Crystal bar pulling speed optimization method, device and equipment and computer storage medium
By obtaining historical crystal rod quality assessment results and actual pulling speeds, an optimized target pulling speed curve is generated, which solves the problem of unstable crystal rod quality during the crystal pulling process and achieves better crystal rod production results.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- XIAN ESWIN MATERIAL TECHNOLOGY CO LTD
- Filing Date
- 2025-11-20
- Publication Date
- 2026-04-21
AI Technical Summary
In existing crystal pulling methods, the instability of the crystal pulling process during crystal growth leads to differences in the production quality and stability of crystal rods. These are specific problems that the existing technology has failed to effectively solve or has not effectively addressed in terms of crystal rod quality.
By obtaining historical crystal rod quality assessment results, compensation values are determined, and an optimized target pulling speed curve is generated based on the actual pulling speed and the design pulling speed trend type, thus realizing automated closed-loop control of the crystal pulling equipment.
This improved the consistency of crystal rod quality, reduced high-density defects, and achieved better crystal rod production results.
Smart Images
Figure CN121903418A_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to the field of semiconductor manufacturing technology, and in particular to a method, apparatus, device, and computer storage medium for optimizing the pulling speed of crystal rods. Background Technology
[0002] Monocrystalline silicon is a fundamental material in the semiconductor industry, and its quality directly affects the performance and yield of downstream chips. The Czochralski method (CZ method) is currently the mainstream monocrystalline silicon growth technology. It grows monocrystalline silicon rods (also called ingots) from molten polycrystalline silicon by controlling the pulling rate and rotation rate of the seed crystal.
[0003] In the crystal pulling process, the pulling speed is a key process parameter that determines the quality of the crystal (such as defect density, resistivity distribution, etc.). The ratio of pulling speed to the temperature gradient (G) at the crystal growth interface (V / G) directly determines whether vacancy defects or interstitial atoms dominate in the crystal.
[0004] In existing crystal pulling processes, the pulling speed profile is usually pre-set based on theoretical models and extensive experimental experience. However, in actual production, there are slight fluctuations in the thermal environment and melt convection during each crystal pulling process. This can lead to differences in the quality of the produced crystal rods, and even the presence of high-density defects, even when using the same target pulling speed profile. Summary of the Invention
[0005] This disclosure provides a method, apparatus, device, and computer storage medium for optimizing crystal pulling speed; it addresses the technical problem that the quality of crystal rods produced by existing crystal pulling methods may vary, and even high-density defects may occur.
[0006] The technical solution disclosed herein is implemented as follows: In a first aspect, this disclosure provides a method for optimizing the crystal pulling speed, applied to a crystal pulling system including a crystal pulling equipment, comprising: The compensation value is determined based on the quality assessment results of wafers at multiple sampling locations in the historical ingots. The target pulling speed at the sampling location is determined based on the compensation value and the actual pulling speed at the sampling location. Based on the trend type of the design pulling speed curve of the historical crystal ingots across multiple sampling locations and the target pulling speed, a target pulling speed curve is generated for controlling the crystal pulling equipment to pull subsequent crystal ingots.
[0007] Secondly, this disclosure provides a device for optimizing the pulling speed of crystal ingots, applied in a crystal ingot pulling system including crystal pulling equipment. The device for optimizing the pulling speed of crystal ingots includes: The compensation determination module is used to determine the compensation value based on the quality evaluation results of the wafers at multiple sampling locations in the historical ingots. A pulling speed compensation module is used to determine the target pulling speed at the sampling position based on the compensation value and the actual pulling speed at the sampling position. The pulling speed determination module is used to generate a target pulling speed curve for controlling the crystal pulling equipment to pull subsequent crystal rods based on the trend type of the design pulling speed curve of the historical crystal rod between multiple sampling positions and the target pulling speed.
[0008] Thirdly, this disclosure provides an electronic device comprising: a processor and a memory; the processor being configured to execute instructions stored in the memory to implement the ingot pulling speed optimization method as described in the first aspect.
[0009] Fourthly, this disclosure provides a computer storage medium storing at least one instruction, which is executed by a processor to implement the ingot pulling speed optimization method as described in the first aspect.
[0010] This disclosure provides a method, apparatus, device, and computer storage medium for optimizing crystal ingot pulling speed. By acquiring actual pulling speed and quality evaluation results, and based on the design trend type, different calculation methods are used to interpolate and calculate the target pulling speed curve. This not only utilizes historical crystal ingot data to correct defects (through compensation values), but also ensures that the generated complete target pulling speed curve is smooth and continuous through classification. This provides executable and optimized process parameters for crystal pulling equipment, laying the technical foundation for subsequent automated closed-loop control and quality improvement. Attached Figure Description
[0011] Figure 1 This is a structural block diagram of a crystal rod pulling system provided in this disclosure.
