An oxygen content prediction method and system applied to a single crystal silicon preparation process

Abstract

This application provides a method and system for predicting oxygen content in the process of preparing monocrystalline silicon. During the preparation of monocrystalline silicon in a monocrystalline furnace, multiple key process parameters affecting the oxygen content within the furnace are collected in real time from various sensors installed on the furnace. Based on these key process parameters, target data features reflecting the changing trend of oxygen content are calculated. These target data features are then input into a pre-trained oxygen content prediction model, which outputs the predicted oxygen content. This application enables real-time prediction of oxygen content within the monocrystalline furnace based on key process parameters collected during the monocrystalline silicon preparation process. This not only overcomes the technical limitation of existing technologies that can only passively detect oxygen content after the monocrystalline silicon preparation is complete, but also effectively reduces the reliance on high-energy-consuming magnetic fields in traditional control methods, thus reducing related preparation and control costs and facilitating the promotion and implementation of the solution.

Description

Technical Field

[0001] This application relates to the field of monocrystalline silicon preparation technology, and more specifically, to a method and system for predicting oxygen content in the process of monocrystalline silicon preparation. Background Technology

[0002] Currently, in existing technologies, monocrystalline silicon is often prepared using the CZ method (Czochralski method, a core process for monocrystalline silicon preparation, which mainly involves melting polycrystalline silicon in a quartz crucible and slowly pulling monocrystalline silicon rods from the melt using seed crystals). During the preparation of monocrystalline silicon, oxygen impurities inevitably release into the melt from the quartz crucible, which leads to changes in the oxygen content within the monocrystalline furnace (the equipment for preparing monocrystalline silicon). Both excessively high and low oxygen content will affect the quality of the monocrystalline silicon wafer (excessive oxygen content will induce "concentric circle" defects, while insufficient oxygen content will easily produce thermal donor effects).

[0003] Currently, to effectively control the oxygen content within a reasonable range in a single-crystal furnace, the traditional control method involves increasing the magnetic field to suppress the release of oxygen impurities from the quartz crucible into the melt. However, because this traditional control method cannot obtain real-time information on the oxygen concentration distribution within the melt of the single-crystal furnace, it lacks the ability to dynamically monitor the oxygen content in real time. The actual oxygen content of the single-crystal silicon can only be passively detected after the preparation process is complete by testing a sample of the prepared single-crystal silicon. Furthermore, increasing the magnetic field requires a significant amount of additional electricity, making the preparation cost far higher than the traditional CZ preparation method (i.e., the CZ method without an added magnetic field). Therefore, the traditional control method suffers from high costs and is not conducive to widespread adoption. Summary of the Invention

[0004] In view of this, this application provides a method and system for predicting oxygen content in the process of preparing monocrystalline silicon. Based on the key process parameters collected in real time during the preparation of monocrystalline silicon, the oxygen content in the monocrystalline furnace can be predicted in real time. This not only overcomes the technical defect of the prior art that can only passively detect oxygen content after the preparation of monocrystalline silicon is completed, but also effectively reduces the dependence on high-energy-consuming magnetic fields in traditional control methods. This is conducive to reducing the relevant preparation and control costs and facilitates the promotion and implementation of the solution.

[0005] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings.

[0006] In a first aspect, embodiments of this application provide a method for predicting oxygen content in the process of preparing single-crystal silicon, the method comprising: During the preparation of monocrystalline silicon using a single crystal furnace, multiple key process parameters affecting the oxygen content inside the single crystal furnace are collected in real time from multiple sensors installed on the single crystal furnace. Based on the aforementioned key process parameters, target data features are calculated to reflect the changing trend of the oxygen content. The target data features are input into a pre-trained oxygen content prediction model, and the oxygen content prediction result is obtained by outputting the oxygen content prediction model.

