Multi-material coupled recycling system and multi-material coupled recycling method for silicon carbide crystal growth
Patent Information
- Application Number
- CN202610909094.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-06-23
- Publication Date
- 2026-08-18
AI Technical Summary
[0005]为了解决上述技术问题,本发明提出用于碳化硅晶体生长的多物料耦合循环利用系统及多物料耦合循环利用方法,旨在解决现有技术中因采用回收粉料,导致批次间晶体电阻率一致性差的问题
[0018] Beneficial effects: Compared with existing technologies, the system provided by this invention effectively suppresses resistivity dispersion caused by fluctuations in nitrogen impurities in the powder through feedforward detection of powder nitrogen content and dynamic gas flow compensation. While ensuring crystal quality, it achieves a high proportion of recycled powder, increasing the average utilization rate of recycled material from 30% to 45%-60%, significantly reducing raw material costs. This invention is mainly implemented through software algorithms and control strategies, and can be directly integrated into existing PVT growth systems without requiring expensive hardware modifications to the growth furnace, making it easy to promote in engineering.
Smart Images

Figure CN122588674A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of silicon carbide single crystal preparation technology, and in particular to a multi-material coupling recycling system and method for silicon carbide crystal growth. Background Technology
[0002] Silicon carbide, as a core material of third-generation semiconductors, has become an ideal material for manufacturing high-temperature, high-frequency, and high-power electronic devices due to its excellent properties such as large bandgap, high thermal conductivity, and high breakdown field strength.
[0003] Physical vapor transport (PVT) is the mainstream technology for preparing silicon carbide single crystals. During PVT growth, nitrogen doping is a key method for controlling the electrical properties (especially resistivity) of silicon carbide crystals. Since high-purity silicon carbide powder often contains trace amounts of nitrogen impurities, and the ungrown source powder and byproducts after crystal growth contain a large amount of valuable raw materials, the industry commonly attempts to recycle these powders. However, the nitrogen impurity content in the recycled powder is often difficult to control precisely, posing a significant challenge to the consistency of resistivity between batches. The nitrogen content of the recycled powder is affected by factors such as preceding growth conditions and high-temperature purification processes, resulting in significant batch-to-batch variations. This makes it difficult to guarantee the resistivity consistency of products between batches in the absence of an effective feedforward detection mechanism.
[0004] Therefore, the existing technology still needs further development and improvement. Summary of the Invention
[0005] To address the aforementioned technical problems, this invention proposes a multi-material coupling recycling system and method for silicon carbide crystal growth, aiming to solve the problem of poor batch-to-batch crystal resistivity consistency caused by the use of recycled powder in existing technologies. Specifically: In a first aspect, a multi-material coupling recycling system for silicon carbide crystal growth includes: The powder recycling unit is used to purify the remaining waste after crystal growth to provide recycled powder. The virgin material supply unit is used to store and transport virgin silicon carbide material; The powder detection unit is at least connected to the output of the recycled powder processing unit and is used for online detection of the nitrogen content of the recycled powder. ; The central control unit is connected to the powder detection unit and is used to receive signals detected online. And obtain the nitrogen content of the virgin silicon carbide material. It has a built-in strategy engine layer, which is based on the... , and the preset target nitrogen concentration The nitrogen flow rate correction value used to adjust gas-phase doping was calculated. and instructions for adjusting powder proportions; The mixing and loading unit is connected to the recycled powder processing unit, the new material supply unit and the central control unit respectively, and includes an automatic mixer for mixing the new material and the recycled powder according to the powder proportioning instruction at a set mass ratio; The flow control unit, signal-connected to the central control unit, includes a digital mass flow controller for receiving the nitrogen flow correction value. And adjust the actual nitrogen flow rate input to the crystal growth unit in real time.
[0006] The following are preferred technical solutions of the present invention, but are not intended to limit the technical solutions provided by the present invention. The purpose and beneficial effects of the present invention can be better achieved and realized through the following preferred technical solutions.
[0007] As a preferred technical solution, the multi-material coupling recycling system further includes: The crystal growth unit, which is connected to the mixing and charging unit and the flow control unit, includes a PVT growth furnace for growing silicon carbide single crystals under the conditions of the adjusted actual nitrogen flow rate and the mass ratio of the mixed powder. A gas recovery unit is provided, with its pipeline connected between the exhaust port of the crystal growth unit and the flow control unit. The gas recovery unit is equipped with a purification system and an online purity monitoring function to purify the process gas discharged during the growth process and then send it to the flow control unit for recycling.
[0008] As a preferred technical solution, the multi-material coupling recycling system further includes: The substrate processing unit, including a multi-wire dicing machine and a chemical mechanical polishing device, is used to process grown silicon carbide crystals into standard substrate samples. The substrate detection unit, which is signal-connected to both the substrate processing unit and the central control unit, includes a non-contact resistivity meter and a secondary ion mass spectrometer, used to measure the average resistivity of the standard substrate sample. ρ avg The measured nitrogen concentration is then input to the central control unit as a closed-loop feedback signal; The crystal growth unit is also equipped with a temperature field sensor and a pressure sensor to monitor the axial temperature gradient and chamber background pressure in the PVT growth furnace in real time, and transmit the monitoring data signals to the central control unit.
