Magnetic field dynamic adaptation system and method for improving single crystal oxygen content uniformity

Abstract

The invention belongs to the technical field of semiconductor crystal growth, and particularly relates to a magnetic field dynamic adaptation system and method for improving single crystal oxygen content uniformity, the melt liquid level height H is sensed in real time through a liquid level position sensing unit, thermal field state data W is obtained through a temperature zone monitoring unit, and the single crystal oxygen content uniformity is improved on the basis of the melt liquid level height H and the thermal field state data W. And the control module converts the target position Pz into a driving instruction and sends the driving instruction to the magnetic field execution module to drive the magnetic field execution module to synchronously and dynamically adjust the magnetic field space position of the magnetic field generator, converts the target intensity Btarget into a driving instruction and sends the driving instruction to the magnetic field execution module to drive the magnetic field execution module to synchronously and dynamically adjust the magnetic field intensity of the magnetic field generator. The magnetic field space position of the magnetic field generator is dynamically adjusted to promote oxygen to be conveyed uniformly, so that the radial oxygen concentration CV value is reduced, the magnetic field intensity of the magnetic field generator is dynamically adjusted to avoid nonlinear fluctuation in the oxygen segregation process, and the linearity of an axial oxygen distribution curve is improved.

Description

Technical Field

[0001] This invention belongs to the field of semiconductor crystal preparation technology, specifically relating to a magnetic field dynamic adaptation system and method for improving the uniformity of oxygen content in single crystals. Background Technology

[0002] During the Czochralski process of growing single-crystal silicon, oxygen impurities generated by the reaction between the quartz crucible and the silicon melt enter the crystal, and their radial and axial distribution uniformity directly affects the performance and yield of semiconductor devices. Currently, the industry commonly uses static or semi-static hook-shaped magnetic fields to suppress melt convection and regulate oxygen transport. However, during the growth process, which lasts for tens of hours, the melt level continuously decreases, and the thermal field temperature regions (such as the volume of the high-temperature melt region, the solid-liquid interface temperature gradient, and the thermal field symmetry) also undergo dynamic changes.

[0003] In the early stages of crystal pulling, an excessively strong magnetic field can excessively suppress melt convection, causing oxygen to accumulate locally near the crucible wall and making it difficult to transport it uniformly to the center of the melt, resulting in a decrease in the radial oxygen uniformity of the crystal. In the later stages of crystal pulling, the melt volume decreases and the convection mode changes, and the original magnetic field strength may no longer be sufficient to effectively suppress turbulence, leading to unstable oxygen transport and causing fluctuations in the axial oxygen content of the crystal.

[0004] Therefore, fixed magnetic field parameters cannot adapt to the above dynamic changes, thus causing uneven oxygen content distribution. Summary of the Invention

[0005] In view of this, the present invention provides a magnetic field dynamic adaptation system and method for improving the uniformity of oxygen content in single crystals, so as to solve the technical problem of uneven distribution of silicon-oxygen content in single crystals during crystal pulling in the prior art.

[0006] To achieve the above objectives, this application adopts the following approach:

[0007] A method for dynamic magnetic field adaptation to improve the uniformity of oxygen content in single crystals includes the following steps:

[0008] S10. Real-time acquisition of melt surface height H and thermal field state data W during crystal growth;

[0009] S20. Based on the liquid level height H, calculate the target position P_z of the required magnetic field zero plane; based on the thermal field state data W, calculate the target value B_target of the required magnetic field strength.

[0010] S30. Based on the target position P_z, adjust the spatial position of the magnetic field, and based on the target value B_target, simultaneously adjust the strength of the magnetic field;

[0011] S40. Repeat steps S10 to S30 until crystal growth is complete.

[0012] Preferably, in step S10, the thermal field state data W includes the key point temperature T of the crucible or melt, the volume V of the high-temperature melt zone, and the solid-liquid interface temperature gradient G.

[0013] Preferably, in step S20, the target position P_z is calculated by looking up a preset mapping relationship between the liquid level height and the zero magnetic surface position and combining it with interpolation.

[0014] Preferably, step S30 further includes converting the target position P_z into a current adjustment value I_1 for the upper coil of the hook-shaped magnetic field generator, converting the target value B_target into a current adjustment value I_2 for the lower coil of the hook-shaped magnetic field generator, and adjusting the currents of the upper and lower coils of the hook-shaped magnetic field generator based on I_1 and I_2, thereby adjusting the spatial position and intensity of the magnetic field.

[0015] I_2=B_target / (2.4×E×N_1), I_1=1.2×I_2, N_1 is the number of turns of the coil on the hook-shaped magnetic field generator, and E is the magnetic circuit calibration coefficient.