[0007] Secondly, embodiments of this application provide an oxygen content prediction system applied in the preparation process of single-crystal silicon. The oxygen content prediction system includes at least one single-crystal furnace and a detection platform, wherein the detection platform is used for: During the process of preparing monocrystalline silicon using the monocrystalline furnace, multiple key process parameters affecting the oxygen content inside the monocrystalline furnace are collected in real time from multiple sensors installed on the monocrystalline furnace. Based on the aforementioned key process parameters, target data features are calculated to reflect the changing trend of the oxygen content. The target data features are input into a pre-trained oxygen content prediction model, and the oxygen content prediction result is obtained by outputting the oxygen content prediction model.

[0008] Thirdly, embodiments of this application provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the oxygen content prediction method described above.

[0009] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the oxygen content prediction method described above.

[0010] Fifthly, embodiments of this application provide a single-crystal silicon rod, which is manufactured using the oxygen content prediction method described above.

[0011] The technical solutions provided by the embodiments of this application may include the following beneficial effects: This application provides a method and system for predicting oxygen content in the process of preparing monocrystalline silicon. During the preparation of monocrystalline silicon in a monocrystalline furnace, multiple key process parameters affecting the oxygen content within the furnace are collected in real time from multiple sensors installed on the furnace. Based on these key process parameters, target data features reflecting the changing trend of oxygen content are calculated. These target data features are input into a pre-trained oxygen content prediction model, which outputs the predicted oxygen content. This application enables real-time prediction of oxygen content within the monocrystalline furnace based on key process parameters collected during the monocrystalline silicon preparation process. This not only overcomes the technical deficiency of existing technologies that can only passively detect oxygen content after the monocrystalline silicon preparation is completed, but also effectively reduces the dependence on high-energy-consuming magnetic fields in traditional control methods, thus reducing related preparation and control costs and facilitating the promotion and implementation of the solution. Attached Figure Description

[0012] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This illustration shows a schematic diagram of an oxygen content prediction system applied in the single-crystal silicon preparation process, provided by an embodiment of this application. Figure 2 A schematic flowchart of an oxygen content prediction method applied in the preparation process of single crystal silicon, provided by an embodiment of this application, is shown. Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the accompanying drawings in this application are for illustrative and descriptive purposes only and are not intended to limit the scope of protection of this application. Furthermore, it should be understood that the schematic drawings are not drawn to scale. The flowcharts used in this application illustrate operations implemented according to some embodiments of this application. It should be understood that the operations in the flowcharts may not be implemented in sequence, and steps without logical contextual relationships may be reversed or implemented simultaneously. In addition, those skilled in the art, guided by the content of this application, may add one or more other operations to the flowcharts, or remove one or more operations from the flowcharts.

[0015] Furthermore, the described embodiments are merely some, not all, of the embodiments of this application. The components of the embodiments of this application described and illustrated herein can typically be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0016] It should be noted that the term "comprising" will be used in the embodiments of this application to indicate the presence of the features declared thereafter, but does not exclude the addition of other features.

[0017] Here, the oxygen content prediction method provided in this application embodiment can be applied to an oxygen content prediction system. Figure 1 This illustration shows a schematic diagram of an oxygen content prediction system applied in the single-crystal silicon preparation process, as provided in an embodiment of this application. Figure 1 As shown, the oxygen content prediction system includes at least one single crystal furnace and a detection platform. The detection platform is connected to the sensors installed on each single crystal furnace, and can obtain various key process parameters of the single crystal furnace in the process of preparing single crystal silicon in real time from the sensors. Thus, the detection platform can predict the oxygen content in the single crystal furnace in real time based on the key process parameters collected in real time during the preparation of single crystal silicon.

[0018] Based on this, the oxygen content prediction system provided in this application not only overcomes the technical defect that the existing technology can only passively detect oxygen content after the preparation of single crystal silicon is completed, but also effectively reduces the dependence on high-energy-consuming magnetic fields in traditional control methods, which is conducive to reducing related preparation and control costs and facilitates the promotion and implementation of the solution.

