GaN gradient doping and suppressing leakage epitaxy method for high-power pulsed device

Abstract

This invention discloses a GaN gradient doping drain suppression epitaxial method for high-power pulsed devices, relating to the field of semiconductor manufacturing technology. The method includes: epitaxially growing a lightly doped gallium nitride buffer layer on a substrate; growing a gradient transition layer with continuously increasing doping concentration from bottom to top on top of the buffer layer; synergistically epitaxially growing a main drain suppression layer and a stress relief band at its top; subsequently, forming a termination adapter layer and a sidewall termination region to create an edge sealing structure; and finally, performing steady-state heat treatment and drain suppression capability verification under pulsed conditions. This invention disperses the vertical electric field through the gradient transition layer, absorbs accumulated stress using the stress relief band, and manages the edge field strength through the edge sealing structure, effectively suppressing regional material fatigue and breakdown path formation under repeated pulses, significantly improving the service life and reliability of high-power pulsed devices.

Description

Technical Field

[0001] This invention relates to the field of semiconductor manufacturing technology, and in particular to a GaN gradient doping drain suppression epitaxy method for high-power pulse devices. Background Technology

[0002] Gallium nitride (GaN)-based high-power pulsed devices, such as radio frequency pulse amplifiers and solid-state power switches, occupy a core position in radar, communication, and energy conversion systems due to their superior characteristics such as high frequency and high power density. These devices need to withstand high voltage, high current, and extremely short pulse rise and fall times during operation, and their long-term reliability directly determines the stability of the entire system. The reliability of these devices is largely determined by the quality of their epitaxial material structure, especially the epitaxial layer's ability to manage electric fields and stress.

[0003] In existing technologies, to suppress leakage current under high voltage, a leakage suppression structure with a specific doping concentration is typically introduced into the GaN epitaxial layer of the device. However, traditional designs often employ uniform or abrupt doping distributions. While this structure may meet requirements under steady-state or low-frequency operation, its inherent defects become apparent under the transient cumulative condition of repeated high-speed pulse switching. Specifically, at the instant of pulse turn-off, the electric field in the vertical direction redistributes rapidly, forming highly concentrated electric field peaks at the doping abrupt interface or the top layer of the barrier. Simultaneously, the inherent polarization effect of GaN material generates repetitive polarization stress under the drive of the rapidly changing electric field. Under the long-term high duty cycle pulse stress, these concentrated electric fields and accumulated stresses work synergistically, leading to irreversible regional material fatigue and structural collapse in the leakage suppression structure, especially in the lattice region at the top of the main epitaxial layer. This degradation process is gradual, initially forming undetectable potential breakdown path initiation points within the device, ultimately causing premature device failure or a sharp deterioration in performance. The uniform or simple abrupt doping structures in the existing technology, because they cannot effectively disperse transient electric fields and dissipate accumulated stress, essentially provide an inherent weak link for the fatigue and structural collapse of such materials, making it difficult to meet the stringent requirements of the new generation of high-power pulse devices for long-term dynamic reliability.

[0004] Therefore, there is an urgent need for a GaN epitaxial method that can fundamentally optimize the electric field and stress distribution in the vertical direction, enabling it to adapt to the harsh operating conditions of repeated pulse switching, thereby effectively suppressing regional material fatigue and the formation of potential breakdown paths, and improving the service life and reliability of high-power pulse devices. Summary of the Invention

[0005] The present invention provides a GaN gradient doping drain suppression epitaxial method for high-power pulse devices, which can solve the above-mentioned problems.

[0006] To solve the above problems, the technical solution adopted by the present invention is as follows:

[0007] A GaN gradient doping drain suppression epitaxial method for high-power pulsed devices includes:

[0008] A low-doped gallium nitride buffer layer is epitaxially grown sequentially on a semiconductor substrate to provide a crystal basis with low dislocation density and low initial stress.

[0009] On the low-doped gallium nitride buffer layer, a gradient transition layer with continuously increasing doping concentration from bottom to top is epitaxially grown. The gradient transition layer is used to disperse the electric field distribution in the vertical direction and achieve a smooth transition of interlayer stress.

[0010] On the gradient transition layer, a main leakage suppression layer and a stress relief band located on top of it are epitaxially grown in synergy. The main leakage suppression layer forms a potential barrier, and the stress relief band absorbs and dissipates accumulated stress.

[0011] On top of the main leakage suppression layer and stress relief band, a terminal adapter layer and a sidewall termination region are formed to create an edge sealing structure to manage the stress distribution and electric field in the edge region of the device.

[0012] The structure that has completed epitaxial growth was subjected to steady-state heat treatment under pulsed conditions and its leakage suppression capability was verified. The dynamic stress safety margin and the spatial distribution of leakage current were evaluated by applying a high duty cycle pulse load, and the transfer of the leakage suppression path to the edge dissipation channel was verified.

[0013] Furthermore, the method for sequentially epitaxially growing a lightly doped gallium nitride buffer layer on a semiconductor substrate specifically includes:

[0014] The substrate is chemically cleaned, and the surface roughness and residual content parameters are obtained by monitoring the cleaning time and solution ratio. Degassing and thermal stabilization treatments are then performed, and the average desorption rate of the desorption process is calculated using the Arrhenius equation to guide the heat treatment parameters.

[0015] The lightly doped gallium nitride buffer layer was deposited on the surface of the processed substrate, and the interfacial stress between the epitaxial layer and the substrate was estimated by the substrate curvature method after deposition. ;

[0016] The crystal quality and electrical properties of the lightly doped gallium nitride buffer layer were evaluated, and the dislocation density was measured. With sheet resistor When the measured value is within the preset acceptable range, the final thickness of the low-doped gallium nitride buffer layer is determined. Interfacial stress and preset stress threshold Solidify the gradient boundary conditions for subsequent gradient layer design .

[0017] Furthermore, the method for epitaxially growing a gradient transition layer with continuously increasing doping concentration from bottom to top specifically includes:

[0018] Based on the gradient boundary conditions and sheet resistors Plan a gradient layer structure with multiple segments and linearly increasing doping concentration in each segment, and determine the number of segments. Total target thickness and the target thickness of each segment and target doping concentration ;

[0019] Based on material stress tolerance constraints, a layer-weighted model is used to calculate the cumulative equivalent internal stress of multi-layer structures. While maintaining the overall target thickness While maintaining the same thickness, the thickness of each segment is corrected to ensure the cumulative equivalent internal stress. Not exceeding the stress threshold ;

[0020] The corrected design parameters are converted into epitaxial process instructions, and epitaxial deposition of each segment is executed sequentially. The target doping concentration is achieved by controlling the equivalent supply of the doped precursor and the growth time. The actual deposition thickness of each segment is measured in real time using an online metrology instrument. and actual doping concentration This enables closed-loop control of the deposition process.

[0021] Furthermore, when planning the gradient layer structure, the doping concentration of each segment increases linearly according to the calculation formula:

[0022]

[0023] in, For segment numbers, The doping concentration of the first segment is set as the baseline and is lower than the doping concentration of the subsequent main drain suppressor layer. This represents the doping concentration increment between two adjacent segments;

[0024] Meanwhile, the total target thickness of the gradient layer structure Equal to the target thickness of each segment sum.

