Adaptive power control method for silicon carbide crystal growth

Abstract

A self-adaptive power control method for silicon carbide crystal growth, first according to the actual growth process, preset target temperature, reference thermal efficiency, bias threshold and lower threshold and other basic parameters. The system collects temperature measurement area temperature, equipment input power, working frequency and working current data at a fixed frequency, filters out abnormal temperature data to synthesize effective growth temperature, then calculates the real-time temperature rising rate, and builds a complete system state vector relying on various core parameters. The equivalent thermal efficiency is solved by the temperature rising rate and the input power, the deviation is calculated by comparing the standard reference value, and the output power is intelligently corrected according to the threshold rule. The updated power parameters are transmitted to the power cabinet, and the working frequency is optimized as needed, the optimal working parameters are selected, the whole process is cyclically controlled, and the crystal growth process is completely finished. The method solves the power mismatch problem caused by the dynamic change of thermal coupling in the prior art, significantly improves the stability of the temperature field and the quality of the silicon carbide crystal.

Description

Technical Field

[0001] This invention belongs to the field of semiconductor technology, specifically relating to an adaptive power control method for silicon carbide crystal growth. Background Technology

[0002] Silicon carbide (SiC) crystals are typically grown at high temperatures above 2000°C using the physical vapor transport (PVT) method. Their thermal field is usually achieved through medium-frequency induction heating. The main control parameters include the furnace input power, coil induction frequency, and temperature feedback signal.

[0003] In existing technologies, the control methods for induction heating power mainly fall into two categories: one is constant power control, which uses a constant power for crystal growth during the set process steps; the other is PID closed-loop control based on temperature feedback, where the crystal growth equipment dynamically adjusts the heating power according to the input temperature signal. In actual production, due to process requirements, these two methods are usually combined. However, regardless of the method used, the following shortcomings exist: First, the thermal field coupling is dynamically changing. During crystal growth, the crystal size gradually increases, the density of the powder changes due to sublimation, and the structural components undergo slight deformation due to thermal expansion and contraction. These numerous changing factors alter the induction coupling conditions, causing the heating efficiency to vary over time. If adjustment is still based on fixed power or fixed temperature feedback, a mismatch between the power and the actual thermal field state will occur. Second, temperature control suffers from lag and misjudgment. Traditional PID closed-loop control adjusts based on measured temperature errors, which essentially cannot distinguish whether the temperature change stems from improper power settings or changes in the thermal coupling state; once temperature signal distortion or lag occurs, the PID will make incorrect power adjustments. Third, power exhibits abnormal fluctuations. In a relatively stable process stage, the temperature control target is a constant value, and the power should remain basically unchanged. However, when the temperature sensor is temporarily blocked by floating foreign objects, the temperature signal will be distorted within a few minutes. For example, the temperature reading may rise abnormally and rapidly. To maintain a constant temperature, the equipment will quickly reduce the power. After the foreign object is removed, the temperature measurement will return to normal, resulting in large fluctuations in power. Such large fluctuations in power can easily introduce crystal growth defects and significantly reduce crystal quality.

[0004] In summary, existing constant power control and temperature feedback PID control are both insufficient to cope with the dynamic changes in thermal field coupling. There is an urgent need to provide a control method that can dynamically identify changes in thermal efficiency and adaptively adjust the power output in order to improve the stability of the temperature field and the quality of silicon carbide crystals. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention provides an adaptive power control method for silicon carbide crystal growth. This method solves the power mismatch problem caused by dynamic changes in thermal coupling in the prior art, and significantly improves the temperature field stability and silicon carbide crystal quality.