[0019] To facilitate understanding of the embodiments of this application, the following description is provided in conjunction with... Figure 1 The oxygen content prediction system shown here provides a detailed description of an oxygen content prediction method applied to the preparation process of single crystal silicon, as provided in the embodiments of this application.

[0020] Reference Figure 2 As shown, Figure 2 The diagram illustrates a flowchart of an oxygen content prediction method applied in the preparation process of single-crystal silicon, provided by an embodiment of this application. The oxygen content prediction method includes steps S201-S203; specifically: S201, during the process of preparing monocrystalline silicon using a monocrystalline furnace, multiple key process parameters affecting the oxygen content inside the monocrystalline furnace are collected in real time from multiple sensors installed on the monocrystalline furnace.

[0021] Here, step S201 runs through the entire process of preparing single crystal silicon. That is, during the process of preparing single crystal silicon in a single crystal furnace, the detection platform can collect multiple key process parameters that affect the oxygen content in the single crystal furnace in real time from multiple sensors installed on the single crystal furnace at each acquisition moment, thereby obtaining the time series data corresponding to each key process parameter (equivalent to a sequence composed of key process parameters at each acquisition moment).

[0022] Specifically, the aforementioned key process parameters include at least the heating power and pulling rate of the single crystal furnace. In addition, the aforementioned key process parameters may also include, but are not limited to, the temperature gradient of the single crystal furnace, crucible rotation speed, cooling rate, magnetic field strength, and other process parameters. The specific types and quantities of the aforementioned key process parameters are not subject to mandatory limitations in the embodiments of this application.

[0023] It should be noted that different types of key process parameters require different sensors. For example, the heating power mentioned above can be collected by a furnace power sensor installed on the single crystal furnace, while the pulling rate can be collected by a speed sensor installed on the single crystal furnace.

[0024] S202, Based on the multiple key process parameters, calculate the target data features used to reflect the changing trend of the oxygen content.

[0025] Here, feature engineering can be used to process the collected key process parameters to obtain the derived features corresponding to these key process parameters, which can then be used as target data features for predicting the oxygen content in the single crystal furnace.

[0026] It should be noted that, referring to the content of step S201 above, since the detection platform collects time series data corresponding to each key process parameter in real time (equivalent to a sequence composed of key process parameters at each collection moment), the target data features obtained by performing feature engineering on the time series data corresponding to the key process parameters (i.e., each key process parameter collected in real time) can reflect the characteristics of the key process parameters changing over time. Since the key process parameters are process parameters that affect the oxygen content in the single crystal furnace, the target data features are also equivalent to reflecting the characteristics of the oxygen content changing over time (i.e., reflecting the trend of the oxygen content change).

[0027] Specifically, as an optional embodiment, for the "heating power" among the above-mentioned key process parameters, the detection platform can calculate the heat flow change rate in the single crystal furnace as the derived feature corresponding to the "heating power" (i.e., the above-mentioned target data feature) according to the method shown in steps a1-a2 below: Step a1: For the heating power obtained at each acquisition time, divide the heating power by the effective area of ​​the heating zone in the single crystal furnace to calculate the average heat flux density of the single crystal furnace at that acquisition time.

[0028] Here, heat flux density refers to the amount of heat transferred through a unit area per unit time, and the original calculation formula is shown in Formula 1 below: Formula 1; Where q is the heat flux density, Q is the total heat transferred by the object, s is the area of ​​the object, and t is the time it takes for the object to transfer heat.

[0029] Specifically, when preparing monocrystalline silicon using a single-crystal furnace (e.g., using the CZ method), unlike the original calculation formula shown in Formula 1 above, the detection platform acquires the heating power of the single-crystal furnace at the i-th acquisition moment in real time using a furnace power sensor. Then, the average heat flux density of the single crystal furnace at the i-th sampling time can be calculated according to the following formula 2. : Formula 2; in, It is the heating power of the single crystal furnace at the i-th acquisition moment, which is collected in real time by the furnace power sensor through the detection platform; S is the effective area of ​​the heating zone inside the single crystal furnace (determined according to the specifications and dimensions of the single crystal furnace); It is the average heat flux density of the single crystal furnace at the i-th sampling time.

