A Multi-Segment Co-doping Method and System for Czochralski Single-Crystal Silicon Based on Dopant Volatility Compensation

By real-time detection of the dopant partial pressure above the melt and combining it with a volatilization kinetics model and a multi-stage co-doping strategy, precise control of dopant in Czochralski single-crystal silicon was achieved, solving the problem of low dopant volatilization compensation accuracy and improving the uniformity of crystal rod resistivity and product yield.

CN123108724APending Publication Date: 2026-11-17CHENGDU TECH UNIV +1
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202611288844.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-24
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

The existing Czochralski single-crystal silicon doping technology has low dopant volatilization compensation accuracy and lacks real-time detection methods, making it impossible to achieve multi-dopant synergistic compensation. This results in poor longitudinal resistivity uniformity of the crystal rod, affecting product yield.

Method used

Infrared or laser spectroscopy is used to detect the partial pressure of dopants in the atmosphere above the melt in real time. Combined with a volatilization kinetics model and a multi-segment co-doping strategy library, a micro-doping device is used to achieve segmented, typed, and variable-rate dynamic compensation of various dopants, ensuring precise control of the doping concentration in the melt.

Benefits of technology

Real-time dynamic balance of dopant concentration was achieved, and the longitudinal resistivity fluctuation of the crystal rod was controlled within ±1.5%, which is significantly better than the ±5% fluctuation level of the existing technology, thus improving the resistivity uniformity and yield of the product.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123108724A_ABST
    Figure CN123108724A_ABST
Patent Text Reader

Abstract

This invention discloses a multi-segment co-doping method and system for Czochralski single-crystal silicon based on dopant volatility compensation, belonging to the field of semiconductor material preparation technology. The method includes: setting up a real-time infrared or laser spectroscopy detection device above the melt surface of the single-crystal furnace to directly detect the partial pressure values ​​of each dopant in the atmosphere above the melt and convert them into actual concentrations; the control module has built-in volatilization kinetic models for each dopant, a multi-segment co-doping strategy library, and a closed-loop control algorithm. Based on the deviation between the real-time detection data and the target concentration, and combined with the amount of volatilization loss, a replenishment doping amount decision is generated, driving the corresponding hopper and quantitative feeder in the micro-doping device to perform precise replenishment doping. This invention achieves segmented, typed, and variable-rate dynamic collaborative compensation for multiple dopants at different pulling stages, controlling the longitudinal resistivity fluctuation of the crystal rod within ±1.5%, effectively solving the problem of uneven resistivity distribution caused by high-temperature volatilization of dopants during the Czochralski single-crystal silicon growth process.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of semiconductor material preparation technology, and in particular to a method and system for multi-segment co-doping of Czochralski single-crystal silicon based on dopant volatility compensation. Background Technology

[0002] The Czochralski method is one of the main methods for preparing single-crystal silicon. Its basic principle involves melting high-purity polycrystalline silicon in a quartz crucible, then immersing a seed crystal into the melt surface. At the solid-liquid interface between the seed crystal and the melt, silicon atoms are slowly pulled and rotated, causing them to grow in an orderly arrangement along the lattice direction of the seed crystal, ultimately forming a large-diameter single-crystal silicon rod. During the Czochralski single-crystal silicon growth process, dopants are added to the silicon melt to control the conductivity and resistivity of the rod. For N-type single-crystal silicon, commonly used dopants are Group 5 elements, such as phosphorus (P), antimony (Sb), and arsenic (As). The amount of dopant added determines the carrier concentration of the rod, and thus the resistivity.

[0003] However, the aforementioned dopants such as phosphorus, antimony, and arsenic exhibit high volatility in the high-temperature environment of silicon melt (typically above 1410℃). Taking antimony as an example, its volatility coefficient is approximately 700 times that of phosphorus, resulting in severe volatilization losses during crystal pulling. As the crystal pulling time increases, the effective doping concentration in the melt continuously decreases, leading to a gradual increase in resistivity from beginning to end of the same crystal rod. The significant difference in resistivity between the beginning and end severely affects the uniformity of the longitudinal resistivity of the crystal rod and the product yield.

[0004] To overcome the above problems, several solutions have been developed in the prior art. Chinese patent CN121593167A discloses a method for co-doping N-type Czochralski single-crystal silicon, which controls the resistivity range by adding a first dopant and a second dopant at different stages of the feeding process. However, this method is an open-loop segmented feeding scheme, and the volatilization loss of the dopant in the melt after its addition is not monitored and dynamically compensated in real time. Its compensation accuracy depends on the operator's experience and judgment, making it difficult to achieve high-precision control of the resistivity of the entire ingot.

