A method, device, equipment and medium for growing indium phosphide crystal by using VGF-VB combined method

Abstract

The application discloses an indium phosphide crystal growth method and device using a VGF-VB combined method, a device, and a medium. The method comprises the following steps: collecting first furnace condition data of an indium phosphide crystal growth furnace from the beginning of a VGF shoulder growth stage; controlling a trained indium phosphide crystal growth control model to determine first control parameters based on the first furnace condition data and control the temperature of each temperature region in the indium phosphide crystal growth furnace based on the first control parameters; collecting second furnace condition data of the indium phosphide crystal growth furnace when the indium phosphide crystal growth enters a VB constant diameter growth stage; controlling the growth control model to determine second control parameters based on the second furnace condition data and control the indium phosphide crystal growth furnace based on the second control parameters, so that each temperature region in the indium phosphide crystal growth furnace is maintained in a preset temperature interval and the crucible works according to the crucible speed data. The application realizes precise closed-loop control of the VGF shoulder growth stage and the VB constant diameter growth stage, reduces the dependence on artificial experience, avoids the solid-liquid interface fluctuation caused by untimely artificial adjustment and inaccurate stage switching time, effectively improves the stability of the indium phosphide crystal growth process, and reduces the probability of defects such as twinning and dislocation.

Description

Technical Field

[0001] This application relates to the field of crystal material growth technology, and in particular to a method, apparatus, equipment and medium for growing indium phosphide crystals using the VGF-VB combination method. Background Technology

[0002] Indium phosphide ( Indium phosphide (IPT) single crystals are important second-generation compound semiconductor materials, primarily used in optoelectronics, microwave technology, and space satellite communications. Due to IPT's low stacking fault energy (17 meV / atom) and the tendency to generate twins and dislocations during growth, the fabrication of large-size, high-quality IPT single crystals is extremely challenging. Currently, industrial production mainly employs the vertical gradient solidification (VGF) and vertical Bridgman (VB) methods.

[0003] The VGF method achieves bottom-up crystal growth by controlling the temperature distribution of a multi-temperature zone heater, causing the temperature gradient to slowly move upwards. The advantages of this method are the absence of moving mechanical parts, low thermal stress, and low dislocation density (<300) in the grown crystal. However, the VGF method requires high precision in thermal field control, and it is difficult to ensure the uniformity of the radial temperature gradient when growing large-sized crystals (such as 4 inches and above).

[0004] The VB method involves a mechanical mechanism that slowly lowers and rotates a crucible within a fixed temperature gradient. The advantages of this method are that it maintains the central symmetry of the thermal and flow fields, resulting in good uniformity of the electrical parameters of the grown crystal (such as carrier concentration and electron mobility). However, the mechanical motion of the VB method can easily introduce minute vibrations, and the thermal convection changes during the crucible's descent can lead to fluctuations at the solid-liquid interface, increasing the risk of twinning and dislocation formation.

[0005] In recent years, the industry has begun to explore combining the VGF and VB methods (i.e., the VGF-VB combined method). The VGF method is used in the shoulder formation stage to reduce initial thermal stress, while the VB method is switched to improve crystal uniformity in the constant-diameter growth stage. However, the stage switching in this combined method, as well as the parameter adjustments throughout the entire process from vacuuming, pressurizing, heating and melting, crystal growth to cooling and annealing, mainly rely on manual experience. Inaccurate stage switching timing and untimely parameter adjustments can easily lead to unstable solid-liquid interface morphology, increasing the probability of defects such as twins and dislocations. This makes it difficult to stably prepare large-size, highly uniform, and low-defect-density indium phosphide single crystals, failing to meet the current industry's quality and size requirements for indium phosphide crystals.

[0006] Therefore, existing technologies still need to be improved and enhanced. Summary of the Invention

[0007] The technical problem to be solved by this application is to provide a method, apparatus, equipment and medium for growing indium phosphide crystals using the VGF-VB combination method, which addresses the shortcomings of the prior art.

[0008] To address the aforementioned technical problems, the first aspect of this application provides a method for growing indium phosphide crystals using the VGF-VB combination method, wherein the method specifically includes: Starting from the VGF shoulder formation stage of indium phosphide crystal growth, the first furnace condition data of the indium phosphide crystal growth furnace is collected. The first furnace condition data includes temperature data of each temperature zone in the indium phosphide crystal growth furnace, boiler high gas pressure and cooling water data. The growth control model for controlling the trained indium phosphide crystal determines the first control parameter based on the first furnace condition data, and controls the temperature of each temperature zone in the indium phosphide crystal growth furnace based on the first control parameter. The first control parameter includes the target heating power of each temperature zone in the indium phosphide crystal growth furnace. When the indium phosphide crystal growth enters the VB constant diameter growth stage, the second furnace condition data of the indium phosphide crystal growth furnace is collected. The second furnace condition data includes the first furnace condition data, crucible condition data and the three-dimensional temperature field inside the indium phosphide crystal growth furnace. The growth control model determines the second control parameters based on the second furnace condition data, and controls the indium phosphide crystal growth furnace based on the second control parameters, so that each temperature region in the indium phosphide crystal growth furnace is maintained within a preset temperature range and the crucible operates according to the crucible speed data. The second control parameters include the target heating power and crucible speed data for each temperature region in the indium phosphide crystal growth furnace.

[0009] The indium phosphide crystal growth method using the VGF-VB combination method includes a growth control model comprising a timing prediction module and a reinforcement learning module. The timing prediction module is used to predict a first control parameter based on the first furnace condition data, and the reinforcement learning module is used to predict a second control parameter based on the first control parameter, crucible state data, and a three-dimensional temperature field.

[0010] The indium phosphide crystal growth method utilizing the VGF-VB combination method, wherein the process of obtaining the three-dimensional temperature field within the indium phosphide crystal growth furnace specifically includes: The temperature distribution image of the outer surface of the indium phosphide crystal growth furnace wall, the thermocouple discrete point data of the indium phosphide crystal growth furnace, and the cooling water data were collected. The temperature distribution image, thermocouple discrete point data, and cooling water data are input into a trained visual model, and the three-dimensional temperature field inside the indium phosphide crystal growth furnace is reconstructed through the visual network.

