Crystal pulling system for pulling heavily boron-doped silicon single crystal and control method of crystal pulling system

Abstract

The invention discloses a crystal pulling system for pulling a heavily boron-doped silicon single crystal and a control method thereof, and the crystal pulling system for pulling the heavily boron-doped silicon single crystal comprises a quartz crucible, a boron particle feeder, a heating device, a dual laser-induced breakdown spectroscopy measurement device and a model prediction controller, wherein the boron particle feeder is positioned above the liquid level of melt in the quartz crucible and is used for supplementing boron particles to the surface of the melt; the heating device comprises a temperature adjusting part, and the temperature adjusting part is located above the liquid level of the melt and used for adjusting the temperature gradient of the melt located in the quartz crucible near a solid-liquid interface; the dual-laser-induced breakdown spectroscopy measurement device comprises a first detection probe and a second detection probe, the first detection probe is used for detecting the melt boron concentration in a nearby area where a crystal bar is in contact with the melt liquid level, and the second detection probe is used for detecting the carrier concentration of the grown crystal bar.

Description

Technical Field

[0001] This invention relates to the field of semiconductor technology, and in particular to a crystal pulling system and its control method for pulling heavily boron-doped silicon single crystals. Background Technology

[0002] Heavy boron-doped silicon single crystals (resistivity ρ ≤ 10 mΩ·cm) are core substrate materials for semiconductor devices such as power devices, IGBTs, and superjunction MOSFETs. As semiconductor devices develop towards higher power and higher integration, higher requirements are placed on the axial uniformity of their resistivity.

[0003] In the Czochralski method for growing heavily boron-doped silicon single crystals, the effective segregation coefficient k of boron in silicon is... eff ≈ 0.8, and k eff It is highly sensitive to the crystal pulling rate and the temperature gradient near the solid-liquid interface. Related technologies use a fixed crystal pulling rate in conjunction with subsequent slicing and grading, and the melt temperature field fluctuates greatly during the process, resulting in poor resistivity uniformity of the crystal rod along the axial direction. This not only leads to an increased breakage rate during subsequent wafer processing, but also affects the performance of semiconductor devices. Summary of the Invention

[0004] This invention discloses a crystal pulling system and its control method for pulling heavily boron-doped silicon single crystals, in order to solve the problem of poor resistivity uniformity of the crystal rod in the axial direction during the crystal pulling process in related technologies.

[0005] To solve the above-mentioned technical problems, the present invention is implemented as follows:

[0006] This application discloses a crystal pulling system for pulling heavily boron-doped silicon single crystals. The disclosed crystal pulling system for pulling heavily boron-doped silicon single crystals includes a quartz crucible, a boron particle feeder, a heating device, a dual-laser induced breakdown spectroscopy measurement device, and a model prediction controller, wherein:

[0007] The boron particle feeder is located above the surface of the melt in the quartz crucible and is used to replenish boron particles to the surface of the melt.

[0008] The heating device includes a temperature regulating element located above the surface of the melt, which is used to regulate the temperature gradient of the melt in the quartz crucible near the solid-liquid interface.

[0009] The dual-laser induced breakdown spectroscopy measurement device includes a first detection probe and a second detection probe. The first detection probe is used to detect the melt boron concentration in the vicinity of the contact between the crystal rod and the melt surface, and the second detection probe is used to detect the carrier concentration of the grown crystal rod.

[0010] The boron particle feeder, the temperature regulator, the first detection probe, and the second detection probe are all connected to the model prediction controller. The model prediction controller controls the axial movement speed of the crystal rod, the feeding rate of the boron particle feeder, and the output power of the temperature regulator based on the molten boron concentration detected by the first detection probe and the carrier concentration detected by the second detection probe, so that the deviation between the axial resistivity of the crystal rod and the preset axial resistivity target curve is within a preset range.

[0011] Optionally, the boron particle feeder includes an annular body and a plurality of nozzles. The annular body has an annular inner cavity, and the plurality of nozzles are disposed on the annular body and distributed at intervals along the annular body. The nozzles communicate with the annular inner cavity. The annular body is used to surround the outside of the crystal rod, and the nozzles extend toward the side facing the melt surface.