[0005] Chinese patent CN120193327A discloses a method and system for evaluating silicon dopant volatilization compensation based on process analysis. This method calculates concentration deviations by parsing doping target instructions and monitoring real-time process parameters at each pulling stage, thereby evaluating the volatilization compensation amount. However, this method has the following shortcomings: First, its input data consists of pulling process parameters (such as temperature, furnace pressure, flow rate, etc.), which is an indirect calculation method rather than a direct measurement of the dopant concentration above the melt; its accuracy is limited by the accuracy of the process model. Second, this method only solves the problem of calculating the volatilization amount and does not provide a specific real-time dopant concentration detection device or a doping compensation execution device. Third, this method cannot achieve coordinated dynamic compensation for multiple dopants at multiple pulling stages.

[0006] Chinese patent CN224160739U discloses a doping device and a single crystal furnace, which prevents dopant volatilization loss by setting up a shell, a feeding device, and a driving device. However, this device is a passive anti-vaporization design, and the dopant cannot be dynamically adjusted according to the actual concentration change in the melt after feeding, and it does not have the control function of real-time detection and dynamic replenishment.

[0007] In addition, the academic papers "Research on Oxygen Content Control in Czochralski Single Crystal Silicon Growth" and "Research on Oxygen and Carbon Impurity Transport Behavior in Czochralski Single Crystal Silicon under the Influence of Magnetic Field" studied the oxygen content control and the influence of magnetic field on impurities, respectively, but neither of them involved a control scheme for dopant volatilization compensation.

[0008] In summary, the existing technology lacks a technical solution that can detect the partial pressure of dopant above the melt in real time, dynamically calculate the volatilization loss based on the volatilization kinetics model, and achieve multi-stage co-doping control through a multi-hopper micro-doping device. Summary of the Invention

[0009] To address the shortcomings of existing Czochralski single-crystal silicon doping technologies, such as low accuracy of dopant volatility compensation, lack of real-time detection methods, and inability to achieve synergistic compensation among multiple dopants, this invention provides a multi-segment co-doping method and system for Czochralski single-crystal silicon based on dopant volatility compensation. The aim is to achieve precise control of dopant concentration during the growth of Czochralski single-crystal silicon, keeping the longitudinal resistivity fluctuation of the crystal rod within ±1.5%.

[0010] This invention is achieved using the following technical solution: a multi-segment co-doping method for Czochralski-grown single-crystal silicon based on dopant volatility compensation, comprising the following steps: Step S1: A real-time detection device is installed above the melt surface in the single crystal furnace. Infrared spectroscopy or laser spectroscopy is used to detect the partial pressure values ​​of each dopant in the atmosphere above the melt in real time, and convert them into the actual concentration values ​​of each dopant in the melt. ; Step S2: Establish a volatilization kinetic model for each dopant in the control module. The model is expressed as follows:

[0011] in, The dopant is the amount of dopant lost through volatilization in time Δt, where T is the melt temperature, P is the furnace pressure, t is the time, and dopant is the volatilization coefficient corresponding to the type of dopant. Step S3: Based on the target resistivity requirement of the current pulling stage, retrieve the target concentration values ​​of each dopant corresponding to this stage from the multi-stage co-doping strategy library. and the proportioning schemes of each dopant; Step S4: The control module calculates the concentration deviation. Based on the real-time volatilization loss ΔC calculated by the volatilization kinetics model, a decision on the amount of admixture to be added is generated: ; in, For the volume of the melt, To ensure the effective utilization rate of dopants entering the melt; Step S5: The control module generates an execution command based on the doping amount decision, which drives the hopper and quantitative feeder of the corresponding type of dopant in the micro-doping device to move, and the precisely measured dopant is quantitatively dropped into the melt through the conveying channel.

[0012] Furthermore, the real-time detection device includes an infrared light source or laser emitter, a spectrometer or filter element, and a photodetector. It uses non-contact spectral analysis technology to detect the characteristic spectra of each dopant in the atmosphere above the melt in real time, thereby obtaining the partial pressure value of each dopant.

[0013] Furthermore, the optical window of the real-time detection device is provided with a protective gas purge port, which continuously introduces a protective atmosphere to prevent the optical window from being contaminated by volatiles.

[0014] Furthermore, the micro-doping device includes two or more independent hoppers, a quantitative feeder, and a conveying channel. The independent hoppers store different types of dopants, and the conveying channel is equipped with a water-cooled or air-cooled jacket and a protective atmosphere inlet.

[0015] Furthermore, there are three independent hoppers, which respectively store phosphorus dopant, antimony dopant, and arsenic dopant. Each hopper is equipped with a suspended weighing assembly above it for real-time monitoring of the remaining dopant dosage.

[0016] Furthermore, the quantitative feeder is a screw propulsion type, vibration type or piston type structure, and the output of dopant with microgram accuracy is achieved by precisely controlling the driving parameters; a protective atmosphere is introduced into the conveying channel to prevent the dopant from being heated and volatilized during the conveying process.

[0017] Furthermore, the multi-stage co-doping strategy library pre-stores the dopant types, ratios, and target concentration values ​​for each stage according to the target resistivity requirements of different pulling stages; the pulling stages include the material preparation stage, the stabilization stage, the initial stage of constant diameter growth, the middle stage of constant diameter growth, the final stage of constant diameter growth, and the finishing stage.

