Czochralski method monocrystalline silicon growth control method, device, equipment and medium

Abstract

The invention discloses a growth control method, device and equipment for czochralski monocrystalline silicon and a medium. The method comprises the following steps: acquiring current growth parameter information of the czochralski monocrystalline silicon in a growth process; inputting the current growth parameter information into a pre-trained crystal bar growth prediction model to obtain a crystal bar diameter prediction result after a target preset duration corresponding to the plurality of candidate temperature compensation amounts; and according to the crystal bar diameter prediction result after the target preset duration corresponding to the plurality of candidate temperature compensation amounts, determining a target temperature compensation amount, and adopting the target temperature compensation amount to perform temperature compensation on the czochralski method monocrystalline silicon growth. According to the method, the crystal bar diameter feedback results corresponding to different candidate temperature compensation amounts are obtained through the pre-trained crystal bar growth prediction model, and then the target temperature compensation amount is determined according to the crystal bar diameter feedback results, so that the accuracy of temperature compensation in the monocrystalline silicon growth process is improved, and the generation quality of the monocrystalline silicon crystal bar is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of single crystal silicon manufacturing, and in particular to a method, device, equipment and medium for controlling the growth of single crystal silicon using a Czochralski method. Background Art

[0002] High-purity single crystal silicon is an important semiconductor material that can be made into diodes, triodes, thyristors and various integrated circuits. The growth conditions of high-purity single crystal silicon are multifaceted, including temperature, pressure, atmosphere, silicon raw material purity and growth method. These factors interact with each other and jointly determine the crystal quality of single crystal silicon. In order to meet the needs of high-performance semiconductor devices, it is necessary to produce high-quality single crystal silicon rods with low defects.

[0003] An important factor affecting the defect properties of single crystal silicon is the control of the crystal rod diameter during the growth process. When the actual diameter deviates from the target diameter by a large margin, it is necessary to adjust the pulling speed, causing the pulling speed to deviate from the target pulling speed under the action of PID; at the same time, the diameter deviation will also cause the crystal rod weight deviation, thereby causing the weight deviation of the remaining silicon liquid in the crucible, causing the crucible lifting rate calculated based on the crucible lifting ratio to deviate from the target, which in turn causes the liquid mouth distance deviation, adding great uncertainty to the control of microscopic defects in the crystal. Therefore, diameter control during the growth of single crystal silicon is of great significance to the growth of high-quality single crystal silicon.

[0004] In the growth process of single crystal silicon by the Czochralski method, temperature control has a direct impact on diameter control. When the temperature is lower than the appropriate temperature, the radial growth of the crystal rod is too fast and the diameter will be higher than the target diameter; when the temperature is higher than the appropriate temperature, the radial growth of the crystal rod is too slow and the diameter will be lower than the target diameter. In conventional single crystal silicon growth automation control systems, in addition to the basic temperature compensation curve set by the process, the conventional temperature compensation method is manual temperature compensation by the on-site operator. With the development of semiconductor technology and the improvement of the quality requirements of single crystal silicon, manual temperature compensation relies heavily on the operator's experience and has poor repeatability. In addition, due to the complexity of the single crystal furnace and the thermal field, there is often a delay of 20 to 30 minutes in the actual feedback of temperature compensation to the diameter change. This delay is also one of the difficulties in temperature compensation diameter control.

[0005] Therefore, in the process of manufacturing single crystal silicon rods by the Czochralski method, how to control the temperature in the single crystal furnace and manufacture high-quality single crystal silicon rods with low defects is an important problem that needs to be solved in this field. Summary of the invention

[0006] The present invention provides a method, device, equipment and medium for controlling the growth of single crystal silicon by the Czochralski method, so as to solve the problem that the change of growth conditions during the growth of single crystal silicon by the Czochralski method causes single crystal silicon defects.

[0007] According to one aspect of the present invention, there is provided a method for controlling the growth of single crystal silicon by a Czochralski method, comprising:

[0008] Obtaining current growth parameter information of Czochralski single crystal silicon during its growth process;

[0009] Inputting the current growth parameter information into a pre-trained crystal ingot growth prediction model to obtain a crystal ingot diameter prediction result after a target preset time corresponding to a plurality of candidate temperature compensation amounts; wherein the crystal ingot growth prediction model is trained based on historical growth records of single crystal silicon;

[0010] According to the prediction results of the crystal rod diameter after the target preset time corresponding to the multiple candidate temperature compensation amounts, the target temperature compensation amount is determined, and the target temperature compensation amount is used to perform temperature compensation for the Czochralski method single crystal silicon growth.

[0011] According to another aspect of the present invention, there is provided a growth control device for single crystal silicon using a Czochralski method, comprising:

[0012] A parameter information acquisition module is used to obtain current growth parameter information of Czochralski single crystal silicon during its growth process;

[0013] A model prediction module, used for inputting the current growth parameter information into a pre-trained crystal ingot growth prediction model to obtain a crystal ingot diameter prediction result after a target preset time corresponding to a plurality of candidate temperature compensation amounts; wherein the crystal ingot growth prediction model is trained based on the historical growth record of single crystal silicon;

[0014] The temperature compensation amount determination module is used to determine the target temperature compensation amount according to the crystal rod diameter prediction results after the target preset time corresponding to the multiple candidate temperature compensation amounts, and use the target temperature compensation amount to perform temperature compensation on the Czochralski method single crystal silicon growth.

