced Micro-LED technology. Micro-LEDs have a relatively small equivalent circle diameter in their emitting region, leading to smaller junction capacitance, higher current density, and shorter carrier radiative recombination lifetimes, demonstrating greater modulation bandwidth potential. Currently, researchers are primarily exploring the modulation bandwidth potential of Micro-LEDs by minimizing device size and improving injection efficiency.

[0006] However, existing GaN-based micro-LEDs face numerous challenges in achieving high modulation bandwidth. In traditional InGaN / GaN multi-quantum-well structures, the thickness of the InGaN quantum well layer is typically 2.5 nm to 3 nm. A thicker quantum well layer leads to a longer carrier radiative recombination lifetime, while the quantum-confined Stark effect further reduces radiative recombination efficiency and increases carrier lifetime. When the size of micro-LEDs shrinks to a certain extent, the mesa sidewalls formed by dry etching introduce a large number of surface defects and dangling bonds, resulting in severe non-radiative surface recombination and reducing both the internal and external quantum efficiencies of the device. These problems make it difficult for existing GaN-based micro-LEDs to simultaneously achieve high modulation bandwidth and high photoelectric efficiency at low drive current densities, limiting their practical deployment in high-end applications such as data center optical interconnects, CPO, and high-speed visible light communication. Summary of the Invention

[0007] To address the technical problems in the prior art, this application provides a GaN-based micro-LED epitaxial structure, device, and fabrication method.

[0008] This application provides a GaN-based micro-LED epitaxial structure, device, and fabrication method using the following technical solution:

[0009] A GaN-based micro-LED epitaxial structure, comprising, along the growth direction:

[0010] The substrate is selected from a (111) crystal plane single crystal silicon substrate, a c-axis single crystal sapphire substrate, a c-axis single crystal silicon carbide substrate or a c-axis self-supporting single crystal gallium nitride substrate.

[0011] A buffer layer is formed on the substrate, the buffer layer comprising an unintentionally doped GaN layer and an AlN / Al layer. x Ga 1- x At least one of N / GaN stacks, wherein x is 0.1 to 1.0;

[0012] An N-type doped GaN electron injection layer is formed on the buffer layer, with silicon as the dopant and a doping concentration of 1×10⁻⁶. 18 cm -3 ~5×10 20 cm -3 The thickness is 2μm~8μm;

[0013] An active region is formed above the N-type doped GaN electron injection layer. The active region includes an InGaN-based light-emitting structure, which is at least one of an InGaN / GaN alternating multi-quantum-well structure, a GaN / InGaN / GaN single-quantum-well structure, and an InGaN quantum dot structure. When the InGaN-based light-emitting structure is an InGaN / GaN alternating multi-quantum-well structure or a GaN / InGaN / GaN single-quantum-well structure, the thickness of the InGaN quantum well layer is 0.5 nm to 3 nm. When the InGaN-based light-emitting structure is an InGaN quantum dot structure, the height of the InGaN quantum dot is 0.5 nm to 5 nm.

[0014] Al y Ga 1-y An N electron blocking layer is formed on the active region, with an Al composition y of 0.05~0.20 and a thickness of 10nm~100nm;

[0015] P-type doped GaN hole injection layer, formed in the Al y Ga 1-y Above the N-electron blocking layer, the doping element is magnesium, and the doping concentration is 5 × 10⁻⁶. 18 cm -3 ~5×10 20 cm -3 The thickness is 10nm~100nm.

[0016] In some embodiments, when the active region is an InGaN / GaN alternating multiple quantum well structure, the thickness of the InGaN quantum well layer is 0.5nm~1.8nm, and the number of cycles of the multiple quantum wells is 2~5 pairs;

[0017] When the active region is a GaN / InGaN / GaN single quantum well structure, the thickness of the InGaN quantum well layer is 0.5nm~1.8nm;

[0018] When the active region is an InGaN quantum dot structure, the InGaN quantum dots are nanostructures distributed in the GaN matrix layer, with 1 to 5 layers, a diameter of 2 nm to 10 nm, a height of 1 nm to 5 nm, and an areal density of 1 × 10⁻⁶. 10 / cm 2 ~1×10 11 / cm 2 ;

[0019] The doping concentration of the N-type doped GaN electron-injected layer is 2 × 10⁻⁶. 19 cm -3 ~2×10 20 cm -3 The thickness is 4.5μm~6μm;

[0020] The Al y Ga 1-y The Al composition of the N electron blocking layer is 0.10~0.15, and the thickness is 30nm~60nm;

[0021] The doping concentration of the P-type doped GaN hole injection layer is 1×10⁻⁶. 19 cm -3 ~5×10 19 cm -3 The thickness is 20nm~60nm.

[0022] In some embodiments, a stress relief layer is further provided between the N-type doped GaN electron injection layer and the active region. The stress relief layer is an InGaN / GaN multi-period structure with a period number of 5 to 20.

[0023] In the InGaN / GaN multi-period structure, the InGaN layer and GaN layer of each period are set to the same thickness; or, along the direction close to the active region, the thickness of the InGaN layer in each period of the InGaN / GaN multi-period structure gradually increases, and / or the thickness of the GaN layer gradually decreases.

[0024] And / or, a highly p-type doped contact layer is further provided above the p-type doped GaN hole injection layer, wherein the doping element of the highly p-type doped contact layer is magnesium, and the doping concentration is 1×10⁻⁶. 19 cm -3 ~5×10 20 cm -3 The thickness is 10nm~100nm.

[0025] This application also provides a GaN-based micro-LED device, including the aforementioned epitaxial structure and an electrode structure formed on the epitaxial structure; the light-emitting region of the GaN-based micro-LED device is any one of rectangular, circular, or hexagonal shapes, and the equivalent circle diameter corresponding to the area of ​​the light-emitting region is 1μm~100μm; the emission wavelength of the GaN-based micro-LED device is 365nm~700nm; the operating driving current density of the GaN-based micro-LED device is less than 50A / cm², wherein the driving current density is the operating current divided by the light-emitting area; at the operating driving current density, the -3dB electro-optic modulation bandwidth of the GaN-based micro-LED device is greater than 1GHz.

[0026] This application also provides a GaN-based micro-LED device array, comprising a plurality of the aforementioned micro-LED devices, wherein the plurality of micro-LED devices are arranged in any one of a rectangular array, a concentric ring array, or a hexagonal close-packed array, and the array contains a number of devices greater than or equal to 2.

[0027] This application also provides a method for fabricating a GaN-based micro-LED epitaxial structure, comprising the following steps:

[0028] S1. Provide a substrate and perform pretreatment. The substrate is selected from (111) crystal plane single crystal silicon substrate, c-axis single crystal sapphire substrate, c-axis single crystal silicon carbide substrate or c-axis self-supporting single crystal gallium nitride substrate.

[0029] S2. A buffer layer is grown on the substrate using an epitaxial growth method. The buffer layer comprises an unintentionally doped GaN layer and an AlN / Al layer. x Ga 1-x At least one of N / GaN stacks, wherein x is 0.1 to 1.0;

[0030] S3. An N-type doped GaN electron injection layer is grown on the buffer layer using an epitaxial growth method. The doping element is silicon, and the doping concentration is 1×10⁻⁶. 18 cm -3 ~5×10 20 cm -3 The thickness is 2μm~8μm;

[0031] S4. An active region is grown on the N-type doped GaN electron injection layer by an epitaxial growth method. The active region includes an InGaN-based light-emitting structure. The InGaN-based light-emitting structure is at least one of an InGaN / GaN alternating multi-quantum-well structure, a GaN / InGaN / GaN single-quantum-well structure, and an InGaN quantum dot structure. When the InGaN-based light-emitting structure is an InGaN / GaN alternating multi-quantum-well structure or a GaN / InGaN / GaN single-quantum-well structure, the thickness of the InGaN quantum well layer is controlled to be 0.5 nm to 3 nm. When the InGaN-based light-emitting structure is an InGaN quantum dot structure, the height of the InGaN quantum dot is controlled to be 0.5 nm to 3 nm.

[0032] S5. Al is grown on the active region by an epitaxial growth method. y Ga 1-y The N electron blocking layer has an Al composition of 0.05~0.20 and a thickness of 10nm~100nm.

[0033] S6, in the Al y Ga 1-y A p-type doped GaN hole injection layer is grown on the N-electron blocking layer using epitaxial growth. The doping element is magnesium, and the doping concentration is 5 × 10⁻⁶. 18 cm -3 ~5×10 20 cm -3 The thickness is 10nm~100nm.

[0034] In some embodiments, step S3a is further included between step S3 and step S4: a stress relief layer is grown on the N-type doped GaN electron injection layer by an epitaxial growth method. The stress relief layer is an InGaN / GaN multi-period structure with 5 to 20 periods. The InGaN and GaN layers in each period of the InGaN / GaN multi-period structure are grown with equal thickness. Alternatively, the thickness of the InGaN layer in each period gradually increases and / or the thickness of the GaN layer gradually decreases along the direction close to the active region.

[0035] And / or, following step S6, step S7 is further included: growing a highly p-type doped contact layer on the p-type doped GaN hole injection layer by an epitaxial growth method, wherein the doping element is magnesium and the doping concentration is 1×10⁻⁶. 19 cm -3 ~5×10 20 cm -3 The thickness is 10nm~100nm.

[0036] This application also provides a method for fabricating a GaN-based micro-LED device, comprising the following steps:

[0037] A1. An epitaxial structure is prepared using the preparation method described above;

[0038] A2. Deposit a transparent conductive film on the surface of the P-type layer of the epitaxial structure. The transparent conductive film is an indium tin oxide thin film with a thickness of 30nm~300nm, and is deposited by any one of magnetron sputtering, electron beam evaporation or chemical vapor deposition.

