Single crystal crystallization process monitoring method and system

Abstract

The application provides a single crystal crystallization process monitoring method and system, which comprises the following steps: collecting a thermal radiation signal of a crystallization area of a single crystal furnace; calculating a corresponding temperature signal based on the thermal radiation signal; performing time threshold transformation analysis on the temperature signal to obtain a temperature change rate; and judging a time when a crystallization interface in the single crystal furnace is formed based on the temperature change rate. Since a single crystal melt releases crystallization latent heat and produces obvious exothermic mutation at the moment of crystallization phase change, the single crystal crystallization process monitoring method can accurately capture the transient mutation characteristics by collecting the thermal radiation signal and performing time threshold transformation analysis on the thermal radiation signal to obtain the temperature change rate. The method realizes objective, real-time and high-precision automatic monitoring of the time when the crystallization interface is formed, provides a reliable basis for accurate control of subsequent crystal pulling and other process parameters, and effectively improves the automation level of single crystal growth and the final crystal quality.

Description

Technical Field

[0001] This invention relates to the field of single crystal crystallization state monitoring technology, and in particular to a method and system for monitoring the single crystal crystallization process. Background Technology

[0002] Single-crystal materials (such as sapphire, silicon, silicon carbide or gallium nitride, piezoelectric single crystals, etc.) are fundamental core materials for the modern semiconductor, optoelectronics, and microelectronics industries. During single-crystal growth, the crystallization state within the single-crystal furnace (such as the formation of the solid-liquid interface and the occurrence of crystallization phase transitions) directly determines the final crystal quality, defect density, and yield. Therefore, real-time and accurate monitoring of the temperature evolution and crystallization state within the crystallization region of the single-crystal furnace is crucial.

[0003] Currently, in the actual production of single crystal furnaces, the monitoring of the crystallization interface formation time mainly relies on the following traditional methods: First, experienced operators make manual visual observation through the observation window; second, they indirectly estimate the temperature by measuring the temperature outside the heater or crucible using contact temperature measuring elements such as thermocouples; and third, they use ordinary infrared thermometers to obtain the surface temperature and set a fixed "absolute temperature value" as the judgment threshold.

[0004] However, the aforementioned existing technologies have obvious technical defects: First, manual observation is limited by the operator's subjective experience, and the intense light and severe heat radiation inside the single crystal furnace can easily cause visual fatigue and misjudgment, making it impossible to achieve automated and standardized production.

[0005] Secondly, the contact temperature measurement method has a large heat conduction hysteresis effect, and the measured temperature cannot directly and in real time reflect the true transient thermodynamic changes in the crystallization region.

[0006] Finally, even with non-contact temperature measurement and a fixed temperature threshold, the thermal environment inside a single crystal furnace is extremely complex and easily affected by background radiation, argon flow, and other thermal interferences. At the moment of crystallization interface formation, the thermodynamic characteristics of the crystallization region undergo subtle dynamic changes along with the release of latent heat of crystallization. Relying solely on "absolute temperature values" often fails to accurately capture this transient characteristic, leading to a significant lag or false alarm rate in judging the moment of crystallization interface formation.

[0007] Therefore, a new method for monitoring the single-crystal crystallization process is needed. Summary of the Invention

[0008] The purpose of this invention is to provide a method and system for monitoring the single crystal crystallization process that can accurately and timely capture the dynamic characteristics of the instantaneous release of latent heat of crystallization, so as to accurately determine the precise moment of crystallization interface formation.

[0009] To achieve the above objectives, the present invention provides a method for monitoring the single crystal crystallization process, comprising: Collect thermal radiation signals from the crystallization region of the single crystal furnace; Based on the thermal radiation signal, the corresponding temperature signal is calculated. Perform time threshold transformation analysis on the temperature signal to obtain the temperature change rate; Based on the temperature change rate, the moment when the crystallization interface forms inside the single crystal furnace is determined.