[0015] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0016] at least one processor; and

[0017] a memory communicatively connected to the at least one processor; wherein,

[0018] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the CZ-pull single crystal silicon growth control method described in any embodiment of the present invention.

[0019] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the method for controlling the growth of single-crystal silicon by the CZ process as described in any embodiment of the present invention when executed.

[0020] The technical solution of the embodiment of the present invention uses a pre-trained crystal rod growth prediction model to learn the current growth parameter information corresponding to any time during the growth process of Czochralski single crystal silicon, obtains crystal rod diameter feedback results corresponding to different candidate temperature compensation amounts, and then determines the target temperature compensation amount based on the crystal rod diameter feedback results, thereby improving the accuracy of temperature compensation during the growth process of single crystal silicon, and thus improving the generation quality of single crystal silicon crystal rods.

[0021] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0023] Figure 1 is a flow chart of a method for controlling growth of single crystal silicon by a Czochralski method provided in accordance with an embodiment of the present invention;

[0024] Figure 2 is a flow chart of a method for training a crystal ingot growth prediction model provided according to an embodiment of the present invention;

[0025] Figure 3 1 is a schematic structural diagram of a Czochralski silicon single crystal growth control device provided according to an embodiment of the present invention;

[0026] Figure 4 It is a schematic diagram of the structure of an electronic device for implementing the method for controlling the growth of single crystal silicon using the Czochralski method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0027] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0028] It should be noted that the terms "candidate", "target", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0029] Figure 1 A flow chart of a method for controlling the growth of single-crystal silicon by a CZ process is provided for an embodiment of the present invention. This embodiment is applicable to the case where the growth parameters of single-crystal silicon by a CZ process are dynamically controlled. The method can be executed by a growth control device for single-crystal silicon by a CZ process. The growth control device for single-crystal silicon by a CZ process can be implemented in the form of hardware and / or software. The growth control device for single-crystal silicon by a CZ process can be configured in a server with computing capabilities. Figure 1 As shown, the method includes:

[0030] S110, obtaining current growth parameter information of single crystal silicon during the growth process of the Czochralski method.

[0031] Among them, Czochralski single crystal silicon refers to single crystal silicon rods manufactured by the Czochralski method. The growth of Czochralski single crystal silicon is carried out in a single crystal furnace. The modern single crystal furnace is equipped with a complete central control system. During the growth of the single crystal silicon rod, the central control system records the current parameters of the single crystal furnace at regular intervals (usually several seconds or tens of seconds), including but not limited to heater temperature, seed crystal lifting speed, crucible rotation speed, furnace pressure, crystal rod diameter and crystal rod length, etc., which also includes the record of manual temperature compensation.

[0032] Parameter information related to temperature and crystal rod diameter is selected from various parameters of the single crystal furnace obtained by the single crystal furnace central control system as the current generated parameter information. Exemplarily, the correlation between all parameter information obtained by the central control system and the temperature and crystal rod diameter is determined, and parameter information with a higher correlation is selected from all parameter information as the generated parameter information according to the correlation. The specific determination and selection process of the correlation degree can select the corresponding determination method according to the actual scenario, which is not limited here.

[0033] Specifically, the growth parameter information is determined from all parameter information obtained from the single crystal furnace, and the current value corresponding to the generation parameter information is obtained during the growth of the single crystal silicon at a preset time interval as the current growth parameter information.

[0034] In a feasible embodiment, the growth parameter information includes at least the temperature of the single crystal furnace heater, the seed crystal lifting speed, and the current crystal rod diameter parameter; the crystal rod diameter prediction result includes the crystal rod diameter and / or the crystal rod diameter change rate.

[0035] Specifically, all parameter information acquired by the single crystal furnace central control system is used as candidate parameter information, the correlation coefficient between each candidate parameter information and the crystal rod diameter is determined, and the correlation coefficients are sorted from high to low, and a preset number of parameter information is determined as generated parameter information. For example, the generated parameter information includes the temperature of the single crystal furnace heater, the seed crystal lifting speed, and the current crystal rod diameter parameter. Optionally, the generated parameter information may also include other parameter information, which is not limited in the embodiment of the present invention.

[0036] The crystal rod diameter prediction result is used to represent the change of the crystal rod diameter. Therefore, the crystal rod diameter prediction result can directly include the crystal rod diameter, or include the crystal rod diameter change rate, and determine the overall change of the crystal rod diameter in the current growth process through the crystal rod diameter change rate, or include the crystal rod diameter and the crystal rod diameter change rate, and determine the current direct prediction result of the crystal rod diameter through the crystal rod diameter, and represent the overall growth process of the crystal rod in the current stage through the crystal rod diameter change rate to improve the control accuracy of the crystal rod diameter.