[0011] The indium phosphide crystal growth method using the VGF-VB combination method, wherein, in the process of obtaining the three-dimensional temperature field, the method further includes: Detect whether there are any abnormalities in the three-dimensional temperature field; When the three-dimensional temperature field is abnormal, the three-dimensional temperature field is compared with the preset standard temperature field to determine the abnormal data. The abnormal data includes one or more of the following: hot spot location, temperature deviation value, and solid-liquid interface shape. Supplementary control parameters are generated based on the abnormal data, and the supplementary control parameters are used to supplement the control of the indium phosphide crystal growth furnace. The supplementary control parameters include the compensation temperature and power adjustment amount for the temperature region where the abnormality occurs.

[0012] The indium phosphide crystal growth method utilizing the VGF-VB combination method, wherein, during the process of controlling the indium phosphide crystal growth furnace based on the second control parameter, the method further includes: If supplementary control parameters are received, the control process of controlling the indium phosphide crystal growth furnace based on the second control parameters is interrupted. The indium phosphide crystal growth furnace is supplemented with the supplementary control parameters until there are no abnormalities in the three-dimensional temperature field, and then the control process based on the second control parameters to control the indium phosphide crystal growth furnace is resumed.

[0013] The indium phosphide crystal growth method using the VGF-VB combination method, wherein, during the VGF shoulder formation stage, the method further includes: The growth control model is controlled to identify the indium phosphide crystal state, which includes the seed crystal weld length and / or shoulder diameter. The VGF shoulder formation progress of indium phosphide crystal is determined based on the crystal state of indium phosphide. When the VGF shoulder formation progress reaches a preset progress threshold, it is determined that the indium phosphide crystal growth has entered the VB constant diameter growth stage, so as to automatically switch the temperature control of each temperature zone in the indium phosphide crystal growth furnace based on the first control parameter to control the indium phosphide crystal growth furnace based on the second control parameter.

[0014] The indium phosphide crystal growth method utilizing the VGF-VB combination method further includes: Historical data on multiple indium phosphide crystal growth processes are aggregated through a cloud server; The growth control model is retrained and / or fine-tuned using the historical data, and the retrained and / or fine-tuned growth control model is deployed on an edge industrial control computer so that the edge industrial control computer can use the retrained and / or fine-tuned growth control model to control the indium phosphide crystal growth process using the VGF-VB combination method.

[0015] A second aspect of this application provides an indium phosphide crystal growth apparatus utilizing the VGF-VB combination method, wherein the indium phosphide crystal growth apparatus utilizing the VGF-VB combination method specifically includes: The first furnace condition data acquisition module is used to acquire the first furnace condition data of the indium phosphide crystal growth furnace when the indium phosphide crystal growth enters the VGF shoulder formation stage. The first furnace condition data includes temperature data of each temperature zone in the indium phosphide crystal growth furnace, crucible height gas pressure and cooling water data. The first control parameter determination module is used to control the growth control model of the trained indium phosphide crystal to determine the first control parameter based on the first furnace condition data, and to control the temperature of each temperature zone in the indium phosphide crystal growth furnace based on the first control parameter. The first control parameter includes the target heating power of each temperature zone in the indium phosphide crystal growth furnace. The second furnace condition data acquisition module is used to acquire the second furnace condition data of the indium phosphide crystal growth furnace after the indium phosphide crystal growth enters the VB constant diameter growth stage. The second furnace condition data includes the first furnace condition data, crucible status data and the three-dimensional temperature field inside the indium phosphide crystal growth furnace. The second control parameter determination module is used to control the growth control model to determine the second control parameters based on the second furnace condition data, and to control the indium phosphide crystal growth furnace based on the second control parameters, so that each temperature region in the indium phosphide crystal growth furnace is maintained within a preset temperature range and the crucible operates according to the crucible speed data. The second control parameters include the target heating power and crucible speed data of each temperature region in the indium phosphide crystal growth furnace.

[0016] A third aspect of this application provides a computer-readable storage medium storing one or more programs that can be executed by one or more processors to implement the steps in the indium phosphide crystal growth method using the VGF-VB combination method as described above.

[0017] A fourth aspect of this application provides a terminal device, which includes: a processor and a memory; The memory stores a computer-readable program that can be executed by the processor; When the processor executes the computer-readable program, it implements the steps in the indium phosphide crystal growth method using the VGF-VB combination method as described above.

[0018] Beneficial effects: 1. This application constructs a growth control model that includes VGF stage control and VB stage control. By combining this growth control model with multi-zone thermocouple data, the optimal heating power, crucible descent speed and rotation speed of each zone are predicted and output in real time. This achieves precise closed-loop control of the VGF shoulder-forming stage and the VB constant-diameter growth stage, reducing reliance on human experience and avoiding solid-liquid interface fluctuations caused by untimely manual adjustments or inaccurate stage switching. This effectively improves the stability of the indium phosphide crystal growth process and reduces the probability of defects such as twins and dislocations.

[0019] 2. This application identifies the crystal state of indium phosphide through a growth control model and automatically switches between VGF stage control and VB stage control based on the crystal state. This achieves a smooth switch from static thermal field control (VGF shoulder formation) to dynamic mechanical motion compensation (VB equal diameter), eliminating the need for segmented control based on human experience. This ensures accurate timing of stage switching, further improves the stability of the crystal growth process and the quality of the finished crystal, and can meet the requirements for stable preparation of large-size indium phosphide single crystals.

[0020] 3. This application breaks through the limitations of high-pressure closed observation in indium phosphide crystal growth furnaces. It reconstructs the three-dimensional temperature field inside the furnace by using temperature distribution images of the outer surface of the furnace wall and thermocouple discrete point data, thereby realizing indirect visualization of the solid-liquid interface under high-pressure closed growth environment of indium phosphide and improving the accuracy of indium phosphide crystal growth control.