[0025] Furthermore, after the gradient transition layer is deposited, a consistency determination of the gradient structure is also included:

[0026] Calculate the thickness uniformity index of the gradient transition layer and doping uniformity index :

[0027]

[0028]

[0029] in, This represents the maximum measured thickness of all segments in the gradient transition layer. The minimum measured thickness of all segments in the gradient transition layer. For the final total thickness, The average doping concentration, This represents the maximum measured doping concentration across all segments of the gradient transition layer. This represents the minimum measured doping concentration across all segments of the gradient transition layer.

[0030] Verify whether the measured doping concentration of each segment maintains a monotonically increasing relationship from bottom to top, and whether the total thickness and the thickness of each segment are within the acceptable tolerance range;

[0031] When uniformity, monotonicity, and thickness all meet the requirements, the top boundary coordinates of the gradient transition layer will be... Final doping concentration of the top section and total thickness Solidification serves as the geometric lower boundary and parameter basis for the deposition of the main leakage suppression layer.

[0032] Furthermore, the method for synergistic epitaxial growth of the main leakage suppression layer and the stress relief band located on top of it specifically includes:

[0033] Based on the top boundary coordinates of the gradient transition layer The design thickness of the main leakage suppression layer is set based on the lower stress accumulation margin. With the target doping range The main leakage suppression layer is epitaxially deposited on the gradient transition layer;

[0034] The top boundary coordinates of the gradient transition layer The lower geometric boundary of the main leakage suppression layer is locked; the design thickness of the main leakage suppression layer is determined based on the lower stress accumulation margin. The preset top boundary coordinates of the main leakage suppression layer are derived based on the lower geometric boundary and the designed thickness. To satisfy Geometric relationships;

[0035] After the main leakage suppression layer is deposited, its actual thickness is measured. Based on this measured thickness The design includes a stress relief band at its top, with a doping concentration of [missing information]. The doping concentration is set to be lower than that of the main drain suppressor layer to enhance lattice compliance;

[0036] The thickness of the stress relief band Through the proportionality coefficient The measured thickness of the main leakage suppression layer Association, satisfying the relation: ;

[0037] Deposit the stress relief zone to form a coating structure covering the main leakage suppression layer, and record the coordinates of the top of the coating structure. This serves as the starting surface for the lower geometric boundary of the terminal adaptation layer.

[0038] Furthermore, the method for forming the terminal adaptation layer and the sidewall termination region specifically includes:

[0039] An epitaxial terminal adapter layer is deposited on the stress relief zone, the designed thickness of the terminal adapter layer being... Through the proportionality coefficient With the total thickness of the covering structure Association, satisfying the relation: ;

[0040] After the terminal adapter layer is deposited, the residual stress margin after deposition is evaluated. And based on this remaining stress margin Determine the geometric unfolding radius of the sidewall termination region ;

[0041] The sidewall termination region is formed by extending outward along the sidewall direction, so that the equivalent thickness of the sidewall termination region is consistent with the terminal adapter layer. Thus, the sidewall termination region and the terminal adapter layer are used to close at the outer edge of the device to form the edge sealing structure, which transforms the potential edge leakage path into an energy dissipation channel.

[0042] Furthermore, the geometric unfolding radius of the sidewall termination region The calculation formula is:

[0043]

[0044] in, The measured thickness of the terminal adapter layer. The stress-geometric conversion factor is used to diffuse the accumulated polarization energy from the edge point to the curve consumption zone through geometric expansion.

[0045] Furthermore, the method for performing the stabilization heat treatment and leakage suppression capability verification specifically includes:

[0046] The epitaxial structure was subjected to steady-state heat treatment in a controlled environment, and the steady-state temperature was set. and steady-state time This stabilizes the polarization residues in the material and reduces the residual stress after heat treatment. It is determined by both the stress allowance before heat treatment and the stress attenuation coefficient during the thermal equilibrium process;

[0047] A high duty cycle pulse load is applied to the heat-treated structure, wherein the pulse duty cycle is... Set between 30% and 80%, the dynamic stress safety margin is calculated. Assess whether material fatigue has been triggered;

[0048] The dynamic stress safety margin The calculation is based on the thermal stability fatigue strength of the material. Residual stress after heat treatment and equivalent instantaneous stress under pulsed conditions Satisfying the relation: ;

[0049] Simultaneously, the spatial distribution of leakage current is analyzed, and the sidewall suppression coefficient is calculated. To verify the effect of the leakage suppression path transfer.

[0050] Furthermore, the sidewall inhibition coefficient Defined as the discharge density of the longitudinal central region Discharge density in the edge termination region The ratio; when the sidewall inhibition coefficient Less than 1 and the dynamic stress safety margin When the value is greater than 1, the epitaxial structure is deemed qualified, confirming that the main path of leakage current has been transferred from the vertical direction to the edge termination region, and the long-term stability of the leakage suppression function under repeated pulse stress has been achieved.

[0051] Compared with the prior art, the beneficial effects of the present invention are:

[0052] (1) This invention fundamentally changes the electric field distribution in the vertical direction of the device by constructing a gradient transition layer with continuously increasing doping concentration from bottom to top on the buffer layer. This structure disperses the electric field peaks that are concentrated on a single interface in traditional abrupt design into a thicker transition region, achieving a smooth transition and effective dissipation of the electric field, and significantly reducing the risk of material breakdown caused by the rapid redistribution of the electric field at the moment of high-power pulse turn-off.

[0053] (2) In this invention, a stress relief band is synergistically constructed on the top of the main leakage suppression layer, forming a composite functional structure of "potential barrier + stress dissipation". The main leakage suppression layer is responsible for establishing a robust potential barrier to block leakage current, while the flexible stress relief band on the top can effectively absorb and release the accumulated stress generated by rapid polarization effect and thermal mismatch, preventing stress from returning to the main layer, thereby significantly suppressing material fatigue and structural disintegration in the surface area under repeated pulse action.

[0054] (3) A closed edge sealing structure was constructed around the device by synergistic layering of the terminal adapter layer and the sidewall termination region. This structure enables fine-grained management of the electric field and stress in the edge region of the device, eliminates the lateral high electric field region and stress concentration point, transforms potential edge leakage paths into controllable energy dissipation channels, avoids premature failure caused by pulse loads in the edge region of the device, and ensures the integrity of leakage suppression capability in three-dimensional space.

[0055] (4) This invention introduces steady-state heat treatment and leakage suppression capability verification steps under pulsed operating conditions, forming a closed-loop quality control system from material growth to performance verification. This method ensures that each functional layer can be constructed under strict geometric and electrical specifications, and directly verifies the long-term leakage suppression capability and reliability of the epitaxial structure under dynamic stress through tests simulating real service conditions, providing a reliable epitaxial foundation for manufacturing high-reliability high-power pulsed devices.

[0056] (5) This invention corrects the segmented thickness and doping concentration of the gradient layer based on the stress tolerance constraint of the material, and calculates the cumulative equivalent internal stress. This design ensures the mechanical stability during the epitaxial growth process, avoids the introduction of new stress concentration points due to improper local thickness or doping, and guarantees the crystal quality of the material from the growth source.