[0037] S120, inputting the current growth parameter information into a pre-trained crystal ingot growth prediction model to obtain a crystal ingot diameter prediction result after a target preset time period corresponding to a plurality of candidate temperature compensation amounts.

[0038] Among them, the crystal rod growth prediction model is pre-trained according to the historical growth records of single crystal silicon. The input of the crystal rod growth prediction model is the current growth parameter information of single crystal silicon and multiple different candidate temperature compensation amounts. The output is the crystal rod diameter prediction result corresponding to different candidate temperature compensation amounts. The crystal rod diameter prediction result is the prediction result after the target preset time corresponding to the acquisition moment of the current growth parameter information.

[0039] Specifically, a training sample set is determined in advance based on the historical growth records of single crystal silicon, and a crystal rod growth prediction model is trained based on the training sample set, so that the crystal rod growth prediction model can learn the relationship between the growth parameter information and the crystal rod diameter after the target preset time corresponding to different temperature compensation amounts, and then based on this relationship, the output is obtained as a prediction result of the crystal rod diameter after the target preset time corresponding to multiple candidate temperature compensation amounts according to the input current growth parameter information.

[0040] Exemplarily, multiple candidate temperature compensation amounts can be determined based on the growth temperature range of single crystal silicon and the temperature compensation accuracy of the single crystal furnace, and dynamically adjusted according to the current scenario. For example, taking the temperature compensation range of [-0.5℃, +0.5℃] and the temperature compensation accuracy of 0.1℃ as an example, the multiple candidate temperature compensation amounts are -0.5℃, -0.4℃, ..., 0℃, ..., 0.4℃, and 0.5℃ respectively. If the temperature compensation accuracy of the single crystal furnace is finer, the candidate temperature compensation amounts can be further refined, and the temperature compensation range can be divided according to the single temperature adjustment step of the single crystal furnace to obtain multiple candidate temperature compensation amounts. The current growth parameter information is input into the pre-trained crystal rod growth prediction model to obtain the crystal rod diameter prediction results after the target preset time corresponding to the 11 candidate temperature compensation amounts.

[0041] Due to the complexity of the single crystal furnace and the thermal field, there is often a delay of 20 to 30 minutes in the actual feedback of temperature compensation to the diameter change. This delay is also one of the difficulties in controlling the diameter of the compensation temperature. Therefore, the crystal rod growth prediction model directly determines the predicted result of the crystal rod diameter after the target preset time after the compensation temperature is compensated according to the candidate compensation temperature, so as to avoid the inaccurate compensation temperature caused by the feedback delay. The target preset time is determined according to the feedback delay of the temperature compensation. The specific value is not limited and is determined according to the actual situation of the single crystal furnace. For example, it is set to 30 minutes.

[0042] S130, determining a target temperature compensation amount according to the prediction results of the crystal rod diameter after the target preset time corresponding to the multiple candidate temperature compensation amounts, and using the target temperature compensation amount to perform temperature compensation for the Czochralski method single crystal silicon growth.

[0043] The crystal ingot growth prediction model obtains the crystal ingot diameter prediction result after the target preset time corresponding to each candidate temperature compensation amount input, so according to the comparison of the crystal ingot diameter prediction results of multiple target preset time, the suitable crystal ingot diameter is determined, and the temperature compensation amount corresponding to the crystal ingot diameter is used as the temperature compensation amount, and the target temperature compensation amount is used to perform temperature compensation for the growth of single crystal silicon in the CZ method. Among them, the suitable crystal ingot diameter determined from multiple crystal ingot diameter prediction results can be determined according to the current growth stage, or according to the single crystal silicon growth plan, that is, the crystal ingot diameter of the single crystal silicon is expected to be obtained, and the prediction result of the crystal ingot diameter closest to the expected single crystal silicon is determined from multiple crystal ingot diameter prediction results as the suitable crystal ingot diameter.

[0044] The prediction accuracy of the crystal ingot growth prediction model has reached the expected level. After the model training is completed, the crystal ingot growth prediction model is connected to the central control system of the single crystal furnace. During the crystal growth process, the system will periodically perform model prediction. Taking the compensation temperature range of [-0.5℃, +0.5℃] and the compensation temperature accuracy of 0.1℃ as an example, the system will generate 11 test samples, in which the compensation temperature is -0.5℃, -0.4℃, ..., 0℃, ..., 0.4℃, 0.5℃ respectively. The other characteristic values ​​are the same and are determined by the various parameters of the single crystal furnace in the current central control system. These data are input into the crystal ingot growth prediction model to obtain the predicted value of the crystal ingot diameter after 30 minutes when the compensation temperature is different. Based on this series of prediction results, the system selects the most appropriate compensation temperature (the predicted value of the diameter after compensation temperature is closest to the target diameter setting value) and automatically performs temperature compensation.