[0009] Secondly, a method for multi-material coupling recycling based on the multi-material coupling recycling system described in the first aspect, comprising the steps of: Before the powder is loaded into the furnace, the nitrogen content of the current batch of recycled powder is determined online by the powder detection unit, and the nitrogen content of the new material is obtained. The central control unit according to the The above Target nitrogen concentration and reference nitrogen flow rate The nitrogen flow correction value was calculated. And generate powder mixing instructions; The mixing and loading unit responds to the powder proportioning command by mixing the new material output from the new material supply unit with the recycled powder output from the recycled powder processing unit according to a set mass ratio and loading it into the crystal growth unit. The flow control unit responds to the Adjust the actual nitrogen flow rate input to the crystal growth unit, and run the PVT crystal growth process under the adjusted actual nitrogen flow rate and the mass ratio of the mixed powder. Thirdly, a multi-material coupling recycling method, wherein the method is applied to the central control unit described in the first aspect, and includes the following steps: Receive the nitrogen content of the current batch of recovered powder, measured online by the powder detection unit. And read the nitrogen content of the current batch of new material. ; Combined with the preset target nitrogen concentration and reference nitrogen flow rate The nitrogen flow correction value is calculated using the built-in doping compensation model. ; Run the internal proportioning decision engine, based on the above Or the equivalent nitrogen concentration obtained by calculation Match the corresponding material mixing ratio to generate powder mixing instructions; The powder proportioning command is sent to the mixing and loading unit to drive the automatic mixer to perform the mixing and loading of new material and recycled powder; The The signal is sent to the flow control unit to drive the digital mass flow controller to perform gas phase doping flow regulation.
[0010] As a preferred technical solution, the multi-material coupling recycling method includes calculating the nitrogen flow correction value. The doping compensation model is a linear compensation model, and the central control unit calculates according to the following formula:
[0011] in, The nitrogen compensation coefficient for the recovered powder is set to a value ranging from 0.005 to 0.02. The nitrogen compensation coefficient for the new material ranges from 0.002 to 0.01.
[0012] As a preferred technical solution, the multi-material coupling recycling method includes calculating the nitrogen flow correction value. The doping compensation model is a weighted summation model, and the central control unit calculates according to the following formula:
[0013]
[0014] in, The nitrogen weighting coefficient for recovered powder; This is the nitrogen weighting coefficient for the new material, and , : This is the total compensation coefficient.
[0015] As a preferred technical solution, in the multi-material coupling recycling method, the decision logic for the central control unit to generate the powder proportioning instruction is: directly based on the nitrogen content of the recovered powder. The weight ratio of recycled powder to virgin material in the specified range is adjusted as follows: when When the concentration is ≤30ppm, the proportion of recycled powder is set at 60% and the proportion of virgin material is set at 40%. when When the concentration is ≤50ppm, the proportion of recycled powder is set at 45% and the proportion of virgin material is set at 55%. when When the concentration is ≤70ppm, the proportion of recycled powder is set at 30% and the proportion of virgin material is set at 70%. when When the concentration is >70ppm, the mass ratio of recycled powder is set at 10% and the mass ratio of virgin material is set at 90%.
[0016] As a preferred technical solution, in the multi-material coupling recycling method, the decision logic for the central control unit to generate the powder proportioning instruction is as follows: based on the calculated equivalent nitrogen concentration... The weight ratio of recycled powder to virgin material in the specified range is adjusted as follows: when When the concentration is ≤30ppm, the proportion of recycled powder is set at 60% and the proportion of virgin material is set at 40%. when When the concentration is ≤50ppm, the proportion of recycled powder is set at 45% and the proportion of virgin material is set at 55%. when When the concentration is ≤70ppm, the proportion of recycled powder is set at 30% and the proportion of virgin material is set at 70%. when When the concentration is >70ppm, the mass ratio of recycled powder is set at 10% and the mass ratio of virgin material is set at 90%.
[0017] As a preferred technical solution, the multi-material coupling recycling method further includes a cross-furnace closed-loop learning and optimization step: The central control unit acquires the average resistivity of the standard substrate sample from the previous growth furnace, as measured by the substrate detection unit described above. ρ avg And calculate its deviation from the target resistivity; When the deviation value of two consecutive furnaces both exceed a preset threshold of 10% At 15%, an emergency risk blocking mechanism is activated, forcibly switching the proportion of recycled powder in the powder mixing instruction to 0%; simultaneously, the built-in fully connected neural network model is invoked for assisted online inference, and the... , The historical temperature field curves of the PVT growth furnace monitored by the aforementioned temperature and pressure sensors, along with the chamber background pressure and the current number of crucible uses, are input into the fully connected neural network model. The output layer then outputs a nonlinear flow rate to compensate for the deviation Δ. F N2 ; The central control unit utilizes the Δ F N2 For the original Algebraic superposition correction is performed to generate the final target flow command, which is then sent to the flow control unit to execute secondary closed-loop fine-tuning control for gas-phase doping.
[0018] Beneficial effects: Compared with existing technologies, the system provided by this invention effectively suppresses resistivity dispersion caused by fluctuations in nitrogen impurities in the powder through feedforward detection of powder nitrogen content and dynamic gas flow compensation. While ensuring crystal quality, it achieves a high proportion of recycled powder, increasing the average utilization rate of recycled material from 30% to 45%-60%, significantly reducing raw material costs. This invention is mainly implemented through software algorithms and control strategies, and can be directly integrated into existing PVT growth systems without requiring expensive hardware modifications to the growth furnace, making it easy to promote in engineering. Attached Figure Description
[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or related technologies, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0020] Figure 1 A diagram of a multi-material coupling and recycling system for silicon carbide crystal growth provided in an embodiment of the present invention.
[0021] Figure 2 This is a schematic diagram of a multi-material coupling recycling method for silicon carbide crystal growth provided in an embodiment of the present invention.