[0030] Step a2: Calculate the difference in average heat flux density of the single crystal furnace at adjacent acquisition times, and divide the difference by the time interval between adjacent acquisition times to obtain the heat flux change rate of the single crystal furnace as the target data feature.

[0031] Here, the rate of change of heat flux is the rate at which the average heat flux density changes over time, and it can be used to monitor the stability of heat flux in a single crystal furnace.

[0032] Specifically, when performing step a2, the adjacent acquisition time is the (i-1)th acquisition time. and the i-th acquisition time For example, the rate of change of heat flux between the (i-1)th and ith sampling times of a single crystal furnace can be calculated using the following formula 3: Formula 3; in, It is the rate of change of heat flux of the single crystal furnace between the (i-1)th and the ith sampling time. It is the average heat flux density of the single crystal furnace at the i-th sampling time; It is the average heat flux density of the single crystal furnace at the (i-1)th sampling time. This is the i-th data acquisition time. This is the (i-1)th data collection moment.

[0033] Specifically, as an optional embodiment, for the "lifting rate" among the above-mentioned key process parameters, the detection platform can calculate the differential term of the "lifting rate" as the derived feature (i.e., the above-mentioned target data feature) according to the method shown in steps b1-b4 below: Step b1: Obtain the pulling rate corresponding to the single crystal furnace at multiple consecutive acquisition times to obtain the pulling rate sequence corresponding to the single crystal furnace.

[0034] Here, the pulling rate is a key process parameter that affects the formation of crystal defects. The detection platform can obtain the pulling rate corresponding to the single crystal furnace at multiple consecutive acquisition times from multiple process parameters acquired in real time, and obtain the pulling rate sequence corresponding to the single crystal furnace.

[0035] Step b2: Calculate the first derivative of the pulling rate sequence to obtain the first data feature that reflects the pulling rate change rate of the single crystal furnace.

[0036] Here, the lifting rate sequence is a discrete time series composed of lifting rates at multiple acquisition moments. Calculating the differential term (e.g., derivative) of the lifting rate sequence can reflect the trend of lifting rate change over time. Therefore, using the calculated differential term as the target data feature is more beneficial for subsequent oxygen content prediction than using the static instantaneous value of the lifting rate (i.e., the lifting rate at each acquisition moment).

[0037] Specifically, when calculating the first derivative, the rate difference between the lifting rates at two adjacent acquisition times can be calculated first. Then, the calculated rate difference can be divided by the time interval between the two adjacent acquisition times to obtain the calculation result as the first data feature.

[0038] It should be noted that since the first data feature is the first derivative of the pulling rate sequence, the first data feature can be used to reflect the acceleration or deceleration of the pulling rate in the single crystal furnace. When the first data feature is greater than zero, it can be determined that the pulling rate in the single crystal furnace is accelerating, which may lead to stress concentration in the crystal. When the first data feature is less than zero, it can be determined that the pulling rate in the single crystal furnace is decelerating, which may lead to uneven crystal formation.

[0039] Step b3: Calculate the second derivative of the pulling rate sequence to obtain a second data feature that reflects the rate of change of the pulling rate acceleration of the single crystal furnace.

[0040] Here, by calculating the second derivative of the pulling rate sequence, the result of the second derivative calculation can be used as the rate of change of the pulling rate acceleration of the single crystal furnace (i.e., the second data feature). Since the second data feature is the rate of change of the pulling rate acceleration of the single crystal furnace, the second data feature can be used to reflect the smoothness of the pulling rate change.

[0041] Step b4: Use the first data feature and the second data feature as the target data feature.

[0042] Specifically, since both the first data feature and the second data feature are differential terms of the aforementioned "lifting rate", they can be used as derived features corresponding to the "lifting rate" (i.e., the aforementioned target data features).