[0012] Figure 2 A flowchart of a method for optimizing the pulling speed of a crystal rod provided in this disclosure.
[0013] Figure 3 This is a schematic diagram illustrating the determination of a crystal rod sampling location provided in this disclosure.
[0014] Figure 4 This is a schematic diagram comparing the actual pulling speed and the design pulling speed provided in this disclosure.
[0015] Figure 5 This is a schematic diagram of a sampling location speed compensation method provided in this disclosure.
[0016] Figure 6 This is a schematic diagram of a monotonically changing computational logic provided in this disclosure.
[0017] Figure 7 This is a schematic diagram of the calculation logic for stabilizing the pulling speed provided in this disclosure.
[0018] Figure 8 This is a schematic diagram of a non-monotonic computational logic provided in this disclosure.
[0019] Figure 9 This is a schematic diagram showing the quality results of sampling positions before and after speed adjustment, as provided in this disclosure.
[0020] Figure 10 This is a structural block diagram of a device for optimizing the pulling speed of a crystal rod, as provided in this disclosure.
[0021] Figure 11 This is a schematic diagram of the structure of an electronic device provided in this disclosure. Detailed Implementation
[0022] The technical solutions in this disclosure will now be clearly and completely described with reference to the accompanying drawings.
[0023] In this disclosure, unless otherwise stated, the following abbreviations have the following meanings: COP refers to Crystal Originated Particle, a micro-defect formed during the growth of single-crystal silicon due to the accumulation of vacancies. LDP (Large Dislocation Pit) refers to a regularly shaped pit formed on the surface of a single-crystal silicon wafer due to the preferential and rapid etching of dislocations intersecting the surface under specific chemical etching conditions. Pv (P-vacancy) represents a quality region dominated by vacancy defects. Pi (P-interstitial) represents a quality region dominated by interstitial defects. Pv-Pi represents the boundary region where the concentrations of vacancy and interstitial defects reach equilibrium; this region is generally considered the optimal quality region of "defect-free". P-Band is a specific defect region with a high risk of COP. B-Band is another specific defect region with a high risk of LDP. Lifetime mapping, or minority carrier lifetime imaging, is a non-destructive testing technique that characterizes the defects and impurities in a material by measuring the distribution of minority carrier lifetimes on the sample surface. Different defect types (such as COP and LDP) and concentrations have different effects on carrier lifetimes, thus exhibiting different characteristics in the mapping. NT refers to the target pulling speed. P refers to the actual pulling speed. T refers to the design pulling speed. O refers to the offset.
[0024] This embodiment describes an exemplary architecture of a crystal ingot pulling system to which this disclosure applies. (See attached document.) Figure 1 This is a schematic diagram of a crystal ingot pulling system 100 provided in an embodiment of this disclosure. The crystal ingot pulling system 100 aims to achieve automated closed-loop optimization of the crystal ingot pulling process. (Refer to the attached diagram.) Figure 1The ingot pulling system 100 may include a crystal pulling device 110, a testing device 120, a big data platform 130, and an ingot pulling speed optimization unit 140. The crystal pulling device 110 is the core equipment for performing physical growth, such as a Czochralski single crystal furnace. The crystal pulling device 110 internally includes (not shown) a crucible for containing polycrystalline silicon melt, a heater for heating, a pulling mechanism (e.g., a seed crystal rod) for pulling the ingot, and a controller for controlling these components. The controller of the crystal pulling device 110 can receive an externally input design pulling speed curve and precisely control the pulling mechanism to pull the ingot upwards according to the rate of the curve. The controller can also record the actual pulling speed and the target pulling speed used during the pulling process. The testing device 120 is used to perform quality inspection on the pulled ingot or its sample. In this embodiment, the testing device 120 may be a lifetime mapping tester used to obtain the minority carrier lifetime spectrum of the ingot sample. In other embodiments, other defect detection devices may also be used, such as optical microscopes, scanning electron microscopes (SEM), or laser scattering tomography (LST) devices. The big data platform 130 is a server-based computing platform. The big data platform 130 internally stores a large number of historical test maps, for example, a sample library of "lifetime mapping" maps from the test device 120. Furthermore, each historical map in this sample library is associated with a known, manually or machine-verified ingot quality classification, such as P-Band, Pv-Pi, etc., and defect information. The ingot pulling speed optimization unit 140 is the core computing unit of this disclosure embodiment, and its function is to execute the ingot pulling speed optimization method in subsequent embodiments.