[0022] Figure 3 A linear compensation model diagram provided for an embodiment of the present invention.
[0023] Figure 4 This is a diagram of a weighted summation model provided in an embodiment of the present invention.
[0024] Figure 5 The schematic diagram of the central control unit provided in the embodiment of the present invention.
[0025] The labels in the attached diagram represent the following: 101, gas station; 102, gas recovery unit; 103, flow control unit; 104, central control unit; 105, powder detection unit; 106, recycled powder processing unit; 107, new material supply unit; 108, mixing and loading unit; 109, crystal growth unit; 110, substrate processing unit; 111, substrate detection unit. Detailed Implementation
[0026] The embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and should not be construed as limiting the scope of the invention.
[0027] In the description of the embodiments of the present invention, it should be noted that the terms "center," "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing the embodiments of the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the embodiments of the present invention. In addition, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.
[0028] In the description of the embodiments of the present invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "connected" and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms in the embodiments of the present invention based on the specific circumstances.
[0029] In embodiments of the present invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "on top of," and "over" the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.
[0030] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.
[0031] like Figure 1As shown in the embodiment of the present invention, a multi-material coupling recycling system for silicon carbide crystal growth includes: a gas station 101 for providing a mixed gas of high-purity argon and nitrogen. Its components include an argon source, a nitrogen source, a mass flow controller (MFC), and a mixing chamber. The gas station is characterized by its nitrogen flow rate being precisely adjustable within the range of 2 to 15 sccm, with a response time of no more than 1 second, enabling rapid response to flow correction commands from the central control unit. Under baseline operating conditions, the argon flow rate is set to 100 sccm as the carrier gas, and the nitrogen flow rate is set to 8 sccm as the dopant source. The mixed gas output from this module enters the crystal growth unit, providing the required atmospheric environment for PVT growth. A gas recovery unit 102 is used to collect, purify, and recycle the process gas discharged during the growth process. Its components include a gas collection pipeline, a filter device, a purification system, and a compressed storage tank. Its technical features include a recovery rate ≥85%, a purified gas purity ≥99.999%, and online purity monitoring functionality. This unit significantly reduces the consumption of argon and nitrogen, achieving green production. Simultaneously, the nitrogen content information in the recovered gas serves as an auxiliary reference for doping control. While recycling the gas, the gas recovery unit analyzes the nitrogen concentration in the recovered gas in real time through an online purity monitoring system. This data indirectly reflects the changing trend of nitrogen content released from the powder during growth and can serve as a cross-validation basis for the measurement results of the powder detection unit. When significant deviations occur between the data from the powder detection unit and the gas recovery unit, the system can automatically trigger a calibration procedure or issue an early warning signal. This multi-source data cross-validation mechanism further enhances the reliability of nitrogen doping control.
[0032] The flow control unit 103 receives nitrogen flow correction commands from the central control unit 104 and adjusts the actual nitrogen flow in real time, while simultaneously controlling the mixing ratio of the recovered gas. It comprises a digital mass flow controller, a PID controller, a flow sensor, and a mixing ratio regulating valve. The technical advantage of this unit is that it achieves millisecond-level precise control of the nitrogen flow rate, with flow fluctuations ≤ ±0.1 sccm, and supports precise ratio control of the gas station gas and the recovered gas. The PID controller adjusts the valve opening of the mass flow controller in real time based on the deviation between the flow setpoint and the measured value from the flow sensor, ensuring the stability and accuracy of the nitrogen flow rate. It is readily understood that the central control unit and the flow control unit can be integrated in this invention to form an integrated intelligent flow controller.
[0033] The central control unit 104 is the core decision-making hub of the system. It executes the nitrogen doping control algorithm and generates flow correction instructions, powder proportioning instructions, and gas recovery control instructions.
[0034] Specifically, such as Figure 5As shown, its components include a data acquisition layer. During the data acquisition process, the nitrogen content of the recovered powder and the nitrogen content of the new material, measured by the material detection unit, are transmitted to the data acquisition layer of the central control unit via a data bus. The data acquisition layer performs preprocessing operations such as filtering and outlier removal on the raw data to ensure the accuracy and reliability of the data. The preprocessed data is then passed to the strategy engine layer, where the built-in compensation model calculates the nitrogen flow correction value. The decision output layer generates specific control commands based on the calculation results, including a nitrogen flow correction command (sent to the MFC of the flow control unit in analog or digital form), a powder proportioning command (sent to the automatic mixer of the mixing and charging unit), and a gas recovery ratio command (sent to the proportional control valve of the gas recovery unit).
[0035] Furthermore, it also includes a learning optimization layer responsible for recording the process parameters and detection results of each furnace run, establishing a historical database, and providing a data foundation for cross-furnace learning. (It receives nitrogen content data of powder from the powder detection unit, nitrogen concentration data of gas from the gas station and gas recovery unit, and resistivity and nitrogen concentration data from the substrate detection unit), a strategy engine layer (with a built-in linear compensation model or weighted summation model, performing doping concentration calculation and flow compensation), a decision output layer (generating nitrogen flow correction instructions, powder proportioning instructions, and gas recovery ratio instructions), and a learning optimization layer (adaptively adjusting model parameters based on cross-furnace feedback data).
[0036] The powder detection unit 105 is used for online detection of the nitrogen content of recycled powder and virgin material, providing data support for powder proportioning decisions and nitrogen flow compensation. By obtaining nitrogen contamination information in the powder in advance, it provides a basis for subsequent flow compensation. Unlike existing methods that rely solely on gas doping, this invention achieves end-to-end control from the powder source to gas doping through this unit, avoiding uncontrolled doping during the growth process.