[0016] Preferably, in step S30, calculating the target value B_target includes:

[0017] S31. Calculate the baseline intensity B_base based on the current thermal field state data W;

[0018] S32. Select the stage correction coefficient k based on the baseline strength B_base and the current growth stage, and calculate the intermediate strength B_mid, where B_mid = B_base × k;

[0019] S33. Calculate the dynamic correction ΔB_T based on the temperature change rate in the current thermal field state data W;

[0020] S34. Calculate the target value B_target based on the intermediate intensity B_mid and the dynamic correction amount ΔB_T, where B_target = B_mid + ΔB_T.

[0021] Preferably, in step S31, the base strength B_base is calculated based on the critical point temperature T of the crucible or melt and the critical point temperature threshold X of the crucible or melt.

[0022] When T ≥ X ℃, B_base = 0.8T - 1100;

[0023] When T < X℃, B_base = 1.2T - 1600 + 50 × (V_0 - V) / V_0;

[0024] In the formula, V_0 is the initial melt volume, and X is 1450.

[0025] Preferably, in step S32, the current growth stage includes the crystal-initiating stage, the mid-stage of equal diameter growth, and the late-stage of equal diameter growth, wherein k is 0.8 for the crystal-initiating stage, k is 1.0 for the mid-stage of equal diameter growth, and k is 1.2 for the late-stage of equal diameter growth.

[0026] A magnetic field dynamic adaptation system for improving the uniformity of oxygen content in single crystals, applied to the aforementioned magnetic field dynamic adaptation method for improving the uniformity of oxygen content in single crystals, includes a data sensing module, a magnetic field execution module, and a control module, wherein the data sensing module and the magnetic field execution module are both communicatively connected to the control module;

[0027] The data sensing module is used to collect dynamic process parameters in real time during crystal growth, including at least a liquid surface position sensing unit for monitoring the liquid surface height H of the melt and a temperature zone monitoring unit for acquiring thermal field state data W.

[0028] The magnetic field execution module includes at least one magnetic field generator, which is capable of generating a magnetic field with adjustable spatial position and intensity at the zero magnetic field plane.

[0029] The control module is configured to: receive real-time process parameters from the data sensing module, calculate and output the target position P_z of the magnetic field zero plane based on the liquid level height H of the melt, convert the target position P_z into a drive command, and send it to the magnetic field execution module to drive the magnetic field execution module to adjust the spatial position of the magnetic field zero plane of the magnetic field generator; calculate and output the target value B_target of the magnetic field strength based on the thermal field state data W, convert the target strength B_target into a drive command, and send it to the magnetic field execution module to drive the magnetic field execution module to synchronously adjust the magnetic field strength of the magnetic field generator.

[0030] Preferably, the magnetic field generator is a hook-shaped magnetic field generator composed of two independent coils, upper and lower. The two independent coils of the hook-shaped magnetic field generator are symmetrically installed on the outside of the furnace chamber of the single crystal furnace. The spatial position of the zero magnetic field plane is adjusted by adjusting the current ratio of the upper coil and the lower coil, and the magnetic field strength is adjusted by adjusting the total current of the upper coil and the lower coil.

[0031] Preferably, the control module adopts an industrial PLC or a dedicated industrial control computer, with a built-in real-time operating system. It establishes a connection with the data sensing module and the magnetic field execution module through the Profinet or Modbus communication protocol, with a data transmission delay of ≤100ms. It has a built-in "process status - magnetic field parameters - oxygen distribution" mapping relationship model or adaptive control algorithm.

[0032] In the aforementioned magnetic field dynamic adaptation system and method for improving the uniformity of oxygen content in single crystals, the liquid level height H of the melt is sensed in real time by a liquid level position sensing unit, and the thermal field state data W is obtained by a temperature zone monitoring unit. Based on the liquid level height H and the thermal field state data W, the target position P_z and the target value B_target are calculated. The control module converts the target position P_z into a driving command and sends it to the magnetic field execution module to drive the magnetic field execution module to adjust the spatial position of the magnetic field zero plane of the magnetic field generator. The target intensity B_target is converted into a driving command and sent to the magnetic field execution module to drive the magnetic field execution module to synchronously adjust the magnetic field intensity of the magnetic field generator. By dynamically adjusting the spatial position of the magnetic field zero plane of the magnetic field generator, uniform oxygen transport is promoted, and the radial oxygen concentration CV value is reduced. By dynamically adjusting the magnetic field intensity of the magnetic field generator, nonlinear fluctuations in the oxygen segregation process are avoided, and the linearity of the axial oxygen distribution curve is improved, thereby improving the uniformity of the overall oxygen content of the single crystal. Attached Figure Description

[0033] Figure 1 This is a flowchart of this application.

[0034] Figure 2 This is a functional block diagram of this application.