[0043] S203, the target data features are input into a pre-trained oxygen content prediction model, and the oxygen content prediction result is obtained by outputting the oxygen content prediction model.

[0044] Here, during the historical preparation process of monocrystalline silicon, historical process parameters and corresponding historical target data features can be collected. After the preparation of monocrystalline silicon is completed, the prepared monocrystalline silicon sample is tested. The intensity of infrared light transmitted through the monocrystalline silicon sample is measured using a Fourier transform infrared spectrometer to obtain the infrared absorption spectrum. Based on the relationship between the peak value representing the absorption intensity in the infrared absorption spectrum and the oxygen content, the actual oxygen content detection result corresponding to the above historical preparation process is obtained. The above historical target data features and the above actual oxygen content detection result are stored as a set of matched historical detection data in the database.

[0045] It should be noted that the method for obtaining historical target data features is the same as the method for obtaining target data features in the aforementioned steps S201-S202, and the repetition will not be repeated here.

[0046] Specifically, during the model training phase, multiple sets of matching historical detection data can be obtained from the aforementioned database as model training data. For each set of historical detection data, the historical target data features in that set of historical detection data can be input into the oxygen content prediction model to be trained to obtain the oxygen content prediction result from the model. Then, based on the prediction loss between the oxygen content prediction result and the actual oxygen content detection result in that set of historical detection data, the model parameters of the oxygen content prediction model are adjusted until the oxygen content prediction model converges, resulting in a well-trained oxygen content prediction model.

[0047] It should be noted that the specific model structure of the above oxygen content prediction model can be an LSTM (Long Short-Term Memory) model or an XGBoost (Gradient Boosting Tree) model. This application does not impose a mandatory limitation on the specific model structure of the above oxygen content prediction model.

[0048] Specifically, in the model application stage, the target data features obtained in step S202 are input into the pre-trained oxygen content prediction model to obtain the oxygen content prediction result output by the model. Referring to the relevant explanations of the target data features in steps S201-S202, it can be seen that since the target data features reflect the changes of key process parameters over time within a unit time interval (i.e., the time interval between two adjacent acquisition times), for each unit time interval, the oxygen content prediction model can output an oxygen content prediction result corresponding to that unit time interval based on the target data features. Thus, discrete acquisition data (i.e., key process parameters at each acquisition time) can also be converted into continuous oxygen content prediction data (e.g., an oxygen content prediction curve can be plotted based on the oxygen content prediction result of each unit time interval).

[0049] Here, after obtaining the above predicted oxygen content, the detection platform can also actively determine whether there is a risk of excessive oxygen content in the single crystal furnace through the following steps c1-c2: Step c1: Determine whether the predicted oxygen content exceeds a preset safety threshold.

[0050] Here, the specific value of the safety threshold can be flexibly adjusted according to the actual detection requirements for monocrystalline silicon preparation. This application does not impose a mandatory limitation on the specific value of the safety threshold.

[0051] Step c2: If it is determined that the predicted oxygen content exceeds the safety threshold, then a target warning signal is generated.

[0052] Specifically, when the predicted oxygen content is detected to be greater than the aforementioned safety threshold, it can be determined that there is a risk of excessive oxygen content in the single crystal furnace. At this time, a target warning signal can be triggered so that relevant operators can adjust the oxygen content in the single crystal furnace in a timely manner after receiving the target warning signal, so as to ensure that the oxygen content in the single crystal furnace is always within a reasonable range, which is conducive to ensuring the production quality of single crystal silicon.

[0053] It should be noted that the aforementioned target warning signal is a pre-set warning signal used to reflect excessive oxygen content in the single crystal furnace; wherein, the target warning signal can be a text-type prompt message (e.g., a warning message about excessive oxygen content in the single crystal furnace can be displayed on the display screen on one side of the detection platform), or it can be a voice-type warning signal (e.g., a voice prompt about excessive oxygen content in the single crystal furnace can be broadcast through a voice device on one side of the detection platform). The specific signal type of the aforementioned target warning signal is not subject to mandatory limitation in this application embodiment.