[0037] The recycling powder processing unit 106 is used to collect, crush, screen, and purify the waste materials after growth to prepare recyclable recycled powder. It consists of a waste collection device, a crusher, a screening machine, and a high-temperature purification furnace. Its processing capacity is 5-20 kg / batch, the particle size control range is 50 to 200 micrometers, the purification temperature is 1200 to 1500 degrees Celsius, and the recovery rate is ≥90%. High-temperature purification reduces the impurity content in the recycled powder.
[0038] The virgin material supply unit 107 is used to provide high-purity silicon carbide virgin material, which is mixed with recycled powder in a certain proportion. It consists of a virgin material storage tank, a metering device and a conveying system. The purity of the virgin material is ≥99.999%, the nitrogen content is controllable within the range of 10 to 50 ppm, and the metering accuracy is ±0.1g, providing a high-quality raw material basis for powder formulation.
[0039] The mixing and loading unit 108 is used to mix new materials and recycled materials in a set ratio according to the instructions of the central control unit and load them into the growth system. It consists of a precision electronic scale, an automatic mixer, a loading robotic arm, and a mixing uniformity detection device. The mixing accuracy is ±1%, the loading repeatability is ±0.5 g, and the mixing uniformity is ≥95%.
[0040] Crystal growth unit 109 is used to grow silicon carbide single crystals under adjusted gas flow rate, gas recovery ratio, and powder ratio. It consists of a PVT growth furnace, a temperature control system, a pressure control system, and an exhaust system. It employs the physical vapor transport method (PVT), with a growth temperature of 2200–2400 degrees Celsius, a growth rate of 0.2–0.5 mm / h, and a chamber pressure of 10–100 mbar. During PVT growth, maintaining an accurate temperature gradient and supersaturation is crucial for crystal growth. The powder sublimates into a gaseous state in the high-temperature region (approximately 2300 degrees Celsius) and is transported to the seed crystal in the low-temperature region for recrystallization under the influence of the temperature field. Nitrogen gas in the growth chamber decomposes into active nitrogen atoms at high temperatures, which enter the silicon carbide lattice and replace carbon atom sites, achieving n-type doping. The partial pressure of nitrogen directly determines the nitrogen doping concentration in the crystal.
[0041] The substrate processing unit 110 is used for precision processing of grown silicon carbide crystals, including cutting, grinding, and polishing, to prepare standard substrates that meet testing requirements. During substrate processing, the cut silicon carbide wafers undergo rough grinding, fine grinding, and chemical mechanical polishing (CMP) sequentially. Each process has quality inspection points: after cutting, the wafer thickness and warpage are inspected; after grinding, the surface roughness is inspected; and after polishing, surface defects and damage layer depth are inspected. Only wafers that pass all inspections enter the substrate testing unit for quantitative measurement of resistivity and nitrogen concentration. The unit consists of a multi-wire dicing machine, a grinding machine, a CMP machine, a cleaning device, and a surface quality inspector. The cutting accuracy is ±0.1 mm, the surface roughness is ≤0.5 nm, the thickness uniformity is ±5 μm, and the processing yield is ≥95%.
[0042] The substrate detection unit 111 is used to detect the average resistivity and nitrogen concentration of the substrate after processing the grown crystal. It comprises a dicing machine, a polishing machine, a non-contact resistivity meter, and a secondary ion mass spectrometer (SIMS). The resistivity is measured using a non-contact method, with a measurement range of 0.001 to 10000 Ω·cm. The substrate detection unit uses a non-contact resistivity meter to measure resistivity, avoiding surface damage and measurement errors that may be introduced by probe contact. SIMS detection is performed in an ultra-high vacuum environment, using a primary ion beam to bombard the sample surface to generate secondary ions, and mass spectrometry analysis to obtain information on the depth distribution of nitrogen. The two measurement methods complement each other; resistivity measurement reflects the macroscopic electrical homogeneity of the crystal, while SIMS provides precise data on microscopic doping concentration. It should be noted that a distributed control system (DCS) can also be used, where multiple growth furnaces share a central database to achieve global optimization.
[0043] Furthermore, the system provided by this invention can be used not only for nitrogen doping but also for controlling other doping elements (such as aluminum and vanadium). It can also be applied to other crystal growth methods (such as solution methods and vapor phase epitaxy).
[0044] like Figures 2 to 4 As shown in the figure, this embodiment of the invention also provides a method for multi-material coupling recycling of a multi-material coupling recycling system, comprising: S101. Before the powder is loaded into the furnace, the nitrogen content of the current batch of recycled powder is determined online by the powder detection unit, and the nitrogen content of the new material is obtained.
[0045] Specifically, the gas station provides 100 sccm of argon and 8 sccm of reference nitrogen to create the protective atmosphere required for growth. Before the powder is loaded into the furnace, the powder detection unit monitors the nitrogen content of the recovered powder and the nitrogen content of the new material online. This step is a feedforward control point, providing a basis for subsequent flow compensation by obtaining information on nitrogen contamination in the powder in advance.
[0046] S102, the central control unit according to the... The above Target nitrogen concentration and reference nitrogen flow rate The nitrogen flow correction value was calculated. It also generates powder mixing instructions.