[0057] (6) By determining the geometric unfolding radius of the sidewall termination region based on the residual stress margin, this invention utilizes geometric unfolding to diffuse the accumulated polarization energy from the edge point to the curve dissipation zone. This physically-based geometric design effectively achieves active dissipation of stress and energy, further enhancing the fatigue resistance of the device edge.

[0058] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, embodiments of the present invention are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0059] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0060] Figure 1 A flowchart of the GaN gradient doping drain suppression epitaxial method for high-power pulse devices provided in an embodiment of the present invention is shown.

[0061] Figure 2 A cross-sectional schematic diagram of a GaN gradient-doped drain-suppressing epitaxial stack structure for high-power pulse devices provided by an embodiment of the present invention is shown. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0063] like Figure 1 As shown, the process may include the following steps:

[0064] S1: A lightly doped gallium nitride buffer layer is epitaxially grown sequentially on a semiconductor substrate.

[0065] This step aims to thoroughly pretreat the semiconductor substrate and epitaxially grow a high-quality, lightly doped gallium nitride buffer layer. This buffer layer provides a reliable foundation for subsequent gradient structures with low dislocation density and low initial stress, mitigating inherent weaknesses caused by substrate variations or surface contamination at their source.

[0066] S1.1: Chemical cleaning and surface condition assessment.

[0067] First, a raw substrate undergoes a systematic chemical cleaning process. This process consists of three stages: organic decontamination, weak oxidation etching to remove metal residues, and removal of the surface oxide layer. Each stage is followed by rinsing with ultrapure water, and finally, nitrogen gas is used to dry the substrate to obtain a clean and dry surface.

[0068] During the cleaning process, by monitoring the cleaning time and solution ratio, and using surface analysis instruments to measure, multiple parameters characterizing the surface state can be obtained, including surface roughness. Surface residual content Surface oxidation ratio and the performance records of current process equipment These parameters are recorded and collectively referred to as parameter set S1. Changes in surface residual content can be estimated using the following relationship: ,in, This represents the initial residual content before cleaning (unit: atoms / cm³). 2 ), and These are the cleaning and removal rate constants for the organic and weakly oxidizing sections, respectively, and their values ​​typically range from 1 to 10 atom / (cm²). 2 ·min) and 0.1 to 5 atom / (cm 2 ·min), and This corresponds to the cleaning time (in minutes). This step removes all organic and metallic contaminants, resulting in a repeatable and uniform starting surface, thereby avoiding the formation of localized nucleation hotspots during subsequent epitaxial growth.

[0069] S1.2: Degassing and thermal stabilization treatment.

[0070] Based on the obtained surface parameter set S1, the cleaned substrate is placed in a high vacuum or inert gas-filled reaction chamber for thermal stabilization. The baking temperature for this process is... and duration The settings should be referenced from the device capability records. Ensure that it does not exceed its safety limit.

[0071] During this heat treatment process, water molecules and other volatile residues physically adsorbed on the substrate surface will steadily desorb. The average desorption rate of this desorption process is... (Unit: s) -1 This can be described by the Arrhenius formula: ,in, It is the pre-desorption factor, and its value range is usually 1×10. 13 Up to 1×10 15 s -1 , The apparent energy for desorption (unit: eV) ranges from 1.0 to 2.5 eV. The gas constant (value 8.617 × 10⁻⁶) -5 eV / K), The baking temperature (in K) is used to effectively reduce uncontrolled nucleation in the early stages of epitaxial growth, thereby suppressing the generation of early crystal defects and stress concentration sources. After completion, the baking temperature is set to... Duration and the calculated average desorption rate The data is recorded as parameter set S2 to guide the subsequent epitaxial deposition.

[0072] S1.3: Epitaxial deposition of a low-doped buffer layer.

[0073] After the substrate underwent thermal stabilization, a lightly doped gallium nitride buffer layer was epitaxially deposited on its surface. The target thickness of this buffer layer was... The doping concentration is set between 0.3 and 1 μm. It needs to be controlled at 1×10 16 cm -3 Up to 5×10 17 cm -3 Within the range.

[0074] The thickness of the buffer layer is determined by the epitaxial growth rate. With growth time They jointly decide to satisfy the following relationship: ,in, The value range is typically 0.1 to 1 μm / min. The value range is 5 to 50 min. After deposition, the interfacial stress between the epitaxial layer and the substrate needs to be estimated using methods such as the substrate curvature method. And measure the surface roughness after extension. The formula for calculating interfacial stress is: ,in, The equivalent elastic modulus of the substrate (unit: MPa). Substrate thickness (unit: μm). denoted as Poisson's ratio of the substrate (dimensionless). The measured radius of curvature of the entire disc after deposition (unit: m). The buffer layer thickness is shown in meters (m). The grown, lightly doped buffer layer provides a stable potential reference and effective dislocation filtering for subsequent gradient doped layers, reducing the initial triggering conditions for vertical potential spikes. All key results, including the buffer layer thickness, are presented. Doping concentration achieved through process control Interfacial stress and surface roughness They are integrated into parameter set S3.

[0075] S1.4: Dislocation density and electrical performance evaluation and parameter solidification.

[0076] Based on the obtained buffer layer parameter set S3, a final evaluation of the crystal quality and electrical properties of the buffer layer is required. First, the dislocation density within the buffer layer is assessed using statistical or equivalent observational methods for chemical corrosion pits. The formula for calculating dislocation density is: ,in, In order to observe the area (Unit: cm) 2 The number of corrosion pits obtained from the statistics (dimensionless). The unit is cm -2 .

[0077] Meanwhile, the sheet resistance of the buffer layer was measured using the four-probe method. The relationship between the sheet resistance and material parameters is expressed by the following formula: ,in, It is the fundamental charge constant (unit: C). It is the carrier mobility (unit: cm) 2 / V·s), Carrier concentration (unit: cm⁻¹) -3 ), This is the effective transport thickness of charge carriers (unit: cm), which can be considered as the thickness of the buffer layer. .

[0078] When the measured dislocation density and sheet resistance are both within the preset acceptable range, the system sets a specific buffer layer acceptance flag, Flag_buf, to a passed state. Subsequently, the final thickness of the buffer layer is... Interfacial stress and a preset stress threshold Solidify into explicit gradient boundary conditions, denoted as The stress threshold This is a pre-set safety upper limit based on the material mechanical properties of the substrate and epitaxial layer, used to constrain the total stress accumulation of subsequent gradient layer structures. Its value typically ranges from 500 MPa to 1500 MPa. This boundary condition provides a precise physical lower boundary and mechanical constraint for subsequent gradient layer design steps. At this point, the dislocation density will be obtained. chip resistors Buffer layer qualification flag Flag_buf and gradient boundary conditions These are collectively referred to as parameter set S4 and are directly passed to subsequent process flows.

[0079] In the technical solution of this invention, a high-quality, low-doped buffer layer with low dislocation density and low interfacial stress is established through systematic substrate pretreatment and epitaxial growth. This buffer layer provides a stable and reliable physical and electrical basis for subsequent epitaxial structures, effectively filters crystal defects from the substrate, and establishes a clear gradient doping initiation boundary. This fundamentally weakens the inherent weaknesses caused by poor initial material quality, laying a solid foundation for the long-term reliability of the entire device.