[0012] Optionally, the nozzle is tilted toward the center of the annular body, and the angle between the central axis of the nozzle and the plane containing the melt surface is greater than or equal to 60° and less than or equal to 80°.

[0013] Optionally, the heating device further includes a main heating element and an auxiliary heating element. The main heating element surrounds the quartz crucible and is opposite to the middle region of the quartz crucible. The auxiliary heating element surrounds the quartz crucible and is opposite to the bottom region of the quartz crucible.

[0014] Optionally, the crystal pulling system for pulling heavily boron-doped silicon single crystals further includes a magnetic field adjustment device located outside the quartz crucible, which is used to apply a transverse magnetic field perpendicular to the surface of the melt to the melt located inside the quartz crucible.

[0015] Optionally, the magnetic field strength of the transverse magnetic field generated by the magnetic field adjustment device is between 0.1T and 0.5T.

[0016] Optionally, the crystal pulling system for pulling heavily boron-doped silicon single crystals further includes a furnace body and a cooling device, wherein the quartz crucible, the boron particle feeder, the heating device and the cooling device are all located inside the furnace body.

[0017] This application also discloses a control method for a crystal pulling system for pulling heavily boron-doped silicon single crystals, wherein the crystal pulling system for pulling heavily boron-doped silicon single crystals is the crystal pulling system for pulling heavily boron-doped silicon single crystals described in the first aspect, and the control method includes:

[0018] During the constant diameter growth stage, the first detection probe is controlled to detect the melt boron concentration in the vicinity of the area where the crystal rod contacts the melt surface, and the second detection probe is controlled to detect the carrier concentration of the grown crystal rod.

[0019] The axial movement rate of the crystal rod, the feeding rate of the boron particle feeder, and the output power of the temperature regulating component are controlled based on the molten boron concentration detected by the first detection probe and the carrier concentration detected by the second detection probe, so that the deviation between the axial resistivity of the crystal rod and the target curve is within a preset range.

[0020] Optionally, the built-in process model of the model predictive controller adopts a one-dimensional diffusion-segregation coupling equation:

[0021] ;

[0022] Wherein: CL is the melt boron concentration in the vicinity of the contact area between the crystal rod and the melt surface; t is time; z is the position of the crystal rod in the axial direction from the head of the crystal rod; v is the moving speed of the crystal rod along the axial direction; Deff is the effective diffusion coefficient; Keff is the effective segregation coefficient of boron, which varies with the temperature gradient G(t); R is the feeding rate of the boron particle feeder; and VL is the melt volume.

[0023] Optionally, the preset axial resistivity target curve ρtarget(z) is:

[0024] ;

[0025] Where: ρ0 is the target resistivity at the head of the crystal rod, and z is the position of the crystal rod in the axial direction from the head of the crystal rod. This represents the coefficient of axial resistivity variation.

[0026] The technical solution adopted in this invention can achieve the following technical effects:

[0027] The crystal pulling system for pulling heavily boron-doped silicon single crystals disclosed in this application realizes real-time online monitoring of melt boron concentration and crystal rod carrier concentration through a dual laser-induced breakdown spectroscopy measurement device, providing accurate input signals for closed-loop control; the model predictive controller adjusts the crystal rod's axial movement speed, the boron particle feeder's feeding rate, and the temperature gradient near the solid-liquid interface based on multivariable inputs to achieve precise control of the axial resistivity distribution, thereby improving the uniformity of the crystal rod's axial resistivity. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the crystal pulling system for pulling heavily boron-doped silicon single crystals disclosed in an embodiment of the present invention;

[0029] Figure 2 This is a schematic diagram of the boron particle feeder disclosed in an embodiment of the present invention;

[0030] Figure 3 This is a flowchart of a control method for a crystal pulling system for pulling heavily boron-doped silicon single crystals, as disclosed in an embodiment of the present invention.