[0039] A3. P-meta-type structures are formed through photolithography and etching processes, with the etching depth penetrating the transparent conductive film, the P-type doped GaN hole injection layer, and Al. y Ga 1-y The N-type electron blocking layer and active region expose the N-type doped GaN electron injection layer;

[0040] A4. Perform rapid thermal annealing under an inert atmosphere, with an annealing temperature of 400℃~800℃ and an annealing time of 30s~600s.

[0041] A5. Deposit a passivation layer and form a metal contact window through photolithography and etching;

[0042] A6. Positive and negative electrodes are formed by depositing metal electrodes through photolithography, metal evaporation and lift-off processes.

[0043] In some embodiments, when the substrate is a silicon substrate, step A4a is further included between step A4 and step A5: removing the silicon substrate under the micro-LED device using an alkaline etching solution to form a suspended pixel structure;

[0044] When the substrate is a sapphire substrate, after step A6, a step of separating the micro-LED device from the original substrate by using a laser lift-off process is also included, wherein the photon energy of the laser is greater than the GaN bandgap and less than the sapphire material bandgap.

[0045] When the substrate is a silicon carbide substrate, after step A6, the step of removing the silicon carbide substrate by grinding or chemical etching is also included.

[0046] In some embodiments, at least one of the following steps is included after step A6:

[0047] A7. A transfer printing process is used to transfer micro-LED devices onto a new substrate, wherein the new substrate includes any one of a CMOS driving array substrate, a diamond substrate, or a glass substrate;

[0048] A8. A microlens is fabricated in situ on the light-emitting surface of the micro-LED device using a two-photon lithography process. The material of the microlens is photosensitive resin. When steps A7 and A8 are performed simultaneously, step A7 is performed first, followed by step A8.

[0049] In summary, this application includes at least one of the following beneficial technical effects:

[0050] 1. By employing thin quantum wells (0.5nm~3nm) or quantum dot active region structures, the carrier radiative recombination lifetime is effectively shortened, the quantum confinement Stark effect is weakened, the spatial overlap integral of the electron-hole wave function is increased, and the radiative recombination rate is improved, thereby significantly increasing the modulation bandwidth of the device. The three-dimensional quantum confinement effect in the quantum dot structure further enhances carrier localization and radiative recombination efficiency. Due to the three-dimensional size confinement, the piezoelectric polarization field in the quantum dot is weakened compared to the quantum well structure of the same composition, which is conducive to achieving higher modulation bandwidth.

[0051] 2. By controlling the equivalent circle diameter of the light-emitting area of ​​the micro-LED device within the range of 1μm to 100μm, the junction capacitance of the device is effectively reduced, which weakens the limitation of the RC time constant on the modulation bandwidth. Combined with the optimized active area design, a high-speed light source with a -3dB electro-optic modulation bandwidth of more than 1GHz is achieved under the condition that the operating drive current density is less than 50A / cm², which meets the requirements of high-speed optical communication for the modulation performance of the light source.

[0052] 3. By introducing an InGaN / GaN multi-period stress relief layer between the active region and the N-type doped GaN electron injection layer, the lattice mismatch stress is effectively alleviated, the dislocation density, V-pit density and piezoelectric field in the active region are reduced, and the crystal quality and radiative recombination efficiency of the active region are improved. The design of the stress relief layer with gradually varying thickness can achieve a smoother composition and stress transition, further reducing defects in the active region.

[0053] 4. By setting a highly P-type doped contact layer on top of the P-type doped GaN hole injection layer, the contact resistance between the P-type GaN and the ITO transparent conductive film is reduced, the ohmic contact characteristics are improved, the operating voltage and power consumption of the device are reduced, and the device can operate efficiently at low drive current densities.

[0054] 5. By providing a variety of substrate options and corresponding substrate removal processes, including selective removal of silicon substrates by alkaline etching solution to form suspended pixel structures, laser lift-off of sapphire substrates, and grinding or chemical etching of silicon carbide substrates, micro-LED devices can be flexibly adapted to different application scenarios; among them, the suspended pixel structure of silicon substrates eliminates the absorption of visible light by silicon, significantly improves light extraction efficiency, and reduces the parasitic capacitance of the device.

[0055] 6. By transferring micro-LED devices onto new substrates such as CMOS driver array substrates, diamond substrates, or glass substrates using transfer printing technology, functional integration such as monolithic integration of micro-LEDs and driver circuits, efficient heat dissipation, or transparent display is achieved, expanding the application range of the devices. Specifically, transferring to CMOS driver array substrates enables co-packaging integration with switching chips, meeting the high-bandwidth, low-power light source requirements of data center CPO architectures; transferring to diamond substrates utilizes diamond's ultra-high thermal conductivity to achieve efficient heat dissipation, allowing the devices to operate stably in high-temperature environments without thermoelectric cooling.

[0056] 7. Through the design of various array arrangement methods (rectangular array, concentric ring array or hexagonal close-packed array), the micro-LED device array can be flexibly configured according to different application requirements. When driven independently, it can realize multi-channel parallel optical communication with spatial multiplexing. When driven in parallel, it can improve the total optical power output and increase the communication distance.

[0057] 8. The GaN-based micro-LED device of this application is based on a wide bandgap GaN material system, which has excellent high temperature resistance and chemical stability. It can operate stably in high temperature environments without thermoelectric coolers. Moreover, it adopts a direct modulation method, which has a simple system structure, low power consumption, and controllable cost. It is particularly suitable for application scenarios such as data center optical interconnects and co-packaged optics (CPO) that have strict requirements for high temperature stability, low power consumption and high modulation bandwidth of light sources. Attached Figure Description

[0058] Figure 1 This is a cross-sectional schematic diagram of a GaN-based micro-LED epitaxial structure provided in one embodiment of this application;

[0059] Figure 2 This is a current-voltage (IV) characteristic curve of InGaN-based red micro-LED devices of different sizes in one embodiment of this application;

[0060] Figure 3 This is a graph showing the electro-optic (EO) modulation bandwidth of InGaN-based red micro-LED devices of different sizes as a function of current density in one embodiment of this application.

[0061] Figure 4 (a) is a graph showing the device voltage and data rate changes of 20μm and 30μm devices under different currents in one embodiment of this application;

[0062] Figure 4 (b) is a DC power consumption curve of 20 μm and 30 μm devices under different currents in one embodiment of this application;

[0063] Figure reference numerals: 110—substrate; 120—buffer layer; 130—N-type doped GaN electron injection layer; 140—stress relief layer; 150—active region; 160—Al y Ga 1-y N-electron blocking layer; 170-P-type doped GaN hole injection layer. Detailed Implementation

[0064] The technical solutions in the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. The described embodiments are only possible technical implementations of the present invention, but are not limited thereto. Other embodiments obtained by those skilled in the art in conjunction with the embodiments of the present invention without creative effort are also within the protection scope of the present invention.

[0065] This application mainly employs optimized epitaxial structure and process to fabricate high-bandwidth GaN-based micro-LEDs, achieving the effect of improving the modulation bandwidth and photoelectric efficiency of GaN-based micro-LEDs. The following is a further detailed description of this application.

[0066] Example 1

[0067] Please refer to Figure 1 The GaN-based micro-LED epitaxial structure provided in this application includes a substrate 110, a buffer layer 120, an N-type doped GaN electron injection layer 130, an active region 150, and an Al... y Ga 1-y An N-type electron blocking layer 160 and a P-type doped GaN hole injection layer 170 are included. A buffer layer 120 is grown on the substrate 110. An N-type doped GaN electron injection layer 130 is grown above the buffer layer 120. An active region 150 is grown on the N-type doped GaN electron injection layer 130. Al is grown above the active region 150. y Ga 1-y N electron blocking layer 160, Al y Ga 1-y A P-type doped GaN hole injection layer 170 is grown on the N electron blocking layer 160. This stacked structure provides good electron and hole injection conditions for the active region 150, ensuring normal light emission of the device. Furthermore, through the optimized design of each layer, the modulation bandwidth and photoelectric efficiency of the device can be effectively improved.

[0068] Specifically, the substrate 110 can be selected from various options, such as a (111) crystal plane single-crystal silicon substrate, a c-axis single-crystal sapphire substrate, a c-axis single-crystal silicon carbide substrate, or a c-axis self-supporting single-crystal gallium nitride substrate. Single-crystal silicon substrates have the advantages of low cost and large-scale production, and are highly compatible with CMOS processes, which is conducive to realizing the monolithic integration of micro-LEDs and driving circuits; c-axis single-crystal sapphire substrates have better insulation and thermal stability; c-axis single-crystal silicon carbide substrates have relatively small lattice mismatch with GaN (about 3.5%), which helps to grow high-quality epitaxial layers; c-axis self-supporting single-crystal gallium nitride substrates have the best lattice matching with epitaxial layers, and can obtain the lowest dislocation density, but the cost is relatively high. The selection of substrate 110 should be determined according to the specific application requirements and fabrication process.

[0069] Buffer layer 120 includes an unintentionally doped GaN layer and an AlN / Al layer. x Ga 1-x At least one of the N / GaN stack, where x is 0.1-1.0. The unintentionally doped GaN layer can act as a transition and buffer, reducing lattice and thermal mismatches between the substrate 110 and subsequent epitaxial layers. AlN / Al x Ga 1-x The N / GaN stack can further adjust the lattice constant and stress by combining different layers, thus alleviating the mismatch problem between the substrate 110 and GaN. For example, the AlN layer can serve as a nucleation layer, providing a good crystallization template for subsequent GaN growth; Al x Ga 1-x The N transition layer can gradually change its composition to achieve a gradual transition in lattice constant. When using a single-crystal silicon substrate with a (111) crystal plane, due to the large lattice mismatch (approximately 17%) between silicon and GaN and the significant difference in their thermal expansion coefficients, the buffer layer 120 is preferably made of AlN / Al. x Ga 1-x The AlN / GaN stacked structure can have an AlN layer thickness of 100 nm to 300 nm. x Ga 1-x The thickness of the N-transition layer can be 100 nm to 500 nm, and the thickness of the unintentionally doped GaN layer can be 1 μm to 3 μm. When using a c-axis single-crystal sapphire substrate, the buffer layer 120 can be a combination of a low-temperature AlN nucleation layer (thickness approximately 15 nm to 30 nm) and an unintentionally doped GaN layer (thickness approximately 1 μm to 3 μm). When using a c-axis self-supporting single-crystal gallium nitride substrate, since the substrate 110 is homogeneous with the epitaxial layer, the buffer layer 120 can consist only of a thinner unintentionally doped GaN layer (thickness approximately 200 nm to 1 μm).