[0025] As attached Figure 1As shown, the information and physical flows between these devices form a complete closed loop. First (physical flow), the crystal pulling equipment 110 pulls historical crystal rods according to the initial design pulling speed curve; these historical crystal rods can also be referred to as the first crystal rod. Subsequently, the first crystal rod is sent to the testing equipment 120 for sampling and testing (e.g., lifetime mapping). Then, the testing equipment 120 sends the test results to the big data platform 130. Next, the big data platform 130 analyzes and classifies the test results to obtain quantitative quality assessment results. The optimization unit 140 obtains the actual pulling speed and design pulling speed data during the first crystal rod pulling process from the crystal pulling equipment 110. Simultaneously, the optimization unit 140 obtains the quality assessment results from the big data platform 130 and queries the corresponding compensation value. The optimization unit 140 executes the crystal rod pulling speed optimization method provided in this embodiment of the disclosure to calculate the complete target pulling speed curve for pulling the second crystal rod to be prepared subsequently. Finally, the optimization unit 140 feeds back the target pulling speed curve to the crystal pulling equipment 110. The crystal pulling equipment 110 uses the target pulling speed curve as a new process parameter to pull a second crystal rod (not shown).
[0026] Specifically, refer to Figure 2 The method for optimizing the crystal rod pulling speed may include steps S210 to S230.
[0027] In step S210, a compensation value is determined based on the quality assessment results of the wafers at multiple sampling locations in the historical crystal rods.
[0028] In some exemplary embodiments of this disclosure, after the historical crystal ingot, i.e., the first crystal ingot, is pulled, it needs to be sampled in segments to obtain quality information of the crystal ingot at different growth stages. (See attached...) Figure 3 As shown, a crystal rod typically includes a tapered head, a tapered tail, and a columnar body in the middle. In actual production, the head and tail are usually removed.
[0029] In this disclosure, sampling can be performed on the body portion of the crystal rod. For example... Figure 3 The crystal rod body can be divided into multiple segments along its length. There is no limit to the number of segments; more samples result in more accurate evaluation. In a specific example, assume the effective length of the crystal rod body is 1000 mm. It can be divided into five equal segments, each 200 mm long. This corresponds to six sampling points (S1, S2, S3, S4, S5, S6), with lengths L of L=0 mm for S1, L=200 mm for S2, L=400 mm for S3, L=600 mm for S4, L=800 mm for S5, and L=1000 mm for S6.
[0030] At the beginning and end positions of each segment (i.e., positions S1, S2, S3, S4, S5, and S6), samples are cut from the crystal ingot. For points S2, S3, S4, and S5, which are both the end of the previous segment and the beginning of the next, only one sample is needed. Therefore, there are six samples corresponding to the six positions. These samples will be sent to testing equipment 120 for lifetime mapping testing.
[0031] Test equipment 120 performs lifetime mapping tests on six samples at positions S1-S6. The output of this test is six test maps. A lifetime mapping map is a two-dimensional image where different colors or grayscale values represent the minority carrier lifetime at different locations on the sample surface. As in the background art, different defect types (COP, LDP) and concentrations lead to characteristic lifetime variations, thus presenting different textures, patches, or regions on the map. These six test maps are uploaded to big data platform 130. After receiving map 301, big data platform 130 calls its internal sample library. The sample library stores various historical maps and their corresponding known quality classifications (P-Band, Pv, Pv-Pi, Pi, B-Band). Big data platform 130 can use image recognition algorithms such as Convolutional Neural Networks (CNN) or other pattern matching algorithms to compare the input test maps with the features in the sample library and output a quality assessment result for classification.
[0032] In some examples, the evaluation process of the big data platform 130 is as follows: A test map at position S1 is input; platform 130 finds that the texture features of this map are highly similar to the sample set marked "P-Band" (high COP risk) in sample library 302 (e.g., similarity > 95%); big data platform 130 outputs a quality evaluation result for position S1: "P-Band". The optimization unit 140 collects the evaluation results for all positions. For example, evaluation result 303 is as follows: the map match at position S1 is P-Band, the map match at position S2 is Pv, the map match at position S3 is Pv-Pi, the map match at position S4 is Pv-Pi, the map match at position S5 is Pi, and the map match at position S6 is B-Band. These quality evaluation results are sent to the optimization unit 140 or retrieved by the optimization unit 140 from platform 130 for subsequent calculations.
[0033] In some example implementations of this disclosure, a compensation value is determined based on the quality classification and the mapping relationship between the quality classification and the compensation value.