[0047] Specifically, the strategy engine layer built into the central control unit 104 acquires the nitrogen content of the recovered powder measured online. And the nitrogen content of the current batch of new material. Then, combined with the preset target nitrogen concentration and reference nitrogen flow rate The built-in doping compensation model is used to perform calculations to obtain accurate nitrogen flow correction values. Simultaneously, the internal proportioning decision engine runs to dynamically generate corresponding powder proportioning instructions.
[0048] As an optional embodiment of the present invention, the doping compensation model can be a linear compensation model. In this case, the central control unit calculates the nitrogen flow correction value according to the following formula. :
[0049] in, The nitrogen compensation coefficient for the recovered powder ranges from 0.005 to 0.02, and is preferably 0.01 in this embodiment; The value is the nitrogen compensation coefficient for the new material, ranging from 0.002 to 0.01, and preferably 0.005 in this embodiment.
[0050] The nitrogen reference flow rate In this embodiment, the target nitrogen concentration is set to 8 sccm (or 10 sccm under some baseline conditions). The preferred value is 90±5 ppm. The physical meaning of this linear compensation model is: when the nitrogen content of the recovered powder... greater than the target nitrogen concentration When the second term in the formula is positive, the calculated result is... The nitrogen content in the feedstock is reduced, thus adaptively decreasing the amount of nitrogen replenishment in the gas phase; similarly, when the nitrogen content of the feedstock is reduced... Greater than When the third term is positive, it makes Further reduction. The two factors are independently superimposed, achieving dual feedforward compensation for "background nitrogen" pollution introduced by the solid powder source. The model has simple calculation logic and fast response speed, making it particularly suitable for production lines with relatively fixed powder ratios. Accordingly, when running the linear compensation model, the decision logic for the central control unit to generate powder ratio instructions is: directly based on the nitrogen content of the recovered powder. The range of control is used to gradually adjust the mass ratio of recycled powder to virgin material. The specific control ranges are as follows: when When the concentration is ≤30ppm, the proportion of recycled powder is set at 60% and the proportion of virgin material is set at 40%; when the concentration is ≤30ppm... When the concentration is ≤50ppm, the proportion of recycled powder is set at 45% and the proportion of virgin material is set at 55%; when the concentration is ≤50ppm... When the concentration is ≤70ppm, the proportion of recycled powder is set at 30% and the proportion of virgin material is set at 70%. When the concentration is >70ppm, the mass ratio of recycled powder is set at 10% and the mass ratio of virgin material is set at 90%.
[0051] Under this decision-making strategy, material proportioning and gas phase flow compensation work synergistically: proportioning control addresses the macroscopic range of overall impurity input from the solid phase source, while flow compensation considers both simultaneously. and Fine-tuning of the gas phase flow rate is performed to jointly ensure doping stability.
[0052] As another optional embodiment of the present invention, the doping compensation model can also be a weighted summation model. In this case, the central control unit first uses the weighting factor to calculate the equivalent nitrogen concentration of the powder. To comprehensively reflect the background state of impurities in the mixture: In the formula, The nitrogen weighting coefficient for recovering powder is preferably in the range of 0.6 to 0.8, and more preferably 0.7; The nitrogen weighting coefficient for the new material is preferably in the range of 0.2 to 0.4, more preferably 0.3, and satisfies the following conditions: The central control unit 104 acquires the equivalent nitrogen concentration. Then, based on With target nitrogen concentration To determine the overall deviation, a unified calculation of the nitrogen flow correction value in the gas phase is performed. : In the formula, The total compensation coefficient ranges from 0.005 to 0.015, with 0.01 being the preferred value. This model, by introducing the concept of "equivalent concentration," can more scientifically align with the different nitrogen release kinetics among powders, making the control strategy more closely aligned with engineering practice. Accordingly, when running the weighted summation model, the decision logic for the central control unit to generate powder proportioning instructions changes to: based on the calculated equivalent nitrogen concentration... The weight ratio of recycled powder to virgin material is adjusted according to the corresponding range, and the specific control range is as follows: when When the concentration is ≤30ppm, the proportion of recycled powder is set at 60% and the proportion of virgin material is set at 40%; when the concentration is ≤30ppm... When the concentration is ≤50ppm, the proportion of recycled powder is set at 45% and the proportion of virgin material is set at 55%; when the concentration is ≤50ppm... When the concentration is ≤70ppm, the proportion of recycled powder is set at 30% and the proportion of virgin material is set at 70%. When the concentration is >70ppm, the mass ratio of recycled powder is set at 10% and the mass ratio of virgin material is set at 90%.
[0053] In this control mode, both the proportioning decision and the gas phase flow compensation are based on the same core key parameter. The convergence process is highly consistent with the control logic, achieving deep coupling optimization of solid phase ratio and gas phase doping.
[0054] S103. The mixing and loading unit responds to the powder proportioning command by mixing the new material output from the new material supply unit with the recycled powder output from the recycled powder processing unit according to the set mass ratio and loading it into the crystal growth unit.
[0055] Specifically, after receiving the powder proportioning instruction from the central control unit 104, the precision electronic scale in the mixing and loading unit 108 accurately weighs the high-purity silicon carbide virgin material output from the storage tank of the virgin material supply unit 107 and the recycled powder prepared by the high-temperature purification furnace of the recycled powder processing unit 106, maintaining a measurement accuracy of ±0.1 g. After weighing, the materials enter the automatic mixer for fully automated mixing. The mixed powder is then loaded by the loading robotic arm into the graphite crucible of the PVT growth furnace in the crystal growth unit 109 with a repeatability of ±0.5 g, thus achieving precise physical control of the material at the solid phase source.