[0070] The N-type doped GaN electron-injected layer 130 has silicon as the dopant element and a doping concentration of 1×10⁻⁶. 18 cm -3 ~5×10 20 cm -3 The thickness is 2μm-8μm. Silicon, as an N-type dopant, can provide a large number of electrons, improving electron injection efficiency. Appropriate doping concentration and thickness selection are crucial for ensuring effective electron injection and reducing series resistance. If the doping concentration is too low, insufficient electron injection will affect the device's luminous efficiency; if the doping concentration is too high, excessive impurity scattering may be introduced, increasing resistance. A suitable thickness ensures the uniformity and stability of electrons during transport. In a preferred embodiment, the doping concentration of the N-type doped GaN electron injection layer 130 is 2×10⁻⁶. 19 cm -3 ~2×10 20 cm -3 The thickness ranges from 4.5 μm to 6 μm. Higher doping concentrations help reduce the sheet resistance of the N-type layer, decrease the RC time constant, and thus improve the modulation bandwidth of the device. For example, when the doping concentration is 5 × 10⁻⁶... 19 When the thickness is 5 μm and the sheet resistance of the N-type doped GaN electron injection layer 130 is reduced to below about 10 Ω / □, effectively reducing the limitation of the modulation bandwidth by the series resistance of the device.

[0071] The active region 150 includes an InGaN-based light-emitting structure, which is at least one of an alternating InGaN / GaN multi-quantum-well structure, a GaN / InGaN / GaN single-quantum-well structure, and an InGaN quantum dot structure. When the InGaN-based light-emitting structure is an alternating InGaN / GaN multi-quantum-well structure or a GaN / InGaN / GaN single-quantum-well structure, the thickness of the InGaN quantum well layer is 0.5 nm-3 nm; when the InGaN-based light-emitting structure is an InGaN quantum dot structure, the height of the InGaN quantum dot is 0.5 nm-5 nm. Using thin quantum well or quantum dot structures can effectively shorten the radiative recombination lifetime of charge carriers and weaken the quantum confinement Stark effect. Taking the alternating InGaN / GaN multi-quantum-well structure as an example, the thin InGaN quantum well layer can increase the overlap integral of the electron-hole wave function, improve the radiative recombination rate, and thus increase the modulation bandwidth.

[0072] Specifically, when the active region 150 adopts an alternating InGaN / GaN multi-quantum-well structure, each period includes one InGaN quantum well layer and one GaN quantum barrier layer. The In composition of the InGaN quantum well layer can be adjusted according to the target emission wavelength: an In composition of approximately 0.05 corresponds to near-ultraviolet emission (approximately 380 nm), an In composition of approximately 0.15 corresponds to blue emission (approximately 450 nm), an In composition of approximately 0.25 corresponds to green emission (approximately 520 nm), an In composition of approximately 0.30 corresponds to yellow emission (approximately 570 nm), and an In composition of approximately 0.40 corresponds to red emission (approximately 630 nm). The thickness of the GaN quantum barrier layer is typically 5 nm to 15 nm, serving to confine charge carriers and modulate stress. For long-wavelength micro-LEDs in the green to red light band, the higher In content in the InGaN quantum well leads to greater lattice mismatch stress and a more significant quantum confinement Stark effect. Therefore, it is particularly important to adopt a thin quantum well (0.5nm~1.8nm) design, which can effectively reduce the quantum confinement Stark effect and improve radiative recombination efficiency and modulation bandwidth.

[0073] When the active region 150 adopts a GaN / InGaN / GaN single quantum well structure, only one InGaN quantum well layer participates in light emission. The radiative recombination lifetime of charge carriers in a single quantum well is the shortest, which is beneficial for achieving the highest modulation bandwidth. The single quantum well structure is suitable for applications with extremely high modulation bandwidth requirements but relatively low optical power requirements.

[0074] When the active region 150 employs an InGaN quantum dot structure, the InGaN quantum dots are self-assembled via the Stranski-Krastanov (SK) growth mode. The quantum dots, as nanostructures distributed within the GaN matrix layer, exhibit a three-dimensional quantum confinement effect, which enhances carrier localization and improves radiative recombination efficiency. The discrete energy level structure of the quantum dots allows for faster carrier relaxation and recombination processes, which is beneficial for increasing the modulation bandwidth. Furthermore, the piezoelectric polarization field in the quantum dots is significantly weakened compared to that in quantum wells, further contributing to a higher radiative recombination rate. The quantum dot layers are separated by GaN spacer layers, typically with a thickness of 5 nm to 20 nm.

[0075] Al y Ga 1-y The N electron blocking layer is 160, with an Al composition (y) of 0.05-0.20 and a thickness of 10 nm-100 nm. y Ga 1-yThe function of the N-electron blocking layer 160 is to prevent electrons from overflowing from the active region 150 into the P-type layer, thereby improving injection efficiency. Appropriate Al composition and thickness ensure effective electron blocking without excessively hindering hole injection. For example, if the Al composition is too low, the electron blocking capability is insufficient; if the Al composition is too high, it may affect hole transport. In a preferred embodiment, Al... y Ga 1-y The N-electron blocking layer 160 has an Al composition of 0.10~0.15 and a thickness of 30nm~60nm. This preferred range achieves a good balance between effectively blocking electron overflow and ensuring smooth hole injection.

[0076] The p-type doped GaN hole injection layer 170 is made of magnesium with a doping concentration of 5 × 10⁻⁶. 18 cm -3 ~5×10 20 cm -3 The thickness is 10nm-100nm. Magnesium, as a p-type dopant, can provide holes. A reasonable doping concentration and thickness ensure effective hole injection, allowing for recombination with electrons in the active region 150 to produce light. Inappropriate doping concentration or thickness can lead to uneven hole injection or low injection efficiency, affecting device performance. In a preferred embodiment, the doping concentration of the p-type doped GaN hole injection layer 170 is 1×10⁻⁶. 19 cm -3 ~5×10 19 cm -3 The thickness ranges from 20 nm to 60 nm. It's important to note that magnesium in GaN has a high activation energy (approximately 170 meV), and only about 1% to 10% of the magnesium acceptors are activated at room temperature. Therefore, the actual hole concentration is much lower than the magnesium doping concentration. After growth, annealing under an N2 atmosphere is typically required to break the Mg-H complex and activate the magnesium acceptors.

[0077] The implementation principle of this embodiment is as follows: By reasonably selecting the substrate 110, a good foundation is provided for subsequent epitaxial growth; the buffer layer 120 can alleviate lattice mismatch and thermal mismatch, and improve the crystal quality of the epitaxial layer; the N-type doped GaN electron injection layer 130 and the P-type doped GaN hole injection layer 170 provide sufficient electrons and holes for the active region 150, respectively; the active region 150 adopts a thin quantum well or quantum dot structure, which effectively shortens the carrier radiative recombination lifetime and improves the modulation bandwidth; Al y Ga 1- yThe N-electron blocking layer 160 prevents electron overflow and improves injection efficiency. The synergistic effect of the overall structure enables GaN-based micro-LED epitaxial structures to achieve high modulation bandwidth and high photoelectric efficiency, meeting the requirements of high-speed visible light communication, representing a significant improvement over traditional GaN-based LED structures. Furthermore, the inherent wide bandgap of the GaN material system endows the device with excellent high-temperature resistance, allowing it to operate stably in the high-temperature environment (85℃~105℃) near data center switching chips without thermoelectric cooling—a significant advantage that is difficult to achieve with traditional InP-based lasers.

[0078] The following examples of specific parameter combinations further illustrate the preferred implementation of the active region 150:

[0079] When the active region 150 is an InGaN / GaN alternating multiple quantum well structure, the thickness of the InGaN quantum well layer is 0.5nm-1.8nm, and the number of multiple quantum well pairs is 2-5. When the active region 150 is a GaN / InGaN / GaN single quantum well structure, the thickness of the InGaN quantum well layer is 0.5nm-1.8nm. When the active region 150 is an InGaN quantum dot structure, the InGaN quantum dots are nanostructures distributed in the GaN matrix layer, with 1-5 layers, a diameter of 2nm-10nm, a height of 1nm-5nm, and an areal density of 1×10⁻⁶. 10 / cm 2 ~1×10 11 / cm²; The doping concentration of the N-type doped GaN electron-injected layer 130 is 2×10⁻⁶. 19 cm -3 ~2×10 20 cm -3 Thickness is 4.5μm-6μm; Al y Ga 1-y The N-electron blocking layer 160 has an Al composition of 0.10-0.15 and a thickness of 30nm-60nm; the P-type doped GaN hole injection layer 170 has a doping concentration of 1×10⁻⁶. 19 cm -3 ~5×10 19 cm -3 The thickness is 20nm-60nm.

[0080] Further optimization of the parameters of each layer can better balance carrier injection and recombination efficiency, improving the overall performance of the device. Thinner InGaN quantum well layers can further reduce the quantum confinement Stark effect and increase the radiative recombination rate of carriers; a suitable number of quantum well cycles can ensure the optimal balance between luminescence efficiency and modulation bandwidth. For InGaN quantum dot structures, reasonable size and density can enhance carrier localization and improve radiative recombination efficiency. Optimized parameters of the N-type doped GaN electron injection layer 130 and the P-type doped GaN hole injection layer 170 can ensure effective electron and hole injection while reducing series resistance. y Ga 1-y The optimized parameters of the N-electron blocking layer 160 can effectively block electrons without affecting the transport of holes.