[0006] To achieve the above objectives, the present invention provides an adaptive power control method for silicon carbide crystal growth, comprising the following steps: Step 1: Initialization settings; Based on the crystal growth process, preset the target temperatures for each stage. Each stage and target temperature Corresponding reference thermal efficiency Deviation threshold and lower limit threshold ; Step 2: Data Acquisition; At a preset acquisition frequency, multiple temperature sensors are used to acquire temperature data from the same measurement area, obtaining multiple temperature data streams. Input power is also acquired simultaneously. Operating frequency and operating current Statistical verification was performed on multiple temperature data streams to identify and remove abnormal temperature data, and the effective growth temperature was synthesized based on the remaining effective temperature data. ; Step 3: Calculate the temperature rise rate; based on the effective growth temperature of two consecutive data acquisitions. The temperature rise rate is calculated based on the difference and the corresponding sampling time interval. ; Step 4: Construct the system state vector; based on the effective growth temperature Temperature rise rate Input power Operating frequency and operating current Construct the system state vector ; Step 5: Calculate the equivalent thermal efficiency; based on the temperature rise rate. and input power Calculate equivalent thermal efficiency parameters ; Step 6: Adjust the output power; Equivalent thermal efficiency parameters Compared with the preset reference thermal efficiency Compare and calculate the thermal efficiency deviation. ; when Not exceeding the deviation threshold If the current thermal efficiency is determined to be within the normal range, the current output power will be maintained unchanged. when Exceeding the deviation threshold ,and Greater than the preset lower threshold At that time, according to The input power is corrected to obtain the corrected output power. ,in Set the current power to bring the growth temperature toward the target temperature. ;when Not greater than the preset lower threshold At that time, maintain the current output power unchanged; Step 7: Update the output power; adjust the output power accordingly. The power supply cabinet outputs to the induction heating system; Step 8: Frequency optimization; Setting frequency optimization trigger conditions: equivalent thermal efficiency parameters Continuous decline, corrected output power Exceeding the preset power threshold, or the operating frequency being close to the upper or lower limit of the preset frequency range; Based on system state vector Determine whether the frequency optimization condition has been triggered. If the frequency optimization condition is met, adjust the operating frequency within the preset frequency range. Perform a scan and select parameters that maximize the equivalent thermal efficiency. Frequency of achieving the maximum value It will be updated automatically as a new working frequency. Step 9: Loop control; return to step 2 and repeat steps 2 to 8 until crystal growth is complete.

[0007] As a preferred embodiment, in step 1, the reference thermal efficiency The crystal growth process is pre-defined according to different stages, including a heating stage, a nucleation stage, a growth stage, and a cooling stage. Each stage adopts a reference thermal efficiency corresponding to the stage's process target. .

[0008] As a preferred embodiment, in step 1, the temperature increase during the heating phase is divided into multiple sub-steps: first, the thermal field is heated to the set value using a fixed power value, and then the temperature is gradually increased using a controlled temperature method; each subsequent heating sub-step requires setting a corresponding target temperature. And set the corresponding reference thermal efficiency. .

[0009] As a preferred embodiment, in step 2, the statistical verification of the multi-channel temperature data includes: calculating the arithmetic mean of the multi-channel temperature data. and standard deviation When a certain temperature data point is equal to the arithmetic mean The absolute value of the difference is greater than the discriminant coefficient. with standard deviation When multiplying by , the temperature data from that path is determined to be abnormal and is removed. The preset discrimination coefficients.

[0010] As a preferred embodiment, in step 2, the statistical verification of the multi-channel temperature data includes: calculating the median of the multi-channel temperature data. And the median absolute deviation (MAD); when a certain temperature data differs from the median... The absolute value of the difference is greater than the discriminant coefficient. When the product of the median absolute deviation (MAD) and the median absolute deviation (MAD) is used, the temperature data is determined to be abnormal and discarded. The preset discrimination coefficients.

[0011] As a preferred option, in step 2, a weighted average of the effective growth temperature is used to synthesize the temperature. ,in The remaining data after removing outliers Effective temperature data of the road For the first The weights corresponding to the road temperature sensing devices.

[0012] As a preferred embodiment, step 2 further includes: A minimum effective number of channels threshold is preset; when the number of remaining effective temperature data channels after removing abnormal data is less than the minimum effective number of channels threshold, the temperature data of this frame is determined to be unreliable, steps 3 to 8 are not executed in this frame, the output power of the previous frame remains unchanged, and the process returns to step 2.

[0013] As a preferred embodiment, the method is applicable to the growth of 6-inch, 8-inch, or 12-inch silicon carbide crystals.

[0014] Compared with the prior art, the present invention has the following technical advantages: 1. It can reflect the coupling state of the thermal field in real time. This is achieved by introducing an equivalent thermal efficiency parameter. The coupling efficiency and heating state of the thermal field are characterized in real time by the temperature rise rate corresponding to a unit input power. Compared with the traditional method that only relies on absolute temperature feedback, It directly reflects the dynamic changes in thermal coupling and can promptly detect changes in coupling conditions and heating efficiency caused by factors such as the gradual increase in crystal size, changes in powder density sublimation, and thermal expansion and contraction of structural components, providing a direct and real-time basis for power regulation.