[0045] Exemplarily, multiple candidate heating amounts include a candidate heating amount of 0, that is, when no heating is currently being performed, the prediction result of the crystal rod diameter after the target preset time period is determined. If the prediction result of the crystal rod diameter after the target preset time period corresponding to the candidate heating amount of 0 meets the expected result, it means that the current growth parameter information meets the conditions, that is, the temperature in the single crystal furnace is accurate, the crystal rod diameter is accurate, and no heating is required; if the prediction result of the crystal rod diameter after the target preset time period corresponding to the candidate heating amount of 0 does not meet the expected result, it means that the current growth parameter information does not meet the conditions and heating is required. Therefore, by determining the prediction results of the crystal rod diameter after the target preset time period corresponding to multiple candidate heating amounts, the growth process of single crystal silicon can be monitored in real time, thereby improving the production efficiency of low-defect single crystal silicon.

[0046] In a feasible embodiment, the target temperature compensation amount is determined according to the prediction results of the crystal ingot diameter after the target preset time corresponding to the multiple candidate temperature compensation amounts, including:

[0047] Determine the target crystal rod diameter parameter after the current corresponding target preset time according to the current growth stage of the Czochralski method single crystal silicon;

[0048] According to the comparison result of the crystal rod diameter prediction result after the target preset time period corresponding to multiple candidate temperature compensation amounts and the target crystal rod diameter parameter, the target temperature compensation amount is determined from the candidate temperature compensation amounts.

[0049] According to the current growth stage of single crystal silicon by the Czochralski method, the ideal crystal rod diameter parameter after the current corresponding target preset time is determined as the target crystal rod diameter parameter, wherein the ideal crystal rod diameter parameter can be determined according to the single crystal silicon production requirements and the current growth stage, and the specific process is not limited here.

[0050] The crystal rod diameter prediction results after the target preset time corresponding to multiple candidate temperature compensation amounts are compared with the target crystal rod diameter parameters, and the crystal rod diameter prediction result closest to the target crystal rod diameter parameter is determined as the reference crystal rod diameter prediction result, and the candidate temperature compensation amount corresponding to the reference crystal rod diameter prediction result is used as the target temperature compensation amount.

[0051] By determining a target crystal rod diameter parameter after a target preset time corresponding to the current growth stage, and determining a target temperature compensation amount from a plurality of candidate temperature compensation amounts according to the target crystal rod diameter parameter, the accuracy of determining the target temperature compensation amount is improved.

[0052] In a feasible embodiment, after temperature compensation is performed on the Czochralski method single crystal silicon growth by using the target temperature compensation amount, the method further includes:

[0053] Determine the actual crystal ingot diameter parameter after the target preset time after temperature compensation, and compare the actual crystal ingot diameter parameter with the crystal ingot diameter prediction result corresponding to the target temperature compensation amount;

[0054] If the comparison results are different, the crystal ingot growth prediction model is corrected according to the deviation between the actual crystal ingot diameter parameter and the crystal ingot diameter prediction result corresponding to the target temperature compensation amount.

[0055] During the growth of single crystal silicon, the system will continuously and automatically perform model prediction and temperature compensation. At the same time, the system records the actual diameter parameters of the single crystal silicon ingot after the target preset time period after the target temperature compensation amount, and compares the actual diameter parameters of the ingot with the predicted diameter results of the ingot corresponding to the target temperature compensation amount obtained by the model prediction. Based on the correction algorithm, the ingot growth prediction model can continuously correct the ingot growth prediction model according to the deviation between the actual ingot diameter parameters and the predicted diameter results of the ingot corresponding to the target temperature compensation amount during the operation of the system, thereby continuously improving the prediction accuracy of the model and realizing self-optimization of the model.

[0056] Exemplarily, the deviation between the actual crystal rod diameter parameter and the crystal rod diameter prediction result corresponding to the target temperature compensation amount is input into the model, and the target temperature compensation amount and the corresponding growth parameter information are used as a training sample set to retrain the crystal rod growth prediction model.

[0057] Since there is a delay of 20 to 30 minutes between the actual feedback of the temperature compensation of the single crystal furnace and the change in the diameter of the single crystal silicon ingot, the ingot growth prediction model in the embodiment of the present invention can realize the prediction of the ingot diameter parameters after adjusting a certain target preset time of the temperature compensation amount during the crystal growth process. In the actual crystal growth process, the ingot growth prediction model connected to the central control system of the single crystal furnace can be used to predict the diameter prediction results of the crystal ingot after the target preset time when different temperature compensation amounts are adjusted at any time based on the current growth parameter information. Based on this prediction result, the current temperature compensation result is determined, including whether temperature compensation is required and the specific temperature compensation amount, and the central control system can automatically perform temperature compensation according to the prediction result, thereby realizing automatic control of the diameter of the crystal ingot. In addition, during the continuous operation of the system, the ingot growth prediction model can also continuously collect the actual ingot diameter parameters after temperature compensation with the support of the embodiment of the present invention, compare the difference between the predicted value and the actual value after temperature compensation, continue to learn, optimize by itself, and improve the prediction accuracy of the ingot growth prediction model.