[0010] Preferably, the heating power of the single crystal furnace and / or the seed crystal pulling speed are also adaptively adjusted according to the temperature change rate.

[0011] Preferably, the thermal radiation signal is acquired using optical fiber.

[0012] Preferably, the optical fiber is made of sapphire.

[0013] Preferably, the thermal radiation signal is separated into at least two characteristic signals with different center wavelengths; The two characteristic signals are respectively converted into photoelectric signals to obtain the corresponding electrical signals; The temperature signal is obtained based on the intensity ratio of the two electrical signals.

[0014] Preferably, the two characteristic signals are located in the near-infrared band and the mid-infrared band, respectively.

[0015] Preferably, the thermal radiation signal is separated into different characteristic signals using at least one of a beam splitter, a filter, or a wavelength division multiplexer.

[0016] Preferably, the thermal radiation signal is collected at any one or more of the solid-liquid interface region, solid region, and liquid region in the single crystal growth process within the single crystal furnace.

[0017] Preferably, an early warning message is issued when the rate of temperature change exceeds a preset early warning threshold.

[0018] The present invention also provides a single crystal crystallization process monitoring system, which includes a controller that monitors the single crystal crystallization process based on the single crystal crystallization process monitoring method described above.

[0019] Compared to existing technologies, the single-crystal melt releases latent heat of crystallization and produces a significant exothermic abrupt change during the crystallization phase transition. The single-crystal crystallization process monitoring method of this invention, by collecting thermal radiation signals and performing time-threshold transformation analysis to derive the temperature change rate, can accurately capture this transient abrupt change characteristic. Compared to traditional methods relying on absolute temperature values ​​or manual observation, the judgment based on the temperature change rate effectively eliminates the interference of complex background thermal fields and absolute temperature drift within the single-crystal furnace, overcoming the thermal hysteresis effect. This method achieves objective, real-time, and high-precision automatic monitoring of the crystallization interface formation moment, providing a reliable basis for the precise control of subsequent process parameters such as crystal seeding, thereby effectively improving the automation level of single-crystal growth and the final crystal quality. Attached Figure Description

[0020] Figure 1 This is a flowchart of the single crystal crystallization process monitoring method in an embodiment of the present invention.

[0021] Figure 2 This is a schematic diagram of the single crystal crystallization process monitoring system in an embodiment of the present invention. Detailed Implementation

[0022] To illustrate the technical content, structural features, objectives, and effects of the present invention in detail, the following description is provided in conjunction with the embodiments and accompanying drawings.

[0023] This embodiment discloses a method for monitoring the single crystal crystallization process, used for online monitoring of the single crystal growth process. Specifically, the monitoring method includes the following steps: S1. Collect thermal radiation signals from the crystallization region of the single crystal furnace.

[0024] During single crystal growth, the crystalline region continuously radiates infrared energy outwards, and the intensity of this thermal radiation is related to the local temperature and phase transition state. The detection component for receiving the thermal radiation signal emitted from the crystalline region can be located outside the furnace observation window, or it can be guided to a low-temperature region outside the furnace via a high-temperature resistant light guide before detection. To avoid the strong background radiation inside the furnace overwhelming the effective signal, the field of view of the detection component is preferably limited to the vicinity of the crystalline region, ensuring that the received light spot covers the solid-liquid interface and its adjacent area.

[0025] S2. Calculate the corresponding temperature signal based on the thermal radiation signal.

[0026] The thermal radiation signal received by the detection component is converted into a raw electrical signal through photoelectric conversion. This raw electrical signal can be a voltage signal, a current signal, or a digital quantity. The controller amplifies, filters, and calibrates the raw electrical signal to obtain the temperature signal.