[0035] The diagram shows a data sensing module 100, a liquid level sensing unit 110, a temperature monitoring unit 120, a magnetic field actuation module 200, and a control module 300. Detailed Implementation

[0036] To facilitate understanding of this application, a more comprehensive description will be provided below with reference to the accompanying drawings. Preferred embodiments of this application are also given. However, this application can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to enable a more thorough and complete understanding of the disclosure of this application.

[0037] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein in the specification of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0038] During the entire process of Czochralski single-crystal silicon growth, the applicant's technical personnel discovered that the thermal field temperature range is not stable but dynamically evolves with the crystal growth progress and melt state, as shown in the following analysis:

[0039] The dynamic shrinkage of the high-temperature melt region volume and the nonlinear change in oxygen dissolution rate: During crystal growth, as the single-crystal silicon rod is continuously pulled up, the melt level continuously decreases (the decrease can reach 30-50 cm), and the volume of the high-temperature melt region (silicon melt region with a temperature ≥1400℃) shrinks synchronously—the melt volume can reach more than 100L in the early stage of crystal growth, but only 20-30L remains in the later stage of constant diameter growth. The oxygen dissolution rate is positively correlated with the volume of the high-temperature melt region: in the early stage, the high-temperature region is large, and the reaction contact surface between the quartz crucible and the melt is wide, resulting in an oxygen dissolution rate as high as 1.2×10¹. 7 at / (cm³·h); In the later stages, the high-temperature region is concentrated at the bottom of the crucible, the reaction contact area is reduced, and the oxygen dissolution rate drops to 0.3×10¹ 7 The value is below at / (cm³·h). A fixed magnetic field strength cannot match this change: if a weak magnetic field (1000-1500 Gs) is set according to the initial needs, allowing moderate convection to disperse high concentrations of oxygen, then when the oxygen dissolution rate is low in the later stages, the weak magnetic field cannot suppress the enhanced convection caused by the shrinkage of the melt volume, resulting in insufficient oxygen transport and low oxygen content at the tail of the crystal; if a strong magnetic field (3000-4500 Gs) is set according to the later needs, the strong magnetic field in the initial stage will excessively suppress convection, causing oxygen to accumulate near the crucible wall, widening the radial oxygen concentration difference (edge ​​vs. center) of the crystal, and the CV value will rise to over 15%.

[0040] The dynamic fluctuations of the solid-liquid interface temperature gradient and its sensitive dependence on oxygen segregation efficiency: The solid-liquid interface is the key node for oxygen to merge from the melt into the crystal, and its temperature gradient (G) directly determines the oxygen segregation efficiency (segregation coefficient k0≈0.35; the larger the temperature gradient, the more difficult it is for oxygen to merge into the crystal). During growth, the temperature gradient G exhibits a dynamic characteristic of "first rising and then falling": In the early stage of crystallization, the crystal has just formed, the solid-liquid interface is unstable, and G≈8-10℃ / mm; in the middle stage of constant diameter, the crystal growth rate is stable, the interface is flat, and G rises to 12-15℃ / mm; in the later stage of constant diameter, the heat loss of the melt intensifies, and the interface temperature gradient falls back to 9-11℃ / mm. The oxygen segregation efficiency is extremely sensitive to the magnetic field strength: At high gradients (G≥12℃ / mm), the magnetic field strength needs to be appropriately increased to slow down the melt flow, prolong the residence time of oxygen at the interface, and ensure sufficient segregation; at low gradients (G<10℃ / mm), the magnetic field strength needs to be reduced to avoid oxygen accumulation at the interface, leading to excessive local oxygen content. A fixed magnetic field strength cannot accommodate this fluctuation: if a medium-strength magnetic field (2000-2500 Gs) is set, the magnetic field is insufficient at high gradients, resulting in incomplete oxygen segregation and low axial oxygen content in the crystal; at low gradients, the magnetic field is too strong, leading to excessive oxygen segregation and high axial oxygen content in the crystal, thus creating a nonlinear fluctuation (fluctuation amplitude > ±0.8 × 10¹). 7 The oxygen distribution at / cm³ cannot meet the linearity requirements of axial oxygen distribution for high-end devices (fluctuation amplitude ≤ ±0.3×10¹). 7 The requirement is (at / cm³).