[0034] Specifically, the optimization unit 140 or the big data platform 130 internally stores a mapping table to represent the above mapping relationship. This mapping table establishes a correspondence between the output quality assessment result and a numerical compensation value O. This mapping table can be specifically defined as follows: Quality category P-Band (COP high risk) corresponds to Offset O1; Quality category Pv (COP low risk) corresponds to Offset O2; Quality category Pv-Pi (no defects) corresponds to Offset O3; Quality category Pi (LDP low risk) corresponds to Offset O4; Quality category B-Band (LDP high risk) corresponds to Offset O5. Since Pv-Pi is the optimal quality, no compensation is performed, and we can determine that O3 = 0. For the other four types of quality with defect risk, the Offset value is non-zero.
[0035] In semiconductor crystal growth physics, the ratio of pulling speed (V) to the temperature gradient (G) at the crystal growth interface (V / G) is a key factor determining the defect type (vacancy or interstitial). When the V / G ratio is too high, vacancies cannot diffuse to the crystal surface in time, leading to oversaturation and aggregation of vacancies in the crystal, forming COP defects. This corresponds to the P-Band (high COP) and Pv (low COP) regions. When the V / G ratio is too low, silicon atoms (interstitials) cannot fill the lattice in time, leading to oversaturation and aggregation of interstitial atoms, forming LDP defects. This corresponds to the B-Band (high LDP) and Pi (low LDP) regions. When the V / G ratio is just right, the concentrations of vacancies and interstitials reach a dynamic equilibrium, annihilating each other, resulting in no net defects in the crystal. This corresponds to the Pv-Pi (defect-free) region. Therefore, the offset values (O1, O2, O4, O5) in the mapping table have clear physical meanings. For P-Band (V / G too high), V / G needs to be reduced. With G remaining essentially constant, the pulling speed V needs to be reduced. Therefore, O1 is a negative value, for example, O1 = -0.2 mm / min. For Pv (V / G too high), V / G needs to be slightly reduced. Therefore, O2 is a small negative value, for example, O2 = -0.1 mm / min. For Pv-Pi (V / G optimal), no adjustment is needed, and O3 = 0 mm / min. For Pi (V / G too low), V / G needs to be slightly increased. Therefore, O4 is a small positive value, for example, O4 = 0.1 mm / min. For B-Band (V / G too low), V / G needs to be increased. Therefore, O5 is a positive value, for example, O5 = 0.2 mm / min. This mapping table can be derived based on a physical model and statistical analysis of a large amount of historical data, and is stored in the memory of the optimization unit 140.
[0036] In other examples, the mapping table 400 can also be dynamically adjusted and adaptively optimized by the big data platform 130 through machine learning algorithms based on the optimization results of each round.
[0037] The optimization unit 140 will perform the following matching: "P-Band" in S1 matches O1 = -0.2 mm / min; "Pv" in S2 matches O2 = -0.1 mm / min; "Pv-Pi" in S3 matches O3 = 0 mm / min; "Pv-Pi" in S4 matches O3 = 0 mm / min; "Pi" in S5 matches O4 = +0.1 mm / min; and "B-Band" in S6 matches O5 = +0.2 mm / min. The optimization unit 140 has obtained the compensation values corresponding to all sampling positions (S1-S6), where O3 = O4 = 0, O1 = -0.2, O2 = -0.1, O5 = 0.1, and O6 = 0.2.
[0038] After obtaining the above compensation value, step S220 can be executed.
[0039] In step S220, the target pulling speed at the sampling position is determined based on the compensation value and the actual pulling speed at the sampling position.
[0040] After obtaining the compensation value, the optimization unit 140 needs to calculate the target pulling speed (NT) at each position point from S1 to S6. The optimization unit 140 first obtains the actual pulling speed curves and the designed pulling speed curves corresponding to each position point from S1 to S6 during the pulling process of the first crystal rod from the crystal pulling equipment 110. Among them, referencing... Figure 4 The actual pulling speed is the pulling speed actually measured and recorded by the sensors of the crystal pulling equipment 110 (e.g., a photoelectric encoder that measures the pulling length of the seed crystal rod). The designed pulling speed is the pulling speed curve preset by the system when pulling the first crystal rod. Ideally, the designed pulling speed should be equal to the actual pulling speed, but in reality, due to factors such as thermal field fluctuations and melt level drops, there may be a deviation between the designed pulling speed and the actual pulling speed.