[0054] Here, regarding the aforementioned oxygen content prediction model, as an optional embodiment, the detection platform can also monitor the online performance of the oxygen content prediction model using the methods shown in steps d1-d3 below. This allows for timely fine-tuning of the model parameters when the online performance of the oxygen content prediction model falls short of expectations, ensuring the accuracy of the oxygen content prediction results output by the model. Specifically: Step d1: After detecting that the single crystal furnace has finished preparing single crystal silicon, determine the actual oxygen content of the single crystal silicon during the preparation process from the prepared single crystal silicon by offline detection.

[0055] Here, the specific implementation of step d1 is the same as the method for obtaining the actual oxygen content detection results in the aforementioned model training phase, and the repetitions will not be repeated here.

[0056] Step d2: Determine whether the error between the actual oxygen content and the predicted oxygen content exceeds a preset error threshold.

[0057] Here, the specific value of the error threshold can be flexibly adjusted according to the actual model evaluation requirements. This application does not impose a mandatory limitation on the specific value of the error threshold.

[0058] Specifically, when the error between the actual oxygen content and the predicted oxygen content output by the model is greater than the error threshold, it can be determined that the online performance of the oxygen content prediction model is not up to expectations, and the model parameters of the oxygen content prediction model need to be fine-tuned.

[0059] Step d2: If it is determined that the error exceeds the error threshold, construct new model training data based on the target data features and the actual oxygen content, and fine-tune the model parameters of the oxygen content prediction model based on the new model training data to obtain a new version of the oxygen content prediction model with updated model parameters.

[0060] Here, the specific method for fine-tuning the model parameters of the oxygen content prediction model is similar to the method for adjusting the model parameters in the aforementioned model training phase, and the repetition will not be repeated here.

[0061] Based on the oxygen content prediction method provided in this application, during the preparation of monocrystalline silicon in a monocrystalline furnace, multiple key process parameters affecting the oxygen content within the furnace are collected in real time from multiple sensors installed on the furnace. Target data features reflecting the changing trend of oxygen content are calculated based on these key process parameters. These target data features are then input into a pre-trained oxygen content prediction model, which outputs the predicted oxygen content. Thus, this application can predict the oxygen content within the monocrystalline furnace in real time based on key process parameters collected during the monocrystalline silicon preparation process. This not only overcomes the technical deficiency of existing technologies that can only passively detect oxygen content after the monocrystalline silicon preparation is completed, but also effectively reduces the dependence on high-energy-consuming magnetic fields in traditional control methods, which is beneficial for reducing related preparation and control costs and facilitating the promotion and implementation of the solution.

[0062] Based on the same inventive concept, this application also provides a single-crystal silicon rod, wherein the single-crystal silicon rod is manufactured by the oxygen content prediction method described above in the embodiments of this application.

[0063] Based on the same inventive concept, this application also provides an oxygen content prediction system corresponding to the above-mentioned oxygen content prediction method. Since the principle of solving the problem by the oxygen content prediction system in the embodiments of this application is similar to that of the above-mentioned oxygen content prediction method in the embodiments of this application, the implementation of the oxygen content prediction system can refer to the implementation of the above-mentioned oxygen content prediction method, and the repeated parts will not be described again.

[0064] Reference Figure 1 As shown, Figure 1 This illustration shows a schematic diagram of an oxygen content prediction system applied in the single-crystal silicon preparation process, provided in an embodiment of this application. The oxygen content prediction system includes at least one single-crystal furnace and a detection platform, wherein the detection platform is used for: During the process of preparing monocrystalline silicon using the monocrystalline furnace, multiple key process parameters affecting the oxygen content inside the monocrystalline furnace are collected in real time from multiple sensors installed on the monocrystalline furnace. Based on the aforementioned key process parameters, target data features are calculated to reflect the changing trend of the oxygen content. The target data features are input into a pre-trained oxygen content prediction model, and the oxygen content prediction result is obtained by outputting the oxygen content prediction model.