[0027] The calibration process can employ a segmented calibration method using a blackbody furnace, recording sensor outputs at several known temperature points to establish a mapping relationship between radiation intensity and temperature. Alternatively, a lookup table method can be used, pre-storing sensor responses at different temperatures in the controller and calculating the temperature value through interpolation during runtime. For applications with a wide temperature range and significant background disturbances, emissivity correction parameters, window transmittance correction parameters, and environmental background compensation parameters can be introduced to improve the accuracy of temperature inversion. Any temperature sequence reflecting the evolution of the local thermal state that can stably map the thermal radiation changes in the crystalline region can be used for subsequent processing.

[0028] S3. Perform time threshold transformation analysis on the temperature signal to obtain the temperature change rate.

[0029] The so-called time threshold transformation analysis refers to windowing continuously acquired temperature signals in chronological order, extracting the temperature change per unit time, and comparing the temperature change with a preset threshold to identify transient thermal events.

[0030] Through the aforementioned time-domain processing, latent heat release events that are difficult to identify directly from absolute temperature can be transformed into peak temperature change rates, slope transitions, or local abrupt change points. To further improve robustness, the controller can first determine the mean and standard deviation of the baseline rate of change during the pre-crystallization stabilization phase, and then determine the judgment threshold. When m consecutive sampling points exceed the judgment threshold, a significant thermal event is determined to have occurred.

[0031] S4. Determine the moment of crystallization interface formation in the single crystal furnace based on the temperature change rate.

[0032] When a single crystal melt transforms from a liquid phase to a solid phase, it releases latent heat of crystallization, which disrupts the local thermal equilibrium in the crystallization region. This manifests as abrupt changes in the temperature change rate sequence that differ from the background drift.

[0033] For crystallization regions in a slowly cooling or relatively stable thermal field, the heat released by the phase transition will cause local radiation enhancement in a short period of time, thus forming a peak, inflection point or a sudden change from negative to positive on the temperature change rate curve.

[0034] The controller compares the temperature change rate obtained from the time threshold transformation analysis with the judgment threshold, and then outputs the moment when the crystallization interface forms. This moment can be defined as the moment when the change rate first crosses the judgment threshold, the moment when the change rate reaches its peak, or the confirmation moment after crossing the judgment threshold and continuing for a preset time.

[0035] The criteria can be adjusted accordingly for different material systems and furnace types. As long as they can reflect the thermodynamic abrupt change caused by the release of latent heat of crystallization, they can be used to identify the moment of interface formation.

[0036] Using the above scheme, the monitoring object changes from absolute temperature values ​​to the characteristics of temperature change over time. Since the background thermal field, insulation structure radiation, and airflow disturbances in the furnace often exhibit slow changes or low-frequency drifts, while the release of latent heat of phase change manifests as local, short-term dynamic abrupt changes, analyzing the rate of temperature change can separate phase change events from the complex background. This reduces errors caused by absolute temperature drift, thermal hysteresis, and differences in human judgment, and improves the timeliness and repeatability of identifying the moment of interface formation.

[0037] In some alternative implementations, the temperature signal can be not only a single-point temperature sequence, but also a multi-point temperature sequence, a scanned temperature sequence, or an equivalent temperature feature value sequence; the time threshold transformation analysis is not limited to first-order difference, but can also use second-order difference, local slope fitting, abrupt change detection, derivative threshold comparison, peak search, or piecewise linear fitting method based on time window, as long as the rate of change characteristic representing the strength of transient temperature change can be obtained.

[0038] In another embodiment, to further stabilize the crystal growth process after the formation of the crystallization interface, the monitoring results can also be used to participate in the closed-loop adjustment of the single crystal furnace process parameters. That is, at the same time as outputting the moment of interface formation, the controller also outputs control commands to the heating power supply and the pulling mechanism to adjust the heating power of the single crystal furnace and / or the seed crystal pulling speed.

