Method and device for determining stability of crystal growth

Abstract

The disclosure provides a method and device for determining a stability of crystal growth. The method includes: obtaining a plurality of reference crystal images associated with a specific crystal, and identifying a neck region and a crown region of the specific crystal in each of the reference crystal images; estimating an actual diameter of the crown region of each reference crystal image based on the neck region and the crown region of each reference crystal image; and evaluating a growth stability of the specific crystal based on the actual diameter of the crown region of each reference crystal image.

Description

[Technical Field]

[0001] This invention relates to a crystal observation mechanism, and more particularly to a method and apparatus for determining the stability of crystal growth. [Previous Technology]

[0002] During the growth of silicon single crystals in a crystal growth furnace, the observation of the crystal diameter is an important task. In conventional technology, after acquiring an image of the crystal rod taken by a lens in the crystal growth furnace, the crystal growth system can construct a horizontal distance feedback control system based on the pixels in the image corresponding to parts such as the crystal, aperture, and melt interface. This control system can infer the diameter of the crystal in the crystal growth furnace based on a specific correction conversion formula and the image content of multiple consecutive crystal images taken of the crystal.

[0003] However, this process is highly limited by factors such as the brightness and temperature of the crystal growth furnace, the viewing angle of the lens, and the type of crystal product. Furthermore, since crystal growth in the crystal growth furnace is a dynamic process, when the crystal switches from the crown growth stage to the body growth stage, the visual perception changes from planar to three-dimensional, making it difficult for the aforementioned control system to accurately estimate the crystal diameter. [Summary of the Invention]

[0004] In view of the above, the present invention provides a method and apparatus for determining the stability of crystal growth, which can be used to solve the above-mentioned technical problems.

[0005] The present invention provides a method for determining the stability of crystal growth, comprising: acquiring multiple reference crystal rod images associated with a specific crystal rod, and identifying a crystal neck region and a crystal crown region of the specific crystal rod in each reference crystal rod image; estimating an actual diameter of the crystal crown region of each reference crystal rod image based on the crystal neck region and crystal crown region of each reference crystal rod image; and evaluating the growth stability of the specific crystal rod based on the actual diameter of the crystal crown region of each reference crystal rod image.

[0006] This invention provides a crystal growth stability determination device, including a storage circuit and a processor. The storage circuit stores program code. The processor is coupled to the storage circuit and accesses the program code to execute: acquiring multiple reference crystal rod images associated with a specific crystal rod, and identifying a crystal neck region and a crystal crown region of the specific crystal rod in each reference crystal rod image; estimating an actual diameter of the crystal crown region of each reference crystal rod image based on the crystal neck region and crystal crown region of each reference crystal rod image; and evaluating a growth stability of the specific crystal rod based on the actual diameter of the crystal crown region of each reference crystal rod image.

Implementation Method

[0008] Please refer to Figure 1, which is a schematic diagram of a determination device according to one embodiment of the present invention. In different embodiments, the determination device 100 may be implemented as various intelligent devices / computer devices. In one embodiment, the determination device 100 may also be integrated into a crystal growth system (e.g., it may be implemented as a control system for controlling or managing the crystal growth process / parameters), but is not limited thereto.

[0009] In Figure 1, the determination device 100 includes a storage circuit 102 and a processor 104. The storage circuit 102 may be, for example, any type of fixed or removable random access memory (RAM), read-only memory (ROM), flash memory, hard disk or other similar device or combination of these devices, and may be used to record multiple code or modules.

[0010] The processor 104 is coupled to the storage circuit 102 and may be a general purpose processor, a special purpose processor, a conventional processor, a digital signal processor, multiple microprocessors, one or more microprocessors incorporating a digital signal processor core, a controller, a microcontroller, an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), any other type of integrated circuit, a state machine, a processor based on an advanced reduced instruction set machine (ARM), and the like.

[0011] In an embodiment of the present invention, the processor 104 may access the modules and program code recorded in the storage circuit 102 to implement the crystal growth stability determination method proposed in the present invention, the details of which are described below.