[0065] In one alternative implementation, the plurality of key process parameters include at least: the heating power of the single crystal furnace and the pulling rate.

[0066] In one optional implementation, the step of calculating target data features reflecting the changing trend of oxygen content based on the plurality of key process parameters includes: For the heating power obtained at each acquisition time, the heating power is divided by the effective area of ​​the heating zone in the single crystal furnace to calculate the average heat flux density of the single crystal furnace at that acquisition time. The difference in average heat flux density of the single crystal furnace at adjacent acquisition times is calculated, and the difference is divided by the time interval between adjacent acquisition times to obtain the heat flux change rate of the single crystal furnace as the target data feature.

[0067] In one optional implementation, the step of calculating target data features reflecting the changing trend of oxygen content based on the plurality of key process parameters includes: The pulling rate corresponding to the single crystal furnace at multiple consecutive acquisition times is obtained to obtain the pulling rate sequence corresponding to the single crystal furnace. The first derivative of the pulling rate sequence is calculated to obtain a first data feature that reflects the rate of change of the pulling rate of the single crystal furnace; The second derivative of the pulling rate sequence is calculated to obtain a second data feature that reflects the rate of change of the pulling rate acceleration of the single crystal furnace; The first data feature and the second data feature are used as the target data feature.

[0068] In one optional implementation, the detection platform is further used for: Determine whether the predicted oxygen content exceeds a preset safety threshold; If the predicted oxygen content is determined to exceed the safety threshold, a target warning signal is generated; wherein the target warning signal is a pre-set warning signal used to reflect that the oxygen content in the single crystal furnace exceeds the standard.

[0069] In one optional implementation, the detection platform is further used for: After the single crystal furnace is detected to have finished preparing single crystal silicon, the actual oxygen content of the single crystal silicon during the preparation process is determined from the prepared single crystal silicon by offline detection. Determine whether the error between the actual oxygen content and the predicted oxygen content exceeds a preset error threshold; If it is determined that the error exceeds the error threshold, new model training data is constructed based on the target data features and the actual oxygen content, and the model parameters of the oxygen content prediction model are fine-tuned based on the new model training data to obtain a new version of the oxygen content prediction model with updated model parameters.

[0070] like Figure 3 As shown, this application embodiment also provides an electronic device 300 for executing the oxygen content prediction method in this application (the electronic device 300 is also equivalent to the detection platform in the aforementioned oxygen content prediction system). The electronic device includes a memory 301, a processor 302, and a computer program stored in the memory 301 and executable on the processor 302. The memory 301 and the processor 302 are connected via a bus for communication. When the processor 302 executes the computer program, it implements the steps of the aforementioned oxygen content prediction method.

[0071] Specifically, the memory 301 and processor 302 mentioned above can be general-purpose memory and processor, without any specific limitations. When the processor 302 runs the computer program stored in the memory 301, it can execute the oxygen content prediction method mentioned above.

[0072] Corresponding to the oxygen content prediction method in this application, this application embodiment also provides a computer-readable storage medium storing a computer program, which is executed by a processor to perform the steps of the oxygen content prediction method described above.

[0073] Specifically, the storage medium can be a general-purpose storage medium, such as a removable disk or hard disk, and the computer program on the storage medium, when run, can execute the oxygen content prediction method described above.

[0074] In the embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. The system embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and there may be other division methods in actual implementation. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the coupling or direct coupling or communication connection shown or discussed may be through some communication interface; the indirect coupling or communication connection between systems or units may be electrical, mechanical, or other forms.

[0075] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0076] In addition, the functional units in the embodiments provided in this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0077] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes 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.

[0078] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0079] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The protection scope of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application; and these modifications, changes, 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. All should be covered within the protection scope of this application. Therefore, the protection scope of this application should be determined by the protection scope of the claims.