[0081] Specifically, when the number of quantum well cycles is 2-3 pairs, carriers are distributed in fewer quantum wells, resulting in a higher carrier concentration in each quantum well and a faster radiative recombination rate, which is beneficial for obtaining a higher modulation bandwidth. When the number of cycles is 4-5 pairs, the total luminescent volume increases, and the optical power output is higher. However, the distribution of carriers in each quantum well may be uneven, with higher carrier concentrations in quantum wells near the p-type layer and lower concentrations in those farther away. This may lead to an increase in the overall equivalent carrier lifetime and a decrease in the modulation bandwidth. Therefore, in applications that pursue the highest modulation bandwidth, a number of 2-3 cycles is preferred; in applications that require a balance between optical power and bandwidth, a number of 4-5 cycles can be selected.

[0082] In this embodiment, by optimizing the active region 150, the N-type doped GaN electron injection layer 130, and Al... y Ga 1-y Further optimization of the parameters of each layer, such as the N-electron blocking layer 160 and the P-type doped GaN hole injection layer 170, makes the synergistic operation between the layers more efficient, further improving the carrier injection and recombination efficiency. Thus, while ensuring high photoelectric efficiency, the modulation bandwidth of the device is further improved, which can better meet the requirements of high-speed visible light communication for light source performance. This is a further improvement and optimization of the existing technology.

[0083] Please continue to refer to Figure 1Preferably, a stress relief layer 140 is further provided between the N-type doped GaN electron injection layer 130 and the active region 150. The stress relief layer 140 is an InGaN / GaN multi-period structure with 5-20 periods. The InGaN and GaN layers in each period of the InGaN / GaN multi-period structure are of equal thickness. Alternatively, along the direction close to the active region 150, the thickness of the InGaN layer in each period of the InGaN / GaN multi-period structure gradually increases, and / or the thickness of the GaN layer gradually decreases. And / or, a highly p-type doped contact layer is further provided above the P-type doped GaN hole injection layer 170. The doping element of the highly p-type doped contact layer is magnesium, and the doping concentration is 1×10⁻⁶. 19 cm -3 ~5×10 20 cm -3 The thickness is 10nm-100nm.

[0084] The stress relief layer 140 can alleviate the lattice mismatch stress between the N-type doped GaN electron injection layer 130 and the active region 150, reducing the dislocation density and piezoelectric field in the active region 150. The InGaN / GaN multi-period structure with uniform thickness provides stable stress buffering; while the gradient thickness design achieves a smoother stress transition, further reducing defects in the active region 150. The highly p-type doped contact layer can reduce the contact resistance between the p-type doped GaN hole injection layer 170 and the transparent conductive film, improving ohmic contact characteristics and reducing the device's operating voltage.

[0085] Specifically, the InGaN layer thickness in each cycle of the stress relief layer 140 can be 1 nm to 5 nm, with the In composition typically lower than that in the quantum well of the active region 150 (e.g., In composition of 0.03 to 0.10). The GaN layer thickness can be 3 nm to 20 nm. When a constant thickness setting is used, for example, each cycle includes an InGaN layer with a thickness of 2 nm and a GaN layer with a thickness of 8 nm, for a total of 10 cycles, the total thickness is 100 nm. When a gradient thickness design is used, for example, along the direction close to the active region 150, the InGaN layer thickness gradually increases from 1 nm to 3 nm, and the GaN layer thickness gradually decreases from 12 nm to 6 nm. This gradient design can achieve a smooth transition in composition and stress, further reducing the defect density and V-shaped pit density in the active region 150.

[0086] The doping concentration of highly p-type doped contact layers is typically higher than that of p-type doped GaN hole injection layers by 170, for example, a doping concentration of 1 × 10⁻⁶. 20 cm -3 The thickness is 20 nm. A higher surface doping concentration facilitates the formation of tunneling contacts between the p-type GaN and the ITO transparent conductive film, reducing the contact resistance to 10 Ω·cm. -4This reduces the device's operating voltage and power consumption to the order of Ω·cm² or less.

[0087] The implementation principle of this embodiment is as follows: the addition of the stress relief layer 140 effectively improves the crystal quality of the active region 150 and increases the radiative recombination efficiency; the setting of the highly p-type doped contact layer improves the electrical performance of the device and reduces power consumption. These improvements further enhance the overall performance of the GaN-based micro-LED epitaxial structure, enabling it to perform better in high-speed visible light communication applications. This represents an effective improvement over existing technologies in addressing lattice mismatch and contact resistance issues.

[0088] Example 2

[0089] This application provides a GaN-based micro-LED device, including the epitaxial structure described in the above embodiments and an electrode structure formed on the epitaxial structure. The electrode structure forms a good electrical connection with each layer of the epitaxial structure, which can introduce external current into the epitaxial structure, enabling the device to emit light normally.

[0090] Specifically, the shape of the light-emitting region of a micro-LED device can be any of rectangular, circular, or hexagonal. Different shapes can be selected based on specific application scenarios and design requirements. Rectangular shapes facilitate arrangement and layout in an array; circular shapes offer better symmetry and more uniform light emission; hexagonal shapes can achieve a higher fill factor in close-packed arrays. The equivalent circular diameter corresponding to the area of ​​the light-emitting region ranges from 1μm to 100μm. A smaller equivalent circular diameter can effectively reduce the junction capacitance of the device and increase the modulation bandwidth. The equivalent circular diameter refers to the diameter of a circle with the same area as the light-emitting region. , where S is the area of ​​the luminescent region.

[0091] Micro-LED devices emit light in the 365nm-700nm range. By adjusting the In composition of the InGaN quantum well layer or quantum dots in the active region 150, different colors of light can be emitted, meeting the needs of optical communication for different wavelength light sources. For example, a lower In composition enables near-ultraviolet light emission, while a higher In composition enables red light emission. In data center optical interconnect applications, micro-LEDs in the blue (approximately 450nm) and green (approximately 520nm) bands have particularly prominent application advantages due to the high radiation efficiency of GaN materials in these bands and mature epitaxial processes.

[0092] The operating drive current density of micro-LED devices is below 50 A / cm², where the drive current density is the operating current divided by the emitting area. At this operating drive current density, the -3dB electro-optic modulation bandwidth of micro-LED devices is greater than 1 GHz. This is due to the optimized design of the epitaxial structure, such as the thin quantum well or quantum dot active region 150 structure, the stress relief layer 140, and the highly p-type doped contact layer, as well as reasonable device size control, enabling the device to achieve high modulation bandwidth at a low drive current density, meeting the requirements of high-speed optical communication. Low drive current density operation means extremely low device power consumption, which is particularly critical for the deployment of large-scale light source arrays in data center optical interconnects and CPO architectures—in CPO scenarios, the optical engine is adjacent to the switching chip package, and the heat dissipation space is extremely limited. Low-power light sources can significantly reduce the difficulty of thermal management and the total system power consumption.

[0093] The -3dB electro-optic modulation bandwidth refers to the frequency at which the device's electro-optic frequency response (i.e., the ratio of the AC component of the optical output power to the AC component of the input current) drops by 3dB relative to its low-frequency value. This bandwidth can be tested using the small-signal modulation method: a small radio frequency signal is output from a network analyzer, superimposed with a DC bias current through a bias circuit, and then injected into the micro-LED device. A high-speed photodetector receives the modulated optical signal and converts it into an electrical signal. The network analyzer measures the electro-optic frequency response curve (S21 parameter), and the -3dB cutoff frequency is read from it; this is the -3dB electro-optic modulation bandwidth.

[0094] The physical mechanism for achieving high modulation bandwidth in micro-LED devices can be understood from the following aspects: (a) The modulation bandwidth of the device is mainly limited by the carrier radiative recombination lifetime τ and the RC time constant, with a -3dB bandwidth f -3 dB≈1 / (2πτ_eff), where τ_eff is the equivalent carrier lifetime; (b) using a thin quantum well (0.5nm~3nm) can increase the overlap of the electron-hole wave function, improve the radiative recombination rate, and shorten τ; (c) reducing the device size can reduce the junction capacitance C and decrease the RC time constant; (d) increasing the driving current density can increase the carrier concentration in the active region 150, further accelerating the radiative recombination rate. Through the synergistic optimization of the above-mentioned methods, this invention achieves an electro-optic modulation bandwidth of more than 1GHz of -3dB under the condition of driving current density below 50A / cm².

[0095] The following test data for devices of different sizes illustrate the impact of device size on modulation bandwidth: a micro-LED with an equivalent circle diameter of 5 μm and a junction capacitance of approximately 0.02 pF has a -3 dB bandwidth of 2.3 GHz at a drive current density of 20 A / cm²; a micro-LED with an equivalent circle diameter of 20 μm and a junction capacitance of approximately 0.3 pF has a -3 dB bandwidth of 1.8 GHz at the same drive current density; a micro-LED with an equivalent circle diameter of 50 μm and a junction capacitance of approximately 2 pF has a -3 dB bandwidth of 1.3 GHz at the same drive current density; and a micro-LED with an equivalent circle diameter of 100 μm and a junction capacitance of approximately 8 pF has a -3 dB bandwidth of 1.05 GHz at the same drive current density. These results demonstrate that within the range of equivalent circle diameters from 1 μm to 100 μm, the -3 dB bandwidth of all devices is greater than 1 GHz.

[0096] The implementation principle of this embodiment is as follows: the optimized epitaxial structure provides excellent light-emitting performance and carrier transport conditions; the reasonable selection of the shape and size of the light-emitting region reduces junction capacitance and improves modulation bandwidth; the adjustable emission wavelength meets the needs of different scenarios; and the low operating drive current density and high modulation bandwidth enable the device to have the advantages of low power consumption and high-speed communication. The combined effect of these advantages makes GaN-based micro-LED devices promising for applications in data center optical interconnects, co-packaged optics (CPO), and visible light communication, offering significant performance improvements compared to traditional LED devices.