[0041] The symmetry decay of the thermal field and the dynamic changes in the melt convection mode: In the later stages of crystal growth, the amount of melt in the crucible decreases, and the heat exchange ratio between the crucible wall and the melt changes. The originally symmetrical thermal field gradually shifts to one side (eccentricity can reach 10%-15%), causing the melt convection to change from "axisymmetric circulation" to "one-sided eddy current". The oxygen transport path changes accordingly: Under a symmetrical thermal field, oxygen diffuses uniformly along the crucible wall to the interface; under a shifted thermal field, oxygen is easily concentrated and transported along the eddy current direction, resulting in a higher local oxygen content in the crystal. The spatial distribution of the fixed magnetic field is symmetrical and cannot cope with asymmetrical convection: The suppression effect of the symmetrical fixed magnetic field on the one-sided eddy current is unbalanced, and the oxygen transport in the eddy current region is not effectively controlled, ultimately leading to a "local peak" in the radial oxygen concentration of the crystal, with the concentration difference between the edge and the center reaching up to 1.5×10¹. 7 at / cm³, far exceeding the allowable limit of ≤0.5×10¹ for high-end semiconductor devices. 7 at / cm³ standard.

[0042] The aforementioned defects result in a high coefficient of variation (CV) of radial oxygen concentration in monocrystalline silicon, and the axial oxygen distribution exhibits nonlinear fluctuations.

[0043] Based on the above conclusions, please refer to Figure 1 This application proposes a method for dynamic magnetic field adaptation to improve the uniformity of oxygen content in single crystals, comprising the following steps:

[0044] S10. Real-time acquisition of melt surface height H and thermal field state data W during crystal growth process, with a sampling frequency ≥10Hz, to ensure data timeliness;

[0045] S20. Based on the liquid level height H, calculate the target position P_z of the required magnetic field zero plane; based on the thermal field state data W, calculate the target value B_target of the required magnetic field strength.

[0046] S30. Based on the target position P_z, adjust the spatial position of the magnetic field, and based on the target value B_target, simultaneously adjust the strength of the magnetic field;

[0047] S40. Repeat steps S10 to S30 until crystal growth is complete.

[0048] The magnetic field dynamic adaptation method for improving the uniformity of oxygen content in single crystals provided in this application involves real-time sensing of the melt surface height H and real-time acquisition of thermal field state data W. Based on the melt surface height H and thermal field state data W, the target position P_z and target value B_target are calculated. The spatial position of the magnetic field zero plane of the magnetic field generator is dynamically adjusted based on the target position P_z, and the magnetic field strength of the magnetic field generator is synchronously adjusted based on the target strength B_target. By dynamically adjusting the spatial position of the magnetic field zero plane of the magnetic field generator, uniform oxygen transport is promoted, reducing the radial oxygen concentration CV value. By dynamically adjusting the magnetic field strength of the magnetic field generator, nonlinear fluctuations in the oxygen segregation process are avoided, and the linearity of the axial oxygen distribution curve is improved, thereby enhancing the overall uniformity of oxygen content in the single crystal.

[0049] Furthermore, the thermal field state data W includes the critical point temperature T of the crucible or melt, the volume V of the high-temperature melt zone, and the solid-liquid interface temperature gradient G.

[0050] In step S20, the target position P_z is calculated by looking up the preset mapping relationship between the liquid level height and the zero magnetic surface position and combining it with interpolation.

[0051] In this embodiment, the mapping relationship table data was calibrated through 30 batches of process tests, covering the liquid level height range throughout the entire growth stage, and clarifying the tracking type (linear / nonlinear) for each interval, as shown in Table 1 below:

[0052] Table 1 Mapping Relationship Table

[0053] In this embodiment, the calculation of the target position P_z is exemplified as follows:

[0054] First, determine the interval to which H belongs, identify the tracking type and interpolation reference point. Based on the real-time reading of H (accuracy ≤ ±0.1mm), match two adjacent reference points (H_1, P_z0-1) and (H_2, P_z0-2) in the above mapping table, and at the same time determine the tracking type (linear / nonlinear) of this interval.

[0055] Then, interpolation is performed based on the tracking type, and dynamic corrections are added:

[0056] The linear tracking interval (H≥10cm: crystallization period + mid-stage of constant diameter) is applicable to the interval: H_1≤H≤H_2 (e.g., 25cm≤H≤30cm, 10cm≤H≤25cm). The calculation logic is linear interpolation (simple formula, strong real-time performance): P_z-linear=P_z0-1+\frac(P_z0-2-P_z0-1)(H_2-H_1)×(H-H_1). To ensure that the distance between the zero magnetic surface and the liquid surface is ≤3cm, a constraint correction ΔP_1=max(0,(H-P_z-linear)-3.0) is superimposed. Finally: P_z=P_z-linear+ΔP_1. A visual verification example is shown below:

[0057] For the equal-diameter mid-term (linear tracking), given: real-time liquid level height H = 20cm, the corresponding interval H_1 = 15cm (P_z0-1 = 13.0cm), H_2 = 25cm (P_z0-2 = 23.5cm), the calculation process is as follows: linear interpolation: P_z-linear = 13.0 + (23.5-13.0) / (25-15) × (20-15) = 13.0 + 1.05 × 5 = 18.25cm, constraint correction: H - P_z-linear = 20 - 18.25 = 1.75cm ≤ 3cm, ΔP_1 = 0, final: P_z = 18.25cm (tracking error 0.25cm, meeting the requirement of ≤ ±0.5cm).