[0018] Furthermore, during the material preparation and stabilization stages, the real-time monitoring device performs preheating and calibration, without performing any additional mixing operations; during the constant diameter growth stage, additional mixing control is initiated; and during the final stage, the amount of additional mixing is gradually reduced.

[0019] Furthermore, the constant diameter growth stage includes the initial stage, the middle stage, and the final stage of constant diameter growth. In the initial stage, doping is mainly performed with dopants with low volatility. In the middle stage, multiple dopants are used for synergistic doping. In the final stage, doping is mainly performed with dopants with high volatility, thus achieving precise compensation through multi-segment co-doping with different types and rates.

[0020] A Czochralski-grown single-crystal silicon multi-segment co-doping system based on dopant volatility compensation, used to realize the aforementioned Czochralski-grown single-crystal silicon multi-segment co-doping method, includes: The control module, located outside the single crystal furnace, is a PLC or industrial computer, and integrates a volatilization kinetics model calculation unit, a multi-segment co-doping strategy library, and a control algorithm unit. A real-time detection device is installed above the furnace cover of a single crystal furnace, including a housing sealed and installed on the furnace cover, an infrared light source or laser emitter, a beam splitter or filter element and a photoelectric detector installed inside the housing, and the housing is provided with an optical window and a protective gas purging port; A micro-doping device is installed above the furnace cover of a single crystal furnace. It includes two or more independent hoppers, a quantitative feeder, and a conveying channel. The independent hoppers store different types of dopants. The conveying channel extends from the furnace cover into the furnace and its end is located above the melt surface. The tube wall is equipped with a water-cooled or air-cooled jacket. The control module is connected to the real-time detection device and the micro-doping device respectively, receives the dopant partial voltage value detected by the real-time detection device and outputs the doping command to the micro-doping device.

[0021] The beneficial effects of this invention are as follows: This invention utilizes a real-time detection device positioned above the melt surface to directly measure the partial pressure of each dopant in the atmosphere above the melt using infrared or laser spectroscopy, rather than indirectly calculating it through process parameters as in existing technologies. This results in higher detection accuracy and faster response speed, with response times reaching the second level. Furthermore, this invention inputs the real-time detection data into a volatilization kinetic model, combining it with a multi-segment co-doping strategy library and control algorithm to drive a micro-doping device to perform quantitative doping. This overcomes the shortcomings of open-loop or semi-open-loop control in existing technologies, achieving real-time dynamic balance of dopant concentration.

[0022] This invention sets up multiple independent hoppers to store different types of dopants (such as phosphorus, antimony, and arsenic). At different pulling stages, based on the differences in the volatility characteristics of each dopant (e.g., the volatility coefficient of antimony is about 700 times that of phosphorus), the type and ratio of dopants to be added are dynamically selected, achieving precise and coordinated control of multi-stage co-doping. Through real-time detection and dynamic compensation, the effective doping concentration in the melt can be kept constant throughout the crystal pulling process, and the longitudinal resistivity fluctuation of the crystal rod can be controlled within ±1.5%, which is significantly better than the fluctuation level of about ±5% in the prior art.

[0023] This invention achieves microgram-level dopant output through a quantitative feeder (screw propulsion, vibration, or piston structure). Combined with water-cooled or air-cooled jackets and protective atmosphere purging of the conveying channel, it effectively prevents the dopant from evaporating due to heat during the conveying process, ensuring the accuracy and consistency of doping replenishment. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0025] Figure 1 This is a schematic diagram of the overall system structure of the present invention; Figure 2 This is the timing control logic diagram for the multi-segment co-doping of the present invention; In the diagram, 1—Control module (PLC / industrial computer); 2—Real-time detection device; 21—Infrared light source / laser emitter; 22—Spectrometer / filter element; 23—Photodetector; 3—Micro-doping device; 31—Independent hoppers (hopper A for phosphorus dopant, hopper B for antimony dopant, hopper C for arsenic dopant); 32—Quantitative feeder; 33—Conveying channel (with water-cooled / air-cooled jacket); 4—Single crystal furnace cavity; 41—Quartz crucible; 42—Graphite crucible; 43—Heater; 44—Heat shield; 45—Seed crystal; 46—Single crystal silicon rod. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0027] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0028] The following detailed description of some embodiments of the present invention is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0029] Example 1 like Figure 1 As shown, the Czochralski single-crystal silicon multi-segment co-doping system based on dopant volatility compensation provided by the present invention mainly includes a control module 1, a real-time detection device 2, a micro-doping device 3, and a single-crystal furnace cavity 4.

[0030] In this embodiment, the control module 1 is located outside the single crystal furnace and is a PLC or industrial computer. Internally, it integrates a volatilization kinetics model calculation unit, a multi-segment co-doping strategy library, and a control algorithm unit. The control module 1 is connected to the real-time detection device 2 and the micro-doping device 3 via signal lines to receive detection data and send control commands. The multi-segment co-doping strategy library pre-stores the dopant types, ratios, and target concentration values ​​for each stage of the pulling process, based on the target resistivity requirements of different pulling stages. The pulling stages include the material preparation stage, stabilization stage, initial stage of constant diameter growth, middle stage of constant diameter growth, final stage of constant diameter growth, and finishing stage.