Claims

1. A method for predicting oxygen content in the process of preparing single-crystal silicon, characterized in that, The oxygen content prediction method includes: During the preparation of monocrystalline silicon using a single crystal furnace, multiple key process parameters affecting the oxygen content inside the single crystal furnace are collected in real time from multiple sensors installed on the single crystal furnace. Based on the aforementioned key process parameters, target data features are calculated to reflect the changing trend of the oxygen content. The target data features are input into a pre-trained oxygen content prediction model, and the oxygen content prediction result is obtained by outputting the oxygen content prediction model.

2. The oxygen content prediction method according to claim 1, characterized in that, The key process parameters include at least the heating power of the single crystal furnace and the pulling rate.

3. The oxygen content prediction method according to claim 2, characterized in that, The step of calculating target data features reflecting the changing trend of oxygen content based on the multiple key process parameters includes: For the heating power obtained at each acquisition time, the heating power is divided by the effective area of ​​the heating zone in the single crystal furnace to calculate the average heat flux density of the single crystal furnace at that acquisition time. The difference in average heat flux density of the single crystal furnace at adjacent acquisition times is calculated, and the difference is divided by the time interval between adjacent acquisition times to obtain the heat flux change rate of the single crystal furnace as the target data feature.

4. The oxygen content prediction method according to claim 2, characterized in that, The step of calculating target data features reflecting the changing trend of oxygen content based on the multiple key process parameters includes: The pulling rate corresponding to the single crystal furnace at multiple consecutive acquisition times is obtained to obtain the pulling rate sequence corresponding to the single crystal furnace. The first derivative of the pulling rate sequence is calculated to obtain a first data feature that reflects the rate of change of the pulling rate of the single crystal furnace; The second derivative of the pulling rate sequence is calculated to obtain a second data feature that reflects the rate of change of the pulling rate acceleration of the single crystal furnace; The first data feature and the second data feature are used as the target data feature.

5. The oxygen content prediction method according to claim 1, characterized in that, The oxygen content prediction method also includes: Determine whether the predicted oxygen content exceeds a preset safety threshold; If the predicted oxygen content is determined to exceed the safety threshold, a target warning signal is generated; wherein the target warning signal is a pre-set warning signal used to reflect that the oxygen content in the single crystal furnace exceeds the standard.

6. The oxygen content prediction method according to claim 1, characterized in that, The oxygen content prediction method also includes: After the single crystal furnace is detected to have finished preparing single crystal silicon, the actual oxygen content of the single crystal silicon during the preparation process is determined from the prepared single crystal silicon by offline detection. Determine whether the error between the actual oxygen content and the predicted oxygen content exceeds a preset error threshold; If it is determined that the error exceeds the error threshold, new model training data is constructed based on the target data features and the actual oxygen content, and the model parameters of the oxygen content prediction model are fine-tuned based on the new model training data to obtain a new version of the oxygen content prediction model with updated model parameters.

7. An oxygen content prediction system applied in the preparation process of single-crystal silicon, characterized in that, The oxygen content prediction system includes at least one single crystal furnace and a detection platform, wherein the detection platform is used for: During the process of preparing monocrystalline silicon using the monocrystalline furnace, multiple key process parameters affecting the oxygen content inside the monocrystalline furnace are collected in real time from multiple sensors installed on the monocrystalline furnace. Based on the aforementioned key process parameters, target data features are calculated to reflect the changing trend of the oxygen content. The target data features are input into a pre-trained oxygen content prediction model, and the oxygen content prediction result is obtained by outputting the oxygen content prediction model.

8. An electronic device, characterized in that, include: The device includes a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, they perform the steps of the oxygen content prediction method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the oxygen content prediction method as described in any one of claims 1 to 6.

10. A single-crystal silicon rod, characterized in that, The single-crystal silicon rod is manufactured using the oxygen content prediction method as described in any one of claims 1 to 6.

Images