[0021] 4. This application automatically identifies the growth stage by recognizing the indium phosphide crystal state and automatically switches between growth stages based on the recognition results. This achieves a smooth transition between the VGF and VB stages without manual intervention, further improving the stability of the growth process. This allows the growth of large-size (4 inches and above) indium phosphide single crystals to simultaneously possess the low dislocation density (<300) of the VGF method. High electrical uniformity (deviation < ±6%) of the VB method.

[0022] 5. This application supplements the control by detecting anomalies in the three-dimensional temperature field, which can correct temperature field deviations in a timely manner, avoid solid-liquid interface deformation caused by local temperature anomalies, further reduce the risk of defect generation, and ensure the uniformity and stability of crystal growth.

[0023] 6. This application uses cloud-based aggregation of historical data from multiple furnace runs to iteratively update the control model, enabling it to automatically adapt to long-term drift factors such as aging of furnace insulation materials and seasonal changes in cooling water temperature. Furthermore, as the number of production furnace runs increases, the control accuracy of the model can be continuously optimized, gradually improving the yield and quality of crystal growth. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 A flowchart of an indium phosphide crystal growth method using the VGF-VB combination method provided in an embodiment of this application.

[0026] Figure 2 This is a general framework diagram of the indium phosphide crystal growth method using the VGF-VB combination method provided in the embodiments of this application.

[0027] Figure 3 A flowchart illustrating the collaborative operation of the growth control model and the vision model.

[0028] Figure 4 This is a schematic diagram of the one-click crystal growth process.

[0029] Figure 5 This is a schematic diagram of the indium phosphide crystal growth apparatus using the VGF-VB combination method provided in the embodiments of this application.

[0030] Figure 6 A schematic block diagram of the terminal device provided in the embodiments of this application. Detailed Implementation

[0031] This application provides a method, apparatus, equipment, and medium for growing indium phosphide crystals using the VGF-VB combination method. To make the objectives, technical solutions, and effects of this application clearer and more explicit, the following detailed description, with reference to the accompanying drawings and embodiments, further illustrates this application. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0032] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this application means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we say an element is “connected” or “coupled” to another element, it can be directly connected or coupled to the other element, or there may be intermediate elements. Furthermore, “connected” or “coupled” as used herein can include wireless connections or wireless coupling. The term “and / or” as used herein includes all or any units and all combinations of one or more associated listed items.

[0033] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0034] It should be understood that the sequence number and size of each step in this embodiment do not imply the order of execution. The execution order of each process is determined by its function and internal logic, and should not constitute any limitation on the implementation process of this application embodiment.

[0035] Research has found that indium phosphide ( Indium phosphide (IPT) single crystals are important second-generation compound semiconductor materials, primarily used in optoelectronics, microwave technology, and space satellite communications. Due to IPT's low stacking fault energy (17 meV / atom) and the tendency to generate twins and dislocations during growth, the fabrication of large-size, high-quality IPT single crystals is extremely challenging. Currently, industrial production mainly employs the vertical gradient solidification (VGF) and vertical Bridgman (VB) methods.

[0036] The VGF method achieves bottom-up crystal growth by controlling the temperature distribution of a multi-temperature zone heater, causing the temperature gradient to slowly move upwards. The advantages of this method are the absence of moving mechanical parts, low thermal stress, and low dislocation density (<300) in the grown crystal. However, the VGF method requires high precision in thermal field control, and it is difficult to ensure the uniformity of the radial temperature gradient when growing large-sized crystals (such as 4 inches and above).

[0037] The VB method involves a mechanical mechanism that slowly lowers and rotates a crucible within a fixed temperature gradient. The advantages of this method are that it maintains the central symmetry of the thermal and flow fields, resulting in good uniformity of the electrical parameters of the grown crystal (such as carrier concentration and electron mobility). However, the mechanical motion of the VB method can easily introduce minute vibrations, and the thermal convection changes during the crucible's descent can lead to fluctuations at the solid-liquid interface, increasing the risk of twinning and dislocation formation.

[0038] In recent years, the industry has begun to explore combining the VGF and VB methods (i.e., the VGF-VB combined method). The VGF method is used in the shoulder formation stage to reduce initial thermal stress, while the VB method is switched to improve crystal uniformity in the constant-diameter growth stage. However, the stage switching in this combined method, as well as the parameter adjustments throughout the entire process from vacuuming, pressurizing, heating and melting, crystal growth to cooling and annealing, mainly rely on manual experience. Inaccurate stage switching timing and untimely parameter adjustments can easily lead to unstable solid-liquid interface morphology, increasing the probability of defects such as twins and dislocations. This makes it difficult to stably prepare large-size, highly uniform, and low-defect-density indium phosphide single crystals, failing to meet the current industry's quality and size requirements for indium phosphide crystals.

[0039] To address the aforementioned issues, in this embodiment, starting from the VGF shoulder formation stage of indium phosphide crystal growth, first furnace condition data of the indium phosphide crystal growth furnace is collected; a trained indium phosphide crystal growth control model is controlled to determine a first control parameter based on the first furnace condition data, and the temperature of each temperature region in the indium phosphide crystal growth furnace is controlled based on the first control parameter; when the indium phosphide crystal growth enters the VB constant diameter growth stage, second furnace condition data of the indium phosphide crystal growth furnace is collected; the growth control model is controlled to determine a second control parameter based on the second furnace condition data, and the indium phosphide crystal growth furnace is controlled based on the second control parameter, so that each temperature region in the indium phosphide crystal growth furnace is maintained within a preset temperature range and the crucible operates according to the crucible speed data. This application constructs a growth control model that includes VGF stage control and VB stage control. By combining this growth control model with multi-zone thermocouple data, it predicts and outputs the optimal heating power, crucible descent speed, and rotation speed for each zone in real time. This achieves precise closed-loop control of the VGF shoulder-forming stage and the VB constant-diameter growth stage, reducing reliance on human experience and avoiding solid-liquid interface fluctuations caused by untimely manual adjustments or inaccurate stage switching. This effectively improves the stability of the indium phosphide crystal growth process, reduces the probability of defects such as twins and dislocations, and improves its yield and batch consistency (especially for large-size indium phosphide single crystals (such as 4 inches and above)). It realizes the digitalization, intelligentization, and continuous self-evolution of the crystal growth process.