[0041] Reference Figure 5 Taking the calculation of the target pulling speed at point S1 as an example, the target pulling speed NT1 is calculated at position S1. First, the actual pulling speed P1 at position S1, the design pulling speed T1 at position S1, and the compensation value O1 = -0.2 mm / min at position S1 are obtained. Next, NT1 is calculated. The calculation of NT1 is based on the actual pulling speed P1 and the compensation value O1, and the formula is NT1 = P1 + O1. Taking position S1 as an example, the Lifetime mapping test result shows P-Band, which indicates that the actual pulling speed at this position is too high and needs to be adjusted downwards. The adjustment amount is O1.
[0042] Assume at position S1 (length L = 0 mm): The designed drawing speed T1 = 1.6 mm / min; The actual drawing speed of the equipment P1 = 1.5 mm / min (the equipment actually did not reach the target, P < T); The quality result = P - Band (high risk of COP); The corresponding Offset O1 = -0.2 mm / min. Calculate the target drawing speed NT1 at position S1 = P1 + O1 = 1.5 mm / min + (-0.2 mm / min) = 1.3 mm / min.
[0043] The optimization unit 140 uses the same principle to calculate the target drawing speed discrete points NT1, NT2, NT3, NT4, NT5, NT6 at all six sampling positions (S1...S6). That is, NT1 = P1 + O1, NT2 = P2 + O2, NT3 = P3 + O3, NT4 = P4 + O4, NT5 = P5 + O5, NT6 = P6 + O6. In some embodiments, if the execution of the crystal pulling equipment 110 is very precise, P and T are almost exactly equal.
[0044] In some examples of the present disclosure, after obtaining the above target drawing speed, step S230 can be executed.
[0045] In step S230, based on the trend type between multiple sampling positions of the designed drawing speed curve of the historical crystal bar and the target drawing speed, a target drawing speed curve for controlling the crystal pulling equipment to pull subsequent crystal bars is generated.
[0046] Six discrete NT points NT1, NT2, NT3, NT4, NT5, NT6 are obtained in step S220. The crystal pulling equipment 110 cannot execute discrete points and requires a continuous and smooth target drawing speed curve. A target drawing speed curve for controlling the crystal pulling equipment to pull subsequent crystal bars can be generated based on the trend type between multiple sampling positions of the above-mentioned designed drawing speed curve of the historical crystal bar and the target drawing speed.
[0047] In some examples, the trend type can include a monotonic change in drawing speed, a stable drawing speed, and a non - monotonic change in drawing speed.
[0048] In some examples, between multiple sampling positions can refer to the area between adjacent two sampling positions. For example, between S2 and S3. When the above trend type is a monotonic change, the monotonic change can include a continuous increase in drawing speed or a continuous decrease in drawing speed. In this trend type, obtain the first difference of the target drawing speed between adjacent sampling positions; determine the change coefficient based on the first difference and the second difference of the designed drawing speed between adjacent sampling positions; determine the target drawing speed curve based on the change coefficient, the designed drawing speed curve, and the target drawing speed at adjacent sampling positions.
[0049] For example, optimization unit 140 analyzes the design pull speed curve T between S2 and S3. Optimization unit 140 calculates the derivative (or difference) of T(x) over the interval [L2, L3] (where L2 and L3 are the lengths of S2 and S3, respectively). If the derivative is always positive within this interval, it is determined to be "continuously rising". If it is always negative, it is determined to be "continuously falling".
[0050] Reference Figure 6 This interval is determined to be monotonically changing. The target pulling speed NT(x) at any point x within this interval (S2-S3) is calculated using a linear interpolation formula: NT （x+1） =NT x +(T x+1 -T x )×(NT3-NT2) / (T3-T2) Among them, NT （x） NT2 represents the target pulling speed at length x, where x is the length corresponding to the pulling speed (L2 <= x <= L3), L2 is the length at position S2, L3 is the length at position S3, NT2 is the target pulling speed at position S2, NT3 is the target pulling speed at position S3, and T... x+1 The design pulling speed at length x+1, T x T1 is the design pulling speed at length x, T2 is the design pulling speed at length S3, T3 is the design pulling speed at length S2, (NT3-NT2) is the first difference, (T3-T2) is the second difference, and (NT3-NT2) / (T3-T2) is the variation coefficient.
[0051] In some examples, when the trend type is stable, refer to Figure 7 The pulling speed adjustment coefficient is determined based on the third difference between the target pulling speeds at the two sampling locations and the distance between the sampling locations; the target pulling speed curve is determined based on the pulling speed adjustment coefficient and the target pulling speed at the sampling locations.