[0056] S104, the flow control unit responds to the Adjust the actual nitrogen flow rate input to the crystal growth unit, and run the PVT crystal growth process under the adjusted actual nitrogen flow rate and the mass ratio of the mixed powder. Specifically, the central control unit 104 will calculate the nitrogen flow correction value. The setpoint is transmitted in real time to the flow control unit 103 in digital or analog signal form. The digital mass flow controller within the flow control unit 103 receives the setpoint and uses its internal PID controller in conjunction with the measured value from the high-precision flow sensor to make millisecond-level response adjustments. Simultaneously, the gas recovery unit 102, connected between the exhaust port of the crystal growth unit 109 and the flow control unit 103, starts operating. Its purification system filters and purifies the emitted process gas and analyzes the nitrogen concentration of the process gas in the exhaust pipeline in real time through online purity monitoring. This data is uploaded to the central control unit 104 in real time as a cross-validation basis for the powder feedforward detection data. Under the condition that the actual gas phase nitrogen flow rate and the solid phase mixed powder are evenly distributed, the PVT growth furnace of the crystal growth unit 109 starts operating and enters the silicon carbide crystal growth stage. The PVT crystal growth process parameters are controlled as follows: growth temperature 2200 to 2400 degrees Celsius (e.g., the solid source powder region is controlled at approximately 2300 degrees Celsius to promote sublimation), chamber background pressure 10 to 100 mbar, and crystal growth rate 0.2-0.5 mm / h. Under this high-temperature and low-pressure environment, the powder sublimates to form a gaseous substance, which is transported to the seed crystal surface at a lower temperature under the drive of the furnace axial temperature gradient for crystallization. Simultaneously, nitrogen gas precisely input by the digital mass flow controller decomposes into active nitrogen atoms at high temperature, entering the silicon carbide lattice to replace carbon atom sites, achieving precise n-type doping and completely overcoming the negative impact of batch-to-batch impurity fluctuations in recycled powder on the consistency of crystal resistivity.
[0057] Furthermore, the method provided by this invention also includes a step of cross-furnace closed-loop learning and optimization to address temperature field drift or component aging during long-term production. The specific process is as follows: After the current furnace crystal growth process is completed, the grown silicon carbide crystal is transported to the substrate processing unit 110, where it is precision-processed using a multi-wire dicing machine and chemical mechanical polishing (CMP) equipment to prepare a standard substrate sample. Subsequently, the substrate testing unit 111 rapidly measures the average resistivity of the standard substrate sample using a non-contact resistivity meter. ρ avg The measured nitrogen concentration is precisely quantitatively determined using a secondary ion mass spectrometer (SIMS), and this measurement data is input as a closed-loop feedback signal to the central control unit 104. The central control unit 104 acquires the average resistivity and calculates the deviation between it and the preset target resistivity. Based on this deviation value, the central control unit 104 initiates the following two-layer closed-loop control logic: When the system detects that the deviation values of two consecutive furnaces both exceed a preset threshold (this threshold is set at 10%-15% to eliminate normal minor temperature field disturbances and measurement errors), it determines that the current system has a significant risk of severe powder contamination or severe hardware drift. The central control unit 104 immediately initiates emergency risk blocking, forcibly switching the proportion of recycled powder in the powder proportioning instruction to 0% in the next furnace (i.e., completely stopping the use of recycled material and forcibly switching to 100% high-purity virgin material), until the process window returns to normal for N consecutive furnaces (preferably 2-5 furnaces).
[0058] During risk blocking or normal model iteration, the central control unit 104 invokes a built-in 4-layer fully connected neural network (DNN) model for assisted online inference. This model includes an input layer, two hidden layers each containing 128 neurons and employing the ReLU activation function, and a single-node output layer. The central control unit 104 will then use the currently measured online data... , The historical temperature field curve characteristic values of the PVT growth furnace (such as axial temperature gradient characteristic parameters), chamber background pressure, and the current cumulative number of uses of the graphite crucible (used to quantitatively characterize the impact of graphite component aging on doping kinetics), all monitored in real time by the temperature and pressure sensors, are used as multi-dimensional input features and input into the fully connected neural network model. Its output layer outputs a nonlinear flow rate compensation deviation Δ for the gas phase flow rate. F N2 .
[0059] The central control unit 104 utilizes the Δ F N2 The original doping compensation model calculated by the aforementioned doping compensation model Algebraic superposition correction is performed to generate the final target flow command, which is then sent to the flow control unit 103 to execute secondary closed-loop fine-tuning control for gas-phase doping. As the number of production furnaces increases, the system can continuously update the model parameters in the optimization layer based on a self-learning mechanism; the weighted model enables the crystal growth system to have an adaptive optimization effect that becomes more accurate with use.
[0060] To make the technical solution and advantages of the present invention clearer, the present invention will be further described in detail below with reference to specific process numerical embodiments: Example 1 Application scenario: Production lines with relatively fixed powder ratios that require high calculation speed.
[0061] Using the multi-material coupling recycling system for silicon carbide crystal growth described above, and following the preparation steps outlined above, the following process parameters were set: , , , .
[0062] Testing the nitrogen content of powder: , Calculate flow correction: Set the powder ratio as follows: 45% recycled material.
[0063] Perform growth and check substrate resistivity. If the standard deviation of resistivity for two consecutive furnaces is >12%, start the pure new material mode and adjust the parameters.
[0064] Example 2 Following the preparation steps of Example 1, adjust , 8. To adapt to different growth conditions.
[0065] Example 3 Applicable scenarios: Production lines where the powder ratio changes dynamically and high control precision is required.
[0066] Following the preparation steps described above, the following process parameters are set: , , , , .