[0015] 2. Improve power regulation accuracy and avoid misjudgments caused by relying solely on temperature feedback. Abnormal temperature data is identified and eliminated through multi-channel temperature measurement statistical verification (Raida criterion or MAD method), and temperature reliability is determined based on a threshold of the number of valid channels. Power is only updated when the temperature data is reliable. Therefore, even if a temperature sensor is blocked by a floating foreign object, causing temperature signal distortion, the abnormal data can be eliminated, and a reliable temperature can be synthesized from the remaining valid channels. This fundamentally avoids power misjudgments caused by single-channel temperature measurement distortion and significantly improves power regulation accuracy.

[0016] 3. Reduce the risk of abnormal power fluctuations and extend equipment lifespan. This is achieved through deviation thresholds. Dead zone triggering (no action is taken due to minor thermal efficiency deviations, avoiding frequent power fine-tuning) and lower threshold. A dual mechanism of protection (avoiding ratio divergence when the steady-state temperature rise rate approaches zero) suppresses frequent power fluctuations and abnormal large fluctuations. This reduces the adverse disturbances of power fluctuations to the crystal growth process and reduces current surges to power devices, thus extending the service life of the equipment.

[0017] 4. Effectively reduces crystal growth defects and improves crystal quality. Through the synergistic effect of power adaptive correction and automatic frequency optimization, a stable and uniform temperature field is maintained, reducing the disturbance of abnormal power fluctuations to the crystal growth interface. This effectively reduces crystal growth defects such as basal plane dislocations (BPD) and microtubes, thereby improving the quality of silicon carbide crystals.

[0018] 5. Strong adaptability, suitable for the growth of crystals of different sizes. This is achieved through staged preset reference thermal efficiency. Combined with real-time efficiency identification and adaptive correction, the same equipment can adapt to different thermal coupling states without changing the control strategy. It is suitable for the growth of silicon carbide crystals of different sizes such as 6-inch, 8-inch, and 12-inch, improving the equipment's versatility and process portability.

[0019] This method solves the power mismatch problem caused by dynamic changes in thermal coupling in existing technologies, and significantly improves temperature field stability and silicon carbide crystal quality. Attached Figure Description

[0020] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0021] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. These embodiments are merely one of the preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention.

[0022] like Figure 1 As shown, this embodiment provides an adaptive power control method for silicon carbide crystal growth, including the following steps: Step 1: Initialize settings; Based on the crystal growth process, target temperatures are preset for different stages. Each stage and target temperature Corresponding reference thermal efficiency Deviation threshold and lower threshold ; Step 2: Collect data; According to the preset acquisition frequency (the acquisition frequency is 1 second to ensure timely adjustment), multiple temperature sensors are used to acquire temperature data from the same temperature measurement area to obtain multiple channels of temperature data, and the input power is acquired simultaneously. Operating frequency and operating current Statistical verification was performed on multiple temperature data streams to identify and remove abnormal temperature data, and the effective growth temperature was synthesized based on the remaining effective temperature data. ; Step 3: Calculate the rate of temperature rise; Based on the effective growth temperature of two consecutive data collections The temperature rise rate is calculated based on the difference and the corresponding sampling time interval. ; Step 4: Construct the system state vector; Based on effective growth temperature Temperature rise rate Input power Operating frequency and operating current Construct the system state vector This is used to characterize the current system state of the thermal field, providing a basis for subsequent equivalent power calculation and frequency optimization; Step 5: Calculate the equivalent thermal efficiency; Based on the rate of temperature rise and input power Calculate equivalent thermal efficiency parameters ; Equivalent thermal efficiency parameters The temperature rise rate corresponding to a unit input power; Step 6: Adjust the output power; Equivalent thermal efficiency parameters Compared with the preset reference thermal efficiency Compare and calculate the thermal efficiency deviation. ; when Not exceeding the deviation threshold If the current thermal efficiency is determined to be within the normal range, the current output power will be maintained unchanged. when Exceeding the deviation threshold ,and Greater than the preset lower threshold At that time, according to The input power is corrected to obtain the corrected output power. ,in Set the current power to bring the growth temperature toward the target temperature. ;when Not greater than the preset lower threshold At that time, maintain the current output power unchanged; Step 7: Update output power; The corrected output power The power supply cabinet outputs to the induction heating system; Step 8: Frequency optimization; Setting frequency optimization trigger conditions: equivalent thermal efficiency parameters Continuous decline, corrected output power Exceeding the preset power threshold, or the operating frequency being close to the upper or lower limit of the preset frequency range; Based on system state vector Determine whether the frequency optimization condition has been triggered. If the frequency optimization condition is met, adjust the operating frequency within the preset frequency range (6–10 kHz). Perform a scan and select parameters that maximize the equivalent thermal efficiency. Frequency of achieving the maximum value It will be updated automatically as a new working frequency. Step 9: Loop control; Return to step 2 and repeat steps 2 through 8 until crystal growth is complete.