[0031] In this embodiment, the real-time detection device 2 is located on the upper left side of the single crystal furnace cover, with its housing sealed to the cover. The device contains an infrared light source or laser emitter 21, a beam splitter or filter element 22, and a photodetector 23. The emitter 21 emits infrared light or laser light towards the molten surface, and the beam passes through an optical window on the furnace cover to illuminate the molten surface. The optical window is a high-temperature resistant window made of sapphire or quartz, with high light transmittance and high temperature resistance. The reflected light carrying the characteristic information of dopant molecules is split by the beam splitter element 22, received by the photodetector 23, converted into an electrical signal, and transmitted to the control module 1 after A / D conversion. This device is a non-contact detection method, does not contaminate the melt, and has a response time in the order of seconds. A protective gas purge port is provided at the optical window, continuously introducing a small amount of argon gas (flow rate of 1-5 L / min) to prevent the window from being contaminated by volatiles and affecting the detection accuracy.

[0032] In this embodiment, the micro-doping device 3 is located on the right side above the furnace cover of the single crystal furnace, and includes three independent hoppers 31, a quantitative feeder 32, and a conveying channel 33. The three independent hoppers respectively store phosphorus dopant, antimony dopant, and arsenic dopant. Each hopper is equipped with a suspended weighing component for real-time monitoring of the remaining dopant dosage. The quantitative feeder 32 adopts a screw propulsion structure. By precisely controlling the number of rotations and the rotation speed of the screw, microgram-level dopant output can be achieved, with a single feeding accuracy of ±2%. The conveying channel 33 extends into the furnace from the furnace cover, with its end located approximately 50-100 mm above the melt surface. The tube wall is equipped with a water-cooled jacket, and circulating cooling water is introduced to maintain the tube wall temperature below 100°C. A protective atmosphere (argon) is introduced into the channel to prevent the dopant from evaporating due to heat during the conveying process. A weighing sensor is installed at the end of the conveying channel 33 to provide real-time feedback on the actual feed amount to the control module 1. The control module 1 compares the actual feed amount with the target feed amount and corrects the error during the next remixing to ensure long-term remixing accuracy.

[0033] In this embodiment, the single crystal furnace cavity 4 is equipped with a quartz crucible 41, a graphite crucible 42, a heater 43, a heat shield 44, a seed crystal 45, and a single crystal silicon rod 46. The quartz crucible 41 contains the silicon melt, the graphite crucible 42 supports the quartz crucible 41, the heater 43 is located around the crucible to provide heat (it is a resistance heater, arranged around the crucible), the heat shield 44 is located above the crucible to provide heat insulation (it is an inverted trapezoidal structure, made of molybdenum or graphite material), the seed crystal 45 extends from the top of the furnace cavity to contact the melt and induce single crystal growth, and the single crystal silicon rod 46 gradually grows at the lower end of the seed crystal 45, forming a core region for crystal growth between the melt surface and the solid-liquid interface.

[0034] Example 2 This embodiment takes the co-doping of phosphorus and antimony in N-type Czochralski single crystal silicon as an example, combined with Figure 2 The timing control logic diagram shown illustrates the specific implementation process of the multi-segment co-doping method based on dopant volatility compensation of this invention. Figure 2 The horizontal axis represents the drawing process, sequentially progressing through the material preparation stage, stabilization stage, initial stage of constant diameter growth, middle stage of constant diameter growth, final stage of constant diameter growth, finishing stage, and furnace shutdown stage. The vertical axis represents the replenishment rate of each dopant. The material preparation and stabilization stages are concentration calibration periods, during which no replenishment is performed. In the initial stage of constant diameter growth, replenishment is primarily performed using dopant A, which has a lower volatility coefficient. The middle stage of constant diameter growth is a period of synergistic replenishment using dopant A and dopant B. In the final stage of constant diameter growth, replenishment is primarily performed using dopant B, which has a higher volatility coefficient. In the finishing stage, the replenishment amount is gradually reduced. Replenishment is stopped in the furnace shutdown stage. This timing control strategy achieves precise compensation for multi-segment co-doping by segment, type, and variable rate.

[0035] In this embodiment, the target crystal rod has a diameter of 300 mm, a target resistivity of 1.5-2.5 mΩ·cm, and a target crystal rod length of 1000 mm.

[0036] like Figure 2 As shown, the multi-stage co-doping timing control logic of this invention divides the entire crystal pulling process into seven timing stages: material preparation stage, stabilization stage, initial stage of constant diameter growth, middle stage of constant diameter growth, final stage of constant diameter growth, finishing stage, and furnace shutdown stage. Different doping strategies are adopted in different stages, and the doping rate of each dopant changes dynamically with the pulling process. The specific operation of each stage is described in detail below.