[0039] In one implementation, when the temperature change rate first exceeds a trigger threshold, the controller determines that a significant phase transition has begun in the crystallization region and initially adjusts the process parameters by a small margin. For example, the current heating power is gradually reduced by 0.5% to 5% over 2 to 20 seconds, while the seed crystal pulling speed is reduced by 2% to 15%, allowing the heat removal rate near the solid-liquid interface and the latent heat release rate of crystallization to rebalance. If the temperature change rate subsequently falls back to a stable range, the new process parameters are maintained; if the temperature change rate continues to rise and exceeds a higher confirmation threshold, the heating power is further reduced or the pulling speed is further decreased to suppress interface fluctuations and excessively rapid growth.

[0040] In another implementation, proportional or segmented adjustment can be performed based on the deviation of the temperature change rate. For example, the target change rate is set as vref. When the real-time change rate is v, the controller calculates the adjustment amount based on the deviation e = v - vref. A segmented lookup table method can be used: when e is in the first deviation range, the lifting speed is reduced by 3%; when e is in the second deviation range, the lifting speed is reduced by 6%; when e exceeds the third deviation range, in addition to reducing the lifting speed, a control limiting command is issued to limit the acceleration of the lifting mechanism within a predetermined time. The above adjustment method is not limited to proportional-integral-derivative control; it can also be fuzzy rule control, model predictive control, or empirical curve control, as long as it can correct the thermal field and growth rate based on the temperature change rate.

[0041] For example, in a process of forming a single crystal using seed crystal pulling, the system collects temperature data at a frequency of 20Hz and calculates the temperature change rate using a 1s sliding window. During the stable phase, the temperature change rate is maintained between -0.05℃ / s and 0.08℃ / s. When the temperature change rate is detected to be higher than 0.35℃ / s for five consecutive sampling points, the controller gradually reduces the heating power from 32kW to 31.2kW, while simultaneously adjusting the pulling speed from 2.5mm / min to 2.2mm / min. If the temperature change rate falls below 0.10℃ / s within 10s, the parameters are maintained for continued growth. If the change rate continues to rise above 0.8℃ / s, the heating power is further reduced by 0.6kW and the pulling speed is reduced to 2.0mm / min. This method prevents localized heat release after interface formation from causing drastic fluctuations at the interface front in a short period.

[0042] The advantage of this approach is that interface formation is not an isolated event, but rather it further influences the thermal field distribution, the morphology of the crystallization front, and the crystal growth rate. By directly incorporating the rate of temperature change as a control variable into the closed-loop regulation, the control action is based on the actual thermal state changes in the crystallization region, rather than on the lagging temperature outside the furnace or empirical time points. This shortens the process response chain, reduces overshoot, and improves the stability of the growth process.

[0043] In another embodiment, the thermal radiation signal can be acquired via optical fiber.

[0044] One end of the optical fiber faces the target observation area inside the single-crystal furnace, while the other end is connected to the processing device located outside the furnace. In this way, the radiant energy from the high-temperature zone can be guided to a location with a lower temperature, where electronic devices can be easily deployed, thereby reducing the risk of thermal damage to the sensor caused by direct exposure to a high-temperature environment.

[0045] In terms of specific structure, optical fibers can be arranged through the observation channels reserved on the side wall or top of the furnace body. The outer periphery of the channel is equipped with a sealing structure and a heat insulation structure to maintain the stability of the atmosphere and thermal field inside the furnace.

[0046] The fiber optic front end can be equipped with a collimating component, focusing lens, light-limiting aperture, or protective sleeve to limit the field of view and suppress stray light; the fiber optic back end can be connected to the photodetector by mechanical connection, flange connection, or standard fiber optic connector.

[0047] The fiber material can be selected based on the furnace temperature, radiation band, and chemical resistance requirements. Sapphire fiber is preferred for medium-to-high temperature single crystal growth applications. Sapphire fiber has a high melting point, good high-temperature mechanical strength, and a wide infrared transmission range. It is not easily softened, carbonized, or rapidly aged in high-temperature, high-radiation, and highly corrosive environments, making it suitable for long-term placement near the crystallization zone. An alumina tube, boron nitride tube, or metal heat insulation sleeve can be installed on the outer side of the sapphire fiber to enhance installation stability and impact resistance.