[0040] It should be noted that the execution subject of each embodiment of this application can be a computing service system with data processing, network communication, and program execution functions, such as an electronic system capable of realizing the above functions, an indium phosphide crystal growth system using the VGF-VB combination method, etc. The following uses an indium phosphide crystal growth system using the VGF-VB combination method (hereinafter referred to as "the system") as an example to describe the following embodiments.

[0041] The application content will be further explained below with reference to the accompanying drawings and the description of the embodiments.

[0042] This embodiment provides a method for growing indium phosphide crystals using the VGF-VB combination method, such as... Figure 1 and Figure 2 As shown, the indium phosphide crystal growth method using the VGF-VB combination method specifically includes: S10. Starting from the VGF shoulder formation stage of indium phosphide crystal growth, the first furnace condition data of the indium phosphide crystal growth furnace is collected.

[0043] Specifically, the VGF shoulder formation stage refers to the growth stage after the indium phosphide crystal growth is completed and the crystal gradually expands radially to the target size. At this time, the solid-liquid interface in the furnace is located in the crystal shoulder formation section. During the VGF shoulder formation stage, the indium phosphide crystal growth furnace is deployed radially with multiple independent temperature zones (such as 6 independent temperature zones). Each temperature zone corresponds to an independent temperature-controlled thermocouple, and the temperature data of each temperature zone is obtained through the thermocouples corresponding to each temperature zone.

[0044] During the VGF shoulder formation stage, growth control involves monitoring temperature data, boiler pressure, and cooling water data for each temperature zone. Specifically, the first furnace condition data includes temperature data for each temperature zone within the indium phosphide crystal growth furnace, boiler pressure, and cooling water data. Temperature data for each zone reflects the current thermal distribution within the furnace. Boiler pressure reflects the saturated vapor pressure of phosphorus within the furnace, maintaining the stoichiometric ratio of phosphorus in the melt and preventing phosphorus loss due to melt decomposition. Cooling water data ensures stable temperature gradient and heat dissipation within the furnace. The cooling water data may include cooling water flow rate and temperature, which may include inlet and / or outlet water temperatures.

[0045] S20. The growth control model of the trained indium phosphide crystal determines the first control parameter based on the first furnace condition data, and controls the temperature of each temperature zone in the indium phosphide crystal growth furnace based on the first control parameter.

[0046] Specifically, the growth control model is a trained deep neural network model that can predict and output the heating power and crucible speed data of each temperature zone in real time based on the furnace condition data in the indium phosphide crystal growth furnace. This enables smooth switching decisions between the VGF shoulder stage, the VB constant diameter growth stage, and the VGF shoulder stage and the VB constant diameter growth stage, so as to achieve one-click crystal growth control of the indium phosphide crystal growth process.

[0047] The growth control model may include a time-series prediction module, which predicts a first control parameter based on first furnace condition data. The time-series prediction module may include a time-series prediction unit and a feedforward neural network unit. The time-series prediction unit predicts the thermal field evolution trend over a predetermined time period, and the feedforward neural network unit determines the first control parameter at the current moment based on the predicted thermal field evolution trend over the predetermined time period. The time-series prediction unit may employ an LSTM network, etc.

[0048] It is understood that when acquiring the first furnace condition data, a historical first furnace condition data sequence is obtained. This historical first furnace condition data sequence can include all first furnace condition data from the start of indium phosphide crystal growth to the current moment, or it can include the first furnace condition data at the current moment and several historical first furnace condition data before the current moment. Then, this historical first furnace condition data sequence is input into the time-series prediction module in the growth control model. The time-series prediction unit in the time-series prediction module predicts the thermal field evolution trend for a future preset duration. Then, the feedforward neural network unit in the time-series prediction module outputs the first control parameter based on the predicted thermal field evolution trend for the future preset duration.

[0049] Furthermore, since there are no moving mechanical parts during the VGF shoulder formation stage, only the temperature distribution of the multi-temperature zone heaters needs to be controlled. Therefore, when determining the first control parameters using the first furnace condition data, it is only necessary to obtain the heating power of each temperature zone to control the temperature of each zone. Thus, the first control parameters include the target heating power of each temperature zone in the indium phosphide crystal growth furnace. By controlling the temperature of each temperature zone within a preset temperature range using the target heating power, the uniform and stable thermal field distribution is ensured, reducing thermal stress in the initial growth stage and minimizing the generation of twins and dislocations.

[0050] It is understandable that after obtaining the first control parameter, the temperature deviation is calculated based on the current actual temperature of each temperature zone and the target temperature corresponding to the target heating power. Then, the actual output power of each heating zone is adjusted in real time through a PID algorithm. Based on the actual output power, the voltage signal is converted by the PLC to drive the thyristor heating, achieving precise closed-loop control of the temperature in each temperature zone, maintaining a stable thermal field distribution, and adapting to the crystal growth requirements of the shoulder formation stage. In addition, to avoid crystal defects caused by additional vibration, the crucible is kept stationary during the control of the indium phosphide crystal growth furnace based on the first control parameter, further ensuring the crystal growth quality during the shoulder formation stage.

[0051] In one embodiment, during the VGF shoulder formation stage, the VGF shoulder formation process is monitored to obtain the VGF shoulder formation completion time. Then, based on the VGF shoulder formation completion time, a switching command is automatically triggered to switch to the VB equal diameter growth stage, thus avoiding solid-liquid interface fluctuations caused by inaccurate stage switching timing.