[0037] (I) Material preparation stage – preheating and calibration High-purity polycrystalline silicon material is loaded into a quartz crucible 41, and the heater 43 is activated to heat and melt the silicon material, maintaining the melt temperature at 1410–1420°C. During the melting stage, the real-time detection device 2 is preheated to stabilize the infrared light source or laser emitter 21, ensuring that the light source output reaches its rated power and remains stable. Simultaneously, the spectroscopic or filtering element 22 and the photodetector 23 perform dark current correction and wavelength calibration, using standard samples to calibrate the spectral detection baseline and eliminate the influence of environmental noise and device drift on detection accuracy. No additional doping is performed during this stage. The control module 1 records the initial operating parameters (melt temperature, furnace pressure, argon flow rate, etc.) as the initial conditions for the volatilization kinetic model. The melting stage lasts approximately 4–6 hours until the silicon material is completely melted and a stable melt surface is formed.

[0038] (II) Stabilization Phase – Initial Concentration Calibration After the silicon material is completely melted, it enters a stable stage. The melt surface tends to stabilize, and the temperature and flow fields gradually become homogenized. The real-time detection device 2 performs the first spectral detection on the atmosphere above the melt surface: the emitter 21 emits infrared spectrum or laser light towards the melt surface. The reflected light carrying the characteristic information of dopant molecules is dispersed by the spectrometer 22, received by the photodetector 23, and converted into an electrical signal. After A / D conversion and concentration calculation, the initial actual concentration values ​​of phosphorus and antimony in the melt are obtained. and .

[0039] Based on the target resistivity requirement (1.5-2.5 mΩ·cm), control module 1 retrieves the target concentration values ​​of phosphorus and antimony corresponding to the stable stage from the multi-stage co-doping strategy library. and Based on the melt volume (calculated from the charge amount, approximately 100L in this embodiment), an initial dopant replenishment scheme is calculated. Simultaneously, control module 1 retrieves the volatilization coefficients λP and λSb of phosphorus and antimony under the current operating conditions from the dopant volatilization parameter library based on the currently measured melt temperature T, furnace pressure P, and argon flow rate Q, preparing for dynamic doping replenishment during the subsequent constant-diameter growth stage. The stabilization phase lasts approximately 1–2 hours. No doping replenishment is performed during this phase; only concentration calibration and parameter initialization are completed.

[0040] (III) Initial stage of constant diameter growth – supplementary doping mainly with low-volatility dopants The seed crystal descends to contact the melt, initiating crystal growth and shoulder formation, followed by the constant diameter growth stage. In the early stage of constant diameter growth (corresponding to a crystal rod length of 0–300 mm), since the volatilization loss of antimony is not yet significant, phosphorus, a dopant with a low volatility coefficient, is mainly used for supplementary doping during this stage.

[0041] Control module 1 uses the phosphorus partial pressure value fed back by real-time detection device 2, combined with the phosphorus volatilization kinetic model, to determine the phosphorus volatilization kinetics value. Calculate the real-time volatilization loss of phosphorus. For doped phosphorus, its volatilization loss in the melt satisfies the following kinetic equation: ; in, This represents the initial concentration of phosphorus in the melt (obtained from the initial spectral analysis). The volatility coefficient of phosphorus (in s) -1 ), where t is time (in seconds).

[0042] Control module 1 calculates concentration deviation And the decision on the amount of phosphorus to be added: ; in, This represents the current melt volume (which gradually decreases as crystal pulling progresses). The effective utilization rate of phosphorus entering the melt (obtained through pre-calibration, in this embodiment) is... (Approximately 0.85).

[0043] Control module 1 generates execution commands based on the doping amount decision, driving the A hopper (phosphorus) and quantitative feeder 32 of the micro-doping device 3 to actuate, quantitatively dropping precisely measured phosphorus dopant into the melt via conveying channel 33. The doping frequency is approximately once every 5-10 minutes, and the amount of doping per batch is dynamically adjusted based on real-time detection results. This constant-diameter growth phase continues until the ingot length reaches approximately 300 mm, lasting about 8-10 hours. During this period, the antimony doping rate is zero.

[0044] (iv) Mid-stage of constant diameter growth - synergistic dynamic ratio doping with multiple dopants As crystal pulling progresses (crystal rod length 300–800 mm), the volatilization loss of antimony in the melt gradually increases. During the middle stage of constant diameter growth, control module 1 initiates synergistic doping of phosphorus and antimony, with the two dopants being added simultaneously in a dynamic ratio.

[0045] Since the volatility coefficient of antimony is approximately 700 times that of phosphorus, the antimony replenishment rate is significantly higher than that of phosphorus in this stage. Control module 1 calculates the volatilization losses of phosphorus and antimony separately. and : ; ; in, .