[0052] Optimization unit 140 analyzes the design casting speed curve T between S3 and S4. Optimization unit 140 calculates the derivative of T(x) over the interval [L3, L4]. If the derivative is close to 0 throughout the interval, the casting speed is considered stable. A linear interpolation formula is used: NT （x） =NT3+×(x-L3)×(NT4-NT3) / (L4-L3).
[0053] Where L3 and L4 are the lengths of S3 and S4, respectively; NT3 and NT4 are the target pulling speeds of S3 and S4, respectively. NT4-NT3 is the third difference, and (NT4-NT3) / (L4-L3) is the pulling speed adjustment coefficient.
[0054] Assumptions: S3 position L3 = 400mm, NT3 = 1.4mm / min (mass = Pv - Pi, O3 = 0, assume P3 = 1.4); S4 position L4 = 600mm, NT4 = 1.4mm / min (mass Pv - Pi, O4 = 0, assume P4 = 1.4). The calculation process is as follows: Slope k = (NT4 - NT3) / (L4 - L3) = (1.4 - 1.4) / (600 - 400) = 0 / 200 = 0. （x） =1.4 + 0 × (x - 400) = 1.4 mm / min. The logic is that if the original T... （x） It is horizontal, and the quality of both points S3 and S4 is optimal (Pv-Pi, Offset=0), then the new NT （x） The curve should also maintain this optimal horizontal line.
[0055] In some example embodiments of this disclosure, when the trend type is non-monotonic change in pulling speed, the method further includes: dividing the interval between sampling positions into multiple sub-intervals according to the design pulling speed curve; wherein the trend type of the design pulling speed curve in the multiple sub-intervals is stable pulling speed or monotonically changing pulling speed.
[0056] For example, refer to Figure 8 The optimization unit 140 analyzes the design casting speed curve between S1 and S2. The optimization unit 140 calculates the derivative of T(x) over the interval [L1, L2]. If the derivative changes sign within this interval (e.g., from negative to positive, or from positive to negative), it means that there is an extreme point (minimum or maximum). This interval is determined to be a non-monotonic casting speed range.
[0057] The specific method for obtaining the target pulling speed curve is as follows: First, the interval is divided. Within the interval S1~S2 ([L1, L2]), the optimization unit 140 finds the extreme point of the design pulling speed T(x) (e.g., the "lowest pulling speed" point, whose length is L_min). The interval S1~S2 is divided into three sub-intervals: the first sub-interval is the part to the left of the lowest pulling speed (i.e., [L1, L_min]); the second sub-interval (Interval2) is the part with the lowest pulling speed (i.e., the L_min point itself, or a small interval near L_min [L_min_a, L_min_b]); the third sub-interval is the part to the right of the lowest pulling speed (i.e., [L_min, L2] or [L_min_b, L2]).
[0058] The first and third sub-intervals represent single-point changes in pulling speed, and the corresponding calculation methods can be used to determine the pulling speed curve according to the aforementioned technical solution for monotonic pulling speed changes. The second sub-interval represents a stable pulling speed, and the corresponding curve can be calculated using the aforementioned technical solution for stable pulling speeds. Finally, integrating the intervals yields the target pulling speed curve. The specific technical logic has been detailed above and will not be repeated here.
[0059] After the above steps, the optimization unit 140 obtains the target pulling speed curves for each segment: S1-S2, S2-S3, S3-S4, S4-S5, and S5-S6. Specifically, these can be NT_seg1, NT_seg2, NT_seg3, NT_seg4, and NT_seg5. Among them, NT_seg1 is in the interval [L1, L2], NT_seg2 is in the interval [L2, L3], NT_seg3 is in the interval [L3, L4], and so on.
[0060] In some examples, optimization unit 140 performs curve stitching, stitching together all segmented curves in order of ingot length to form a complete target pulling speed curve. Since the calculation of each segmentation algorithm ensures that the endpoints are continuous, the stitched complete NT curve is also globally continuous.
[0061] Finally, the optimization unit 140 executes the feedback step, sending the complete target pulling speed curve data to the controller of the crystal pulling equipment 110. The crystal pulling equipment 110 will use this target pulling speed curve as the process parameters for pulling the second crystal ingot. This embodiment demonstrates the final step of the closed loop. It transforms the physical defect data of the historical crystal ingots into the physical control parameters of the next crystal ingot through an algorithm. This solves the problems of inability to calculate NT and inability to provide fast feedback in the prior art.