[0080] S2: On the low-doped gallium nitride buffer layer, a gradient transition layer with doping concentration continuously increasing from bottom to top is epitaxially grown.

[0081] Using epitaxial growth technology, a gallium nitride multilayer structure with continuously increasing doping concentration from bottom to top is constructed on a pre-fabricated buffer layer. This structure, through its gradient electrical properties, disperses the electric field spikes concentrated at a single interface in traditional devices into a thicker transition region, thereby significantly reducing the electric field stress during high-power pulse turn-off and improving device reliability.

[0082] S2.1: Gradient segmentation and doping level planning.

[0083] The system reads parameter set S4 and confirms that the buffer layer qualification flag Flag_buf is in a passed state. Based on this, the system will utilize the fixed gradient boundary conditions in parameter set S4. And combined with the measured sheet resistance The underlying electrical characteristics are used to plan the structure of the entire gradient transition layer. The planning process first determines the number of segments. This value is typically an integer between 3 and 5. Next, the total target thickness of the gradient layer is set. Its value ranges from 1 to 3 μm. Subsequently, a target thickness is assigned to each segment. and a target doping concentration The sum of the target thicknesses of all segments must equal the total target thickness, i.e., satisfy the following relationship: ,in, This represents the total thickness of the gradient layer. The number of segments, Let be the target thickness of the i-th segment. The doping concentration of each segment follows a strict linear increasing law and is determined by the following formula: Where i is the segment number, and its value is an integer from 1 to 1. Let i be the target doping level for the i-th segment. This is the doping concentration baseline for the first segment, the bottommost segment, which is set to be lower than the level of the subsequent main suppressor / drain layer. It is the doping concentration increment between two adjacent segments, and its value is usually controlled within... Between 1% and 20%. All these parameters, including the number of segments... Total thickness Thickness set of each segment and the doped sets of each segment All need to be in boundary conditions and sheet resistors The parameters were selected based on a comprehensive trade-off within acceptable limits. Ultimately, these planning parameters were integrated into parameter set G1. This step transforms the vertically incremental electric field management concept into a concrete, actionable list of geometric layer thicknesses and doping levels.

[0084] S2.2: Interface continuity and thickness correction under stress constraints.

[0085] After obtaining the preliminary design parameters, the stress accumulation effect during actual epitaxial growth needs to be considered, and the geometric parameters need to be fine-tuned to ensure the mechanical stability of the structure. Based on parameter set G1, while maintaining the total thickness... While maintaining the same thickness, the thickness of each segment is corrected. The purpose of this correction is to ensure a smooth transition at the interfaces of each segment within the material's stress tolerance, avoiding the introduction of new stress concentration points due to improper local thickness. The correction process requires calculating the cumulative equivalent internal stress of the entire multi-layer structure. For prediction, a layer-weighted model is used for calculation, and the formula is as follows: ,in, The cumulative equivalent internal stress across multiple layers (unit: MPa); It is the equivalent internal stress coefficient of the i-th segment of material (unit: MPa / μm). Its specific value depends on the physical state of the material segment and the temperature history during the deposition process. It is usually obtained through experimental standards and the value ranges from 100 to 1000 MPa / μm. This is the corrected thickness of the i-th segment (unit: μm). The calculated cumulative equivalent internal stress... Must not exceed the gradient boundary conditions The given stress threshold After optimization calculations under stress constraints, the corrected thickness set for each segment was obtained. The target doping concentration for each segment remains constant throughout the process and is recorded as follows: All corrected thicknesses Doping concentration and the calculated cumulative internal stress value This is recorded as parameter set G2. This step ensures that the design of the gradient structure meets the reliability requirements of material mechanics while pursuing electrical performance.

[0086] S2.3: Segmented deposition execution and online metering.

[0087] The corrected design parameter set G2 is transformed into a solid epitaxial layer structure, and real-time monitoring ensures consistency between process execution and design objectives. Based on the parameters in G2, the system sequentially completes the process from segment 1 to segment 2. Epitaxial deposition of segments. For each segment, the target thickness is achieved by controlling the average growth rate and growth time of that segment, with the following relationship: ,in, Let be the measured thickness of the i-th segment. The average growth rate of the i-th segment is typically controlled within the range of 0.1 to 1 μm / min. The growth time for the i-th segment is adjusted from 1 to 60 minutes based on the target thickness. Simultaneously, the doping concentration of each segment is controlled by precisely controlling the equivalent supply of the doped precursor. and growth time To achieve this, the relationship is described by the following formula: ,in, The measured doping concentration of the i-th segment (unit: cm⁻¹) -3 ), It is the doping introduction ratio factor (unit: cm). -3 ·min -1 ·sccm -1 ), which is a positive value, and its specific value is determined by the equipment system and the deposition temperature, and needs to be pre-calibrated before the process; The equivalent supply of the precursor for the i-th segment (unit: sccm); Growth time (unit: min).

[0088] After each deposition segment is completed, the actual deposition thickness of that segment is immediately measured using an online metering instrument. and actual doping concentration Subsequently, the thickness and doping concentration of this segment were calculated relative to the design target values ​​in parameter set G2. and The relative deviation is calculated using the following formula: , ,in, Let be the relative deviation of the thickness of the i-th segment. denoted as the relative deviation of the doping concentration in the i-th segment.

[0089] Set of measured thicknesses for all segments Measured doping concentration set Total measured thickness and the set of deviations for each segment This is recorded as parameter set G3. This step enables closed-loop control and quantitative traceability from design parameters to physical deposition results.

[0090] S2.4: Gradient consistency determination and segment boundary solidification.

[0091] The final quality assessment and data archiving of the deposited gradient layer are performed. The system verifies the quality of the entire gradient structure based on measured data from parameter set G3. First, the final total measured thickness is calculated. ,in, For the final total thickness, Let be the final thickness of the i-th segment after correction within acceptable tolerance. Then, evaluate the uniformity of the gradient structure and calculate the thickness uniformity index. and doping uniformity index ,in, As an index of thickness uniformity, These are indicators of doping uniformity, and all are dimensionless numbers. and These represent the maximum and minimum values ​​of the total segment thickness; and These represent the maximum and minimum doping concentrations across the entire fragment. This represents the average doping concentration across the entire wafer.

[0092] The system verifies two core indicators: first, whether the measured doping concentration of each segment maintains a strict bottom-up monotonically increasing relationship; and second, whether the total thickness and the thickness of each segment are within the acceptable tolerance range. When uniformity, monotonicity, and thickness all meet the requirements, the gradient validity flag Flag_grad is set to pass.

[0093] Then, the boundary coordinates of each segment Final thickness set of each segment The final doped set of each segment Total thickness The gradient validity identifier Flag_grad is fixed into parameter set G4, ensuring that the subsequent main leakage suppression layer can be deposited on a rigorously validated base structure with excellent gradient characteristics.