[0012] Please refer to Figure 2, which is a flowchart illustrating a method for determining crystal growth stability according to one embodiment of the present invention. The method of this embodiment can be executed by the determination device 100 in Figure 1. The details of each step in Figure 2 are explained below with reference to the components shown in Figure 1.

[0013] In step S210, the processor 104 obtains multiple reference crystal rod images (hereinafter referred to as M1~MK) associated with a specific crystal rod (hereinafter referred to as C1), and identifies the crystal neck region and crystal crown region of the specific crystal rod C1 in each reference crystal rod image M1~MK.

[0014] In one embodiment, a specific crystal rod C1 is, for example, a crystal rod grown in a crystal growth furnace, and the reference crystal rod images M1~MK are, for example, multiple images continuously captured by the lens of the crystal growth furnace of the specific crystal rod C1 during growth.

[0015] In one embodiment, after obtaining the reference crystal ingot images M1~MK, the processor 104 can input each reference crystal ingot image M1~MK into the identification model MD1, wherein the identification model MD1 can identify the corresponding crystal neck region and crystal crown region in each reference crystal ingot image M1~MK in accordance with each reference crystal ingot image M1~MK.

[0016] For example, after acquiring the reference image M1, the processor 104 can feed the reference image M1 into the recognition model MD1, and the recognition model MD1 can accordingly identify the crystal neck region and crystal crown region of a specific crystal rod C1 in the reference image M1 when it was captured as the reference image M1. As another example, after acquiring the reference image MK, the processor 104 can feed the reference image MK into the recognition model MD1, and the recognition model MD1 can accordingly identify the crystal neck region and crystal crown region of a specific crystal rod C1 in the reference image MK when it was captured as the reference image MK.

[0017] In different embodiments, the recognition model MD1 may include one or more object recognition models. In some embodiments, each object recognition model may be implemented as at least one of a you only look once (YOLO) model and a mask region-based convolutional neural network (mask R-CNN) model, but is not limited thereto. To enable the recognition model MD1 to have the above-mentioned capabilities, the processor 104 may obtain multiple first ingot images CM1~CMN corresponding to a specific growth stage (e.g., crown growth stage) during the training of the recognition model MD1, wherein each first ingot image CM1~CMN is marked with a reference crystal neck region and a reference crystal crown region.

[0018] In one embodiment, assuming that the specific growth stage under consideration is the crown growth stage, the first crystal rod images CM1~CMN are obtained, for example, by taking pictures of various crystal rods in the crown growth stage by one or more lenses disposed in one or more crystal growth furnaces.

[0019] In one embodiment, the reference crystal neck region and the reference crystal crown region in each of the first crystal ingot images CM1~CMN may be marked by relevant personnel in each of the first crystal ingot images CM1~CMN. In one embodiment, the reference crystal neck region is, for example, an image region determined by relevant personnel to correspond to the crystal neck of the crystal ingot, and the reference crystal crown region is, for example, an image region determined by relevant personnel to correspond to the crystal crown of the crystal ingot.

[0020] Please refer to Figure 3, which is a crystal ingot image marked with a reference crystal neck region and a reference crystal crown region according to one embodiment of the present invention. In Figure 3, the crystal ingot image 300 is, for example, one of the first crystal ingot images CM1 to CMN, but is not limited to this. As can be seen from Figure 3, the crystal ingot image 300 is obtained by taking a picture of a crystal ingot A1 in the crystal crown growth stage through a lens inside the crystal growth furnace, and the crystal ingot image 300 may be marked by relevant personnel with a reference crystal neck region 301 corresponding to the crystal neck of crystal ingot A1 (for example, corresponding to the crystal neck boundary of crystal ingot A1), wherein the mark 301a can be used, for example, to indicate the label corresponding to the reference crystal neck region 301 (for example, the class 1 corresponding to the crystal neck).