[0097] Example 3

[0098] This application provides a GaN-based micro-LED device array, including multiple micro-LED devices as described in Embodiment 2 above. The multiple micro-LED devices are arranged in any one of the following ways: rectangular array, concentric ring array, or hexagonal close-packed array. The number of devices included in the array is greater than or equal to 2.

[0099] Specifically, the array arrangement can be selected according to application requirements:

[0100] (a) Rectangular array: Multiple micro-LED devices are arranged in a rectangular grid, for example, 16 devices in a 4×4 grid, with a device spacing of 30μm~100μm. Rectangular arrays facilitate interfacing with the rectangular pixel arrangement of CMOS driving circuits and are suitable for display and communication integration applications. In data center CPO applications, rectangular arrays can be coupled one-to-one with waveguide arrays on silicon photonics chips to achieve multi-channel parallel optical interconnects.

[0101] (b) Concentric Ring Array: Consists of a central device and several concentric rings of devices, for example, 1 central device plus 3 concentric rings, totaling 19 devices. Concentric ring arrays are suitable for integrated lighting and communication applications that require circular light spots.

[0102] (c) Hexagonal close-packed array: Multiple micro-LED devices are arranged in a hexagonal close-packed pattern, such as a honeycomb array consisting of 7 devices. Hexagonal close-packing can achieve the highest area fill factor and is suitable for applications requiring high light power density output.

[0103] Each device in the array can be driven independently or in parallel. When driven independently, it enables spatially multiplexed multi-channel parallel optical communication, with each device acting as an independent communication channel. The total communication capacity is the product of the single-channel capacity and the number of channels. When driven in parallel, it increases the total optical power output and extends the communication distance. In data center optical interconnect scenarios, independently driven micro-LED arrays can achieve multi-channel wavelength division multiplexing (WDM) or spatial division multiplexing (SDM), significantly improving the total transmission capacity of a single optical fiber or waveguide.

[0104] Example 4

[0105] This application provides a method for fabricating a GaN-based micro-LED epitaxial structure, including the following steps:

[0106] S1. A substrate 110 is provided and pretreated. The substrate 110 is selected from a (111) crystal plane single-crystal silicon substrate, a c-axis single-crystal sapphire substrate, a c-axis single-crystal silicon carbide substrate, or a c-axis self-supporting single-crystal gallium nitride substrate. The pretreatment includes cleaning and surface treatment of the substrate 110 to remove surface contaminants and oxide layers. Cleaning can be performed by a combination of chemical reagent cleaning and deionized water rinsing to ensure that the surface of the substrate 110 is clean. Surface treatment can be performed by high-temperature annealing or plasma treatment to improve the flatness and activity of the substrate 110 surface, providing a good foundation for subsequent epitaxial growth.

[0107] S2, A buffer layer 120 is grown on the substrate 110 by an epitaxial growth method. The buffer layer 120 includes an unintentionally doped GaN layer and an AlN / Al layer. x Ga 1-x At least one of the N / GaN stacks, where x is 0.1-1.0. Epitaxial growth methods can include metal-organic chemical vapor deposition (MOCVD), molecular beam epitaxy (MBE), etc. Unintentionally doped GaN layers can be grown using MOCVD at appropriate temperatures and gas flow rates; AlN / Al x Ga 1-xThe growth of N / GaN stacks can be achieved by controlling the introduction order and ratio of different precursors. The buffer layer 120 serves to alleviate lattice and thermal mismatches between the substrate 110 and GaN, improving the crystal quality of subsequent epitaxial layers. When using the MOCVD method, trimethylgallium (TMGa) is used as the Ga source, trimethylaluminum (TMAl) as the Al source, high-purity ammonia (NH3) as the N source, and high-purity hydrogen and / or nitrogen as the carrier gas. The growth temperature of the AlN nucleation layer is typically 500℃~600℃, and the growth temperature of the unintentionally doped GaN layer is typically 1000℃~1100℃.

[0108] S3, An N-type doped GaN electron injection layer 130 is grown on the buffer layer 120 using an epitaxial growth method. The doping element is silicon, and the doping concentration is 1×10⁻⁶. 18 cm -3 ~5×10 20 cm -3 The thickness ranges from 2 μm to 8 μm. The MOCVD method can also be used, with the doping concentration adjusted by controlling the flow rate of silane (SiH4). Suitable growth temperature and gas environment ensure good crystal quality and uniform doping distribution. The growth temperature is typically 1000℃~1100℃.

[0109] In some embodiments, step S3a is further included between step S3 and step S4: a stress relief layer 140 is grown on the N-type doped GaN electron injection layer 130 by an epitaxial growth method. The stress relief layer 140 is an InGaN / GaN multi-period structure with 5 to 20 periods. In the InGaN / GaN multi-period structure, the InGaN layer and GaN layer in each period are grown with equal thickness; or, along the direction close to the active region 150, the thickness of the InGaN layer in each period gradually increases, and / or the thickness of the GaN layer gradually decreases. The growth temperature of the InGaN layer in the stress relief layer 140 is typically 750°C to 850°C, and the growth temperature of the GaN layer is typically 850°C to 950°C. The stress relief layer 140 serves to provide a gradual transition of composition and stress between the N-type doped GaN electron injection layer 130 and the active region 150, thereby alleviating lattice mismatch stress, reducing the dislocation density and V-pit density in the active region 150, and thus improving the crystal quality and radiative recombination efficiency of the active region 150.

[0110] S4. An active region 150 is grown on the N-type doped GaN electron injection layer 130 via epitaxial growth. The active region 150 includes an InGaN-based light-emitting structure, which is at least one of an InGaN / GaN alternating multi-quantum-well structure, a GaN / InGaN / GaN single-quantum-well structure, and an InGaN quantum dot structure. When the InGaN-based light-emitting structure is an InGaN / GaN alternating multi-quantum-well structure or a GaN / InGaN / GaN single-quantum-well structure, the thickness of the InGaN quantum well layer is controlled to be 0.5 nm-3 nm; when the InGaN-based light-emitting structure is an InGaN quantum dot structure, the height of the InGaN quantum dots is controlled to be 0.5 nm-3 nm. Precise control of the growth temperature, gas flow rate, and growth time is required during the growth process to obtain the desired structure and size. The growth temperature of the InGaN quantum well layer is typically 700℃~800℃, and the growth temperature of the GaN quantum barrier layer is typically 800℃~900℃. Using trimethylindium (TMIn) as the In source, the In composition of the InGaN layer is controlled by adjusting the flow ratio of TMI to TMGa. When using an InGaN quantum dot structure, the self-assembly growth of InGaN quantum dots is achieved using the Stranski-Krastanov (SK) growth mode by adjusting parameters such as growth temperature, V / III ratio, and growth rate.

[0111] S5, Al is grown on the active region 150 by epitaxial growth method. y Ga 1-y The N-electron blocking layer is 160, with an Al composition of 0.05-0.20 and a thickness of 10nm-100nm. The Al composition is adjusted by controlling the flow rate ratio of trimethylaluminum and trimethylgallium, and the growth time is precisely controlled to obtain the desired thickness. y Ga 1-y The growth of the N-electron blocking layer 160 must ensure good interface quality with the active region 150 to avoid introducing defects. The growth temperature is typically 900℃~1000℃.

[0112] S6, in Al y Ga 1-y A p-type doped GaN hole injection layer 170 is grown on the N-electron blocking layer 160 via epitaxial growth. The doping element is magnesium, and the doping concentration is 5 × 10⁻⁶. 18 cm -3 ~5×10 20 cm -3 The thickness ranges from 10 nm to 100 nm. The MOCVD method is used, and the doping concentration is adjusted by controlling the flow rate of magnesia-diocene (Cp₂Mg). After growth, annealing may be required to activate the magnesium dopant and increase the hole concentration. The growth temperature is typically 900℃~1000℃.

[0113] In some embodiments, step S7 is further included after step S6: a highly p-type doped contact layer is grown on the p-type doped GaN hole injection layer 170 by epitaxial growth, wherein the doping element is magnesium and the doping concentration is 1×10⁻⁶. 19 cm -3 ~5×10 20 cm -3 The thickness ranges from 10 nm to 100 nm. The growth temperature of the highly p-type doped contact layer is typically 850℃ to 950℃; a lower growth temperature is beneficial for improving the magnesium doping efficiency. The function of the highly p-type doped contact layer is to reduce the contact resistance between p-type GaN and the subsequently deposited transparent conductive film, thereby improving ohmic contact characteristics.

[0114] The implementation principle of this embodiment is as follows: Pre-treatment of the substrate 110 ensures favorable starting conditions for subsequent epitaxial growth. Epitaxial growth is employed to grow each layer sequentially, allowing for precise control of the thickness, composition, and doping concentration of each layer. The orderly growth and rational design of each layer result in a GaN-based micro-LED epitaxial structure with excellent crystal quality and electrical performance, achieving high modulation bandwidth and high photoelectric efficiency. The systematic and precise nature of this fabrication method ensures the stability and consistency of device performance, representing an effective improvement over existing GaN-based LED fabrication technologies and meeting the stringent requirements of high-speed visible light communication for light source fabrication.

[0115] Example 5

[0116] This application provides a method for fabricating a GaN-based micro-LED device, comprising the following steps:

[0117] A1. An epitaxial structure was prepared using the preparation method described in Example 4.

[0118] A2. A transparent conductive film is deposited on the surface of the p-type layer of the epitaxial structure. The transparent conductive film is an indium tin oxide (ITO) film with a thickness of 30 nm to 300 nm, deposited using any one of magnetron sputtering, electron beam evaporation, or chemical vapor deposition. The ITO film, as a p-type current spreading layer, has high transmittance (visible light transmittance greater than 85%) and low sheet resistance (typically 10 to 100 Ω / □), enabling uniform current spreading on the p-type layer surface without significantly affecting light extraction. In one specific embodiment, the ITO film is deposited using magnetron sputtering under the following conditions: Ar atmosphere, power 80 to 150 W, substrate temperature 150°C to 300°C, and an ITO film thickness of 100 nm.