[0058] The technical solution of the embodiment of the present invention uses a pre-trained crystal rod growth prediction model to learn the current growth parameter information corresponding to any time during the growth process of Czochralski single crystal silicon, obtains crystal rod diameter feedback results corresponding to different candidate temperature compensation amounts, and then determines the target temperature compensation amount based on the crystal rod diameter feedback results, thereby improving the accuracy of temperature compensation during the growth process of single crystal silicon, and thus improving the generation quality of single crystal silicon crystal rods.

[0059] Figure 2 This is a flow chart of a method for training a crystal ingot growth prediction model provided by an embodiment of the present invention. This embodiment further refines the training process of the crystal ingot growth prediction model in the above embodiment. Figure 2 As shown, the method includes:

[0060] S210, obtaining historical growth records of single crystal silicon.

[0061] The single crystal silicon historical growth record is a statistical result obtained by recording the parameter information and the amount of temperature compensation during the single crystal silicon growth process in the single crystal furnace within the historical time period. Each single crystal silicon historical growth record includes the growth parameter information and the total amount of temperature compensation corresponding to the historical moment.

[0062] S220, selecting a first sample record from the historical growth record of single crystal silicon, and determining the sample growth parameter information, the sample temperature compensation amount, and the sample crystal rod diameter parameter after the corresponding target preset time corresponding to the first sample record.

[0063] Among them, the total amount of temperature compensation recorded in the first sample is different from the total amount of temperature compensation recorded in the corresponding previous record, that is, a manual temperature compensation adjustment was performed from the time of the previous record corresponding to the first sample record to the time corresponding to the first sample record.

[0064] Specifically, the total amount of heating compensation for each historical growth record of single crystal silicon is identified in chronological order, and records with a total amount of heating compensation different from the previous record are screened out one by one as the first sample record, and the corresponding sample growth parameter information, sample heating compensation amount and the sample crystal rod diameter parameter after the corresponding target preset time are read from the first sample record, wherein the sample heating compensation amount corresponding to the first sample record is the difference between the total amount of heating compensation of the first sample record and the total amount of heating compensation of the corresponding previous record, and the sample crystal rod diameter parameter after the target preset time is determined from the historical growth record of single crystal silicon according to the recording time corresponding to the first sample record.

[0065] Optionally, determine the first recording time of the first sample record, determine the second recording time after adding the target preset time to the first recording time, and determine the third recording time after subtracting the target preset time from the previous recording time corresponding to the first recording time, determine all intermediate single crystal silicon historical growth records from the first recording time to the second recording time, and from the third recording time to the previous record corresponding to the first recording time, and if the total amount of temperature compensation corresponding to any record in the intermediate single crystal silicon historical growth records is different from the total amount of temperature compensation corresponding to the previous record, then delete the first sample record. Specifically, if there is another temperature compensation amount in the intermediate time after the first sample record to the target preset time, and in the intermediate time of the target preset time before the previous record corresponding to the first sample record, it is impossible to determine that the sample crystal rod diameter parameter after the target preset time corresponding to the first sample record is completely compensated by the sample temperature compensation amount corresponding to the first sample record, and therefore the first sample record is deleted and not used as a training sample for the model.

[0066] S230, determining a training sample according to the sample growth parameter information, the sample temperature compensation amount, and the sample crystal rod diameter parameter after the corresponding target preset time corresponding to the first sample record, and training the crystal rod growth prediction model according to the training sample.

[0067] The first sample record obtained by screening is used as a training sample to train the crystal rod growth prediction model. Specifically, the records whose total amount of temperature compensation is different from the previous record are screened one by one (which means that a manual temperature compensation adjustment was made from the time of the previous record to the time of this record). From this record, a series of parameters such as the heater temperature, seed crystal lifting speed, crystal rod diameter, and the amount of temperature compensation for this manual temperature compensation can be read as features for model training. At the same time, the crystal rod diameter at this moment is read from the record corresponding to 30 minutes after this moment as the target property. Based on this rule, the historical single crystal silicon rod growth process records are screened, and all the screened records are summarized into a sample set. The selection of target properties is scalable, for example, the crystal rod diameter parameters such as the diameter change rate after 30 minutes are selected as the target properties, or based on the multi-target training function of the artificial intelligence model, multiple parameters such as the crystal rod diameter and the crystal rod diameter change rate after 30 minutes are selected for model training. The selection of features is scalable, such as selecting more basic features, or generating more extended features based on the basic features according to certain rules as training sample sets to participate in model training.

[0068] Exemplarily, the first sample record is divided into a training set and a test set, a suitable intelligent model is selected, and the samples of the training set are used to train the intelligent model. After the training is completed, the samples of the test set are applied to the model to obtain the predicted value of the test set. The predicted value of the test set is compared with the actual value to evaluate the performance of the model. The smaller the error between the predicted value and the actual value means the more accurate the prediction and the better the model performance. Optionally, the selection of the intelligent model is scalable, including methods such as decision trees, random forests, support vector machines, and deep learning models. The model with the best prediction effect can be selected for use by horizontal comparison. Optionally, the verification method of the intelligent model is scalable, such as using other methods such as K-fold cross validation. Optionally, during the model training process, feature engineering, hyperparameter tuning, and other methods can be used to improve model performance.