[0058] Nonlinear tracking interval (H < 10cm: late stage of constant diameter), applicable interval: H_1≤H≤H_2 (e.g., 2cm≤H≤10cm). Due to the small melt volume and thermal field shift, quadratic interpolation is used (adapted to changes in flow pattern). First, a quadratic curve is fitted based on the reference point, and then P_z is calculated. That is, first, the quadratic curve equation P_z0-nonlinear=aH²+bH+c is fitted (coefficients a, b, and c are fitted through 3 adjacent reference points in the mapping table and fixed in the module): Example fitting (based on data in sequence 3-5-6): a=0.025, b=-0.35, c=1.5. Then, the reference value is calculated by substituting it into the real-time H: P_z0-nonlinear=0.025H²-0.35H+1.5, and the thermal field symmetry correction ΔP_2=0.05×(10-H) is superimposed (to compensate for the thermal field shift in the later stage). Finally: P_z=P_z0-nonlinear+ΔP_2.

[0059] It should be noted that the interpolation calculation delay is ≤50ms (adapted to real-time PLC calculation, implemented using the FC function block of S7-1500, and the SCALE_X instruction is called to assist in interpolation); linear tracking is used by default (adapted to 80% growth scenario), and nonlinear tracking is automatically switched in the later stage of equal diameter (H<10cm), with no abrupt change in switching (difference of P_z before and after ≤0.3cm); the liquid level measurement accuracy is ≤±0.1mm, the interpolation calculation error is ≤±0.2cm, and the total error after adding correction terms is ≤±0.5cm (compliant with patent requirements); if H exceeds the range of the reference table (e.g., H>30cm or H<2cm), the boundary value is automatically used (P_z=28.0cm when H>30cm, P_z=0.8cm when H<2cm) to avoid calculation overflow.

[0060] Furthermore, step S30 further includes converting the target position P_z into a current adjustment value I_1 for the upper coil of the hook-shaped magnetic field generator, and converting the target value B_target into a current adjustment value I_2 for the lower coil of the hook-shaped magnetic field generator. Based on I_1 and I_2, the currents of the upper and lower coils of the hook-shaped magnetic field generator are adjusted to adjust the spatial position and intensity of the magnetic field.

[0061] I_2=B_target / (2.4×E×N_1), I_1=1.2×I_2, N_1 is the number of turns of the coil on the hook-shaped magnetic field generator. The magnetic circuit calibration coefficient E is obtained by gradient sampling test covering the full range of 0-5000Gs, repeated measurement 3 times and average value, and fitted by least squares method to obtain E=0.83Gs / (A·turn) (example value).

[0062] Based on the coil characteristic parameters of the magnetic field generator, P_z and B_target are converted into specific current adjustment values ​​I_1 and I_2 of the coil on the hook-shaped magnetic field generator, as shown in Table 2:

[0063] Table 2 Current Adjustment Values

[0064] Furthermore, in step S30, calculating the target value B_target includes:

[0065] S31. Calculate the baseline intensity B_base based on the current thermal field state data W;

[0066] S32. Select the stage correction coefficient k based on the baseline strength B_base and the current growth stage, and calculate the intermediate strength B_mid, where B_mid = B_base × k;

[0067] S33. Calculate the dynamic correction ΔB_T based on the temperature change rate in the current thermal field state data W;

[0068] S34. Calculate the target value B_target based on the intermediate intensity B_mid and the dynamic correction amount ΔB_T, where B_target = B_mid + ΔB_T, ΔB_T = 30 × (T_current - T_previous cycle) / T_previous cycle.

[0069] Furthermore, in step S31, the base strength B_base is calculated based on the critical point temperature T of the crucible or melt and the critical point temperature threshold X of the crucible or melt.

[0070] When T ≥ X ℃, B_base = 0.8T - 1100;

[0071] When T < X℃, B_base = 1.2T - 1600 + 50 × (V_0 - V) / V_0;

[0072] In the formula, V_0 is the initial melt volume, and X is 1450.