[0119] A3. P-meta-structures are formed through photolithography and etching processes, with the etching depth penetrating the transparent conductive film, the P-type doped GaN hole injection layer 170, and Al.y Ga 1-y An N-type electron blocking layer 160 and an active region 150 are formed, exposing an N-type doped GaN electron injection layer 130. Photolithography employs either ultraviolet lithography or electron beam lithography, with the appropriate method selected based on the device dimensions. Etching utilizes an inductively coupled plasma (ICP) dry etching process with a Cl2 / BCl3 mixed gas. The etching power, gas flow rate, and etching time are adjusted according to the required etching depth. The mesa shape can be circular, rectangular, or hexagonal, and its size determines the size of the device's light-emitting area.

[0120] A4. Rapid thermal annealing (RTA) is performed in an inert atmosphere at a temperature of 400°C to 800°C for 30 to 600 seconds. The inert atmosphere can be N2 or Ar. The purposes of rapid thermal annealing include: (a) improving the ohmic contact characteristics between ITO and the hole injection layer 170 of P-type doped GaN, and reducing the contact resistance; (b) repairing lattice damage and dangling bonds introduced on the mesa sidewalls during ICP etching, and reducing the surface nonradiative recombination rate; and (c) activating Mg acceptors in P-type GaN and breaking the Mg-H complex. In one specific embodiment, the annealing temperature is 550°C, the annealing time is 120 seconds, and the atmosphere is N2.

[0121] A5. A passivation layer is deposited, and metal contact windows are formed through photolithography and etching. The passivation layer material can be SiO2, Si3N4, Al2O3, or a combination thereof, and the deposition method can be plasma-enhanced chemical vapor deposition (PECVD), atomic layer deposition (ALD), etc. The thickness of the passivation layer is typically 100nm~500nm. The passivation layer covers the entire device surface, including the mesa sidewalls, and is used to protect the mesa sidewalls, reduce the surface state density, and prevent surface nonradiative recombination. Subsequently, metal contact windows are formed on the top of the P-mesa and the surface of the N-type doped GaN electron injection layer 130 through photolithography and wet etching (such as using buffered oxide etchant BOE) or dry etching.

[0122] A6. Positive and negative electrodes are formed by depositing metal electrodes through photolithography, metal evaporation, and lift-off processes. The P electrode is deposited on the ITO contact window at the top of the P-mesa, and a Ni / Au (e.g., 5nm / 100nm) metal stack can be used. The N electrode is deposited on the N-type doped GaN electron injection layer 130 contact window, and a Ti / Al / Ti / Au (e.g., 10nm / 100nm / 10nm / 100nm) metal stack can be used. Metal evaporation can be performed using electron beam evaporation or thermal evaporation methods.

[0123] The implementation principle of this embodiment is as follows: uniform expansion of the P-type current is achieved through the deposition of an ITO transparent conductive film; a precise mesa structure is formed through ICP dry etching, defining the light-emitting area of ​​the device; rapid thermal annealing improves the ohmic contact and repairs etching damage; the deposition of a passivation layer effectively suppresses surface nonradiative recombination on the mesa sidewalls; and the deposition of metal electrodes achieves electrical connection between the device and external circuitry. The synergistic optimization of each step results in a micro-LED device with excellent electrical and optical performance.

[0124] Example 6

[0125] This application provides methods for removing substrates from micro-LED devices for different substrate types 110.

[0126] When substrate 110 is a silicon substrate, step A4a is included between steps A4 and A5: the silicon substrate under the micro-LED device is removed using an alkaline etching solution to form a suspended pixel structure. The alkaline etching solution can be a KOH solution (mass fraction 20%~40%, temperature 60℃~90℃) or a tetramethylammonium hydroxide (TMAH) solution. Utilizing the anisotropic etching characteristics of the alkaline solution on (111) silicon, the silicon substrate is selectively removed without damaging the GaN epitaxial layer, where the AlN layer in the buffer layer 120 serves as an etching stop layer. The suspended pixel structure eliminates the absorption of light by the silicon substrate (silicon has a high absorption coefficient in the visible light band), significantly improving light extraction efficiency, while reducing the parasitic capacitance of the device, which is beneficial for further improving the modulation bandwidth.

[0127] When the substrate 110 is a sapphire substrate, after step A6, a step is further included to separate the micro-LED device from the original substrate using a laser lift-off (LLO) process. The photon energy of the laser is greater than the GaN bandgap (3.4 eV, corresponding to a wavelength of approximately 365 nm) and less than the sapphire bandgap (approximately 9.9 eV, corresponding to a wavelength of approximately 125 nm). For example, a 248 nm KrF excimer laser (photon energy 5.0 eV) or a third harmonic of a 355 nm Nd:YAG laser (photon energy 3.49 eV) can be used. The laser is incident from the back side of the sapphire substrate, passes through the sapphire, and is absorbed by GaN at the GaN / sapphire interface, causing the GaN at the interface to thermally decompose into Ga metal and N2 gas, thereby achieving the separation of the epitaxial layer from the substrate 110. The laser energy density is typically 400~800 mJ / cm².

[0128] When the substrate 110 is a silicon carbide substrate, the step after step A6 includes a step of removing the silicon carbide substrate by grinding or chemical etching. The silicon carbide substrate can be thinned to about 30μm~80μm by mechanical grinding first, and then the remaining SiC material can be removed by SF6-based plasma dry etching or high-temperature molten KOH chemical etching.

[0129] Example 7

[0130] This application provides methods for transferring micro-LED devices and fabricating microlenses.

[0131] After step A6, at least one of the following steps may be performed:

[0132] Step A7: Transfer the micro-LED device onto a new substrate using a transfer printing process. The new substrate can be any one of a CMOS driver array substrate, a diamond substrate, or a glass substrate. The transfer printing process typically uses an elastomeric stamp (such as a PDMS stamp), and the specific steps include: (a) aligning and contacting the PDMS stamp with the micro-LED device, using van der Waals forces to pick the device up from the original substrate; (b) aligning the stamp with the corresponding position on the new substrate; (c) slowly peeling off the stamp to release the device onto the new substrate; and (d) achieving electrical connection between the device and the circuitry on the new substrate through reflow soldering or conductive adhesive bonding. Transferring to a CMOS driver array substrate enables monolithic integration of the micro-LED and driver circuitry, facilitating independent addressing and driving of high-density arrays; transferring to a diamond substrate utilizes the high thermal conductivity of diamond (approximately 2000 W / (m·K)) to improve heat dissipation performance, which is beneficial for stable operation of the device under high current density; transferring to a glass substrate can be used for transparent displays and integrated lighting and communication applications. In data center CPO applications, transferring the micro-LED array onto a CMOS driver array substrate enables tight integration of the light source and driver circuitry, allowing for co-packaging with the switching chip. This significantly shortens the electrical interconnect distance and reduces signal transmission delay and power consumption. Transferring it to a diamond substrate utilizes diamond's ultra-high thermal conductivity to achieve efficient heat dissipation without the need for thermoelectric coolers, ensuring long-term stable operation of the micro-LED in the high-temperature environment (85℃~105℃) near the switching chip.

[0133] Step A8: A microlens is fabricated in situ on the light-emitting surface of the micro-LED device using a two-photon lithography process. The microlens is made of photosensitive resin. Two-photon lithography employs a femtosecond laser (e.g., a 780nm Ti:sapphire femtosecond laser) for two-photon polymerization, achieving a resolution below 100nm and enabling precise control of the microlens' three-dimensional morphology. Commercially available materials such as IP-Dip and IP-S can be used as the photosensitive resin. The microlens can be designed as a hemispherical, aspherical, or freeform surface, with a diameter matching the micro-LED's light-emitting area. The microlens collimates or focuses the Lambertian beam emitted by the micro-LED, increasing light intensity in a specific direction, thereby improving the signal-to-noise ratio and communication distance at the optical communication receiver. In data center optical interconnect applications, the microlens can efficiently couple the light emitted by the micro-LED into optical fibers or silicon waveguides, significantly improving coupling efficiency and reducing optical link loss. After fabrication, post-curing with ultraviolet light enhances the microlens' mechanical strength and durability.

[0134] When steps A7 and A8 are performed simultaneously, step A7 (device transfer) is performed first, followed by step A8 (microlens fabrication). This is because microlens fabrication needs to be performed at the final working position of the device to ensure precise alignment between the microlens and the light-emitting area.

[0135] Example 8

[0136] This application provides specific examples of device fabrication and performance testing.

[0137] InGaN-based red micro-LEDs were transferred onto a substrate to fabricate three sizes of devices with square light-emitting regions of 20 μm, 30 μm, and 40 μm side lengths, respectively.

[0138] First, the electrical characteristics of the three sizes of devices were characterized. Figure 2 The current-voltage (IV) characteristic curves are shown for devices with sizes of 20 μm, 30 μm, and 40 μm. Figure 2 It can be seen that the larger 40 μm device exhibits higher current under the same applied voltage.

[0139] Next, the frequency response of devices of different sizes was measured using a vector network analyzer (VNA, PicoVNA 106, 6 GHz) and a high-speed PIN photodetector (Femto HSPR-XI-1G4-SI, 1.4 GHz). Figure 3The figures show the electro-optical (EO) modulation bandwidth as a function of current density for devices of different sizes. Test results show that at a current density of 30 A / cm², the maximum modulation bandwidths of the 20 μm, 30 μm, and 40 μm devices are 1623.52 MHz, 1334.43 MHz, and 929.81 MHz, respectively. These results fully demonstrate that smaller micro-LED devices have higher modulation bandwidths. This is because the smaller device size results in lower junction capacitance and a smaller RC time constant, thus reducing the limitation of the RC time constant on the modulation bandwidth. The 1623.52 MHz modulation bandwidth achieved by the 20 μm device is currently the highest modulation bandwidth record for red micro-LEDs.