[0069] In a feasible embodiment, before training the crystal ingot growth prediction model according to the training samples, the method further includes:

[0070] A second sample record is selected from the historical growth record of single crystal silicon, and sample growth parameter information, sample temperature compensation amount, and sample crystal rod diameter parameter after the corresponding target preset time corresponding to the second sample record are determined; wherein the total temperature compensation amount of the second sample record is the same as the total temperature compensation amount of the corresponding previous record;

[0071] A supplementary sample record is obtained by screening from the second sample record according to the change information of the sample crystal rod diameter parameter after the target preset time length corresponding to the second sample record;

[0072] The sample growth parameter information, sample temperature compensation amount and sample crystal rod diameter parameter after the corresponding target preset time length recorded by the supplementary sample are used as training samples.

[0073] Since the candidate heating amount includes the case where the heating value is 0 when predicting the crystal ingot diameter, in order to improve the prediction accuracy of the crystal ingot growth prediction model, this embodiment expands the training samples and increases the completeness of the model samples. The total heating amount recorded in the second sample is the same as the total heating amount recorded in the previous record, that is, no heating adjustment is performed from the time of the previous record corresponding to the second sample record to the time corresponding to the second sample record, and it can be regarded that the sample heating amount corresponding to the second sample record is 0.

[0074] Specifically, the total amount of temperature compensation of each single crystal silicon historical growth record is identified in chronological order, and the records with the same total amount of temperature compensation as the previous one are selected one by one as the second sample record. The corresponding sample growth parameter information, sample temperature compensation amount, and sample crystal rod diameter parameter after the corresponding target preset time are read from the second sample record, wherein the sample temperature compensation amount corresponding to the second sample record is 0, and the change information of all sample crystal rod diameter parameters between the recording time and the subsequent target preset time is determined from the single crystal silicon historical growth record according to the recording time corresponding to the second sample record. If the sample crystal rod diameter parameter does not change during the period between the recording time and the subsequent target preset time, the second sample record is used as a supplementary sample record; if the sample crystal rod diameter parameter changes during this period, it means that there are other factors affecting the growth of the crystal rod, and the crystal rod diameter parameter after the target preset time corresponding to the temperature compensation amount of 0 is not referenceable, and the second sample record is deleted.

[0075] After obtaining the supplementary sample record by screening out the second sample record according to the above conditions, the sample growth parameter information, sample heating amount and the sample crystal rod diameter parameter after the corresponding target preset time corresponding to the supplementary sample record and the sample growth parameter information, sample heating amount and the sample crystal rod diameter parameter after the corresponding target preset time corresponding to the first sample record are used as training samples to train the crystal rod growth prediction model.

[0076] In a feasible embodiment, after screening the first sample record from the historical growth record of single crystal silicon, the method further comprises:

[0077] Counting the crystal ingot diameter parameters after a plurality of candidate preset time periods corresponding to the first sample record;

[0078] A target preset time duration is determined from a plurality of candidate preset time durations according to the statistical result.

[0079] Since the target preset time represents the waiting time for feedback after temperature compensation is performed on the growth of single crystal silicon, it is necessary to ensure the accuracy of the target preset time to avoid inaccurate determination of the feedback amount of the crystal rod diameter corresponding to the temperature compensation.

[0080] Specifically, the crystal rod diameter parameters after multiple candidate preset time lengths corresponding to the first sample record are counted, and the crystal rod diameter parameters corresponding to the candidate preset time lengths after the target preset time length no longer change. The target preset time length is determined from multiple candidate preset time lengths according to this condition to ensure that the crystal rod diameter parameter after the target preset time length is a stable result after temperature compensation, thereby improving the accuracy of temperature supplementation.

[0081] Exemplarily, the candidate preset time lengths can be determined based on the statistical period of historical records and a predetermined feedback range. For example, if the statistical period is 5 minutes and the predetermined feedback range is 20 to 30 minutes, the predetermined feedback range is divided with the statistical period as a step size to obtain multiple candidate preset time lengths.

[0082] The technical solution of the embodiment of the present invention improves the accuracy of training samples by screening the first sample record from the historical growth record of single crystal silicon, and predicts the crystal rod growth prediction model according to the first sample record, thereby improving the accuracy of the model prediction result.

[0083] Figure 3 The schematic diagram of the structure of a Czochralski silicon single crystal growth control device provided by an embodiment of the present invention. Figure 3 As shown, the device comprises:

[0084] The parameter information acquisition module 310 is used to acquire the current growth parameter information of the CZ-grown single crystal silicon during its growth process;

[0085] The model prediction module 320 is used to input the current growth parameter information into a pre-trained crystal ingot growth prediction model to obtain a crystal ingot diameter prediction result after a target preset time corresponding to a plurality of candidate temperature compensation amounts; wherein the crystal ingot growth prediction model is trained based on the historical growth record of single crystal silicon;

[0086] The heating amount determination module 330 is used to determine the target heating amount according to the prediction results of the crystal rod diameter after the target preset time corresponding to the multiple candidate heating amounts, and use the target heating amount to perform temperature compensation for the Czochralski method single crystal silicon growth.