[0094] In the technical solution of this invention, by constructing a gradient transition layer with doping concentration continuously increasing from bottom to top, the electric field distribution in the vertical direction of the device is fundamentally optimized. This structure disperses the high electric field peaks concentrated at a single interface in traditional designs into a thicker transition region, achieving a smooth transition of the electric field and stress dissipation. This significantly reduces the electric field stress impact at the moment of high-power pulse turn-off, thereby providing core electrical management capabilities for suppressing regional material fatigue and the formation of potential breakdown paths.

[0095] S3: On the gradient transition layer, a main leakage suppression layer and a stress relief band located on top of it are epitaxially grown together.

[0096] After completing the deposition and verification of the vertically increasing gradient transition layer, the co-construction stage of the main drain suppression layer and the top stress relief band begins. A main drain suppression layer with specific thickness and doping characteristics is deposited on top of the gradient layer, and a stress relief band is integrated on top of it, thereby forming a composite structure of electric field management and stress dissipation in the vertical direction, effectively suppressing fatigue disintegration and the formation of hidden breakdown nuclei in the surface region under high-power pulse conditions.

[0097] S3.1: Matching the lower boundary of the main leakage suppression layer and generating the thickness target.

[0098] Based on parameter set G4, which includes the segment boundary coordinate set, the final thickness and doping set of each segment, the total thickness, and the gradient validity flag Flag_grad, when Flag_grad is in the pass state, the geometric lower boundary of the main suppressor layer is locked to the top boundary coordinate of the gradient layer. Based on the total thickness of the gradient layer The design thickness of the main leakage suppression layer is determined based on the lower stress accumulation margin. Its value ranges from 0.5 to 1 μm. Meanwhile, the final doping based on the top segment of the gradient layer... Define the target doping region of the main drain suppressor layer. So that it can be in The thickness is appropriately increased based on the existing thickness, but does not exceed the upper limit of the process. The thickness of the main leakage suppression layer is determined by the preset top boundary coordinates. and lower boundary The difference determines whether the relationship is satisfied. This step precisely correlates the leakage suppression function with the geometry, ensuring that the main leakage suppression layer forms an effective potential barrier above the gradient layer. All design parameters, including the design thickness of the main leakage suppression layer, are considered. Lower boundary and target doping range They are integrated into parameter set P1.

[0099] S3.2: Actual deposition and thickness of the main leakage suppression layer—doping coupling calibration.

[0100] Based on the design thickness and doping target of the main drain suppressor layer defined in parameter set P1, the epitaxial growth of the main drain suppressor layer is performed within the reaction chamber. The growth of the main drain suppressor layer is controlled by adjusting the average growth rate. and growth time To achieve this, the thickness is actually measured. From the formula The calculation yielded, where The value range is typically 0.1 to 1 μm / min. Adjustments are made within tens of minutes based on the target thickness. Simultaneously, the target doping concentration is achieved by precisely controlling the supply amount and time of the doping precursor, and the measured doping value is obtained. Immediately after deposition, the actual thickness and doping were measured using online metrology, and the relative thickness deviation was calculated. In addition, the thickness uniformity index within the main leakage suppression layer was evaluated. This step ensures deposition quality. It translates leakage suppression capability into verifiable geometric and electrical parameters, ensuring the primary leakage suppression layer forms a reliable barrier path in the actual structure. All measured values ​​and evaluation results, including the measured thickness of the primary leakage suppression layer, are included. Measured doping concentration relative thickness deviation and thickness uniformity index within the main leakage suppression layer , is recorded as parameter set P2.

[0101] S3.3: Based on the deposition results of the main leakage suppression layer, the design parameters of the top stress relief zone are generated.

[0102] Parameter set P2 provides the actual state of the completed main leakage suppression layer, including its thickness and stress distribution. Based on this, a stress relief zone is planned above the main leakage suppression layer, with a thickness of... The setting is within the range of 0.1 to 0.3 μm, and is achieved through a scaling factor. Measured thickness of the main leakage suppression layer The association satisfies the formula ,in, The value ranges from 0.1 to 0.3. Simultaneously, the doping concentration of the sustained-release band is set. The doping level is set slightly lower than that of the main leakage suppression layer to enhance its lattice compliance without affecting its leakage suppression function. Furthermore, a stress tolerance is defined based on the stress state after the main layer deposition. This step ensures that the sustained-release layer can effectively absorb and dissipate stress in subsequent processes. It establishes the design baseline for the flexible capping layer, which regulates thermal and polarization deformation to prevent surface stress from returning to the main layer and causing fatigue under repeated pulses. All design parameters, including the sustained-release layer thickness, are considered. Doping concentration of the sustained-release band and stress tolerance They are integrated into parameter set P3.

[0103] S3.4: Deposition of the slow-release zone and stabilization of the upper surface.

[0104] Based on the thickness and doping targets defined in parameter set P3, epitaxial growth of stress-relieving bands was sequentially performed on the surface of the main drain suppression layer. During deposition, the growth rate and precursor flow rate were controlled to ensure that the actual parameters of the stress-relieving bands met the design values. After deposition, the actual coordinates of the stress-relieving layer were recorded and aligned with the coordinates of the main drain suppression layer to form a complete coating structure. The total thickness of this coating structure is... It is obtained by summing the measured thickness of the main leakage suppression layer and the thickness of the sustained-release zone, i.e. Simultaneously, a flag `Flag_relax` is set to indicate that the structure is ready. Finally, the thickness of the complete overlay structure is determined. , Flag_relax (signature indicating completion of the slow-release layer), and the top coordinates of the overlay structure. and the coordinate alignment record of the main layer and the sustained-release layer This is solidified into parameter set P4. This step provides a stable and continuous geometric upper boundary for the subsequent construction of the terminal adaptation and sidewall termination region, ensuring that the entire epitaxial stack has excellent fatigue resistance under pulsed loads.

[0105] In the technical solution of this invention, a composite structure combining efficient leakage suppression and active stress management is formed by collaboratively constructing a stress relief band on top of the main leakage suppression layer. The main leakage suppression layer forms a clear and robust potential barrier above the gradient layer, while the flexible relief band on top effectively absorbs and dissipates the stress generated by thermal mismatch and polarization effects, preventing it from backtracking to the main leakage suppression layer under repeated pulse loads, thereby significantly improving the fatigue resistance and long-term dynamic stability of the device surface area.

[0106] S4: On top of the main leakage suppression layer and the stress relief zone, a terminal adaptation layer and a sidewall termination zone are formed.

[0107] After completing the synergistic construction of the main drain suppression layer and the top stress relief band, the next stage is the layering of the termination layer and the sidewall termination region. By depositing the equivalent epitaxial bands of the termination layer and the sidewall termination region, the stress distribution and electric field management in the device edge region are further optimized, thereby effectively suppressing the formation of lateral free paths and fatigue points induced by pulsed loads at the device edge. The termination layer is mainly used to balance the stress backflow that may occur in the thermal history of subsequent processes, while the sidewall termination region diffuses the edge energy distribution to the low-stress region by forming a localized doping and thickness fine-tuning region in the upper region of the outer edge of the device, avoiding material fatigue from occurring first on the surface or at the edge.

[0108] S4.1: Determining the lower bound and thickness planning of the terminal adaptation layer.