[0046] Control module 1 calculates the concentration deviations of phosphorus and antimony respectively. and And generate supplementary doping amount decisions respectively: ; ; Based on the above decisions, control module 1 drives hopper A (phosphorus) and hopper B (antimony) to perform supplementary doping. The supplementary doping ratio of the two dopants... The ratio of phosphorus to antimony changes dynamically during the crystal pulling process: In the first half of the constant-diameter growth stage (crystal length 300-550 mm), the ratio of phosphorus to antimony is approximately 2:1 to 3:1; in the second half of the constant-diameter growth stage (crystal length 550-800 mm), as the cumulative volatilization loss of antimony in the melt increases, the ratio of phosphorus to antimony gradually increases to 5:1 to 8:1. This dynamic ratio doping strategy ensures that the concentrations of phosphorus and antimony are maintained near the target values. The constant-diameter growth stage lasts until the crystal length reaches approximately 800 mm, for a duration of approximately 12-15 hours.

[0047] (v) End of constant diameter growth stage – supplementary doping mainly with highly volatile dopants As crystal pulling enters its final stage (ingot length 800-1000 mm), the melt volume decreases significantly (the remaining melt volume is approximately 30-40% of the initial volume), and the dopant concentration fluctuates more intensely. During the final stage of constant-diameter growth, control module 1 primarily uses antimony, which has a higher volatility coefficient, for supplementary doping, while gradually reducing the phosphorus doping rate.

[0048] The antimony replenishment rate reaches its highest level in the entire crystal pulling process during this stage, primarily to compensate for two types of losses: firstly, the continuous volatilization loss of antimony in the high-temperature melt; and secondly, the decrease in the absolute amount of dopant per unit concentration as the melt volume decreases, requiring a higher frequency of replenishment to maintain the target concentration. Control module 1 performs antimony replenishment every 3-5 minutes based on the real-time detected antimony partial voltage value to prevent the tail resistivity from exceeding the limit. Phosphorus replenishment is gradually reduced in this stage because the tail resistivity of the crystal rod is mainly determined by the antimony concentration, and phosphorus only serves as an auxiliary dopant to maintain the basic N-type doping level. The constant-diameter growth phase continues until the crystal rod length reaches approximately 1000 mm, lasting about 5-8 hours.

[0049] (vi) Final stage – gradual reduction After the crystal rod grows to the target length, it enters the finishing stage. In the finishing stage, control module 1 gradually reduces the amount of replenishment for all dopants. Specifically, control module 1 reduces the amount of replenishment for each dopant in a gradient every 10-15 minutes, with the reduction gradient being 10-15% of the replenishment amount in the previous cycle, until the replenishment amount drops to less than 10% of the initial replenishment amount.

[0050] The finishing stage lasts about 1-2 hours, during which the crucible is raised and the temperature is adjusted (the crucible temperature is gradually increased by about 20-50°C) so that the diameter of the crystal rod gradually shrinks until it is separated from the melt, thus completing the crystal rod finishing.

[0051] (vii) Shutdown phase – Stop adding additives After the crystal rod completely detaches from the melt, the furnace is shut down. During the shutdown phase, control module 1 stops all doping operations. Real-time monitoring device 2 enters standby mode, and heater 43 gradually cools down (the cooling rate is controlled at 50-100℃ / hour to avoid thermal stress causing the crucible to crack), completing the pulling of the entire crystal rod.

[0052] Tests showed that, through the above seven-stage multi-stage co-doping timing control, the longitudinal resistivity fluctuation of the entire crystal rod (1000mm in length) was controlled within ±1.5%, and the resistivity difference between the head and tail was less than 0.08Ω·cm, which is significantly better than the ±5% fluctuation level of the traditional single-stage doping method.

[0053] Example 3 This embodiment illustrates the specific method for constructing the volatilization kinetics model in this invention.

[0054] For dopant i, its volatilization loss in the melt The following dynamic equations are satisfied: ; in, The initial concentration of dopant i in the melt (obtained from the first spectral detection). The volatility coefficient of dopant i (in s) -1 ), where t is time (in seconds).

[0055] For different dopants such as phosphorus, antimony, and arsenic, the volatility coefficient Pre-determined through the following steps: The first step is to operate under standard conditions (set temperature T0 = 1420℃, furnace pressure P0 = 1.2 × 10⁻⁶). 4 Under the conditions of Pa and argon flow rate Q0 = 100 L / min, calibration rods containing only a single dopant were drawn, such as a P-type calibration rod containing only phosphorus and an Sb-type calibration rod containing only antimony. The second step is to take samples at equal intervals along the longitudinal direction of the calibration rod (one sample every 50 mm) and measure the resistivity at each location. By converting the resistivity to the doping concentration (such as the ASTM F723 standard), the dopant concentration at each location is obtained. The third step is to use the least squares method to fit the curve of concentration change with pulling time to obtain the volatility coefficient λ_i0 of each dopant under standard working conditions. The fourth step involves testing at different temperatures (1380-1450℃) and furnace pressures (0.5×10⁻⁶). 4 ~2.0×10 4 Repeat the above steps under conditions of Pa and argon flow rate (50-200 L / min) to establish a database of mapping relationships between the volatility coefficient λ_i and operating parameters (T, P, Q).