[0087] The technical solution of the embodiment of the present invention uses a pre-trained crystal rod growth prediction model to learn the current growth parameter information corresponding to any time during the growth process of Czochralski single crystal silicon, obtains crystal rod diameter feedback results corresponding to different candidate temperature compensation amounts, and then determines the target temperature compensation amount based on the crystal rod diameter feedback results, thereby improving the accuracy of temperature compensation during the growth process of single crystal silicon, and thus improving the generation quality of single crystal silicon crystal rods.

[0088] Optionally, the device further includes a crystal ingot growth prediction model training module, including:

[0089] A temperature compensation record acquisition unit is used to acquire the historical growth record of single crystal silicon;

[0090] A sample record screening unit, used to screen a first sample record from the historical growth record of the single crystal silicon, and determine the sample growth parameter information, the sample temperature compensation amount, and the sample crystal rod diameter parameter after the corresponding target preset time corresponding to the first sample record; wherein the total temperature compensation amount of the first sample record is different from the total temperature compensation amount of the corresponding previous record;

[0091] The model training unit is used to determine the training sample according to the sample growth parameter information, sample temperature compensation amount and the sample crystal rod diameter parameter after the corresponding target preset time corresponding to the first sample record, and train the crystal rod growth prediction model according to the training sample.

[0092] Optionally, the device further includes a training sample supplement unit, which is used, before training the crystal ingot growth prediction model according to the training sample, to:

[0093] A second sample record is selected from the historical growth record of the single crystal silicon, and sample growth parameter information, sample temperature compensation amount, and sample crystal rod diameter parameter after the corresponding target preset time corresponding to the second sample record are determined; wherein the total temperature compensation amount of the second sample record is the same as the total temperature compensation amount of the corresponding previous record;

[0094] A supplementary sample record is obtained by screening from the second sample record according to the change information of the sample crystal ingot diameter parameter after the target preset time period corresponding to the second sample record;

[0095] The sample growth parameter information, sample temperature compensation amount and the sample crystal rod diameter parameter after the corresponding target preset time length recorded by the supplementary sample are used as training samples.

[0096] Optionally, the device further includes a preset time length determination unit, which is used, after screening the first sample record from the historical growth record of the single crystal silicon, to:

[0097] Collecting statistics on the crystal ingot diameter parameters after a plurality of candidate preset time periods corresponding to the first sample records;

[0098] The target preset time duration is determined from a plurality of candidate preset time durations according to the statistical result.

[0099] Optionally, the growth parameter information includes at least the temperature of the single crystal furnace heater, the seed crystal lifting speed, and the current crystal rod diameter parameter; and the crystal rod diameter prediction result includes the crystal rod diameter and / or the crystal rod diameter change rate.

[0100] Optionally, the temperature compensation amount determination module is specifically used for:

[0101] Determining a target crystal rod diameter parameter after a current target preset time period corresponding to the current growth stage of the Czochralski method single crystal silicon;

[0102] According to the comparison result of the crystal ingot diameter prediction result after the target preset time period corresponding to the multiple candidate temperature compensation amounts and the target crystal ingot diameter parameter, the target temperature compensation amount is determined from the candidate temperature compensation amounts.

[0103] Optionally, the device further includes a model correction module, which is used, after the target temperature compensation amount is used to perform temperature compensation on the Czochralski method single crystal silicon growth, to:

[0104] Determine an actual crystal ingot diameter parameter after a target preset time after temperature compensation, and compare the actual crystal ingot diameter parameter with a crystal ingot diameter prediction result corresponding to the target temperature compensation amount;

[0105] If the comparison results are different, the crystal ingot growth prediction model is corrected according to the deviation between the actual crystal ingot diameter parameter and the crystal ingot diameter prediction result corresponding to the target temperature compensation amount.

[0106] The Czochralski silicon single crystal growth control device provided in the embodiment of the present invention can execute the Czochralski silicon single crystal growth control method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0107] The acquisition, storage, use, and processing of data in the technical solution of this application comply with the relevant provisions of national laws and regulations and do not violate public order and good morals.

[0108] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0109] Figure 4A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0110] like Figure 4 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0111] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0112] The processor 11 may be a variety of general and / or dedicated processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as a method for controlling the growth of single crystal silicon by a CZ process.

[0113] In some embodiments, the growth control method of CZ-pull single crystal silicon can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the growth control method of CZ-pull single crystal silicon described above can be executed. Alternatively, in other embodiments, the processor 11 can be configured to execute the growth control method of CZ-pull single crystal silicon in any other appropriate manner (for example, by means of firmware).