[0067] Testing the nitrogen content of powder: , Calculate the equivalent nitrogen concentration: .
[0068] Calculate flow correction: Set the powder ratio: 30% recycled material (because...) Perform growth and check substrate resistivity. If the standard deviation of resistivity for two consecutive furnaces is >12%, start the pure new material mode and adjust. , Based on this embodiment, the weighting coefficients can also be... , It is not fixed, but dynamically calculated based on the actual powder ratio (e.g.) The proportion of recycled materials, ).
[0069] Example 4 Applicable scenarios: Intelligent production lines with big data accumulation.
[0070] Technical solution: Step 1: Database Construction (Data Engineering) The system first extracts historical data from the past three years from the database to construct a "feature matrix".
[0071] Input feature (X): : Nitrogen content of recycled powder (ppm); Nitrogen content of new materials (ppm); Characteristic values of the historical temperature field curve of the growth furnace (such as axial temperature gradient) ); Background pressure in the growth chamber; : The number of times the crucible has been used (considering the effect of graphite component aging on doping); Output label (Y): The optimal nitrogen flow rate correction value for achieving the target resistivity in actual production.
[0072] : The actual measured resistivity.
[0073] Step Two: Model Selection and Training The engineers chose a 4-layer fully connected neural network (DNN) as the base model: Input layer: 8 nodes (corresponding to the 8 feature parameters mentioned above).
[0074] Hidden layers: 2 layers, 128 neurons per layer, using the ReLU activation function.
[0075] Output layer: 1 node (predicted) ).
[0076] Training process: 1. Divide the 2000 sets of data into training and test sets in an 8:2 ratio.
[0077] 2. The model learns the complex nonlinear mapping relationship between powder impurities and final resistivity through the backpropagation algorithm.
[0078] 3. Key Innovations: The model not only learned the relationship between "powder and flow rate" but also the coupling effect of "temperature field and doping". For example, the model found that when the temperature at the bottom of the furnace is high, the nitrogen release rate in the recovered powder will increase by 15%, thus requiring an additional reduction in nitrogen flow rate.
[0079] Step 3: Online Inference and Closed-Loop Control Before the 2001st heat of growth, the system automatically executes the following logic: 1. Detection: The powder detection unit measures the following: , .
[0080] 2. Data Acquisition Environment: The MES system reads the real-time temperature field data of the current furnace platform (displaying that the bottom temperature is 5°C higher than the standard value).
[0081] 3. Prediction: The central control unit calls the trained DNN model and inputs the above parameters.
[0082] Traditional linear models suggest: ; DNN model prediction: Considering temperature anomalies, the model output... (Automatic compensation of 0.7 sccm).
[0083] 4. Execution: The flow controller executes the 5.8 sccm setting.
[0084] Experimental Example The nitrogen doping control method for silicon carbide single crystal growth of the present invention and the conventional method were used to control and process different batches. The statistical data are shown in Table 1 below: Table 1
[0085] As can be seen, resistivity stability is significantly improved: the standard deviation of resistivity between batches decreases from ±20.4% in the traditional method to ±6.7% (weighted summation model) or ±4.1% (linear compensation model). The utilization rate of recycled materials is improved: while ensuring doping stability, the average utilization rate of recycled materials increases from 30% to 45-60%.
[0086] Finally, it should be noted that the above embodiments are only for illustrating the present invention and not for limiting the present invention. Although the present invention has been described in detail with reference to the embodiments, those skilled in the art should understand that various combinations, modifications, or equivalent substitutions of the technical solutions of the present invention do not depart from the spirit and scope of the technical solutions of the present invention and should be covered within the scope of the claims of the present invention.
Claims
1. A multi-material coupling recycling system for silicon carbide crystal growth, characterized in that, include: The powder recycling unit is used to purify the remaining waste after crystal growth to provide recycled powder. The virgin material supply unit is used to store and transport virgin silicon carbide material; The powder detection unit is at least connected to the output of the recycled powder processing unit and is used for online detection of the nitrogen content of the recycled powder. ; The central control unit is connected to the powder detection unit and is used to receive signals detected online. And obtain the nitrogen content of the virgin silicon carbide material. ; It has a built-in strategy engine layer, which is based on the... , and the preset target nitrogen concentration The nitrogen flow rate correction value used to adjust gas-phase doping was calculated. and instructions for adjusting powder proportions; The mixing and loading unit is connected to the recycled powder processing unit, the new material supply unit and the central control unit respectively, and includes an automatic mixer for mixing the new material and the recycled powder according to the powder proportioning instruction at a set mass ratio; The flow control unit, signal-connected to the central control unit, includes a digital mass flow controller for receiving the nitrogen flow correction value. And adjust the actual nitrogen flow rate input to the crystal growth unit in real time.
2. The multi-material coupling recycling system according to claim 1, characterized in that, Also includes: The crystal growth unit, which is connected to the mixing and charging unit and the flow control unit, includes a PVT growth furnace for growing silicon carbide single crystals under the conditions of the adjusted actual nitrogen flow rate and the mass ratio of the mixed powder. A gas recovery unit is provided, with its pipeline connected between the exhaust port of the crystal growth unit and the flow control unit. The gas recovery unit is equipped with a purification system and an online purity monitoring function to purify the process gas discharged during the growth process and then send it to the flow control unit for recycling.