[0021] In addition, the crystal ingot image 300 may also be marked by relevant personnel with a reference crystal crown region 302 corresponding to the crystal crown of the crystal ingot A1 (which, for example, corresponds to the crystal crown boundary of the crystal crown of the crystal ingot A1), wherein the mark 302a may be used, for example, to indicate the label corresponding to the reference crystal crown region 302 (for example, the class 2 corresponding to the crystal crown).

[0022] In the embodiments of the present invention, each first crystal rod image CM1~CMN may have a form / state similar to that in Figure 3, but may not be limited thereto.

[0023] In one embodiment, the processor 104 may input the first crystal rod images CM1~CMN marked with the reference crystal neck region and the reference crystal crown region as training data to the recognition model MD1, so that the recognition model MD1 can learn the features in the image regions corresponding to the reference crystal neck region and the reference crystal crown region.

[0024] Accordingly, when the trained recognition model MD1 receives a crystal rod image (e.g., a specific crystal rod C1), it can find the crystal neck region and crystal crown region of the crystal rod in the crystal rod image, but it is not limited to this.

[0025] After identifying the corresponding crystal neck region and crystal crown region in each reference crystal ingot image M1~MK, in step S220, the processor 104 estimates the actual diameter of the crystal crown region of each reference crystal ingot image M1~MK based on the crystal neck region and crystal crown region of each reference crystal ingot image M1~MK.

[0026] Please refer to Figures 4A and 4B, where Figure 4A is a schematic diagram of a reference crystal rod image with an identified crystal neck region and crystal crown region according to one embodiment of the present invention, and Figure 4B is a schematic diagram of the mechanism for obtaining the actual diameter of the crystal crown region according to Figure 4A. In Figures 4A and 4B, the reference crystal rod image 400 is, for example, one of the reference crystal rod images M1 to MK, and after being identified by the identification model MD1, the identification model MD1 can find the crystal neck region 401 and the crystal crown region 402 in the reference crystal rod image 400.

[0027] In this case, the processor 104 can estimate a specific ratio RR between the diameter 401a of the neck region 401 and the diameter 402b of the crown region 402 in the reference crystal bar image 400.

[0028] In one embodiment, after the identification model MD1 identifies the crystal neck region 401 and the crystal crown region 402 in the reference crystal bar image 400, the identification model MD1 can also directly estimate a specific ratio RR between the diameter 401a of the crystal neck region 401 and the diameter 402a of the crystal crown region 402, and provide the specific ratio RR to the processor 104, but it is not limited to this.

[0029] In one embodiment, after identifying the crystal neck region 401, the processor 104 (or the recognition model MD1) can use the x-coordinate difference between the leftmost and rightmost pixels of the crystal neck region 401 as the diameter 401a. Similarly, after identifying the crystal crown region 402, the processor 104 (or the recognition model MD1) can use the x-coordinate difference between the leftmost and rightmost pixels of the crystal crown region 402 as the diameter 402a, but this is not limited to this method.

[0030] Subsequently, the processor 104 can estimate the actual diameter DI2 of the crown region 402 of each reference crystal ingot image 400 based on the specific ratio RR corresponding to the reference crystal ingot image 400 and the actual diameter DI1 of the crystal neck region 401. In one embodiment, since the actual diameter of the crystal neck region 401 is a parameter preset in the crystal growth system, the actual diameter DI1 of the crystal neck region 401 can be considered known and can be obtained directly from the crystal growth system by the processor 104, but it is not limited to this.

[0031] In one embodiment, assuming that the specific ratio RR is defined as the diameter 401a divided by the diameter 402a, after obtaining the actual diameter DI1 of the crystal neck region 401, the processor 104 can estimate the actual diameter DI2 of the crystal crown region 402 by dividing the actual diameter DI1 by the specific ratio RR.

[0032] In another embodiment, assuming that the specific ratio RR is defined as the diameter 402a divided by the diameter 401a, after obtaining the actual diameter DI1 of the crystal neck region 401, the processor 104 can estimate the actual diameter DI2 of the crystal crown region 402 by multiplying the actual diameter DI1 by the specific ratio RR, but it is not limited to this.