[0073] Furthermore, in step S32, the current growth stage includes the crystal-initiating stage, the mid-stage of constant diameter growth, and the late-stage of constant diameter growth. The k value for the crystal-initiating stage is 0.8, for the mid-stage of constant diameter growth is 1.0, and for the late-stage of constant diameter growth is 1.2. Examples of magnetic field adaptation and parameter calculation for different growth stages are as follows:

[0074] Initial stage of crystal development (time t1)

[0075] The liquid level is high (H≥25cm), the high-temperature melt zone has a large volume (V≈100L), and the oxygen dissolution rate is high. The system controls the zero-position plane of the magnetic field to be close to the liquid surface (≤3cm away from the liquid surface). At this time, T=1480℃, B_base=0.8×1480-1100=74Gs, the stage correction coefficient k=0.8, B_mid=74×0.8=60Gs, the temperature change rate is 0.5℃ / s, ΔB_T=+10Gs, and finally B_target=70Gs. The corresponding currents are I_2=70 / (2.4×0.83×1000)≈0.035A and I_1=1.2×0.035≈0.042A. Applying this magnetic field allows for moderate melt convection, promoting the uniform transport of oxygen from the crucible wall to the solid-liquid interface.

[0076] Mid-term of isodiameter (time t2)

[0077] As the liquid level drops to a moderate height (10cm≤H<25cm), the high-temperature zone becomes relatively concentrated, and the oxygen source weakens. The system controls the magnetic field zero-level to move downward to track the liquid level. At this point, T=1455℃, V≈50L, B_base=0.8×1455-1100=64Gs, the stage correction coefficient k=1.0, B_mid=64×1.0=64Gs, the temperature change rate is 0.4℃ / s, ΔB_T=+8Gs, and finally B_target=72Gs. The corresponding currents are I_2=72 / (2.4×0.83×1000)≈0.036A and I_1=1.2×0.036≈0.043A, further suppressing convection and stabilizing the axial condensation process of oxygen.

[0078] Late stage of isodiameter (time t3)

[0079] With a low liquid level (H < 10 cm), the melt volume decreases, and the thermal field morphology changes again. The system continues to adjust the magnetic field position to move down synchronously with the liquid level. At this point, T = 1440℃, V ≈ 35L, B_base = 1.2 × 1440 - 1600 + 50 × (100 - 35) / 100 = 160.5 Gs, the stage correction coefficient k = 1.2, B_mid = 160.5 × 1.2 ≈ 192.6 Gs, the temperature change rate is 0.4℃ / s, ΔB_T = +8 Gs, and finally B_target = 200.6 Gs. The corresponding currents are I_2 = 200.6 / (2.4 × 0.83 × 1000) ≈ 0.101 A and I_1 = 1.2 × 0.101 ≈ 0.121 A, ensuring a smooth transition of oxygen distribution along the entire crystal axis.

[0080] Please refer to Figure 2 This application also provides a magnetic field dynamic adaptation system for improving the uniformity of oxygen content in single crystals, applied to the above-mentioned magnetic field dynamic adaptation method for improving the uniformity of oxygen content in single crystals, including a data sensing module 100, a magnetic field execution module 200 and a control module 300, wherein the data sensing module 100 and the magnetic field execution module 200 are both communicatively connected to the control module 300.

[0081] The data sensing module 100 is used to collect dynamic process parameters in real time during crystal growth, including at least a liquid surface position sensing unit 110 for monitoring the liquid surface height H of the melt and a temperature zone monitoring unit 120 for acquiring thermal field state data W.

[0082] The magnetic field execution module 200 includes at least one magnetic field generator, which is capable of generating a magnetic field with adjustable spatial position and intensity.

[0083] The control module 300 is configured to: receive real-time process parameters from the data sensing module 100, calculate and output the target position P_z of the magnetic field zero plane based on the liquid level height H of the melt, convert the target position P_z into a drive command, and send it to the magnetic field execution module 200 to drive the magnetic field execution module 200 to adjust the spatial position of the magnetic field zero plane of the magnetic field generator; calculate and output the target value B_target of the magnetic field strength based on the thermal field state data W, convert the target strength B_target into a drive command, and send it to the magnetic field execution module 200 to drive the magnetic field execution module 200 to synchronously adjust the magnetic field strength of the magnetic field generator.

[0084] In this embodiment, the liquid level position sensing unit 110 can employ a laser ranging device, a visual recognition system, or a model calculation module based on the lifting length and melt consumption to monitor or calculate the liquid level height at the melt-crystal solid-liquid interface in real time, with a measurement accuracy of ≤±0.1mm. The laser ranging device is installed outside the observation window on the top of the single crystal furnace, vertically aligned with the melt surface, or the visual recognition system acquires liquid level images through the observation window and performs algorithm analysis. The model calculation module communicates with the encoder of the single crystal furnace lifting mechanism to infer melt consumption and liquid level height through the lifting length. The temperature zone monitoring unit 120 can be a thermocouple or an infrared temperature measuring device, distributed on the crucible wall, heater, and insulation cylinder, used to acquire the temperature T, preset thermal field temperature zone model parameters, or power data of multiple key points of the crucible and melt, with a temperature measurement accuracy of ≤±1℃. For example, 3-5 thermocouples are evenly distributed at different heights on the crucible wall, and 1-2 infrared temperature measuring devices are aligned with the solid-liquid interface area. All temperature measuring elements are connected to the control module 300 through a data acquisition card.