[0140] Subsequently, the data transmission rates of 20 μm and 30 μm devices were tested under low-current conditions. OFDM (Orthogonal Frequency Division Multiplexing) modulation technology, combined with pre-equalization and bit loading algorithms, was employed to improve the data rate. OFDM data was generated by a MATLAB program and output through an arbitrary waveform generator (AWG, Keysight Technologies M8190A, 12 GS / s). The OFDM signal was amplified by an electrical amplifier (EA, Mini-Circuits ZHL-42W+, 4200 MHz), and the output signal was combined with a DC signal through a bias converter (Mini-Circuits ZFBT-6GW+, 6 GHz) to drive the micro-LED. The light signal emitted by the modulated micro-LED was collimated and focused onto an avalanche photodiode (APD, Hamamatsu C5658, 1 GHz) through two lenses. The light signal was converted into an electrical signal and acquired by a digital signal analyzer (DSA, Agilent DSA90604A Infiniium, 20 GSa / s), and finally demodulated by the receiving module. The signal bandwidth was set to 1.5 GHz, the drive current varied from 1 mA to 25 μA, and the transmission distance was 0.04 m.

[0141] Figure 4 (a) Curves showing the device voltage and data rate variations of 20 μm and 30 μm devices under different currents. Figure 4 (a) It can be seen that at a current of 1 mA, the data rates of the 20 μm and 30 μm devices are 2.25 Gbps and 2.12 Gbps, respectively, and both achieve bit error rate (BER) below the 3.8 × 10⁻³ forward error correction (FEC) threshold. Figure 4 (b) shows the DC power consumption curves of the 20 μm and 30 μm devices at different currents. Figure 4As shown in (b), when the current drops below 500 μA, the DC power consumption is consistently below 1 pJ / bit. The 20 μm device achieves a data rate of 1.54 Gbps at 100 μA with a DC power consumption of only 0.22 pJ / bit. These results demonstrate that higher modulation bandwidth offers significant advantages for achieving high-speed, low-power wireless optical interconnects.

[0142] Please refer to Figure 1 Taking a c-axis single-crystal sapphire substrate as an example, a blue micro-LED device with a circular light-emitting area of ​​20 μm diameter was fabricated using the epitaxial structure containing a stress-relieving layer 140 and a highly p-type doped contact layer as described in Example 1. The specific parameters of the epitaxial structure are as follows: substrate 110 is a c-axis single-crystal sapphire substrate; buffer layer 120 includes a low-temperature AlN nucleation layer (approximately 20 nm thick, growth temperature 550 °C) and an unintentionally doped GaN layer (approximately 2 μm thick, growth temperature 1050 °C); the doping concentration of the N-type doped GaN electron injection layer 130 is 5 × 10⁻⁶. 19 cm -3 The thickness is 5 μm; the stress relief layer 140 is an InGaN / GaN multi-period structure with a total of 10 periods, each period including an InGaN layer with a thickness of 2 nm (In composition of approximately 0.06) and a GaN layer with a thickness of 8 nm; the active region 150 adopts an InGaN / GaN alternating multi-quantum-well structure with a total of 3 periods, the InGaN quantum well layer has a thickness of 1.5 nm (In composition of approximately 0.15, corresponding to an emission wavelength of approximately 450 nm), and the GaN quantum barrier layer has a thickness of 8 nm; Al y Ga 1-y The N-electron blocking layer 160 has an Al composition of 0.12 and a thickness of 40 nm; the P-type doped GaN hole injection layer 170 has a doping concentration of 2 × 10⁻⁶. 19 cm -3 The thickness is 40 nm; the doping concentration of the highly p-type doped contact layer is 1 × 10⁻⁶. 20 cm -3 The thickness is 20nm.

[0143] Device fabrication process: A 100 nm thick ITO transparent conductive film was deposited by magnetron sputtering; a 20 μm diameter circular P-mesa was formed by photolithography and ICP etching. The etching gas was a Cl2 / BCl3 mixed gas, and the etching depth penetrated the transparent conductive film, the highly p-type doped contact layer, the p-type doped GaN hole injection layer 170, and Al. y Ga 1-yAn N-type electron blocking layer 160 and an active region 150 are formed, exposing an N-type doped GaN electron injection layer 130. The layer is annealed at 550°C for 120 seconds in an N2 atmosphere. A 200 nm thick SiO2 passivation layer is deposited by PECVD, and contact windows are opened by BOE wet etching. Ni / Au (5 nm / 100 nm) is deposited as the P electrode, and Ti / Al / Ti / Au (10 nm / 100 nm / 10 nm / 100 nm) is deposited as the N electrode.

[0144] The fabricated micro-LED devices underwent -3dB electro-optic modulation bandwidth testing. The test system included a network analyzer (e.g., Keysight N5227B), a high-speed photodetector (bandwidth > 12 GHz), a bias transformer (Bias-Tee), and a DC power supply. The test method was small-signal modulation. The test results are as follows:

[0145] At a driving current density of 10 A / cm², the -3 dB electro-optic modulation bandwidth is 1.2 GHz;

[0146] At a driving current density of 20 A / cm², the -3 dB electro-optic modulation bandwidth is 1.8 GHz;

[0147] With a driving current density of 50 A / cm², the -3 dB electro-optic modulation bandwidth is 2.5 GHz.

[0148] The above results demonstrate that the micro-LED device of this invention exhibits a -3dB electro-optic modulation bandwidth greater than 1GHz under operating drive current densities below 50A / cm², meeting the requirements of high-speed visible light communication. With increasing drive current density, the modulation bandwidth further improves, due to the higher injected carrier concentration leading to a faster radiative recombination rate and a shorter carrier lifetime.

[0149] In contrast, a conventional micro-LED device of the same size using a 2.5nm thick quantum well and a 5-cycle multi-quantum-well structure exhibits a -3dB electro-optic modulation bandwidth of only 0.4GHz to 0.8GHz at the same driving current density, significantly lower than the thin quantum well design of this invention. This fully demonstrates the technical effectiveness of this invention in shortening carrier radiative recombination lifetime and improving modulation bandwidth through the thin quantum well design.

[0150] In data center optical interconnect applications, the above performance indicators mean that, taking a drive current density of 20A / cm² as an example, the -3dB bandwidth of a single micro-LED device is 1.8GHz, and when using PAM-4 modulation format, the single-channel data rate can reach approximately 5.4Gbps; if a 16-channel parallel transmission with a 4×4 rectangular array is used, the total transmission capacity can reach approximately 86.4Gbps, which can meet the bandwidth requirements of short-distance optical interconnects in data centers. Meanwhile, at a low drive current density of 20A / cm², the power consumption of a single micro-LED device is only in the sub-milliwatt range, far lower than traditional laser solutions, and requires no thermoelectric cooling, resulting in significant advantages in total system power consumption and cost.

[0151] Example 9

[0152] This application provides examples of micro-LED device fabrication based on (111) crystal plane single-crystal silicon substrates.

[0153] The specific parameters of the epitaxial structure are as follows: the substrate 110 is a (111) crystal plane single crystal silicon substrate; the buffer layer 120 includes an AlN layer (approximately 200 nm thick), Al 0.5 Ga 0.5 An N-type transition layer (approximately 200 nm thick) and an unintentionally doped GaN layer (approximately 1.5 μm thick); the N-type doped GaN electron injection layer 130 has a doping concentration of 1 × 10⁻⁶. 20 cm -3 The thickness is 6 μm; the active region 150 adopts an InGaN quantum dot structure, with three layers of InGaN quantum dots distributed in the GaN matrix layer. The average diameter of the quantum dots is about 5 nm, the average height is about 2 nm, and the areal density is about 5 × 10⁻⁶. 10 / cm 2 The quantum dots in each layer are separated by GaN spacers with a thickness of approximately 10 nm; Al y Ga 1-y The N-electron blocking layer 160 has an Al composition of 0.15 and a thickness of 50 nm; the P-type doped GaN hole injection layer 170 has a doping concentration of 3 × 10⁻⁶. 19 cm -3 The thickness is 50nm.

[0154] The device fabrication steps A1-A3 are similar to those in Example 8, but the mesa shape is a regular hexagon with a diagonal length of 10 μm (equivalent circle diameter approximately 9.3 μm). Step A4 involves annealing at 500°C for 180 s under a N2 atmosphere. Step A4a uses a 30% KOH solution to etch at 80°C for approximately 2 hours to selectively remove the silicon substrate beneath the micro-LED device, forming a suspended pixel structure. The AlN layer in buffer layer 120 serves as an etching stop layer. Steps A5-A6 are similar to those in Example 8.

[0155] Micro-LED devices fabricated on silicon substrates offer the advantage of inherent compatibility with CMOS processes. In data center CPO applications, silicon-based micro-LEDs can be directly heterogeneously integrated on silicon-based CMOS driver circuit wafers, eliminating the complex alignment and packaging processes between lasers and silicon photonic chips found in traditional solutions. This is expected to significantly reduce the manufacturing cost and packaging complexity of the optical engine. The suspended pixel structure eliminates light absorption on the silicon substrate while also significantly reducing parasitic capacitance, further improving modulation bandwidth and making it more suitable for high-speed data transmission applications.

[0156] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Any other corresponding changes and modifications made based on the technical concept of this application should be included within the scope of protection of this application.