[0023] As a preferred embodiment, in step 1, the reference thermal efficiency The crystal growth process is pre-defined according to different stages, including a heating stage, a nucleation stage, a growth stage, and a cooling stage. Each stage adopts a reference thermal efficiency corresponding to the stage's process target. .

[0024] As a preferred embodiment, in step 1, the temperature increase during the heating phase is divided into multiple sub-steps: first, the thermal field is heated to a set value (e.g., 900℃) with a fixed power value, and then the temperature is gradually increased using a controlled temperature method; each subsequent heating sub-step requires setting a corresponding target temperature. And set the corresponding reference thermal efficiency. .

[0025] As a preferred embodiment, in step 2, the statistical verification of the multi-channel temperature data includes: calculating the arithmetic mean of the multi-channel temperature data. and standard deviation When a certain temperature data point is equal to the arithmetic mean The absolute value of the difference is greater than the discriminant coefficient. with standard deviation When multiplying by , the temperature data from that path is determined to be abnormal and is removed. The preset discrimination coefficients.

[0026] As a preferred embodiment, in step 2, the statistical verification of the multi-channel temperature data includes: calculating the median of the multi-channel temperature data. And the median absolute deviation (MAD); when a certain temperature data point deviates from the median... The absolute value of the difference is greater than the discriminant coefficient. When the product of the median absolute deviation (MAD) and the median absolute deviation (MAD) is used, the temperature data is determined to be abnormal and discarded. The preset discrimination coefficients.

[0027] As a preferred option, in step 2, a weighted average of the effective growth temperature is used to synthesize the temperature. ,in The remaining data after removing outliers Effective temperature data of the road For the first The weights corresponding to the temperature sensing devices along the route are determined based on the historical measurement stability of those devices.

[0028] As a preferred embodiment, step 2 further includes: A minimum effective number of channels threshold is preset; when the number of remaining effective temperature data channels after removing abnormal data is less than the minimum effective number of channels threshold, the temperature data of this frame is determined to be unreliable, steps 3 to 8 are not executed in this frame, the output power of the previous frame remains unchanged, and the process returns to step 2.

[0029] As a preferred embodiment, the method is applicable to the growth of 6-inch, 8-inch, or 12-inch silicon carbide crystals.

[0030] This invention uses the equivalent thermal efficiency parameter as the core to construct an adaptive control system that includes real-time efficiency identification, reliable temperature assurance, deviation dead zone correction, automatic frequency optimization, and phased adaptation. It solves the power mismatch problem caused by dynamic changes in thermal coupling in existing technologies and significantly improves temperature field stability and silicon carbide crystal quality.