[0062] See attached document Figure 9 This is a schematic diagram comparing the quality optimization effects of crystal rods according to an embodiment of this disclosure. The first crystal rod was drawn using a designed pulling speed curve. Its "Cu Haze" test results show that the crystal rod has large defect areas (e.g., P-Band, B-Band) at positions S1, S2, S5, and S6, while the "defect-free" Pv-Pi area is very narrow. The second crystal rod was drawn using the complete target pulling speed curve calculated according to an embodiment of this disclosure. The defect areas of the second crystal rod are significantly reduced or eliminated, while the "defect-free" Pv-Pi area is significantly expanded, penetrating the main body of the crystal rod. This embodiment, through... Figure 9 The comparison between the first and second crystal rods in the process confirms that the present disclosure can effectively control the crystal pulling equipment 110 to pull crystal rods of higher quality and fewer defects.
[0063] Furthermore, this disclosure also provides an optimization device for crystal rod pulling speed, with reference to... Figure 10 The crystal rod pulling speed optimization device 1000 may include a compensation determination module 1010, a pulling speed compensation module 1020, and a pulling speed determination module 1030.
[0064] In some examples, the ingot pulling speed optimization device 1000 can also be used to segment and sample historical ingots to obtain multiple samples; and to perform minority carrier lifetime imaging tests on the multiple samples to determine the quality assessment results.
[0065] In some examples, the quality assessment results include multiple quality categories; the compensation determination module 0 can be used to determine the compensation value based on the quality category and the mapping relationship between the quality category and the compensation value.
[0066] In some examples, the trend types include monotonically changing pull speed, stable pull speed, and non-monotonic change in pull speed.
[0067] In some examples, when the trend type is monotonically changing pulling speed, the pulling speed determination module 1030 can be used to obtain a first difference in the target pulling speed between adjacent sampling positions; determine a change coefficient based on the first difference and a second difference in the design pulling speed between adjacent sampling positions; and determine a target pulling speed curve based on the change coefficient, the design pulling speed curve, and the target pulling speed at adjacent sampling positions.
[0068] In some examples, when the trend type is stable, the pulling speed determination module 1030 can be used to determine the pulling speed adjustment coefficient based on the third difference between the target pulling speeds at two sampling locations and the distance between the sampling locations; and to determine the target pulling speed curve based on the pulling speed adjustment coefficient and the target pulling speed at the sampling locations.
[0069] In some examples, the ingot pulling speed optimization device 1000 can also be used to divide the interval between sampling positions into multiple sub-intervals according to the design pulling speed curve; wherein the trend type of the design pulling speed curve in the multiple sub-intervals is either stable pulling speed or single-point change in pulling speed.
[0070] Please refer to Figure 11 This diagram illustrates a structural block diagram of a computing device provided in an exemplary embodiment of this disclosure. In some examples, the computing device 110 may be at least one of devices such as a smartphone, smartwatch, desktop computer, laptop, virtual reality terminal, augmented reality terminal, wireless terminal, and laptop computer. The computing device 110 has communication capabilities and can access wired or wireless networks. The computing device 110 may refer to one of a plurality of terminals, and those skilled in the art will understand that the number of such terminals may be more or less. It is understood that the computing device 110 undertakes the computation and processing work of the technical solution of this disclosure, and this disclosure does not limit it in this respect.
[0071] like Figure 11 As shown, the computing device in this disclosure may include one or more of the following components: processor 1110 and memory 1120.
[0072] Optionally, the processor 1110 connects various parts within the computing device using various interfaces and lines. It executes various functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in the memory 1120, and by calling data stored in the memory 1120. Optionally, the processor 1110 can be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 1110 can integrate one or more of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), Neural-network Processing Unit (NPU), and baseband chip. Specifically, the CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display on the touch screen; the NPU is used to implement Artificial Intelligence (AI) functions; and the baseband chip is used for wireless communication. It is understandable that the aforementioned baseband chip may not be integrated into the processor 1110, but may be implemented using a separate chip.
[0073] The memory 1120 may include random access memory (RAM) or read-only memory (ROM). Optionally, the memory 1120 may include a non-transitory computer-readable storage medium. The memory 1120 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 1120 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the various method embodiments described above, etc.; the data storage area may store data created according to the use of the computing device, etc.
[0074] In addition, those skilled in the art will understand that the structure of the computing device shown in the above figures does not constitute a limitation on the computing device. The computing device may include more or fewer components than shown, or combine certain components, or have different component arrangements. For example, the computing device may also include a display screen, camera assembly, microphone, speaker, radio frequency circuit, input unit, sensors (such as accelerometer, angular velocity sensor, light sensor, etc.), audio circuit, WiFi module, power supply, Bluetooth module, etc., which will not be described in detail here.
[0075] This disclosure also provides a computer-readable storage medium storing at least one instruction, which is executed by a processor to implement the ingot pulling speed optimization method of the various embodiments described above.