[0052] Therefore, in the VGF shoulder formation stage, the indium phosphide crystal growth method using the VGF-VB combination method further includes: The growth control model is used to identify the indium phosphide crystal state; The VGF shoulder formation progress of indium phosphide crystal is determined based on the crystal state of indium phosphide. When the VGF shoulder formation progress reaches a preset progress threshold, it is determined that the indium phosphide crystal growth has entered the VB constant diameter growth stage, so as to automatically switch the temperature control of each temperature zone in the indium phosphide crystal growth furnace based on the first control parameter to control the indium phosphide crystal growth furnace based on the second control parameter.

[0053] Specifically, the indium phosphide crystal state includes the seed crystal weld length and / or shoulder diameter. The seed crystal weld length is the length of the shoulder already grown after the crystal begins shoulder growth following the completion of the seeding process. The shoulder diameter is the radial dimension of the current shoulder section of the crystal. The growth control model can analyze the advancement rate of the crystal shoulder growth based on the historical trend of the first furnace condition data and the crystal growth process parameters, and identify the current seed crystal weld length and shoulder diameter based on the existing growth time. Then, based on the indium phosphide crystal state, the VGF shoulder growth progress of the indium phosphide crystal (such as the shoulder diameter or the seed crystal weld length) is determined. When the VGF shoulder growth progress reaches a preset progress threshold, the shoulder growth is considered complete, automatically triggering a switching command. This switching command triggers the growth control to enter the VB constant diameter growth stage, eliminating the need for manual judgment of the switching timing and ensuring the timeliness and accuracy of the stage switching.

[0054] S30. After the indium phosphide crystal growth enters the VB constant diameter growth stage, the second furnace condition data of the indium phosphide crystal growth furnace is collected.

[0055] Specifically, after the indium phosphide crystal growth enters the VB constant diameter growth stage, it is necessary to control the temperature regions of the indium phosphide crystal growth furnace to maintain them within the preset temperature range, and control the crucible operation to achieve uniform crystal growth. The second furnace condition data collected at this time also includes temperature data for each temperature region, boiler pressure, and cooling water data, used to reflect the furnace state in real time during the VB constant diameter growth stage, providing real-time decision-making basis for the growth control model. Simultaneously, the collected second furnace condition data also needs to include crucible state data and the three-dimensional temperature field inside the indium phosphide crystal growth furnace. The crucible state data includes the crucible position and rotation speed, reflecting the current movement state of the crucible and providing data support for adjusting the crucible descent and rotation speeds. The three-dimensional temperature field reflects the overall thermal field distribution inside the furnace, enabling the growth control model to accurately determine the solid-liquid interface shape characteristics (such as the solid-liquid interface position and morphology). The crucible state data can be collected via an encoder.

[0056] Furthermore, since indium phosphide must be grown under closed high pressure (>2.75 MPa), the solid-liquid interface cannot be directly observed from the indium phosphide crystal growth furnace. Therefore, this application deploys a thermal imaging acquisition device (such as an infrared thermal imager) outside the indium phosphide crystal growth furnace. This thermal imaging acquisition device acquires real-time temperature distribution images of the outer surface of the furnace wall, and then uses these temperature distribution images to reconstruct the three-dimensional temperature field inside the indium phosphide crystal growth furnace, so as to indirectly reflect the shape characteristics of the solid-liquid interface through this three-dimensional temperature field.

[0057] In one embodiment, the process of obtaining the three-dimensional temperature field inside the indium phosphide crystal growth furnace specifically includes: The temperature distribution image of the outer surface of the indium phosphide crystal growth furnace wall, the thermocouple discrete point data of the indium phosphide crystal growth furnace, and the cooling water data were collected. The temperature distribution image, thermocouple discrete point data, and cooling water data are input into a trained visual model, and the three-dimensional temperature field inside the indium phosphide crystal growth furnace is reconstructed through the visual network.

[0058] Specifically, the temperature distribution image is an infrared thermal image of the outer surface of the furnace wall acquired by the thermal imaging acquisition device, which can intuitively reflect the temperature distribution differences at different locations on the furnace wall; the thermocouple discrete point data is the temperature of key locations inside the furnace directly acquired by thermocouples in each temperature measurement area, which is used to calibrate and constrain the reconstruction process; and the cooling water data is used to correct the temperature deviation caused by the heat dissipation of the furnace body.

[0059] The visual model is a convolutional neural network based on the ResNet architecture. The visual model takes temperature distribution images, thermocouple discrete point data and cooling water data as input data, and uses physical information neural network (PINN) technology to solve the heat conduction equation in reverse to reconstruct the continuous three-dimensional temperature field distribution inside the furnace cavity. This solves the problem that the temperature cannot be directly measured in a closed high-pressure furnace and provides reliable data support for judging the solid-liquid interface state.

[0060] Furthermore, after acquiring the three-dimensional temperature field, in addition to using it for solid-liquid interface identification, it can also be used for anomaly warning to prevent the generation of twins and dislocations. That is, the acquisition process of the three-dimensional temperature field can also include an anomaly warning process, which specifically includes: Detect whether there are any abnormalities in the three-dimensional temperature field; When the three-dimensional temperature field is abnormal, the three-dimensional temperature field is compared with a preset standard temperature field to determine the abnormal data. Supplementary control parameters are generated based on the abnormal data, and the supplementary control parameters are used to perform supplementary control on the indium phosphide crystal growth furnace.

[0061] Specifically, abnormal conditions refer to the presence of abnormal hot spots or asymmetrical distributions. In other words, when acquiring a 3D temperature field, the visual model performs anomaly detection (i.e., identifies whether abnormal hot spots or asymmetrical distributions exist) to ensure that the output 3D temperature field carries an anomaly category. This category can be abnormal hot spots, asymmetrical distributions, or normal conditions. Then, after acquiring the 3D temperature field, the presence of abnormal conditions can be detected based on the anomaly category carried by the 3D temperature field.