Claims

1. A GaN-based micro-LED epitaxial structure, characterized in that, Along the growth direction, the following are included in sequence: The substrate (110) is selected from a single-crystal silicon substrate with a (111) crystal plane, a c-axis single-crystal sapphire substrate, a c-axis single-crystal silicon carbide substrate or a c-axis self-supporting single-crystal gallium nitride substrate. A buffer layer (120) is formed on the substrate (110), the buffer layer (120) comprising an unintentionally doped GaN layer and an AlN / Al layer. x Ga 1-x At least one of N / GaN stacks, wherein x is 0.1 to 1.0; An N-type doped GaN electron-injected layer (130) is formed on the buffer layer (120), the dopant element being silicon, and the doping concentration being 1×10⁻⁶. 18 cm -3 ~5×10 20 cm -3 The thickness is 2μm~8μm; An active region (150) is formed above the N-type doped GaN electron injection layer (130). The active region (150) includes an InGaN-based light-emitting structure, which is at least one of an InGaN / GaN alternating multi-quantum-well structure, a GaN / InGaN / GaN single-quantum-well structure, and an InGaN quantum dot structure. When the InGaN-based light-emitting structure is an InGaN / GaN alternating multi-quantum-well structure or a GaN / InGaN / GaN single-quantum-well structure, the thickness of the InGaN quantum well layer is 0.5 nm to 3 nm. When the InGaN-based light-emitting structure is an InGaN quantum dot structure, the height of the InGaN quantum dot is 0.5 nm to 5 nm. Al y Ga 1-y An N electron blocking layer (160) is formed on the active region (150), with an Al composition y of 0.05~0.20 and a thickness of 10nm~100nm; A p-type doped GaN hole injection layer (170) is formed on the Al y Ga 1-y Above the N electron blocking layer (160), the dopant element is magnesium, and the doping concentration is 5 × 10⁻⁶. 18 cm -3 ~5×10 20 cm -3 The thickness is 10nm~100nm.

2. The GaN-based micro-LED epitaxial structure according to claim 1, characterized in that, When the active region (150) is an InGaN / GaN alternating multiple quantum well structure, the thickness of the InGaN quantum well layer is 0.5nm~1.8nm, and the number of cycles of the multiple quantum wells is 2~5 pairs; When the active region (150) is a GaN / InGaN / GaN single quantum well structure, the thickness of the InGaN quantum well layer is 0.5nm~1.8nm; When the active region (150) is an InGaN quantum dot structure, the InGaN quantum dots are nanostructures distributed in the GaN matrix layer, with 1 to 5 layers, a diameter of 2 nm to 10 nm, a height of 1 nm to 5 nm, and an areal density of 1 × 10⁻⁶. 10 / cm 2 ~1×10 11 / cm²; The doping concentration of the N-type doped GaN electron-injected layer (130) is 2 × 10⁻⁶. 19 cm -3 ~2×10 20 cm -3 The thickness is 4.5μm~6μm; The Al y Ga 1-y The Al composition of the N electron blocking layer (160) is 0.10~0.15, and the thickness is 30nm~60nm; The doping concentration of the P-type doped GaN hole injection layer (170) is 1×10⁻⁶. 19 cm -3 ~5×10 19 cm -3 The thickness is 20nm~60nm.

3. The GaN-based micro-LED epitaxial structure according to claim 1, characterized in that, A stress relief layer (140) is further provided between the N-type doped GaN electron injection layer (130) and the active region (150). The stress relief layer (140) is an InGaN / GaN multi-period structure with a period number of 5 to 20. In the InGaN / GaN multi-period structure, the InGaN layer and GaN layer of each period are set to the same thickness; or, along the direction close to the active region (150), the thickness of the InGaN layer of each period in the InGaN / GaN multi-period structure gradually increases, and / or the thickness of the GaN layer gradually decreases. And / or, a highly p-type doped contact layer is further provided above the p-type doped GaN hole injection layer (170), wherein the doping element of the highly p-type doped contact layer is magnesium, and the doping concentration is 1×10⁻⁶. 19 cm -3 ~5×10 20 cm -3 The thickness is 10nm~100nm.

4. A GaN-based micro-LED device, characterized in that, The device includes the epitaxial structure as described in any one of claims 1 to 3 and the electrode structure formed on the epitaxial structure; the shape of the light-emitting region of the GaN-based micro-LED device is any one of rectangular, circular, or hexagonal, and the equivalent circle diameter corresponding to the area of ​​the light-emitting region is 1 μm to 100 μm; the emission wavelength of the GaN-based micro-LED device is 365 nm to 700 nm; the operating driving current density of the GaN-based micro-LED device is less than 50 A / cm², wherein the driving current density is the operating current divided by the light-emitting area; at the operating driving current density, the -3 dB electro-optic modulation bandwidth of the GaN-based micro-LED device is greater than 1 GHz.

5. A GaN-based micro-LED device array, characterized in that, The array includes multiple micro-LED devices as described in claim 4, wherein the multiple micro-LED devices are arranged in any one of a rectangular array, a concentric ring array, or a hexagonal close-packed array, and the array contains a number of devices greater than or equal to 2.

6. A method for fabricating a GaN-based micro-LED epitaxial structure, characterized in that, Includes the following steps: S1. Provide a substrate (110) and perform pretreatment. The substrate (110) is selected from (111) crystal plane single crystal silicon substrate, c-axis single crystal sapphire substrate, c-axis single crystal silicon carbide substrate or c-axis self-supporting single crystal gallium nitride substrate. S2. A buffer layer (120) is grown on the substrate (110) by an epitaxial growth method. The buffer layer (120) includes an unintentionally doped GaN layer and an AlN / Al layer. x Ga 1-x At least one of N / GaN stacks, wherein x is 0.1 to 1.0; S3. An N-type doped GaN electron injection layer (130) is grown on the buffer layer (120) by an epitaxial growth method. The doping element is silicon, and the doping concentration is 1×10⁻⁶. 18 cm -3 ~5×10 20 cm -3 The thickness is 2μm~8μm; S4. An active region (150) is grown on the N-type doped GaN electron injection layer (130) by an epitaxial growth method. The active region (150) includes an InGaN-based light-emitting structure. The InGaN-based light-emitting structure is at least one of an InGaN / GaN alternating multi-quantum-well structure, a GaN / InGaN / GaN single-quantum-well structure, and an InGaN quantum dot structure. When the InGaN-based light-emitting structure is an InGaN / GaN alternating multi-quantum-well structure or a GaN / InGaN / GaN single-quantum-well structure, the thickness of the InGaN quantum well layer is controlled to be 0.5 nm to 3 nm. When the InGaN-based light-emitting structure is an InGaN quantum dot structure, the height of the InGaN quantum dot is controlled to be 0.5 nm to 3 nm. S5. Al is grown on the active region (150) by an epitaxial growth method. y Ga 1-y The N electron blocking layer (160) has an Al composition of 0.05~0.20 and a thickness of 10nm~100nm. S6, in the Al y Ga 1-y A p-type doped GaN hole injection layer (170) is grown on the N electron blocking layer (160) using an epitaxial growth method. The doping element is magnesium, and the doping concentration is 5 × 10⁻⁶. 18 cm -3 ~5×10 20 cm -3 The thickness is 10nm~100nm.

7. The method for fabricating a GaN-based micro-LED epitaxial structure according to claim 6, characterized in that, Between step S3 and step S4, step S3a is also included: growing a stress relief layer (140) on the N-type doped GaN electron injection layer (130) by an epitaxial growth method. The stress relief layer (140) is an InGaN / GaN multi-period structure with a period number of 5 to 20. In the InGaN / GaN multi-period structure, the InGaN layer and GaN layer of each period are grown with equal thickness; or, along the direction close to the active region (150), the thickness of the InGaN layer of each period gradually increases, and / or the thickness of the GaN layer gradually decreases. And / or, following step S6, step S7 is further included: growing a highly p-type doped contact layer on the p-type doped GaN hole injection layer (170) by an epitaxial growth method, wherein the doping element is magnesium and the doping concentration is 1×10⁻⁶. 19 cm -3 ~5×10 20 cm -3 The thickness is 10nm~100nm.

8. A method for fabricating a GaN-based micro-LED device, characterized in that, Includes the following steps: A1. An epitaxial structure is prepared using the preparation method described in claim 6 or 7; A2. Deposit a transparent conductive film on the surface of the P-type layer of the epitaxial structure. The transparent conductive film is an indium tin oxide thin film with a thickness of 30nm~300nm, and is deposited by any one of magnetron sputtering, electron beam evaporation or chemical vapor deposition. A3. P-meta-structures are formed through photolithography and etching processes, with the etching depth penetrating the transparent conductive film, the P-type doped GaN hole injection layer (170), and Al. y Ga 1-y The N-type electron blocking layer (160) and the active region (150) expose the N-type doped GaN electron injection layer (130); A4. Perform rapid thermal annealing under an inert atmosphere, with an annealing temperature of 400℃~800℃ and an annealing time of 30s~600s. A5. Deposit a passivation layer and form a metal contact window through photolithography and etching; A6. Positive and negative electrodes are formed by depositing metal electrodes through photolithography, metal evaporation and lift-off processes.

9. The method for fabricating a GaN-based micro-LED device according to claim 8, characterized in that, When the substrate (110) is a silicon substrate, step A4a is also included between step A4 and step A5: the silicon substrate under the micro-LED device is removed by using an alkaline etching solution to form a suspended pixel structure; When the substrate (110) is a sapphire substrate, after step A6, a step is also included to separate the micro-LED device from the original substrate using a laser lift-off process, wherein the photon energy of the laser is greater than the GaN bandgap and less than the sapphire material bandgap; When the substrate (110) is a silicon carbide substrate, the step after step A6 further includes a step of removing the silicon carbide substrate by grinding or chemical etching.

10. The method for fabricating a GaN-based micro-LED device according to claim 8 or 9, characterized in that, Following step A6, at least one of the following steps is also included: A7. A transfer printing process is used to transfer micro-LED devices onto a new substrate, wherein the new substrate includes any one of a CMOS driving array substrate, a diamond substrate, or a glass substrate; A8. A microlens is fabricated in situ on the light-emitting surface of the micro-LED device using a two-photon lithography process. The material of the microlens is photosensitive resin. When steps A7 and A8 are performed simultaneously, step A7 is performed first, followed by step A8.

Images