[0114] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific reference products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0115] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0116] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0117] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0118] The systems and techniques described herein may be implemented in a computing system that includes a backend component (e.g., as a data server), or a computing system that includes a switch component (e.g., an application server), or a computing system that includes a frontend component (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described herein), or a computing system that includes any combination of such backend components, switch components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0119] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0120] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0121] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for controlling the growth of single crystal silicon by a Czochralski method, characterized in that: The method includes: Obtaining current growth parameter information of Czochralski single crystal silicon during its growth process; Inputting the current growth parameter information into a pre-trained crystal ingot growth prediction model to obtain a crystal ingot diameter prediction result after a target preset time corresponding to a plurality of candidate temperature compensation amounts; wherein the crystal ingot growth prediction model is trained based on historical growth records of single crystal silicon; According to the prediction results of the crystal rod diameter after the target preset time corresponding to the multiple candidate temperature compensation amounts, the target temperature compensation amount is determined, and the target temperature compensation amount is used to perform temperature compensation for the Czochralski method single crystal silicon growth.

2. The method according to claim 1, characterized in that The training process of the crystal ingot growth prediction model includes: Obtain historical growth records of single crystal silicon; A first sample record is selected from the historical growth record of the single crystal silicon, and sample growth parameter information, sample temperature compensation amount, and a sample crystal rod diameter parameter after a corresponding target preset time corresponding to the first sample record are determined; wherein the total temperature compensation amount of the first sample record is different from the total temperature compensation amount of the corresponding previous record; A training sample is determined according to the sample growth parameter information, the sample temperature compensation amount and the sample crystal rod diameter parameter after the corresponding target preset time length recorded in the first sample, and the crystal rod growth prediction model is trained according to the training sample.

3. The method according to claim 2, characterized in that Before training the crystal ingot growth prediction model according to the training samples, the method further includes: A second sample record is selected from the historical growth record of the single crystal silicon, and sample growth parameter information, sample temperature compensation amount, and sample crystal rod diameter parameter after the corresponding target preset time corresponding to the second sample record are determined; wherein the total temperature compensation amount of the second sample record is the same as the total temperature compensation amount of the corresponding previous record; A supplementary sample record is obtained by screening from the second sample record according to the change information of the sample crystal ingot diameter parameter after the target preset time period corresponding to the second sample record; The sample growth parameter information, sample temperature compensation amount and the sample crystal rod diameter parameter after the corresponding target preset time length recorded by the supplementary sample are used as training samples.

4. The method according to claim 2 or 3, characterized in that: After screening the first sample record from the historical growth record of the single crystal silicon, the method further comprises: Collecting statistics on the crystal ingot diameter parameters after a plurality of candidate preset time periods corresponding to the first sample records; The target preset time duration is determined from a plurality of candidate preset time durations according to the statistical result.

5. The method according to claim 1, characterized in that in, The growth parameter information includes at least the temperature of the single crystal furnace heater, the seed crystal lifting speed, and the current crystal rod diameter parameter; the crystal rod diameter prediction result includes the crystal rod diameter and / or the crystal rod diameter change rate.

6. The method according to claim 1, characterized in that The step of determining the target heating amount according to the prediction results of the crystal ingot diameter after the target preset time period corresponding to the multiple candidate heating amounts comprises: Determining a target crystal rod diameter parameter after a current target preset time period corresponding to the current growth stage of the Czochralski method single crystal silicon; According to the comparison result of the crystal ingot diameter prediction result after the target preset time period corresponding to the multiple candidate temperature compensation amounts and the target crystal ingot diameter parameter, the target temperature compensation amount is determined from the candidate temperature compensation amounts.

7. The method according to claim 1, characterized in that After the target temperature compensation amount is used to perform temperature compensation on the Czochralski method single crystal silicon growth, the method further includes: Determine an actual crystal ingot diameter parameter after a target preset time after temperature compensation, and compare the actual crystal ingot diameter parameter with a crystal ingot diameter prediction result corresponding to the target temperature compensation amount; If the comparison results are different, the crystal ingot growth prediction model is corrected according to the deviation between the actual crystal ingot diameter parameter and the crystal ingot diameter prediction result corresponding to the target temperature compensation amount.

8. A Czochralski silicon single crystal growth control device, characterized in that: The device includes: A parameter information acquisition module is used to obtain current growth parameter information of Czochralski single crystal silicon during its growth process; A model prediction module, used for inputting the current growth parameter information into a pre-trained crystal ingot growth prediction model to obtain a crystal ingot diameter prediction result after a target preset time corresponding to a plurality of candidate temperature compensation amounts; wherein the crystal ingot growth prediction model is trained based on the historical growth record of single crystal silicon; The temperature compensation amount determination module is used to determine the target temperature compensation amount according to the crystal rod diameter prediction results after the target preset time corresponding to the multiple candidate temperature compensation amounts, and use the target temperature compensation amount to perform temperature compensation on the Czochralski method single crystal silicon growth.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can perform the Czochralski silicon single crystal growth control method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the CZ-pull single crystal silicon growth control method according to any one of claims 1 to 7 when executed.

Images