[0056] During the actual crystal pulling process, control module 1 retrieves the corresponding volatility coefficient from the database based on the real-time monitored operating parameters (T, P, Q). Based on the real-time detected dopant partial pressure, the volatilization loss at the current moment is calculated according to the above kinetic equation. .

[0057] Example 4 This embodiment illustrates the specific working process of the trace doping device 3 in this invention.

[0058] After the control module 1 generates a doping instruction based on the doping amount decision, the micro-doping device 3 performs the doping operation according to the following steps: The first step is for control module 1 to determine the type and amount of dopant to be added based on the doping amount decision. The second step is that the control module 1 sends a drive signal to the quantitative feeder 32 below the corresponding hopper 31. This signal includes the target number of rotations of the screw and the rotation speed parameters. The third step is that the quantitative feeder 32 starts the screw propulsion mechanism according to the drive signal, and pushes the dopant from the bottom of the hopper to the inlet of the conveying channel 33 at a precisely controlled speed. The amount pushed is precisely determined by the geometric parameters of the screw and the number of rotations. In the fourth step, the dopant is cooled and protected by the water-cooled jacket (the tube wall temperature is maintained below 100°C) and carried by the argon protective atmosphere in the transport channel 33, and moves downward along the channel. Throughout the transport process, the temperature of the dopant is always below its volatilization threshold. In the fifth step, the dopant is quantitatively discharged into the melt from the discharge port at the end of the conveying channel 33 (located about 50-100 mm above the melt surface). Due to the surface tension of the melt, the dopant quickly dissolves in the melt and is evenly dispersed. The sixth step involves the weighing sensor located at the end of the conveying channel 33 feeding back the actual feed amount to the control module 1 in real time. The control module 1 compares the actual feed amount with the target feed amount and corrects the error during the next remixing to ensure that the long-term remixing accuracy is better than ±2%.

[0059] In summary, the Czochralski single-crystal silicon multi-segment co-doping system and method based on dopant volatility compensation provided by this invention directly measures the partial pressure of each dopant in the atmosphere above the melt by setting up a real-time infrared or laser spectroscopy detection device above the melt surface of the single-crystal furnace and converting it into actual concentration. This overcomes the shortcomings of the prior art, which relies on indirect calculation of volatilization loss through process parameters, resulting in insufficient accuracy. Through the volatilization kinetic model of each dopant, the multi-segment co-doping strategy library, and the closed-loop control algorithm built into the control module, combined with the precise metering and quantitative feeding of the micro-doping device, dynamic collaborative compensation by segment, type, and rate is achieved in different pulling processes such as the material preparation stage, stabilization stage, early stage of equal diameter growth, middle stage of equal diameter growth, late stage of equal diameter growth, and finishing stage. At the same time, multiple independent hoppers store different dopants such as phosphorus, antimony, and arsenic, and the conveying channel with water-cooled or gas-cooled jackets and protective atmosphere effectively prevents the dopant from volatilizing due to heat during the conveying process, ensuring the accuracy and consistency of the doping. Through actual crystal pulling verification, the technical solution of this invention can control the longitudinal resistivity fluctuation of the crystal rod within ±1.5%, and the resistivity difference between the head and tail is less than 0.08Ω·cm, which is significantly better than the ±5% fluctuation level of the prior art. It effectively solves the technical problem of uneven resistivity distribution caused by high-temperature volatilization of dopants during the Czochralski single crystal silicon growth process, and has extremely high industrial application value and promotion prospects.

[0060] It should be noted that, for the sake of simplicity, the foregoing embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, because according to this application, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions involved are not necessarily essential to this application.

[0061] The above embodiments describe the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Modifications and variations made by those skilled in the art without departing from the spirit and scope of the invention should be within the protection scope of the appended claims.

Claims

1. A multi-segment co-doping method for Czochralski-grown single-crystal silicon based on dopant volatility compensation, characterized in that, Includes the following steps: Step S1: A real-time detection device is installed above the melt surface in the single crystal furnace. Infrared spectroscopy or laser spectroscopy is used to detect the partial pressure values ​​of each dopant in the atmosphere above the melt in real time, and convert them into the actual concentration values ​​of each dopant in the melt. ; Step S2: Establish a volatilization kinetic model for each dopant in the control module. The model is expressed as follows: ; in, The dopant is the amount of dopant lost through volatilization in time Δt, where T is the melt temperature, P is the furnace pressure, t is the time, and dopant is the volatilization coefficient corresponding to the type of dopant. Step S3: Based on the target resistivity requirement of the current pulling stage, retrieve the target concentration values ​​of each dopant corresponding to this stage from the multi-stage co-doping strategy library. and the proportioning schemes of each dopant; Step S4: The control module calculates the concentration deviation. Based on the real-time volatilization loss ΔC calculated by the volatilization kinetics model, a decision on the amount of admixture to be added is generated: ; in, For the volume of the melt, To ensure the effective utilization rate of dopants entering the melt; Step S5: The control module generates an execution command based on the doping amount decision, which drives the hopper and quantitative feeder of the corresponding type of dopant in the micro-doping device to move, and the precisely measured dopant is quantitatively dropped into the melt through the conveying channel.