[0048] By positioning the sapphire fiber front end in an area that allows direct visibility of the target interface while avoiding highly reflective surfaces, the quality of effective thermal radiation acquisition can be significantly improved. Since the radiation signal is transmitted through the fiber to the outside of the furnace for processing, the photodetector does not need to withstand the high temperatures inside the furnace, resulting in more stable operating conditions for the detection link. This, in turn, helps reduce measurement errors caused by temperature drift in electronic components.

[0049] In another embodiment, the thermal radiation signal can be separated into at least two characteristic signals with different center wavelengths, and each signal can be photoelectrically converted. The temperature signal can then be obtained based on the intensity ratio of the two electrical signals.

[0050] Specifically, thermal radiation from the crystallization region first enters the spectrometer, which separates the original radiation into a first characteristic signal and a second characteristic signal.

[0051] The first and second characteristic signals correspond to different center wavelengths or different bandwidths, respectively. The beam splitting unit can be a beam splitter prism, a filter, a wavelength division multiplexer, or a combination thereof.

[0052] Taking a beam splitter as an example, the radiation beam is separated into two or more optical paths according to wavelength after passing through the beam splitter; taking a filter as an example, narrowband filters with different center wavelengths can be set to make the detector receive only the target band; taking a wavelength division multiplexer as an example, different wavelengths can be directly assigned to different channels in optical fiber transmission scenarios.

[0053] The two characteristic signals enter the corresponding photodetectors, which convert the light intensity into a first electrical signal and a second electrical signal.

[0054] During the temperature calculation process, the controller collects the first electrical signal strength I1 and the second electrical signal strength I2, and calculates their ratio R=I1 / I2.

[0055] Since the radiation intensity of the same target area in two adjacent or different infrared bands is controlled by temperature, and certain common disturbance factors will have an approximately proportional effect on the two signals, the ratio calculation can, to a certain extent, offset the effects of small changes in target emissivity, slow attenuation of window transmittance, and overall attenuation of the optical path on absolute intensity.

[0056] The controller can obtain the correspondence between the ratio R and the temperature T through pre-calibration, and then inversely determine the temperature based on this correspondence. This correspondence can be obtained by fitting blackbody calibration data or by establishing it based on the radiation model of the selected band, and then corrected according to the actual furnace type.

[0057] In a preferred embodiment, the two characteristic signals are located in the near-infrared band and the mid-infrared band, respectively.

[0058] For example, the center wavelength of the first channel can be selected between 0.9 μm and 1.6 μm, and the center wavelength of the second channel can be selected between 2.0 μm and 5.0 μm. The near-infrared channel is sensitive to high-temperature targets, while the mid-infrared channel is stronger for mid-to-high temperature radiation energy. By combining different wavelengths, the applicable temperature range can be broadened, and the contrast of temperature change recognition can be improved.

[0059] To reduce time registration errors caused by differences in detector response speeds, synchronous sampling of the two channels is preferred, or time alignment processing can be performed in the controller. When the number of spectroscopic elements increases, a multicolor temperature measurement structure with three or more channels can be formed. In this case, the controller can randomly select two signals from multiple channels to calculate the ratio, or construct multiple ratio features and perform weighted fusion to adapt to the radiation characteristics of different materials in different temperature regions. Any implementation is feasible as long as the final temperature signal can stably characterize the thermal state changes in the crystalline region and meets the input requirements for subsequent time threshold transformation analysis.

[0060] In another preferred embodiment, the thermal radiation signal acquisition point can be located in the solid-liquid interface region of single crystal growth, or in the solid region or liquid region.

[0061] The thermal response characteristics vary depending on the sampling location, and the selection can be made based on the furnace structure, field of view conditions, and control objectives.

[0062] When the sampling point is set in the solid-liquid interface region, thermal radiation most directly reflects the impact of the release of latent heat of crystallization on the local thermal state, and therefore responds fastest to the moment of interface formation. This method is suitable for scenarios with good observation conditions and predictable interface location.