[0031] Explanation of reference numerals in the attached figures:

[0032] 100-quartz crucible,

[0033] 200 - Boron particle feeder, 210 - Annular body, 220 - Nozzle

[0034] 310 - Temperature regulating components

[0035] 400-Dual Laser-Induced Breakdown Spectrometry Measurement Device

[0036] 500-Magnetic field adjustment device

[0037] 600-furnace body,

[0038] 700 - Cooling device; 800 - Model predictive controller. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0040] Before describing the embodiments of this application, let's explain the main reasons for the poor axial resistivity uniformity of the crystal rod during the existing heavy boron-doped silicon single crystal pulling process. First, the effective segregation coefficient k of boron in silicon... eff The coefficient is approximately 0.8, and this coefficient is highly sensitive to the crystal pulling rate and the temperature gradient near the solid-liquid interface; even minor process fluctuations can alter the segregation behavior. Secondly, existing technologies lack real-time online monitoring methods for the melt boron concentration (the concentration of boron in the melt) and the charge carrier concentration of the crystal rod, making it impossible to promptly detect and compensate for boron losses caused by segregation and volatilization. Furthermore, traditional processes typically employ a fixed pulling speed combined with post-processing slicing and grading, lacking the ability to coordinate and adjust the pulling speed, boron replenishment rate, and temperature gradient. These factors combined result in a gradient distribution of the axial resistivity of the crystal rod.

[0041] Axial non-uniformity of resistivity not only affects the consistency of electrical properties but also leads to the degradation of mechanical properties. Since resistivity is directly related to doping concentration, and changes in doping concentration cause slight variations in the lattice constant, lattice mismatch and differences in thermal expansion coefficients occur at different axial positions of the crystal rod. These differences can easily induce microcrack propagation during subsequent slicing, leading to an increased breakage rate. To address the problem of poor axial resistivity uniformity in the crystal rod, this application discloses a crystal pulling system and its control method for pulling heavily boron-doped silicon single crystals. The technical solutions disclosed in the various embodiments of this invention are described in detail below with reference to the accompanying drawings.

[0042] Please refer to Figures 1 to 2 This invention discloses a crystal pulling system for pulling heavily boron-doped silicon single crystals. The disclosed crystal pulling system for pulling heavily boron-doped silicon single crystals includes a quartz crucible 100, a boron particle feeder 200, a heating device, a dual laser-induced breakdown spectroscopy measurement device 400, and a model prediction controller 800.

[0043] The quartz crucible 100 is used to contain the melt (mainly silicon melt) for growing silicon single crystal rods. The boron particle feeder 200 is located above the melt surface within the quartz crucible 100 and is used to replenish boron particles to the melt surface. By supplementing boron online, the boron concentration in the melt can be adjusted in real time to compensate for boron loss caused by segregation and volatilization.

[0044] The heating device includes a temperature regulating element 310, which is located above the melt surface. The temperature regulating element 310 is used to adjust the temperature gradient of the melt near the solid-liquid interface within the quartz crucible 100. By adjusting the temperature gradient of the melt near the solid-liquid interface within the quartz crucible 100, the effective segregation coefficient k of boron can be finely adjusted. eff This affects the amount of boron incorporated into the crystal. The temperature gradient can be adjusted from 1.5 K / mm to 2.5 K / mm, and the response time can be ≤20 s.

[0045] Specifically, on the melt side of the crystal growth interface, there exists a nearly static diffusion boundary layer. Boron atoms must pass through this diffusion boundary layer to reach the growth interface and be absorbed by the crystal. Increasing the temperature gradient of the melt near the solid-liquid interface enhances melt convection or alters the convection pattern, which thins the diffusion boundary layer. A thinner diffusion boundary layer means that boron atoms can diffuse to the growth interface faster and more smoothly, resulting in a boron concentration at the interface that is closer to the bulk melt concentration, thus increasing the "effective segregation coefficient." Conversely, decreasing the temperature gradient of the melt near the solid-liquid interface weakens convection and thickens the diffusion boundary layer. A thick diffusion boundary layer acts as a "barrier" for boron atoms to reach the interface, causing boron consumption at the interface without timely replenishment, thereby decreasing the "effective segregation coefficient."