2. The method for multi-segment co-doping of Czochralski single-crystal silicon based on dopant volatility compensation according to claim 1, characterized in that, The real-time detection device includes an infrared light source or laser emitter, a spectrometer or filter element, and a photodetector. It uses non-contact spectral analysis technology to detect the characteristic spectra of each dopant in the atmosphere above the melt in real time, thereby obtaining the partial pressure value of each dopant.

3. The method for multi-segment co-doping of Czochralski single-crystal silicon based on dopant volatility compensation according to claim 1, characterized in that, The optical window of the real-time detection device is equipped with a protective gas purge port, which continuously introduces a protective atmosphere to prevent the optical window from being contaminated by volatile substances.

4. The method for multi-segment co-doping of Czochralski single-crystal silicon based on dopant volatility compensation according to claim 1, characterized in that, The micro-doping device includes two or more independent hoppers, a quantitative feeder, and a conveying channel. The independent hoppers store different types of dopants, and the conveying channel is equipped with a water-cooled or air-cooled jacket and a protective atmosphere inlet.

5. The method for multi-segment co-doping of Czochralski single-crystal silicon based on dopant volatility compensation according to claim 4, characterized in that, There are three independent hoppers, which respectively store phosphorus dopant, antimony dopant, and arsenic dopant. Each hopper is equipped with a suspended weighing component above it for real-time monitoring of the remaining dopant dosage.

6. The method for multi-segment co-doping of Czochralski single-crystal silicon based on dopant volatility compensation according to claim 4, characterized in that, The quantitative feeder is a screw propulsion type, vibration type or piston type structure, and the output of dopant with microgram accuracy is achieved by precisely controlling the driving parameters; a protective atmosphere is introduced into the conveying channel to prevent the dopant from being heated and volatilized during the conveying process.

7. The method for multi-segment co-doping of Czochralski single-crystal silicon based on dopant volatility compensation according to claim 1, characterized in that, The multi-stage co-doping strategy library pre-stores the dopant types, ratios, and target concentration values ​​for each stage according to the target resistivity requirements of different pulling stages; the pulling stages include the material preparation stage, the stabilization stage, the initial stage of constant diameter growth, the middle stage of constant diameter growth, the final stage of constant diameter growth, and the finishing stage.

8. The method for multi-segment co-doping of Czochralski single-crystal silicon based on dopant volatility compensation according to claim 7, characterized in that, During the material preparation and stabilization stages, the real-time monitoring device performs preheating and calibration, and does not perform supplementary mixing operations; supplementary mixing control is initiated during the constant diameter growth stage; and the amount of supplementary mixing is gradually reduced during the final stage.

9. The method for multi-segment co-doping of Czochralski single-crystal silicon based on dopant volatility compensation according to claim 8, characterized in that, The constant diameter growth stage includes the initial stage, the middle stage, and the final stage of constant diameter growth. In the initial stage, doping is mainly carried out with dopants with low volatility. In the middle stage, multiple dopants are used for synergistic doping. In the final stage, doping is mainly carried out with dopants with high volatility, so as to achieve precise compensation of multi-segment co-doping by segment, type, and variable rate.

10. A Czochralski-grown single-crystal silicon multi-segment co-doping system based on dopant volatility compensation, used to implement the Czochralski-grown single-crystal silicon multi-segment co-doping method according to any one of claims 1 to 9, characterized in that, include: The control module (1) is located outside the single crystal furnace. It is a PLC or industrial computer and integrates a volatilization kinetic model calculation unit, a multi-segment co-doping strategy library and a control algorithm unit. The real-time detection device (2) is set above the furnace cover of the single crystal furnace and includes a housing sealed and installed on the furnace cover, an infrared light source or laser emitter (21) set in the housing, a beam splitter or filter element (22) and a photodetector (23). The housing is provided with an optical window and a protective gas purging port. The micro-doping device (3) is set above the furnace cover of the single crystal furnace. It includes two or more independent hoppers (31), a quantitative feeder (32) and a conveying channel (33). The independent hoppers (31) store different types of dopants respectively. The conveying channel (33) extends from the furnace cover into the furnace and its end is located above the melt surface. The pipe wall is provided with a water-cooled or air-cooled jacket. The control module (1) is connected to the real-time detection device (2) and the micro-doping device (3) respectively, receives the dopant partial voltage value detected by the real-time detection device (2) and outputs the doping instruction to the micro-doping device (3).

Citation Information

Patent Citations

  • Silicon dopant volatilization compensation evaluation method and system based on process analysis

    CN120193327A

  • N-type Czochralski silicon dopant co-doping method

    CN121593167A

  • Doping device and single crystal furnace

    CN224160739U