[0063] When the sampling point is set in the solid phase region, the thermal response conducted along the crystal direction after the interface is formed is detected. Although the response speed is slightly delayed relative to the interface point, the solid surface state is usually relatively stable with less reflection interference, making it suitable for working conditions where the interface is blocked or the liquid surface fluctuates greatly.

[0064] When the sampling point is set in the liquid phase region, the monitored object mainly reflects the changes in the thermal state of the melt near the interface, which is suitable for sensing the trend of interface formation in advance through liquid phase thermal disturbance. For furnace types with more obvious melt flow, the signal in the liquid phase region may contain more flow field disturbance components. Therefore, the accuracy of discrimination can be enhanced by narrowing the field of view, increasing the sampling rate, or using a multi-point comparison method.

[0065] In some implementations, multiple sampling points can be set, located in the interface region, solid phase region, and liquid phase region, respectively. The controller combines and analyzes the temperature change rates of multiple sampling points, for example, by comparing the order and amplitude differences of the peak values ​​of the three change rates, to further improve the reliability of interface identification. If the interface region and solid phase region show synchronous abrupt changes, while the liquid phase region only shows slow changes, it further indicates that a phase change has actually occurred rather than simply background thermal disturbance.

[0066] Multi-point configuration can also be used to estimate local temperature gradients and help determine whether the interface morphology is stable. By using the different acquisition locations described above, monitoring perspectives can be flexibly constructed according to site conditions. Since thermodynamic changes in the crystallization region propagate within the neighborhood through radiation, conduction, and convection, acquisition points are not limited to strict geometric interface locations. As long as the acquired signal has a stable correlation with interface formation, it can be used for time determination and process monitoring.

[0067] In another embodiment, in order to intervene in a timely manner when there are abnormalities in interface formation, excessive thermal disturbances, or deviations from the expected process, an early warning mechanism can be set according to the temperature change rate. That is, when the temperature change rate exceeds the preset early warning threshold, an early warning message is output.

[0068] The warning threshold can be set offline based on normal process data, or it can be updated online adaptively by the controller.

[0069] When setting up offline, the temperature change rate sequence can be statistically analyzed during the qualified growth process of several batches to obtain its average fluctuation range, and the upper warning threshold, lower warning threshold, or absolute warning threshold can be set accordingly. For example, in the stable growth stage, the change rate is mostly between -0.15℃ / s and 0.15℃ / s, so the first-level warning threshold can be set to 0.30℃ / s, and the second-level warning threshold can be set to 0.60℃ / s.

[0070] During online updates, the controller can perform rolling calculations on the mean and variance of the rate of change over a recent period, allowing the warning threshold to be adjusted slowly according to the furnace condition, thereby avoiding mismatch of fixed thresholds caused by different batches, different materials, or different thermal field conditions.

[0071] The output of early warning information can take the form of audible and visual alarms, display terminal prompts, historical data marking, process log recording, sending signals to the host computer, or directly linking with control and protection programs.

[0072] To suppress false alarms caused by transient noise, it can be specified that an alarm will only be triggered when the rate of change exceeds the alarm threshold and remains so for a preset time or exceeds a preset number of times. For example, when the sampling frequency is 20Hz, it can be set that a first-level alarm is triggered only when eight consecutive sampling points exceed the first-level alarm threshold; a second-level alarm is triggered immediately and protective actions are executed when five consecutive sampling points exceed the second-level threshold.