[0046] The dual laser-induced breakdown spectroscopy measurement device 400 includes a first detection probe and a second detection probe. The first detection probe is used to detect the boron concentration in the melt near the contact area between the crystal rod and the molten metal surface, and the second detection probe is used to detect the carrier concentration in the grown crystal rod. Under heavy boron doping conditions, the carrier concentration is approximately equal to the boron concentration in the crystal rod; therefore, the second detection probe essentially monitors the actual amount of boron incorporated into the crystal rod. It should be noted that laser-induced breakdown spectroscopy measurement devices are existing technology, and the structure and principle of such devices will not be described in detail in this application.

[0047] Let me further explain the relationship between the axial movement rate of the crystal rod and the effective segregation coefficient k. eff According to the Burton-Prim-Slichter (BPS) boundary layer theory, the effective segregation coefficient K... eff The relationship between the crystal rod's axial movement speed v and the speed v is shown in Formula 1:

[0048] (Formula 1)

[0049] k eff δ is the effective segregation coefficient of boron; k0 is the equilibrium segregation coefficient of boron in silicon (approximately 0.8); δ is the diffusion boundary layer thickness; and D is the diffusion coefficient of boron in the melt.

[0050] As the axial movement speed v of the crystal rod increases, the effective segregation coefficient k eff As the axial movement speed v of the crystal rod decreases, the effective segregation coefficient k increases. eff This decreases, thus affecting the amount of boron incorporated into the crystal.

[0051] The boron particle feeder 200, temperature regulator 310, first detection probe, and second detection probe are all connected to the model prediction controller 800. The model prediction controller 800 controls the axial movement speed of the crystal rod (i.e., the pulling speed of the crystal rod), the feeding rate of the boron particle feeder 200, and the output power of the temperature regulator 310 based on the melt boron concentration detected by the first detection probe and the charge carrier concentration detected by the second detection probe, so that the deviation between the axial resistivity of the crystal rod and the preset axial resistivity target curve is within a preset range.

[0052] Specifically, the model predictive controller 800 can employ a model predictive control algorithm to perform rolling optimization with a preset time period (e.g., 30 seconds, or other possible periods). Within each control cycle, it performs the following operations: receiving the melt boron concentration detected in real-time by the first detection probe and the carrier concentration detected in real-time by the second detection probe; based on a built-in process model, predicting the change in the axial resistivity of the crystal rod over a future period; and, under constraints of adjusting the crystal rod's axial movement rate, feeding rate, and the temperature gradient of the melt near the solid-liquid interface, solving for the optimal control sequence that minimizes the deviation between the predicted resistivity and the preset axial resistivity target curve; and outputting the current control quantity in the optimal control sequence as an instruction to the crystal pulling machine, the boron particle feeder 200, and the temperature regulator 310, respectively, to synchronously adjust the crystal rod's axial movement rate, feeding rate, and the temperature gradient of the melt near the solid-liquid interface.

[0053] The crystal pulling system for pulling heavily boron-doped silicon single crystals disclosed in this application realizes real-time online monitoring of melt boron concentration and crystal rod carrier concentration through a dual laser-induced breakdown spectroscopy measurement device 400, providing accurate input signals for closed-loop control; the model predictive controller 800 adjusts the axial moving speed of the crystal rod, the feeding rate of the boron particle feeder 200, and the temperature gradient near the solid-liquid interface based on multivariable inputs, thereby achieving precise control of the axial resistivity distribution and improving the uniformity of the axial resistivity of the crystal rod.

[0054] Optionally, the boron particle feeder 200 includes an annular body portion 210 and a plurality of nozzles 220. The annular body portion 210 may have an annular inner cavity. The plurality of nozzles 220 may be disposed on the annular body portion 210 and may be spaced apart along the annular body portion 210. The nozzles 220 may communicate with the annular inner cavity. The annular body portion 210 may be used to surround the outside of the crystal rod. The nozzles 220 may extend toward one side of the melt surface.

[0055] Preferably, the nozzle 220 can be tilted toward the center of the annular body 210, and the angle between the central axis of the nozzle 220 and the plane where the melt surface is located can be greater than or equal to 60° and less than or equal to 80°.