[0076] This disclosure also provides a computer program product including computer instructions stored in a computer-readable storage medium; a processor of a computing device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computing device to perform the method for optimizing the ingot pulling speed of the various embodiments described above.
[0077] Those skilled in the art will recognize that the functions described in this disclosure in one or more of the examples above can be implemented using hardware, software, firmware, or any combination thereof. When implemented in software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or code on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any medium that facilitates the transfer of a computer program from one place to another. Storage media can be any available medium accessible to a general-purpose or special-purpose computer.
[0078] It should be noted that the technical solutions described in this disclosure can be combined arbitrarily as long as they do not conflict.
[0079] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A method for optimizing the crystal ingot pulling speed, applied to a crystal ingot pulling system including a crystal pulling equipment, characterized in that, The method includes: The compensation value is determined based on the quality assessment results of wafers at multiple sampling locations in the historical ingots. The target pulling speed at the sampling location is determined based on the compensation value and the actual pulling speed at the sampling location. Based on the trend type of the design pulling speed curve of the historical crystal ingots across multiple sampling locations and the target pulling speed, a target pulling speed curve is generated for controlling the crystal pulling equipment to pull subsequent crystal ingots.
2. The method for optimizing the crystal pulling speed according to claim 1, characterized in that, The method further includes: The historical crystal rods were sampled in segments to obtain multiple sample pieces; Minority carrier lifetime imaging tests were performed on the multiple samples to determine the quality assessment results.
3. The method for optimizing the crystal pulling speed according to claim 1, characterized in that, The quality assessment results include multiple quality categories; Based on the quality assessment results of wafers at multiple sampling locations in historical ingots, compensation values are determined, including: Based on the quality classification, the compensation value is determined according to the mapping relationship between the quality classification and the compensation value.
4. The method for optimizing the crystal pulling speed according to claim 1, characterized in that, The trend types include monotonic change in pulling speed, stable pulling speed, and non-monotonic change in pulling speed.
5. The method for optimizing the crystal pulling speed according to claim 4, characterized in that, When the trend type is a monotonically changing pulling speed, the trend type of the design pulling speed curve based on the historical crystal ingot among multiple sampling positions, and the target pulling speed generation for controlling the crystal pulling equipment to pull subsequent crystal ingots, include: Obtain the first difference in target pulling speed between adjacent sampling locations; The variation coefficient is determined based on the first difference and the second difference in the design pulling speed between adjacent sampling locations; The target pulling speed curve is determined based on the variation coefficient, the designed pulling speed curve, and the target pulling speed at adjacent sampling positions.
6. The method for optimizing the crystal pulling speed according to claim 4, characterized in that, When the trend type is stable pulling speed, the trend type of the design pulling speed curve based on the historical crystal ingot among multiple sampling positions, and the target pulling speed generation for controlling the crystal pulling equipment to pull subsequent crystal ingots, include: The pulling speed adjustment coefficient is determined based on the third difference between the target pulling speeds at the two sampling locations and the distance between the sampling locations; The target pulling speed curve is determined based on the pulling speed adjustment coefficient and the target pulling speed at the sampling position.
7. The method for optimizing the crystal pulling speed according to claim 4, characterized in that, When the trend type is a non-monotonic change in pulling speed, the method further includes: Based on the designed pulling speed curve, the interval between the sampling positions is divided into multiple sub-intervals; Among them, the trend type of the design pulling speed curve in multiple sub-intervals is either stable pulling speed or single-point change in pulling speed.
8. A device for optimizing the pulling speed of crystal rods, characterized in that, An optimization device for the ingot pulling speed, applicable to an ingot pulling system including a crystal pulling equipment, comprises: The compensation determination module is used to determine the compensation value based on the quality evaluation results of the wafers at multiple sampling locations in the historical ingots. A pulling speed compensation module is used to determine the target pulling speed at the sampling position based on the compensation value and the actual pulling speed at the sampling position. The pulling speed determination module is used to generate a target pulling speed curve for controlling the crystal pulling equipment to pull subsequent crystal rods based on the trend type of the design pulling speed curve of the historical crystal rod between multiple sampling positions and the target pulling speed.
9. An electronic device, characterized in that, The electronic device includes a processor and a memory; the processor is configured to execute instructions stored in the memory to implement the method for optimizing the ingot pulling speed as described in any one of claims 1 to 7.
10. A computer storage medium, characterized in that, The computer storage medium stores at least one instruction, which is executed by a processor to implement the method for optimizing the ingot pulling speed as described in any one of claims 1 to 7.