[0109] Based on parameter set P4, which includes the thickness of the complete upper covering structure. , Flag_relax (signature indicating completion of the slow-release layer), and the top coordinates of the overlay structure. and the coordinate alignment record of the main layer and the sustained-release layer When Flag_relax is in the pass state, the top coordinates of the wrapper structure will be used. The lower geometric boundary of the terminal adaptation layer, i.e., the lower boundary coordinates of the terminal adaptation layer, confirms that the structure meets the conditions for entering the terminal layer construction. The design thickness of the terminal adaptation layer... The setting is within the range of 0.1 to 0.3 μm, and is achieved through a scaling factor. Related to the thickness of the upper covering structure, it satisfies the formula ,in The value ranges from 0.1 to 0.3. Simultaneously, the design goal of this adapter layer is to effectively buffer stress changes caused by thermal history in subsequent processes; this stress buffering capacity will serve as one of the reference criteria for the design of the sidewall termination area. This step completes the final surface finishing in the vertical direction, providing a low-stress, highly compatible interface for the sidewall termination area. All planning parameters, including the design thickness of the terminal adapter layer, are considered. Lower bound coordinates They are integrated into parameter set T1.

[0110] S4.2: Actual deposition and parameter solidification of the terminal adaptation layer.

[0111] Based on the thickness and lower bound coordinates set in parameter set T1 With a stress-buffered target, epitaxial growth of the terminal adaptation layer is performed within the reaction chamber. The growth of this layer is controlled by adjusting the average growth rate. and sedimentation time To achieve this, the thickness is actually measured. From the formula The calculation yielded that, It is usually controlled within the range of 0.1 to 1 μm / min. Adjustments are made within tens of minutes based on the target thickness. Immediately after deposition, the actual thickness is measured using online metering methods, and the relative thickness deviation is calculated. Simultaneously, a membrane stress conversion algorithm was used to evaluate the remaining stress margin after deposition. This ensures that the stress value is below the upper limit of the load-bearing capacity of the buffer layer to avoid excessive stress accumulation. This step translates the geometric and stress characteristics of the terminal adaptation layer into verifiable physical parameters, providing a reliable stress boundary for the subsequent construction of the sidewall termination zone. All measured values ​​and evaluation results, including measured thickness, are supported. Thickness deviation and residual stress margin , is recorded as parameter set T2.

[0112] S4.3: Based on the deposition results of the terminal adaptation layer, the geometric unfolding and expansion radius of the sidewall termination zone are determined.

[0113] Parameter set T2 provides the actual thickness and stress state after the adaptation layer is completed. Based on this, the termination region is geometrically unfolded along the sidewall direction, with the unfolding radius... (Unit: μm) From residual stress margin Measured thickness of the adapter layer Jointly determined, satisfying the formula ,in The stress-geometric conversion factor (unit: μm / (MPa·μm)) ranges from 0.05 to 0.2 μm / (MPa·μm). The sidewall termination region is an equivalent extensional structure with an equivalent thickness of... Maintain consistency with the adaptation layer, i.e. This ensures geometric continuity without introducing new cross-sectional gradients. This step diffuses accumulated polarization energy from the edge points to the curve dissipation zone through geometric resolution, effectively reducing local stress peaks. All design parameters, including the design development radius of the termination zone, are considered. and design equivalent thickness They are integrated into parameter set T3.

[0114] S4.4: Depositional closure and boundary solidification in the termination zone.

[0115] Based on the geometric parameters set in parameter set T3, the termination region is extended along the sidewall direction above the terminal adapter layer. During deposition, growth conditions are controlled to ensure that the actual contour of the termination region matches the design radius and maintains its thickness continuity with the adapter layer. After deposition, the measured unfolded radius of the termination region is obtained through contour metrology. Equivalent thickness as measured The complete geometric profile of the termination region, obtained from measured parameters. and The final sealing curve set and boundary coordinates are recorded and solidified into a closed termination structure (i.e., edge sealing structure). Simultaneously, a boundary solidification flag (Flag_term) is set to indicate that the termination area has formed a sealing profile without exposed stress peaks. This step completely suppresses the potential discharge outlet of the device into a stress passivation effect zone, achieving the final closure of the suppression structure from the vertical to the sidewall. All results, including the complete geometric profile of the termination area and the sealing curve set composed of measured parameters (…), are recorded and solidified into a closed termination structure (…). , The boundary solidification identifier Flag_term, along with the other two, are integrated into parameter set T4, providing a complete and stable edge structure foundation for subsequent pulse condition verification steps.

[0116] Thus, the following is generated: Figure 2 The diagram shows a cross-sectional schematic of a GaN gradient-doped drain suppression epitaxial layer stack structure for high-power pulse devices.

[0117] In the technical solution of this invention, the electric field and stress management of the device edge region are optimized by depositing a terminal adapter layer and a sidewall termination region. The terminal adapter layer balances the stress backflow that may be introduced by subsequent processes, while the sidewall termination region dissipates the lateral leakage paths and high-stress areas that may be concentrated at the device edge to the low-stress zone through geometric expansion, effectively suppressing the formation of free breakdown paths at the edge and achieving complete closure of the leakage suppression structure from the vertical to the sidewall.

[0118] S5: Verify the steady-state heat treatment and leakage suppression capability of the structure after epitaxial growth under pulsed conditions.

[0119] After completing the terminal adaptation and layering of the sidewall termination region, the process enters the stage of steady-state heat treatment under pulsed operating conditions and leakage suppression capability verification. Through heat treatment and electrical tests simulating actual service conditions, the long-term stability and leakage suppression performance of the aforementioned epitaxial structure under repeated pulsed stress are verified.

[0120] S5.1: Perform pre-pulse stabilization heat treatment to bring the epitaxial structure into its actual service state.

[0121] Based on parameter set T4, which includes the complete geometric profile of the termination region, the final edge-sealing curve set composed of measured parameters, and the boundary curing marker Flag_term, when Flag_term is in the pass state, it is confirmed that the device edge structure has been completely closed. The sample is then placed in a controlled thermal treatment environment. Steady-state temperature. Set the stabilization time within a safe range below the highest thermal history of the back-end process of the device. The heat treatment process, controlled within the range of 10 to 30 minutes, promotes the redistribution of interfacial stress between the epitaxial layer and the substrate, and stabilizes the residual polarization in the material to a repeatable state. (The residual stress after heat treatment is also mentioned.) Stress allowance before heat treatment Stress attenuation coefficient during thermal equilibrium process Joint decision-making to satisfy the relationship ,in The value typically ranges from 0.5 to 0.9. This step transforms the epitaxial structure from a geometrically completed state to a stress equilibrium state close to actual service conditions, preventing false leakage suppression performance data from subsequent pulse tests due to lack of thermal equilibrium. All heat treatment parameters and results, including stabilization temperature, are included. steady-state time and residual stress after thermal stabilization , is recorded as parameter set V1.

[0122] S5.2: High duty cycle pulse loading and approaching the structural fatigue threshold.