[0062] Furthermore, when anomalies exist in the three-dimensional temperature field, it indicates that the actual distribution of the current three-dimensional temperature field deviates from the preset standard distribution, leading to instability in the solid-liquid interface shape and increasing the risk of twinning and dislocation formation. Therefore, the three-dimensional temperature field is compared with the preset standard temperature field to identify abnormal data, and supplementary control parameters are generated based on this abnormal data to supplement the control of the indium phosphide crystal growth furnace. The abnormal data includes one or more of the following: hotspot locations, temperature deviation values, and solid-liquid interface shapes. The supplementary control parameters include the compensation temperature power adjustment amount for the temperature region where the anomaly occurs. Conversely, when there are no anomalies in the three-dimensional temperature field, it indicates that the current thermal field distribution meets the process requirements and no additional adjustments are needed.

[0063] In one embodiment, such as Figure 2 and Figure 3 As shown, the visual model (i.e., the CNN visual model) and the growth control model (i.e., the DNN process model) work synchronously and in parallel through a multi-threaded mechanism. That is, during the control process utilizing the growth control model, a supplementary control process is also executed synchronously. The specific parallel execution process of the control process utilizing the growth control model and the supplementary control process is as follows: During the process of controlling the indium phosphide crystal growth furnace based on the second control parameter, it is monitored whether supplementary control parameters are received. If supplementary control parameters are received, the control process of controlling the indium phosphide crystal growth furnace based on the second control parameters is interrupted. The indium phosphide crystal growth furnace is supplemented with the supplementary control parameters until there are no abnormalities in the three-dimensional temperature field, and then the control process based on the second control parameters to control the indium phosphide crystal growth furnace is resumed.

[0064] Specifically, during the process of controlling the indium phosphide crystal growth furnace based on the control parameters generated by the growth control model, if no supplementary control parameters are received, the furnace will be directly controlled using the control parameters determined by the growth control model. If supplementary control parameters are received, they will be used as high-priority interrupt / compensation inputs. Then, local power fine-tuning will be performed based on the supplementary control parameters (e.g., selectively reducing the heating power of the temperature zone corresponding to the abnormal hot spot). Alternatively, the supplementary control parameters can be used as input data for the growth control model to fine-tune the crucible rotation speed through the reinforcement learning module in the growth control model to enhance melt convection and thus create heat dissipation points. The abnormal state detection process of the three-dimensional temperature field will be re-executed until there are no abnormalities in the three-dimensional temperature field, and the process of controlling the indium phosphide crystal growth furnace based on the control parameters generated by the growth control model will be resumed.

[0065] The embodiments of this application, through three-dimensional temperature field anomaly detection and real-time adjustment, can promptly correct thermal field deviations during the constant diameter growth stage, suppress the generation of twins and dislocations from the source, and improve the overall crystallization quality of indium phosphide crystals.

[0066] S40. The growth control model determines the second control parameters based on the second furnace condition data, and controls the indium phosphide crystal growth furnace based on the second control parameters, so that each temperature region in the indium phosphide crystal growth furnace is maintained within a preset temperature range and the crucible operates according to the crucible speed data.

[0067] Specifically, the growth control model also includes a reinforcement learning module, which predicts second control parameters based on the first control parameters, crucible state data, and the three-dimensional temperature field. The second control parameters include target heating power and crucible speed data for each temperature region in the indium phosphide crystal growth furnace. The target heating power for each temperature region is used to maintain each temperature region in the indium phosphide crystal growth furnace within a preset temperature range (e.g., 1030-1080℃) to ensure uniform heat distribution. The crucible speed data includes the crucible descent speed (e.g., within 1.5-2.5 mm / h) and the crucible rotation speed (e.g., within 1-5 r / h), used to control the crucible to descend at a uniform speed and drive the solid-liquid interface to move at a uniform speed, achieving stable, constant-diameter crystal growth. Simultaneously, adjusting the crucible rotation speed can adjust the melt convection state, optimize the solid-liquid interface shape, and reduce impurity segregation.

[0068] It is understandable that when obtaining the second control parameter, the timing prediction module in the growth control model will first use the first working condition data to determine the first control parameter (i.e., the target heating power of each temperature region), and then the reinforcement learning module will use the first control parameter, crucible state data and three-dimensional temperature field to determine the crucible speed data. Finally, the first control parameter and crucible speed data will be output as the second control parameter to adapt to the small changes in the furnace thermal field during the constant diameter growth process and ensure the stability of the crystal growth process.

[0069] Furthermore, after obtaining the second control parameter, the heating power of each temperature zone in the indium phosphide crystal growth furnace is adjusted based on this parameter. Simultaneously, the descent and rotation speeds of the crucible are updated (e.g., by controlling the lifting and rotating motors via servo drives to update the crucible's descent and rotation speeds), ensuring that the thermal field distribution within the furnace always conforms to the crystal growth process requirements and avoiding crystallization defects caused by thermal field deviation or crucible motion parameter mismatch. Through real-time decision-making by the growth control model, minute state changes during the constant-diameter growth process can be dynamically adapted without manual intervention, maintaining a stable solid-liquid interface shape and ensuring uniform indium phosphide crystal growth.

[0070] In one embodiment, to improve computational efficiency and meet the real-time control requirements of the indium phosphide crystal growth process using the VGF-VB combination method, the method of this application is implemented based on a cloud-edge collaborative architecture. This cloud-edge collaborative architecture includes: An edge industrial control computer deploys a trained growth control model (.onnx or .tflite format) and a vision model, acquiring control parameters through these models. For example, the growth control model and vision model are deployed on an edge industrial computer or high-performance PLC corresponding to an indium phosphide crystal growth furnace. This edge industrial control computer executes real-time closed-loop control according to a preset control cycle (e.g., 5s, 6s) to achieve a one-click crystal growth process. Figure 4 As shown, this one-click crystal growth process automatically completes vacuuming and filling after the operator clicks "One-click Start" on the interface. The entire process, from protective gas > heating and melting > seed crystal orientation > VGF shoulder growth > automatic triggering of switching commands > VB constant diameter growth > annealing, requires no manual intervention.