[0073] For negative and positive mutations, different warning thresholds can be set to distinguish between different operating conditions such as supercooling, sudden cooling, or abnormal exothermic reactions. For example, in a certain growth batch, the controller uses a 1-second window to calculate the rate of change, and sets the first-level warning threshold to 0.40℃ / s and the second-level warning threshold to 0.90℃ / s. When the rate of change is detected to be between 0.45℃ / s and 0.60℃ / s for 10 consecutive sampling points, the system marks "increased thermal disturbance" on the monitoring screen to alert the operator. When the rate of change rapidly jumps to 1.10℃ / s within 0.5s and persists for more than 3 sampling points, the system immediately outputs a second-level warning, automatically limits the change in lifting speed, and lowers the heating power by a preset value to prevent further expansion of interface instability. If the rate of change subsequently falls back below the first-level warning threshold and remains below it for a preset time, the alarm is deactivated and the system enters observation mode.

[0074] By setting early warning thresholds, the system can not only determine the normal process node of interface formation, but also identify abnormal states such as excessive thermal disturbance, abnormal phase transition, and crystallization interruption trends. Since the early warning is based on the rate of temperature change in the crystallization region, which directly reflects the local thermal state, the early warning timing is closer to the actual source of the anomaly compared to relying on furnace wall temperature or manual visual observation.

[0075] In summary, this invention discloses a method for monitoring the single crystal crystallization process. The following description uses the single crystal pulling growth process as an example: A heating component is installed inside the single crystal furnace. The seed crystal is inserted into the melt from top to bottom through a lifting mechanism and then lifted upward to form a single crystal.

[0076] An observation channel is provided on the side wall of the furnace body, and a sapphire fiber optic probe is installed on the outside of the channel, with the front end of the fiber optic pointing towards the predicted solid-liquid interface region. The rear end of the fiber optic is connected to a beam splitter, which divides the received thermal radiation into a near-infrared channel and a mid-infrared channel, each connected to a corresponding detector. The detector output is amplified and converted from analog to digital before being sent to the controller.

[0077] Before growth begins, the controller calls the calibration data to establish a mapping relationship between the dual-channel ratio and temperature, and sets the sampling frequency to 50Hz, the temperature smoothing window to 0.2s, and the rate of change calculation window to 1s.

[0078] The controller continuously records the temperature sequence from the time the seed crystal approaches the melt surface until crystallization begins, first determining the baseline rate of change distribution before the interface forms. As the seed crystal contacts the melt and enters the crystallization stage, a phase transition exothermics occur near the interface, and the local radiation intensity deviates from its original stable trend in a short period of time. After obtaining the temperature sequence through ratio conversion, the controller further calculates the temperature rate of change sequence.

[0079] When the rate of change first exceeds the judgment threshold and remains there for an extended period of time, the controller outputs an event and records this time point in the process log. Under this condition, the judgment threshold can be set to 0.30℃ / s, the confirmation threshold to 0.50℃ / s, and the continuous holding time to 0.5s.

[0080] When the real-time rate of change reaches 0.34℃ / s, the system enters a state of pending confirmation; if the rate of change remains between 0.52℃ / s and 0.68℃ / s for the next 0.6s, the interface is considered to have been formed.

[0081] After the interface is formed, the controller sends a power fine-tuning command to the heating power source, reducing the heating power by about 2% within 8 seconds and adjusting the lifting speed from 2.8 mm / min to 2.5 mm / min, so that the heat release near the interface is rematched with the external heat input.

[0082] If the rate of change subsequently falls below 0.10℃ / s and remains stable, the current parameters are maintained to enter the normal growth stage. If an abnormal thermal disturbance occurs during the normal growth stage, such as the rate of change suddenly rising above 1.0℃ / s, the system triggers a level two warning and executes protective actions: temporarily limiting the change in lifting speed, reducing the slope of the heating power change, and simultaneously displaying an abnormal prompt on the display interface.

[0083] If the anomaly is resolved, the system can return to monitoring and normal closed-loop control; if the anomaly persists, it can enter the manual intervention or automatic shutdown process according to the preset strategy.

[0084] In this way, the same system can both identify the moment of interface formation and perform process control and anomaly warning functions. The above parameters can be adjusted accordingly for different material systems.