[0123] Based on the post-thermal equilibrium stress state provided by parameter set V1, the sample is placed on a pulse testing platform, and electrical stress conditions matching the target application are applied. Pulse duty cycle. Set within the typical operating range of 30% to 80%, pulse rise time Under stringent conditions controlled to no more than 100 nanoseconds, the turn-off voltage was close to the device's design limit. The fatigue threshold of the epitaxial structure was progressively approached by gradually increasing the pulse load. During this process, the dynamic stress safety margin was evaluated. This parameter is determined by the material's thermal stability fatigue strength. Residual stress after heat treatment Equivalent instantaneous stress induced under pulsed operating conditions The formula is obtained through joint calculation. .when When the value exceeds the preset safety threshold, it indicates that the epitaxial structure has sufficient safety margin under the current pulse conditions and has not triggered the fatigue critical point. The system sets the fatigue-not-triggered flag Flag_dura to the pass state. This step confirms the ability of the edge-sealing structure to withstand dynamic stress before fracture occurs, effectively preventing fatigue-induced latent breakdown cores from tracing back to the main leakage suppression layer. All pulse test parameters and evaluation results, including pulse duty cycle, are included. Rising edge of the pulse Stress response ratio The fatigue-untriggered flag Flag_dura is integrated into parameter set V2.

[0124] S5.3: Confirmation of whether the leakage suppression path has been transferred to the edge sealing dissipation channel.

[0125] Based on the pulse test data in parameter set V2, the spatial distribution characteristics of leakage current are analyzed in detail. The leakage current density in the longitudinal central region is accurately measured under pulsed operating conditions. Discharge density in the edge termination region By calculating the sidewall inhibition coefficient To quantify the transfer effect of the leakage suppression path. If A value less than 1 indicates that the main path of the leakage current has been successfully transferred from the vertical direction to the edge termination region, proving that the leakage suppression mechanism is no longer simply reducing the leakage value, but has achieved geometric reconstruction of the leakage path. At this time, the system sets the leakage suppression effective flag Flag_block to the pass state. This step verifies the energy dissipation function of the sidewall termination region from the perspective of current distribution and confirms the synergistic leakage suppression effect of gradient epitaxy and the termination structure. All leakage characteristic parameters, including post-pulse leakage current density, are recorded. Sidewall inhibition coefficient And the effective flag for leakage suppression, Flag_block, is recorded as parameter set V3.

[0126] S5.4: The ability to ultimately suppress closed-loop control is established.

[0127] Based on all validation data in parameter set V3, the system performs a final pass / fail determination. Passability requires simultaneously meeting two core indicators: sidewall inhibition coefficient. Less than 1, and the ratio of structural fatigue resistance The value is greater than 1. When both conditions are met, the system sets the final structure qualification flag Flag_final to qualified and generates a flag Flag_pkg indicating that the epitaxial structure can safely enter the subsequent metallization and device fabrication process without additional compensation. Simultaneously, the system outputs a complete closed-loop verification report, detailing the entire process data and judgment results from heat treatment to pulse testing. This step transforms the leakage suppression capability of the epitaxial layer under pulse fatigue from theoretical speculation to a physically verified closed loop, providing a solid guarantee for the high reliability of the device. All final judgment results and report data are integrated into parameter set V4, marking the successful completion of the entire process of the GaN gradient doping leakage suppression epitaxial method for high-power pulsed devices.

[0128] In the technical solution of this invention, the long-term leakage suppression capability and mechanical stability of the epitaxial structure were finally verified through pulsed steady-state heat treatment and electrical testing simulating actual service conditions. Heat treatment induces the material stress state to tend towards equilibrium, while high duty cycle pulse loading directly verifies the fatigue resistance threshold of the structure under dynamic stress. By analyzing the spatial distribution of leakage current, it was confirmed that the leakage suppression path has been successfully transferred to the edge dissipation channel, thus experimentally confirming the leakage suppression effectiveness and reliability jointly constructed by all previous epitaxial steps.

[0129] In summary, this invention fundamentally alters the vertical electric field distribution of the device by introducing a gradient transition layer with continuously increasing vertical doping concentration. This structure disperses the electric field peaks concentrated at a single interface in traditional uniform or abrupt doping designs into a thicker transition region, achieving a smooth electric field transition and effective dissipation. This significantly reduces the risk of material breakdown caused by rapid electric field redistribution during high-power pulse turn-off. By collaboratively constructing a stress relief band on top of the main drain suppressor layer, a composite functional structure for electric field management and stress dissipation is formed. Not only does the main drain suppressor layer establish a robust potential barrier, but the flexible layer at the top effectively absorbs and releases accumulated stress generated by rapid polarization and thermal mismatch, thus significantly suppressing material fatigue and structural collapse in the surface region under repeated pulse stress. Through the synergistic layering of the termination adapter layer and the sidewall termination region, refined management of the electric field and stress in the device's edge region is achieved. This structure effectively eliminates the lateral high electric field region and stress concentration points at the device's peripheral edges, transforming potential edge leakage paths into controllable energy dissipation channels. This avoids premature failure caused by pulsed loads in the device's edge regions, ensuring the integrity of leakage suppression capability in three-dimensional space. Through continuous parametric process control and final pulsed operating condition verification, a closed-loop quality control system from material growth to performance confirmation is formed. This method ensures that each functional layer can be constructed under strict geometric and electrical specifications. Furthermore, through testing simulating real-world service conditions, the long-term leakage suppression capability and reliability of the epitaxial structure under dynamic stress are directly verified, providing a reliable epitaxial foundation for manufacturing high-reliability, high-power pulsed devices.

[0130] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A GaN gradient doping drain suppression epitaxial method for high-power pulsed devices, characterized in that, include: A lightly doped gallium nitride buffer layer is epitaxially grown sequentially on a semiconductor substrate; On the low-doped gallium nitride buffer layer, a gradient transition layer with continuously increasing doping concentration from bottom to top is epitaxially grown. The gradient transition layer is used to disperse the electric field distribution in the vertical direction and achieve a smooth transition of interlayer stress. On the gradient transition layer, a main leakage suppression layer and a stress relief band located on top of it are epitaxially grown in synergy. The main leakage suppression layer forms a potential barrier, and the stress relief band absorbs and dissipates accumulated stress. On top of the main leakage suppression layer and stress relief band, a terminal adapter layer and a sidewall termination region are formed to create an edge sealing structure to manage the stress distribution and electric field in the edge region of the device. The structure that has completed epitaxial growth was subjected to steady-state heat treatment under pulsed conditions and its leakage suppression capability was verified. The dynamic stress safety margin and the spatial distribution of leakage current were evaluated by applying a high duty cycle pulse load, and the transfer of the leakage suppression path to the edge dissipation channel was verified.

2. The GaN gradient doping drain suppression epitaxial method for high-power pulsed devices according to claim 1, characterized in that, The method for sequentially epitaxially growing a lightly doped gallium nitride buffer layer on a semiconductor substrate specifically includes: The substrate is chemically cleaned, and the surface roughness and residual content parameters are obtained by monitoring the cleaning time and solution ratio. Degassing and thermal stabilization treatments are then performed, and the average desorption rate of the desorption process is calculated using the Arrhenius equation to guide the heat treatment parameters. The lightly doped gallium nitride buffer layer was deposited on the surface of the processed substrate, and the interfacial stress between the epitaxial layer and the substrate was estimated by the substrate curvature method after deposition. ; The crystal quality and electrical properties of the lightly doped gallium nitride buffer layer were evaluated, and the dislocation density was measured. With sheet resistor When the measured value is within the preset acceptable range, the final thickness of the low-doped gallium nitride buffer layer is determined. Interfacial stress and preset stress threshold Solidify the gradient boundary conditions for subsequent gradient layer design .