[0071] The cloud server (serving as a training server) is responsible for collecting historical growth data of indium phosphide crystals grown using the VGF-VB combination method. Based on the collected historical growth data, the growth control model and vision model are optimized offline, and the optimized growth control model and vision model are pushed to the edge industrial control computer through a secure HTTPS channel.

[0072] Based on this, in one embodiment, the method further includes: Historical data on multiple indium phosphide crystal growth processes are aggregated through a cloud server; The growth control model is retrained and / or fine-tuned using the historical data, and the retrained and / or fine-tuned growth control model is deployed on an edge industrial control computer so that the edge industrial control computer can use the retrained and / or fine-tuned growth control model to control the indium phosphide crystal growth process using the VGF-VB combination method.

[0073] Specifically, such as Figure 4 As shown, after the growth cycle is completed, the extracted indium phosphide crystals are subjected to quality testing (dislocation density EPD test, Hall effect test, carrier concentration and mobility test, PL spectrum test). If the test results meet the standards (e.g., tail-end EPD < 300), the crystal is considered acceptable. The electrical parameter uniformity deviation is <±6%). Complete time-series data of the indium phosphide crystal (including temperature data for each temperature region, boiler high-pressure gas, cooling water data, and crucible state data) and infrared visual data (i.e., temperature distribution image of the furnace wall outer surface) are marked as positive samples to obtain historical data. Then, the historical data is reported to a cloud server so that the cloud server can aggregate historical data from multiple indium phosphide crystal growth processes. Finally, at preset time intervals, the aggregated historical data is used to retrain and / or fine-tune the growth control model and the visual model, achieving continuous iterative optimization of the process model.

[0074] Based on the above-described indium phosphide crystal growth method using the VGF-VB combination method, this embodiment provides an indium phosphide crystal growth apparatus using the VGF-VB combination method, such as... Figure 5 As shown, the indium phosphide crystal growth apparatus utilizing the VGF-VB combination method specifically includes: The first furnace condition data acquisition module is used to acquire the first furnace condition data of the indium phosphide crystal growth furnace when the indium phosphide crystal growth enters the VGF shoulder formation stage. The first furnace condition data includes temperature data of each temperature zone in the indium phosphide crystal growth furnace, boiler high gas pressure and cooling water data. The first control parameter determination module is used to control the growth control model of the trained indium phosphide crystal to determine the first control parameter based on the first furnace condition data, and to control the temperature of each temperature zone in the indium phosphide crystal growth furnace based on the first control parameter. The first control parameter includes the target heating power of each temperature zone in the indium phosphide crystal growth furnace. The second furnace condition data acquisition module is used to acquire the second furnace condition data of the indium phosphide crystal growth furnace after the indium phosphide crystal growth enters the VB constant diameter growth stage. The second furnace condition data includes the first furnace condition data, crucible status data and the three-dimensional temperature field inside the indium phosphide crystal growth furnace. The second control parameter determination module is used to control the growth control model to determine the second control parameters based on the second furnace condition data, and to control the indium phosphide crystal growth furnace based on the second control parameters, so that each temperature region in the indium phosphide crystal growth furnace is maintained within a preset temperature range and the crucible operates according to the crucible speed data. The second control parameters include the target heating power and crucible speed data of each temperature region in the indium phosphide crystal growth furnace.

[0075] Based on the above-described indium phosphide crystal growth method using the VGF-VB combination method, this embodiment provides a computer-readable storage medium storing one or more programs that can be executed by one or more processors to implement the steps in the indium phosphide crystal growth method using the VGF-VB combination method as described in the above embodiment.

[0076] Based on the above-described indium phosphide crystal growth method utilizing the VGF-VB combination method, this application also provides a terminal device, such as... Figure 6 As shown, it includes at least one processor 20; a display screen 21; and a memory 22, and may also include a communications interface 23 and a bus 24. The processor 20, display screen 21, memory 22, and communications interface 23 can communicate with each other via the bus 24. The display screen 21 is configured to display a preset user guide interface in the initial setup mode. The communications interface 23 can transmit information. The processor 20 can invoke logical instructions in the memory 22 to execute the methods described in the above embodiments.

[0077] Furthermore, the logical instructions in the aforementioned memory 22 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium.

[0078] The memory 22, as a computer-readable storage medium, can be configured to store software programs, computer-executable programs, such as program instructions or modules corresponding to the methods in the embodiments of this disclosure. The processor 20 executes functional applications and data processing by running the software programs, instructions, or modules stored in the memory 22, thereby implementing the methods in the above embodiments.

[0079] The memory 22 may include a program storage area and a data storage area. The program storage area may store the operating system and application programs required for at least one function; the data storage area may store data created based on the use of the terminal device. Furthermore, the memory 22 may include high-speed random access memory (RAM) and non-volatile memory. Examples include various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks, as well as transient storage media.

[0080] Furthermore, the specific process of loading and executing multiple instruction processors in the aforementioned storage medium and terminal device has been described in detail in the above method, and will not be repeated here.

[0081] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for growing indium phosphide crystals using the VGF-VB combination method, characterized in that, The indium phosphide crystal growth method using the VGF-VB combination method specifically includes: Starting from the VGF shoulder formation stage of indium phosphide crystal growth, the first furnace condition data of the indium phosphide crystal growth furnace is collected. The first furnace condition data includes temperature data of each temperature zone in the indium phosphide crystal growth furnace, furnace height gas pressure and cooling water data. The growth control model for controlling the trained indium phosphide crystal determines the first control parameter based on the first furnace condition data, and controls the temperature of each temperature zone in the indium phosphide crystal growth furnace based on the first control parameter. The first control parameter includes the target heating power of each temperature zone in the indium phosphide crystal growth furnace. When the indium phosphide crystal growth enters the VB constant diameter growth stage, the second furnace condition data of the indium phosphide crystal growth furnace is collected. The second furnace condition data includes the first furnace condition data, crucible condition data and the three-dimensional temperature field inside the indium phosphide crystal growth furnace. The growth control model determines the second control parameters based on the second furnace condition data, and controls the indium phosphide crystal growth furnace based on the second control parameters, so that each temperature region in the indium phosphide crystal growth furnace is maintained within a preset temperature range and the crucible operates according to the crucible speed data. The second control parameters include the target heating power and crucible speed data for each temperature region in the indium phosphide crystal growth furnace.