[0085] The magnetic field dynamic adaptation system for improving the uniformity of oxygen content in single crystals provided in this application uses a liquid surface position sensing unit 110 to sense the liquid surface height H of the melt in real time and a temperature zone monitoring unit 120 to acquire thermal field state data W. Based on the liquid surface height H and thermal field state data W, the target position P_z and target value B_target are calculated. The control module 300 converts the target position P_z into a driving command and sends it to the magnetic field execution module 200 to drive the magnetic field execution module 200 to adjust the spatial position of the magnetic field zero plane of the magnetic field generator. The target intensity B_target is also converted into a driving command and sent to the magnetic field execution module 200 to drive the magnetic field execution module 200 to synchronously adjust the magnetic field intensity of the magnetic field generator. By dynamically adjusting the spatial position of the magnetic field zero plane of the magnetic field generator, uniform oxygen transport is promoted, the radial oxygen concentration CV value is reduced, and the nonlinear fluctuations in the oxygen segregation process are avoided by dynamically adjusting the magnetic field intensity of the magnetic field generator. The linearity of the axial oxygen distribution curve is improved, thereby improving the uniformity of the overall oxygen content of the single crystal.

[0086] In one specific embodiment, the magnetic field generator is a hook-shaped magnetic field generator composed of two sets of independent coils. The two sets of independent coils of the hook-shaped magnetic field generator are symmetrically installed on the outside of the furnace chamber of the single crystal furnace. The spatial position of the zero magnetic field plane is adjusted by adjusting the current ratio of the upper coil and the lower coil, and the magnetic field strength is adjusted by adjusting the total current of the upper coil and the lower coil.

[0087] In this embodiment, the turns ratio of the upper independent coil (N_1) to the lower independent coil (N_2) of the hook-shaped magnetic field generator is 1:1.2, where N_1 = 1000 turns and N_2 = 1200 turns. The hook-shaped magnetic field generator is equipped with a high-precision adjustable DC power supply (output range 0-50A, accuracy ±0.1A), possessing the ability to independently control the spatial magnetic field distribution: by adjusting the relative current ratio of the upper and lower coils (e.g., I_1:I_2 = 6:5), the spatial position of the magnetic field zero-position plane (NS interface) in the melt region is changed, with the adjustment range covering the entire melt height range; by adjusting the total current of the upper and lower coils, the magnetic field strength gradient near the magnetic field zero-position plane is changed, with the magnetic field strength adjustment range continuously adjustable from 0-5000Gs and an adjustment accuracy ≤±10Gs.

[0088] Furthermore, the control module 300 adopts an industrial PLC (such as S7-1500) or a dedicated industrial computer with a built-in real-time operating system. It establishes a connection with the data sensing module 100 and the magnetic field execution module 200 through the Profinet or Modbus communication protocol. The data transmission delay is ≤100ms. It has a built-in "process status - magnetic field parameters - oxygen distribution" mapping relationship model or adaptive control algorithm. The specific configuration is as follows: Signal reception: Real-time reception of liquid level position signal and temperature zone status signal from the data sensing module 100, with a data sampling frequency ≥10Hz; Based on the received real-time signal, magnetic field adjustment command is generated by table lookup, interpolation or real-time model calculation, and sent to the magnetic field execution module 200 to drive it to synchronously adjust the magnetic field position and magnetic field strength of the magnetic field generator.

[0089] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and such modifications or substitutions should all be covered within the scope of protection of the present invention.

Claims

1. A method for dynamically adapting a magnetic field to improve the uniformity of oxygen content in a single crystal, characterized in that, Includes the following steps: S10. Real-time acquisition of melt surface height H and thermal field state data W during crystal growth; S20. Based on the liquid level height H, calculate the target position P_z of the required magnetic field zero plane; based on the thermal field state data W, calculate the target value B_target of the required magnetic field strength. S30. Based on the target position P_z, adjust the spatial position of the magnetic field, and based on the target value B_target, simultaneously adjust the strength of the magnetic field; S40. Repeat steps S10 to S30 until crystal growth is complete.

2. The method for dynamic magnetic field adaptation for improving the uniformity of oxygen content in single crystals according to claim 1, characterized in that, In step S10, the thermal field state data W includes the key point temperature T of the crucible or melt, the volume V of the high-temperature melt zone, and the solid-liquid interface temperature gradient G.

3. The method for dynamic magnetic field adaptation for improving the uniformity of oxygen content in single crystals according to claim 1, characterized in that, In step S20, the target position P_z is calculated by looking up the preset mapping relationship between the liquid level height and the zero magnetic surface position and combining it with interpolation.