[0056] The crystal pulling system for pulling heavily boron-doped silicon single crystals disclosed in this application improves the uniformity of boron particle dissolution by setting the nozzle 220 to be tilted toward the center of the annular body 210, which facilitates the falling of boron particles into the central region of the melt at a suitable speed and angle.

[0057] Specifically, the feeding rate of the boron particle feeder 200 can be adjusted from 0 to 30 mg / min, thus meeting the boron supplementation requirements at different growth stages. The diameter of the boron particles ejected by the nozzle 220 can be between 50 μm and 200 μm. This particle diameter range ensures good transportability of the particles in the carrier gas (boron particles need to be transported by a carrier gas) and rapid dissolution in the melt. The carrier gas can be argon, and the flow rate can be between 0.5 slm and 2 slm. The boron particle feeder 200 can also have a carrier gas preheating function, with a preheating temperature between 200℃ and 400℃, thus avoiding interference from cold carrier gas on the melt temperature field.

[0058] The crystal pulling system for pulling heavily boron-doped silicon single crystals disclosed in this application can achieve 360° uniform boron supplementation by setting the boron particle feeder 200 to a structure including an annular main body 210 and multiple nozzles 220, thus avoiding excessively high local boron concentration.

[0059] Optionally, the heating device may also include a main heating element and an auxiliary heating element. The main heating element may surround the quartz crucible 100 and may be opposite to the middle region of the quartz crucible 100. The auxiliary heating element may surround the quartz crucible 100 and may be opposite to the bottom region of the quartz crucible 100.

[0060] Specifically, the temperature regulating element 310, the main heating element, and the auxiliary heating element can be independently heated heating coils. The main heating element can be the main body for heating the melt, and the auxiliary heating element can be used to adjust the temperature difference between the melt located in the bottom region of the quartz crucible 100 and the melt located in the middle region of the quartz crucible 100, thereby reducing convection inside the melt (different from the temperature regulating element 310 adjusting the temperature gradient of the diffusion boundary layer).

[0061] The crystal pulling system for pulling heavily boron-doped silicon single crystals disclosed in this application provides a flexible temperature gradient adjustment capability and fast response speed by setting up a three-zone independent heating structure consisting of a temperature regulating component 310, a main heating component, and an auxiliary heating component. It can also respond to control commands in real time and realize dynamic fine-tuning of the effective segregation coefficient, adding an important adjustment dimension to resistivity control.

[0062] The axial spacing between the temperature regulating element 310 and the main heating element, as well as between the main heating element and the auxiliary heating element, can be between 80 mm and 120 mm, and the total power of the heating device can be ≤60 kW.

[0063] Optionally, the crystal pulling system for pulling heavily boron-doped silicon single crystals may further include a magnetic field adjustment device 500. The magnetic field adjustment device 500 may be disposed outside the quartz crucible 100, and may be used to apply a transverse magnetic field perpendicular to the melt surface to the melt located within the quartz crucible 100. Specifically, the magnetic field strength of the transverse magnetic field generated by the magnetic field adjustment device 500 may be between 0.1T and 0.5T.

[0064] It should be noted that when the transverse magnetic field generated by the magnetic field regulating device 500 acts on the conductive silicon melt, it will generate a Lorentz force to suppress melt convection, especially to suppress turbulence and irregular flow in the melt. This flow suppression effect is beneficial to: stabilizing the solid-liquid interface shape and reducing growth streaks; improving the uniform distribution of boron particles after dissolution and avoiding local concentration fluctuations; and reducing temperature fluctuations, which is beneficial to maintaining stable segregation conditions.

[0065] Optionally, the crystal pulling system for pulling heavily boron-doped silicon single crystals may further include a furnace body 600 and a cooling device 700. The furnace body 600 is a sealed container that provides a controllable atmosphere for crystal growth. The quartz crucible 100, boron particle feeder 200, heating device, and cooling device 700 can all be located within the furnace body 600. The cooling device 700 may include an annular cooling body, which may have an annular water-cooling cavity. The cooling device 700 may be located at the top of the furnace body. The cooling device 700 can be used to cool the grown crystal rod, control the temperature gradient of the crystal rod, and avoid excessive thermal stress that could lead to dislocation formation, thereby improving the quality of the crystal rod.