2. The indium phosphide crystal growth method using the VGF-VB combination method according to claim 1, characterized in that, The growth control model includes a time-series prediction module and a reinforcement learning module. The time-series prediction module is used to predict a first control parameter based on the first furnace condition data, and the reinforcement learning module is used to predict a second control parameter based on the first control parameter, crucible state data, and a three-dimensional temperature field.

3. The indium phosphide crystal growth method using the VGF-VB combination method according to claim 1, characterized in that, The process of obtaining the three-dimensional temperature field inside the indium phosphide crystal growth furnace specifically includes: The temperature distribution image of the outer surface of the indium phosphide crystal growth furnace wall, the thermocouple discrete point data of the indium phosphide crystal growth furnace, and the cooling water data were collected. The temperature distribution image, thermocouple discrete point data, and cooling water data are input into a trained visual model, and the three-dimensional temperature field inside the indium phosphide crystal growth furnace is reconstructed through the visual model.

4. The indium phosphide crystal growth method using the VGF-VB combination method according to claim 1 or 3, characterized in that, In the process of acquiring the three-dimensional temperature field, the method further includes: Detect whether there are any abnormalities in the three-dimensional temperature field; When the three-dimensional temperature field is abnormal, the three-dimensional temperature field is compared with the preset standard temperature field to determine the abnormal data. The abnormal data includes one or more of the following: hot spot location, temperature deviation value, and solid-liquid interface shape. Supplementary control parameters are generated based on the abnormal data, and the supplementary control parameters are used to supplement the control of the indium phosphide crystal growth furnace. The supplementary control parameters include the compensation temperature and power adjustment amount for the temperature region where the abnormality occurs.

5. The indium phosphide crystal growth method using the VGF-VB combination method according to claim 4, characterized in that, In the process of controlling the indium phosphide crystal growth furnace based on the second control parameter, the method further includes: If supplementary control parameters are received, the control process of controlling the indium phosphide crystal growth furnace based on the second control parameters is interrupted. The indium phosphide crystal growth furnace is supplemented with the supplementary control parameters until there are no abnormalities in the three-dimensional temperature field, and then the control process based on the second control parameters to control the indium phosphide crystal growth furnace is resumed.

6. The indium phosphide crystal growth method using the VGF-VB combination method according to claim 1, characterized in that, During the VGF shoulder-releasing stage, the method further includes: The growth control model is controlled to identify the indium phosphide crystal state, which includes the seed crystal weld length and / or shoulder diameter. The VGF shoulder formation progress of indium phosphide crystal is determined based on the crystal state of indium phosphide. When the VGF shoulder formation progress reaches a preset progress threshold, it is determined that the indium phosphide crystal growth has entered the VB constant diameter growth stage, so as to automatically switch the temperature control of each temperature zone in the indium phosphide crystal growth furnace based on the first control parameter to control the indium phosphide crystal growth furnace based on the second control parameter.

7. The indium phosphide crystal growth method using the VGF-VB combination method according to claim 1, characterized in that, The method further includes: Historical data on multiple indium phosphide crystal growth processes are aggregated through a cloud server; The growth control model is retrained and / or fine-tuned using the historical data, and the retrained and / or fine-tuned growth control model is deployed on an edge industrial control computer so that the edge industrial control computer can use the retrained and / or fine-tuned growth control model to control the indium phosphide crystal growth process using the VGF-VB combination method.

8. An indium phosphide crystal growth apparatus using the VGF-VB combination method, characterized in that, The indium phosphide crystal growth apparatus utilizing the VGF-VB combination method specifically includes: The first furnace condition data acquisition module is used to acquire the first furnace condition data of the indium phosphide crystal growth furnace when the indium phosphide crystal growth enters the VGF shoulder formation stage. The first furnace condition data includes temperature data of each temperature zone in the indium phosphide crystal growth furnace, boiler high gas pressure and cooling water data. The first control parameter determination module is used to control the growth control model of the trained indium phosphide crystal to determine the first control parameter based on the first furnace condition data, and to control the temperature of each temperature zone in the indium phosphide crystal growth furnace based on the first control parameter. The first control parameter includes the target heating power of each temperature zone in the indium phosphide crystal growth furnace. The second furnace condition data acquisition module is used to acquire the second furnace condition data of the indium phosphide crystal growth furnace after the indium phosphide crystal growth enters the VB constant diameter growth stage. The second furnace condition data includes the first furnace condition data, crucible status data and the three-dimensional temperature field inside the indium phosphide crystal growth furnace. The second control parameter determination module is used to control the growth control model to determine the second control parameters based on the second furnace condition data, and to control the indium phosphide crystal growth furnace based on the second control parameters, so that each temperature region in the indium phosphide crystal growth furnace is maintained within a preset temperature range and the crucible operates according to the crucible speed data. The second control parameters include the target heating power and crucible speed data of each temperature region in the indium phosphide crystal growth furnace.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores one or more programs, which can be executed by one or more processors to implement the steps in the indium phosphide crystal growth method using the VGF-VB combination method as described in any one of claims 1-7.

10. A terminal device, characterized in that, include: Processor and memory; The memory stores a computer-readable program that can be executed by the processor; When the processor executes the computer-readable program, it implements the steps in the indium phosphide crystal growth method using the VGF-VB combination method as described in any one of claims 1-7.

Images