3. The multi-material coupling recycling system according to claim 1, characterized in that, Also includes: The substrate processing unit, including a multi-wire dicing machine and a chemical mechanical polishing device, is used to process grown silicon carbide crystals into standard substrate samples. The substrate detection unit, which is signal-connected to both the substrate processing unit and the central control unit, includes a non-contact resistivity meter and a secondary ion mass spectrometer, used to measure the average resistivity of the standard substrate sample. ρ avg The measured nitrogen concentration is then input to the central control unit as a closed-loop feedback signal; The crystal growth unit is also equipped with a temperature field sensor and a pressure sensor to monitor the axial temperature gradient and chamber background pressure in the PVT growth furnace in real time, and transmit the monitoring data signals to the central control unit.
4. A method for multi-material coupling recycling based on the multi-material coupling recycling system of claim 1, characterized in that, Including the following steps: Before the powder is loaded into the furnace, the nitrogen content of the current batch of recycled powder is determined online by the powder detection unit, and the nitrogen content of the new material is obtained. The central control unit according to the The above Target nitrogen concentration and reference nitrogen flow rate The nitrogen flow correction value was calculated. And generate powder mixing instructions; The mixing and loading unit responds to the powder proportioning command by mixing the new material output from the new material supply unit with the recycled powder output from the recycled powder processing unit according to a set mass ratio and loading it into the crystal growth unit. The flow control unit responds to the Adjust the actual nitrogen flow rate input to the crystal growth unit, and run the PVT crystal growth process under the adjusted actual nitrogen flow rate and the mass ratio of the mixed powder.
5. A multi-material coupled recycling method, characterized in that, The application of the central control unit according to claim 1 includes the following steps: Receive the nitrogen content of the current batch of recovered powder from the online measurement of the powder detection unit. And read the nitrogen content of the current batch of new material. ; Combined with the preset target nitrogen concentration and reference nitrogen flow rate The nitrogen flow correction value is calculated using the built-in doping compensation model. ; Run the internal proportioning decision engine, based on the above Or the equivalent nitrogen concentration obtained by calculation Match the corresponding material mixing ratio to generate powder mixing instructions; The powder proportioning instruction is sent to the mixing and loading unit to drive the automatic mixer to perform the mixing and loading of new material and recycled powder; The The signal is sent to the flow control unit to drive the digital mass flow controller to perform gas phase doping flow regulation.
6. The multi-material coupling recycling method according to claim 4 or 5, characterized in that, Calculate the nitrogen flow correction value The doping compensation model is a linear compensation model, and the central control unit calculates according to the following formula: in, The nitrogen compensation coefficient for the recovered powder is set to a value ranging from 0.005 to 0.
02. The nitrogen compensation coefficient for the new material ranges from 0.002 to 0.
01.
7. The multi-material coupling recycling method according to claim 4 or 5, characterized in that, Calculate the nitrogen flow correction value The doping compensation model is a weighted summation model, and the central control unit calculates according to the following formula: in, The nitrogen weighting coefficient for recovered powder; This is the nitrogen weighting coefficient for the new material, and , : This is the total compensation coefficient.
8. The multi-material coupling recycling method according to claim 6, characterized in that, The decision logic for the central control unit to generate the powder proportioning instruction is as follows: directly based on the nitrogen content of the recovered powder. The weight ratio of recycled powder to virgin material in the specified range is adjusted as follows: when When the concentration is ≤30ppm, the proportion of recycled powder is set at 60% and the proportion of virgin material is set at 40%. when When the concentration is ≤50ppm, the proportion of recycled powder is set at 45% and the proportion of virgin material is set at 55%. when When the concentration is ≤70ppm, the proportion of recycled powder is set at 30% and the proportion of virgin material is set at 70%. when When the concentration is >70ppm, the mass ratio of recycled powder is set at 10% and the mass ratio of virgin material is set at 90%.
9. The multi-material coupling recycling method according to claim 7, characterized in that, The decision logic for the central control unit to generate the powder proportioning instruction is as follows: based on the calculated equivalent nitrogen concentration... The weight ratio of recycled powder to virgin material in the specified range is adjusted as follows: when When the concentration is ≤30ppm, the proportion of recycled powder is set at 60% and the proportion of virgin material is set at 40%. when When the concentration is ≤50ppm, the proportion of recycled powder is set at 45% and the proportion of virgin material is set at 55%. when When the concentration is ≤70ppm, the proportion of recycled powder is set at 30% and the proportion of virgin material is set at 70%. when When the concentration is >70ppm, the mass ratio of recycled powder is set at 10% and the mass ratio of virgin material is set at 90%.
10. The multi-material coupling recycling method according to claim 4 or 5, characterized in that, It also includes cross-furnace closed-loop learning and optimization steps: The central control unit acquires the average resistivity of the standard substrate sample from the previous growth furnace, as measured by the substrate detection unit described in claim 3. ρ avg And calculate its deviation from the target resistivity; When the deviation value of two consecutive furnaces both exceed a preset threshold of 10% At 15%, an emergency risk blocking mechanism is activated, forcibly switching the proportion of recycled powder in the powder mixing instruction to 0%; simultaneously, the built-in fully connected neural network model is invoked for assisted online inference, and the... , The historical temperature field curve feature values of the PVT growth furnace monitored by the temperature field sensor and pressure sensor as described in claim 3, the background pressure of the chamber, and the current number of times the crucible has been used are used as input features into the fully connected neural network model. The output layer outputs a nonlinear flow rate to compensate for the deviation Δ. F N2 ; The central control unit utilizes the Δ F N2 For the original Algebraic superposition correction is performed to generate the final target flow command, which is then sent to the flow control unit to execute secondary closed-loop fine-tuning control for gas-phase doping.