[0066] Please refer to Figure 3 This application also discloses a control method for a crystal pulling system for pulling heavily boron-doped silicon single crystals. The crystal pulling system for pulling heavily boron-doped silicon single crystals is the crystal pulling system for pulling heavily boron-doped silicon single crystals disclosed in the above embodiments. The disclosed control method includes:

[0067] S101, during the constant diameter growth stage, the first detection probe is controlled to detect the melt boron concentration in the vicinity of the crystal rod in contact with the melt surface, and the second detection probe is controlled to detect the carrier concentration of the grown crystal rod.

[0068] S102, based on the melt boron concentration detected by the first detection probe and the carrier concentration detected by the second detection probe, control the axial movement speed of the crystal rod, the feeding rate of the boron particle feeder 200 and the output power of the temperature regulating component 310 so that the deviation between the axial resistivity of the crystal rod and the target curve is within a preset range.

[0069] Specifically, the model predictive controller 800 can employ a model predictive control algorithm to perform rolling optimization with a preset time period (e.g., 30 seconds, or other possible periods). Within each control cycle, it performs the following operations: receiving the melt boron concentration detected in real-time by the first detection probe and the carrier concentration detected in real-time by the second detection probe; based on a built-in process model, predicting the change in the axial resistivity of the crystal rod over a future period; and, under constraints of adjusting the crystal rod's axial movement rate, feeding rate, and the temperature gradient of the melt near the solid-liquid interface, solving for the optimal control sequence that minimizes the deviation between the predicted resistivity and the preset axial resistivity target curve; and outputting the current control quantity in the optimal control sequence as an instruction to the crystal pulling machine, the boron particle feeder 200, and the temperature regulator 310, respectively, to synchronously adjust the crystal rod's axial movement rate, feeding rate, and the temperature gradient of the melt near the solid-liquid interface.

[0070] Specifically, the built-in process model of the Model Predictive Controller 800 can employ a one-dimensional diffusion-segregation coupling equation:

[0071] (Formula 2)

[0072] Where: CL is the melt boron concentration in the region near the contact between the crystal rod and the melt surface; t is time; z is the position of the crystal rod in the axial direction from the head of the crystal rod; v is the moving speed of the crystal rod along the axial direction; Deff is the effective diffusion coefficient; Keff is the effective segregation coefficient of boron, which varies with the temperature gradient G(t); R(t) is the feeding rate of the boron particle feeder; VL is the melt volume.

[0073] The built-in process model of the model predictive controller 800 comprehensively considers three physical processes: boron diffusion in the melt, segregation consumption caused by crystal growth, and feed replenishment. This model can accurately describe the dynamic changes in the melt boron concentration. Based on this model, the model predictive controller 800 predicts concentration changes over a future period, and then predicts the resistivity of the crystal, providing a basis for optimized control.

[0074] Optionally, the preset axial resistivity target curve ρtarget(z) is:

[0075] (Formula 3)

[0076] Where: ρ0 is the target resistivity at the head of the crystal rod, and z is the position of the crystal rod in the axial direction from the head of the crystal rod. This represents the coefficient of axial resistivity variation. The preset axial resistivity target curve ρtarget(z) takes into account the natural concentration gradient caused by boron segregation, making the control target more in line with physical laws and reducing the adjustment burden on the controller.

[0077] The preset range is preferably ±5%, which means that the relative deviation of the resistivity at any position of the crystal rod from the target curve should not exceed 5%.