[0031] The above descriptions are merely preferred embodiments of the present invention and are not intended to limit the invention. Those skilled in the art can make various modifications, substitutions, and variations to these embodiments without departing from the principles and spirit of the invention. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. An adaptive power control method for silicon carbide crystal growth, characterized in that, Includes the following steps: Step 1: Initialization settings; Based on the crystal growth process, preset the target temperatures for each stage. Each stage and target temperature Corresponding reference thermal efficiency Deviation threshold and lower threshold ; Step 2: Collect data; According to the preset acquisition frequency, multiple temperature measuring devices are used to acquire temperature data from the same temperature measuring area to obtain multiple channels of temperature data, and the input power is acquired simultaneously. Operating frequency and operating current Statistical verification was performed on multiple temperature data streams to identify and remove abnormal temperature data, and the effective growth temperature was synthesized based on the remaining effective temperature data. ; Step 3: Calculate the temperature rise rate; based on the effective growth temperature of two consecutive data acquisitions. The temperature rise rate is calculated based on the difference and the corresponding sampling time interval. ; Step 4: Construct the system state vector; based on the effective growth temperature Temperature rise rate Input power Operating frequency and operating current Construct the system state vector ; Step 5: Calculate the equivalent thermal efficiency; based on the temperature rise rate. and input power Calculate equivalent thermal efficiency parameters ; Step 6: Adjust the output power; Equivalent thermal efficiency parameters Compared with the preset reference thermal efficiency Compare and calculate the thermal efficiency deviation. ; when Not exceeding the deviation threshold If the current thermal efficiency is determined to be within the normal range, the current output power will be maintained unchanged. when Exceeding the deviation threshold ,and Greater than the preset lower threshold At that time, according to The input power is corrected to obtain the corrected output power. ,in Set the current power to bring the growth temperature toward the target temperature. ;when Not greater than the preset lower threshold At that time, maintain the current output power unchanged; Step 7: Update the output power; adjust the output power accordingly. The power supply cabinet outputs to the induction heating system; Step 8: Frequency optimization; Setting frequency optimization trigger conditions: equivalent thermal efficiency parameters Continuous decline, corrected output power Exceeding the preset power threshold, or the operating frequency being close to the upper or lower limit of the preset frequency range; Based on system state vector Determine whether the frequency optimization condition has been triggered. If the frequency optimization condition is met, adjust the operating frequency within the preset frequency range. Perform a scan and select parameters that maximize the equivalent thermal efficiency. Frequency of achieving the maximum value It will be updated automatically as a new working frequency. Step 9: Loop control; return to step 2 and repeat steps 2 to 8 until crystal growth is complete.

2. The adaptive power control method for silicon carbide crystal growth according to claim 1, characterized in that, In step 1, the reference thermal efficiency The crystal growth process is pre-defined according to different stages, including a heating stage, a nucleation stage, a growth stage, and a cooling stage. Each stage adopts a reference thermal efficiency corresponding to the stage's process target. .

3. The adaptive power control method for silicon carbide crystal growth according to claim 2, characterized in that, In step 1, the temperature increase during the heating phase is divided into multiple sub-steps: first, the thermal field is heated to the set value with a fixed power value, and then the temperature is gradually increased using a controlled temperature method; each subsequent heating sub-step requires setting a corresponding target temperature. And set the corresponding reference thermal efficiency. .

4. The adaptive power control method for silicon carbide crystal growth according to claim 1, characterized in that, In step 2, the statistical verification of the multi-channel temperature data includes: calculating the arithmetic mean of the multi-channel temperature data. and standard deviation When a certain temperature data point is equal to the arithmetic mean The absolute value of the difference is greater than the discriminant coefficient. with standard deviation When multiplying by , the temperature data from that path is determined to be abnormal and is removed. The preset discrimination coefficients.

5. The adaptive power control method for silicon carbide crystal growth according to claim 1, characterized in that, In step 2, statistical verification of the multi-channel temperature data includes: calculating the median of the multi-channel temperature data. And the median absolute deviation (MAD); when a certain temperature data differs from the median... The absolute value of the difference is greater than the discriminant coefficient. When the product of the median absolute deviation (MAD) and the median absolute deviation (MAD) is used, the temperature data is determined to be abnormal and discarded. The preset discrimination coefficients.

6. The adaptive power control method for silicon carbide crystal growth according to claim 1, characterized in that, In step 2, a weighted average of the effective growth temperature is used to synthesize the temperature. ,in The remaining data after removing outliers Effective temperature data of the road For the first The weights corresponding to the road temperature sensing devices.

7. The adaptive power control method for silicon carbide crystal growth according to claim 1, characterized in that, Step 2 also includes: A minimum effective number of channels threshold is preset; when the number of remaining effective temperature data channels after removing abnormal data is less than the minimum effective number of channels threshold, the temperature data of this frame is determined to be unreliable, steps 3 to 8 are not executed in this frame, the output power of the previous frame remains unchanged, and the process returns to step 2.

8. The adaptive power control method for silicon carbide crystal growth according to claim 1, characterized in that, The method is applicable to the growth of 6-inch, 8-inch, or 12-inch silicon carbide crystals.

Images