4. The method for dynamic magnetic field adaptation to improve the uniformity of oxygen content in single crystals according to claim 1, characterized in that, Step S30 further includes converting the target position P_z into a current adjustment value I_1 for the upper coil of the hook-shaped magnetic field generator, and converting the target value B_target into a current adjustment value I_2 for the lower coil of the hook-shaped magnetic field generator. Based on I_1 and I_2, the currents of the upper and lower coils of the hook-shaped magnetic field generator are adjusted to adjust the spatial position and intensity of the magnetic field. I_2=B_target / (2.4×E×N_1), I_1=1.2×I_2, N_1 is the number of turns of the coil on the hook-shaped magnetic field generator, and E is the magnetic circuit calibration coefficient.

5. The method for dynamic magnetic field adaptation for improving the uniformity of oxygen content in single crystals according to claim 2, characterized in that, In step S30, calculating the target value B_target includes: S31. Calculate the baseline intensity B_base based on the current thermal field state data W; S32. Select the stage correction coefficient k based on the baseline strength B_base and the current growth stage, and calculate the intermediate strength B_mid, where B_mid = B_base × k; S33. Calculate the dynamic correction ΔB_T based on the temperature change rate in the current thermal field state data W; S34. Calculate the target value B_target based on the intermediate intensity B_mid and the dynamic correction amount ΔB_T, where B_target = B_mid + ΔB_T.

6. The method for dynamic magnetic field adaptation for improving the uniformity of oxygen content in single crystals according to claim 5, characterized in that, In step S31, the base strength B_base is calculated based on the critical point temperature T of the crucible or melt and the critical point temperature threshold X of the crucible or melt. When T ≥ X ℃, B_base = 0.8T - 1100; When T < X℃, B_base = 1.2T - 1600 + 50 × (V_0 - V) / V_0; In the formula, V_0 is the initial melt volume, and X is 1450.

7. The method for dynamic magnetic field adaptation for improving the uniformity of oxygen content in single crystals according to claim 5, characterized in that, In step S32, the current growth stage includes the crystal-initiating stage, the mid-stage of equal diameter growth, and the late-stage of equal diameter growth. The k value for the crystal-initiating stage is 0.8, the k value for the mid-stage of equal diameter growth is 1.0, and the k value for the late-stage of equal diameter growth is 1.

2.

8. A magnetic field dynamic adaptation system for improving the uniformity of oxygen content in single crystals, characterized in that, The magnetic field dynamic adaptation method for improving the uniformity of oxygen content in single crystals, as described in any one of claims 1 to 7, includes a data sensing module, a magnetic field execution module, and a control module. Both the data sensing module and the magnetic field execution module are communicatively connected to the control module. The data sensing module is used to collect dynamic process parameters during crystal growth in real time, and includes at least a liquid surface position sensing unit for monitoring the liquid surface height H of the melt and a temperature zone monitoring unit for acquiring thermal field state data W. The magnetic field execution module includes at least one magnetic field generator capable of generating a magnetic field with adjustable spatial position and intensity at the zero-position surface. The control module is configured to: receive real-time process parameters from the data sensing module, calculate and output the target position P_z of the magnetic field zero plane based on the liquid level height H of the melt, convert the target position P_z into a drive command, and send it to the magnetic field execution module to drive the magnetic field execution module to adjust the spatial position of the magnetic field zero plane of the magnetic field generator; calculate and output the target value B_target of the magnetic field strength based on the thermal field state data W, convert the target strength B_target into a drive command, and send it to the magnetic field execution module to drive the magnetic field execution module to synchronously adjust the magnetic field strength of the magnetic field generator.

9. The magnetic field dynamic adaptation system for improving the uniformity of oxygen content in single crystals according to claim 8, characterized in that, The magnetic field generator is a hook-shaped magnetic field generator composed of two sets of independent coils. The two sets of independent coils of the hook-shaped magnetic field generator are symmetrically installed on the outside of the furnace chamber of the single crystal furnace. The spatial position of the zero-position plane of the magnetic field is adjusted by adjusting the current ratio of the upper coil and the lower coil, and the magnetic field strength is adjusted by adjusting the total current of the upper coil and the lower coil.

10. The magnetic field dynamic adaptation system for improving the uniformity of oxygen content in single crystals according to claim 8, characterized in that, The control module adopts an industrial PLC or a dedicated industrial control computer with a built-in real-time operating system. It establishes a connection with the data sensing module and the magnetic field execution module through the Profinet or Modbus communication protocol. The data transmission delay is ≤100ms. It has a built-in "process status - magnetic field parameters - oxygen distribution" mapping relationship model or adaptive control algorithm.

Images