[0078] Optionally, this embodiment provides a detailed description of the implementation process of the control method in conjunction with specific process parameters. Taking the growth of a 12-inch heavily boron-doped silicon single crystal as an example, the target resistivity ρ = 7 mΩ·cm and the crystal rod length is 1000 mm. During the charging stage: 100 kg of polycrystalline silicon is charged into a quartz crucible, and 22 g of boron is pre-doped, corresponding to an initial melt boron concentration C. L0 = 2.8 × 10 19 cm -3 Neck-raising / shoulder-forming stage: Using a standard pulling speed of 1.0 mm / min, activate the magnetic field adjustment device (500), setting the magnetic field strength to 0.3 T to suppress melt convection and stabilize the solid-liquid interface. Constant diameter growth stage: When the first detection probe measures the melt boron concentration C... L(t) When the concentration is 2% below the target value, the model predictive controller 800 outputs a feeding rate R(t) = 12 mg / min, instructing the boron particle feeder 200 to inject boron particles into the melt; when the second detection probe measures the crystal carrier concentration N... A(t) When the concentration is 3% higher than expected, it indicates that the actual amount of admixture is too large. The model predicts that the controller 800 will reduce the pulling speed v(t) by 0.05 mm / min and simultaneously increase the temperature gradient G(t) by 0.08 K / mm to reduce the effective segregation coefficient k. eff The value was reduced from 0.78 to 0.75 to decrease boron incorporation; through iterative cycles, |ρ(z) - ρ target (z)| ≤ 0.3 mΩ·cm, meeting the requirement of deviation ≤5%. Final stage: turn off the feeder, restore fixed pulling speed, and reduce the magnetic field to 0.1 T to prevent interface oscillation.

[0079] The resistivity fluctuation of the 12-inch heavily boron-doped silicon ingot grown using the method of this embodiment was reduced from ±15% to ±4.2% over a length of 1000 mm using the traditional process; the total amount of boron particles fed was reduced by 8%, and the single furnace running time was shortened by 35 min; the breakage rate after slicing was reduced from 1.8% to 0.6%, and the yield was increased by 12%.

[0080] This embodiment further defines the parameters of the dual-laser-induced breakdown spectroscopy measurement device 400. Both the first and second detection probes use a 266 nm Nd:YAG laser source with a repetition frequency ≥ 1 kHz and a measurement accuracy ≤ 2%. The 266 nm ultraviolet laser has high photon energy, effectively exciting the plasma spectrum of silicon and boron, and offers high spatial resolution, making it suitable for micro-area analysis of melt surfaces and crystal surfaces. The high repetition frequency (1 kHz) ensures real-time measurement, enabling the capture of rapid concentration changes. The measurement accuracy ≤ 2% meets the accuracy requirements of closed-loop control, ensuring the reliability of the control system.

[0081] This embodiment further defines the parameters of the magnetic field adjustment device 500. The transverse magnetic field strength generated by the magnetic field adjustment device 500 is adjustable between 0.1T and 0.5T. In the initial stage of constant diameter growth, a higher magnetic field strength (e.g., 0.5T) can be used to suppress convection and stabilize growth conditions; as the length of the crystal rod increases, the magnetic field strength can be appropriately reduced (e.g., 0.3T) to balance convection suppression and oxygen content control; in the final stage, it can be further reduced to 0.1T to prevent interface oscillation. The continuous adjustability of the magnetic field strength provides a flexible process window for different growth stages.

[0082] The above embodiments of the present invention focus on describing the differences between the various embodiments. As long as the different optimization features between the various embodiments are not contradictory, they can be combined to form a better embodiment. For the sake of brevity, they will not be described in detail here.

[0083] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of the present invention.

Claims

1. A crystal pulling system for pulling heavily boron-doped silicon single crystals, characterized in that, It includes a quartz crucible (100), a boron particle feeder (200), a heating device, a dual laser-induced breakdown spectroscopy measurement device (400), and a model prediction controller (800), wherein: The boron particle feeder (200) is located above the melt surface inside the quartz crucible (100) and is used to replenish boron particles to the surface of the melt. The heating device includes a temperature regulating element (310), which is located above the surface of the melt and is used to regulate the temperature gradient of the melt in the quartz crucible (100) near the solid-liquid interface. The dual laser-induced breakdown spectroscopy measurement device (400) includes a first detection probe and a second detection probe. The first detection probe is used to detect the melt boron concentration in the vicinity of the contact between the crystal rod and the melt surface, and the second detection probe is used to detect the carrier concentration of the grown crystal rod. The boron particle feeder (200), the temperature regulator (310), the first detection probe, and the second detection probe are all connected to the model prediction controller (800). The model prediction controller (800) controls the axial movement speed of the crystal rod, the feeding rate of the boron particle feeder (200), and the output power of the temperature regulator (310) based on the molten boron concentration detected by the first detection probe and the carrier concentration detected by the second detection probe, so that the deviation between the axial resistivity of the crystal rod and the preset axial resistivity target curve is within a preset range. The boron particle feeder (200) includes an annular body (210) and a plurality of nozzles (220). The annular body (210) has an annular inner cavity. The plurality of nozzles (220) are disposed on the annular body (210) and are spaced apart along the annular body (210). The nozzles (220) communicate with the annular inner cavity. The annular body (210) is used to surround the outside of the crystal rod. The nozzles (220) extend toward the side facing the melt surface. The nozzle (220) is tilted toward the center side of the annular body (210).