3. The GaN gradient doping drain suppression epitaxy method for high-power pulsed devices according to claim 2, characterized in that, The method for epitaxially growing a gradient transition layer with continuously increasing doping concentration from bottom to top specifically includes: Based on the gradient boundary conditions and sheet resistors Plan a gradient layer structure with multiple segments and linearly increasing doping concentration in each segment, and determine the number of segments. Total target thickness and the target thickness of each segment and target doping concentration ; Based on material stress tolerance constraints, a layer-weighted model is used to calculate the cumulative equivalent internal stress of multi-layer structures. While maintaining the overall target thickness While maintaining the same thickness, the thickness of each segment is corrected to ensure the cumulative equivalent internal stress. Not exceeding the stress threshold ; The corrected design parameters are converted into epitaxial process instructions, and epitaxial deposition of each segment is executed sequentially. The target doping concentration is achieved by controlling the equivalent supply of the doped precursor and the growth time. The actual deposition thickness of each segment is measured in real time using an online metrology instrument. and actual doping concentration This enables closed-loop control of the deposition process.

4. The GaN gradient doping drain suppression epitaxial growth method for high-power pulsed devices according to claim 3, characterized in that, When planning the gradient layer structure, the doping concentration of each segment increases linearly according to the calculation formula: in, For segment numbers, The doping concentration of the first segment is set as the baseline and is lower than the doping concentration of the subsequent main drain suppressor layer. This represents the doping concentration increment between two adjacent segments; Meanwhile, the total target thickness of the gradient layer structure Equal to the target thickness of each segment sum.

5. The GaN gradient doping drain suppression epitaxy method for high-power pulsed devices according to claim 3, characterized in that, After the gradient transition layer is deposited, the consistency of the gradient structure is also determined. Calculate the thickness uniformity index of the gradient transition layer and doping uniformity index : in, This represents the maximum measured thickness of all segments in the gradient transition layer. The minimum measured thickness of all segments in the gradient transition layer. For the final total thickness, The average doping concentration, This represents the maximum measured doping concentration across all segments of the gradient transition layer. This represents the minimum measured doping concentration across all segments of the gradient transition layer. Verify whether the measured doping concentration of each segment maintains a monotonically increasing relationship from bottom to top, and whether the total thickness and the thickness of each segment are within the acceptable tolerance range; When uniformity, monotonicity, and thickness all meet the requirements, the top boundary coordinates of the gradient transition layer will be... Final doping concentration of the top section and total thickness Solidification serves as the geometric lower boundary and parameter basis for the deposition of the main leakage suppression layer.

6. The GaN gradient doping drain suppression epitaxy method for high-power pulsed devices according to claim 5, characterized in that, The method for synergistic epitaxial growth of the main leakage suppression layer and the stress relief band located on top of it specifically includes: Based on the top boundary coordinates of the gradient transition layer The design thickness of the main leakage suppression layer is set based on the lower stress accumulation margin. With the target doping range The main leakage suppression layer is epitaxially deposited on the gradient transition layer; The top boundary coordinates of the gradient transition layer The lower geometric boundary of the main leakage suppression layer is locked; the design thickness of the main leakage suppression layer is determined based on the lower stress accumulation margin. The preset top boundary coordinates of the main leakage suppression layer are derived based on the lower geometric boundary and the designed thickness. To satisfy Geometric relationships; After the main leakage suppression layer is deposited, its actual thickness is measured. Based on this measured thickness The design includes a stress relief band at its top, with a doping concentration of [missing information]. The doping concentration is set to be lower than that of the main drain suppressor layer to enhance lattice compliance; The thickness of the stress relief band Through the proportionality coefficient The measured thickness of the main leakage suppression layer The association satisfies the following relation: ; Deposit the stress relief zone to form a coating structure covering the main leakage suppression layer, and record the coordinates of the top of the coating structure. This serves as the starting surface for the lower geometric boundary of the terminal adaptation layer.

7. The GaN gradient doping drain suppression epitaxial growth method for high-power pulsed devices according to claim 6, characterized in that, The method for forming the terminal adaptation layer and the sidewall termination region specifically includes: An epitaxial terminal adapter layer is deposited on the stress relief zone, the designed thickness of the terminal adapter layer being... Through the proportionality coefficient With the total thickness of the covering structure The association satisfies the following relation: ; After the terminal adapter layer is deposited, the residual stress margin after deposition is evaluated. And based on this remaining stress margin Determine the geometric unfolding radius of the sidewall termination region ; The sidewall termination region is formed by extending outward along the sidewall direction, so that the equivalent thickness of the sidewall termination region is consistent with the terminal adapter layer. Thus, the sidewall termination region and the terminal adapter layer are used to close at the outer edge of the device to form the edge sealing structure, which transforms the potential edge leakage path into an energy dissipation channel.

8. The GaN gradient doping drain suppression epitaxy method for high-power pulsed devices according to claim 7, characterized in that, The geometric unfolding radius of the sidewall termination region The calculation formula is: in, The measured thickness of the terminal adapter layer. The stress-geometric conversion factor is used to diffuse the accumulated polarization energy from the edge point to the curve consumption zone through geometric expansion.

9. The GaN gradient doping drain suppression epitaxy method for high-power pulsed devices according to claim 1, characterized in that, The method for performing the aforementioned stabilization heat treatment and leakage suppression capability verification specifically includes: The epitaxial structure was subjected to steady-state heat treatment in a controlled environment, and the steady-state temperature was set. and steady-state time This stabilizes the polarization residues in the material and reduces the residual stress after heat treatment. It is determined by both the stress allowance before heat treatment and the stress attenuation coefficient during the thermal equilibrium process; A high duty cycle pulse load is applied to the heat-treated structure, wherein the pulse duty cycle is... Set between 30% and 80%, the dynamic stress safety margin is calculated. Assess whether material fatigue has been triggered; The dynamic stress safety margin The calculation is based on the thermal stability fatigue strength of the material. Residual stress after heat treatment and equivalent instantaneous stress under pulsed conditions Satisfying the relation: ; Simultaneously, the spatial distribution of leakage current was analyzed, and the sidewall suppression coefficient was calculated. To verify the effect of the leakage suppression path transfer.

10. The GaN gradient doping drain suppression epitaxy method for high-power pulsed devices according to claim 9, characterized in that, The sidewall inhibition coefficient Defined as the discharge density of the longitudinal central region Discharge density in the edge termination region The ratio; when the sidewall inhibition coefficient Less than 1 and the dynamic stress safety margin When the value is greater than 1, the epitaxial structure is deemed qualified, confirming that the main path of leakage current has been transferred from the vertical direction to the edge termination region, and the long-term stability of the leakage suppression function under repeated pulse stress has been achieved.

Images