[0085] Another example: For the Bridgman process of growing piezoelectric crystals. The probe end of a sapphire optical fiber is pre-embedded in the center of the bottom of a crucible, with the probe end face precisely flush with the inner wall of the crucible. As the crucible moves within the temperature field, or the temperature field itself moves, the solid-liquid interface sweeps across the probe end. When the interface is precisely positioned at the probe end, crystallization occurs at that point, releasing latent heat, and the optical fiber can record an extremely sharp temperature peak. This method can accurately obtain the thermal history of crystallization at a specific location within the crystal.

[0086] In another preferred embodiment of the present invention, a single crystal crystallization process monitoring system is also disclosed, which includes a controller that monitors the single crystal crystallization process based on the single crystal crystallization process monitoring method described in the above embodiments. Specifically, the system is configured as follows: like Figure 2 The system is configured as follows: an optical fiber acquisition unit (using a sapphire temperature probe), a beam splitting unit (using a dual-channel bandpass filter module), a photoelectric detection unit (containing two wavelength-matched high-speed photodiodes), a signal processing unit (embedded in an FPGA or DSP chip to execute a two-color ratio algorithm), and a crystallization state determination unit (a host industrial control computer). Each unit is connected in series via optical fiber jumpers or high-speed digital interfaces, transforming the data flow from a unidirectional optical signal at the front end to a digital instruction set at the back end.

[0087] It should be understood that the structural form, connection method, parameter range, sampling frequency, band configuration, threshold setting, control logic, and material selection in the above embodiments can all be adjusted according to the specific furnace type, crystal material, and process requirements. For those skilled in the art, equivalent substitutions, combinations, and conventional optimizations made without departing from the technical concept are all permissible.

[0088] The above-disclosed embodiments are merely preferred embodiments of the present invention and should not be construed as limiting the scope of the present invention. Therefore, any equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1. A method for monitoring the single crystal crystallization process, characterized in that, include: Collect thermal radiation signals from the crystallization region of the single crystal furnace; Based on the thermal radiation signal, the corresponding temperature signal is calculated. Perform time threshold transformation analysis on the temperature signal to obtain the temperature change rate; Based on the temperature change rate, the moment when the crystallization interface forms inside the single crystal furnace is determined.

2. The method for monitoring the single crystal crystallization process according to claim 1, characterized in that, The heating power of the single crystal furnace and / or the seed crystal pulling speed are also adaptively adjusted according to the temperature change rate.

3. The method for monitoring the single crystal crystallization process according to claim 1, characterized in that, The thermal radiation signal is acquired using optical fiber.

4. The method for monitoring the single crystal crystallization process according to claim 3, characterized in that, The optical fiber is made of sapphire.

5. The method for monitoring the single crystal crystallization process according to claim 1, characterized in that, The thermal radiation signal is separated into at least two characteristic signals with different center wavelengths; The two characteristic signals are respectively converted into photoelectric signals to obtain the corresponding electrical signals; The temperature signal is obtained based on the intensity ratio of the two electrical signals.

6. The method for monitoring the single crystal crystallization process according to claim 5, characterized in that, The two characteristic signals are located in the near-infrared band and the mid-infrared band, respectively.

7. The method for monitoring the single crystal crystallization process according to claim 5, characterized in that, The thermal radiation signal is separated into different characteristic signals using at least one of a beam splitter, a filter, or a wavelength division multiplexer.

8. The method for monitoring the single crystal crystallization process according to claim 1, characterized in that, The thermal radiation signal is collected at any one or more of the solid-liquid interface region, solid region, and liquid region in the single crystal growth process within the single crystal furnace.

9. The method for monitoring the single crystal crystallization process according to claim 1, characterized in that, When the rate of temperature change exceeds a preset warning threshold, a warning message is issued.

10. A single crystal crystallization process monitoring system, characterized in that, The system includes a controller that monitors the single crystal crystallization process based on the single crystal crystallization process monitoring method according to any one of claims 1 to 9.

Images