2. The crystal pulling system for pulling heavily boron-doped silicon single crystals according to claim 1, characterized in that, The angle between the central axis of the nozzle (220) and the plane containing the melt surface is greater than or equal to 60° and less than or equal to 80°.

3. The crystal pulling system for pulling heavily boron-doped silicon single crystals according to claim 1, characterized in that, The heating device further includes a main heating element and an auxiliary heating element. The main heating element surrounds the quartz crucible (100) and is opposite to the middle region of the quartz crucible (100). The auxiliary heating element surrounds the quartz crucible (100) and is opposite to the bottom region of the quartz crucible (100).

4. The crystal pulling system for pulling heavily boron-doped silicon single crystals according to claim 1, characterized in that, The crystal pulling system for pulling heavily boron-doped silicon single crystals also includes a magnetic field adjustment device (500), which is located on the outside of the quartz crucible (100) and is used to apply a transverse magnetic field perpendicular to the surface of the melt to the melt located in the quartz crucible (100).

5. The crystal pulling system for pulling heavily boron-doped silicon single crystals according to claim 4, characterized in that, The magnetic field strength of the transverse magnetic field generated by the magnetic field adjustment device (500) is between 0.1T and 0.5T.

6. The crystal pulling system for pulling heavily boron-doped silicon single crystals according to claim 1, characterized in that, The crystal pulling system for pulling heavily boron-doped silicon single crystals also includes a furnace body (600) and a cooling device (700), wherein the quartz crucible (100), the boron particle feeder (200), the heating device and the cooling device (700) are all located inside the furnace body (600).

7. A control method for a crystal pulling system used in the pulling of heavily boron-doped silicon single crystals, characterized in that, The crystal pulling system for pulling heavily boron-doped silicon single crystals is the crystal pulling system for pulling heavily boron-doped silicon single crystals as described in any one of claims 1 to 6, and the control method includes: During the constant diameter growth stage, the first detection probe is controlled to detect the melt boron concentration in the vicinity of the area where the crystal rod contacts the melt surface, and the second detection probe is controlled to detect the carrier concentration of the grown crystal rod. The axial movement rate of the crystal rod, the feeding rate of the boron particle feeder (200), and the output power of the temperature regulating component (310) are controlled based on the molten boron concentration detected by the first detection probe and the carrier concentration detected by the second detection probe, so that the deviation between the axial resistivity of the crystal rod and the target curve is within a preset range.

8. The control method according to claim 7, characterized in that, The built-in process model of the model predictive controller (800) adopts a one-dimensional diffusion-segregation coupling equation: ； Where: C L t is the melt boron concentration in the region near the contact between the crystal rod and the molten surface; z is the axial distance of the crystal rod from its head; v is the axial moving speed of the crystal rod; D eff K is the effective diffusion coefficient. eff The effective segregation coefficient of boron varies with the temperature gradient G(t); R(t) is the feeding rate of the boron particle feeder; V L This represents the volume of the melt.

9. The control method according to claim 7, characterized in that, The preset axial resistivity target curve ρtarget(z) is: ； Where: ρ0 is the target resistivity at the head of the crystal rod, and z is the position of the crystal rod in the axial direction from the head of the crystal rod. This represents the coefficient of axial resistivity variation.

Images