[0033] For example, in one embodiment, assuming the x-coordinates of the leftmost and rightmost pixels of the crystal neck region 401 are 223 and 253 respectively, the processor 104 can obtain the diameter 401a based on the difference between these two (i.e., 30 pixels). Assuming the x-coordinates of the leftmost and rightmost pixels of the crystal crown region 402 are 49 and 395 respectively, the processor 104 can obtain the diameter 402a based on the difference between these two (i.e., 346 pixels). In this case, the processor 104 can accordingly estimate a specific ratio RR as 11.53 (i.e., 346 / 30). In one embodiment, assuming the actual diameter DI1 of the crystal neck region 401 is known to be 5.48 mm, the processor 104 can accordingly estimate the actual diameter DI2 of the crystal crown region 402 as 63.18 mm (i.e., 5.48 x 11.53).

[0034] In one embodiment, the identification model MD1 initiates a detection to simultaneously obtain diameter 401a and diameter 402a, and estimates a specific ratio RR and calculates the actual diameter DI2 based on the known actual diameter DI1.

[0035] In an embodiment of the present invention, the processor 104 may estimate the actual diameter of the crown region in each reference crystal bar image M1~MK based on the above teachings.

[0036] Subsequently, in step S230, the processor 104 evaluates the growth stability of a specific crystal rod C1 based on the actual diameter of the crown region of each reference crystal rod image M1~MK.

[0037] In one embodiment, the processor 104 can determine the degree of change V1 of the crown diameter of a specific crystal rod C1 based on the actual diameter of the crown region corresponding to each reference crystal rod image M1~MK. As previously mentioned, the reference crystal rod images M1~MK are images of a specific crystal rod C1 captured consecutively. In this case, the processor 104 can, for example, estimate the standard deviation (or other similar statistical value) of the actual diameter of the crown region corresponding to each reference crystal rod image M1~MK as the degree of change V1 of the crown diameter of the specific crystal rod C1, but is not limited to this.

[0038] In one embodiment, the processor 104 can determine whether the degree of change V1 of the crown diameter of a specific crystal rod C1 is within a specified range. For example, the processor 104 can determine whether the standard deviation of the actual diameter of the crown region corresponding to each reference crystal rod image M1~MK is within a specified range (e.g., 1%), but it is not limited to this.

[0039] In one embodiment, the degree of change V1 in the crown diameter of a specific crystal rod C1 is within a specified range, indicating that the diameter of the crown region of the specific crystal rod C1 is relatively stable during its growth and has not changed significantly. In this case, the processor 104 can determine that the specific crystal rod C1 is growing stably.

[0040] On the other hand, if the degree of change V1 in the crown diameter of a specific crystal rod C1 is not within a specified range, it indicates that the diameter of the crown region of the specific crystal rod C1 is relatively unstable during its growth, i.e., its change is relatively drastic. In this case, the processor 104 may determine that the specific crystal rod C1 is not growing stably, but it may not be limited to this.

[0041] In summary, the method proposed in this embodiment of the invention can estimate the actual diameter of the crown region of each reference crystal rod image after identifying the individual crystal neck region and crown region of multiple reference crystal rod images captured for a specific crystal rod, thereby evaluating the growth stability of the specific crystal rod. This eliminates the need for visual assessment of the crystal rod's growth stability, thus improving the efficiency of observing crystal rod growth. Furthermore, after determining that the crystal rod is not growing stably (e.g., an abnormal crystal growth condition occurs), personnel can immediately adjust the corresponding crystal growth parameters, thereby shortening the time required to handle abnormal crystal growth conditions.

[0042] Although the present invention has been disclosed above by way of embodiments, it is not intended to limit the present invention. Anyone skilled in the art can make some modifications and refinements without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims. [Simplified Explanation of the Diagram]

[0007] Figure 1 is a schematic diagram of a determination device according to one embodiment of the present invention. Figure 2 is a flowchart of a crystal growth stability determination method according to one embodiment of the present invention. Figure 3 is an image of a crystal rod marked with a reference crystal neck region and a reference crystal crown region according to one embodiment of the present invention. Figure 4A is a schematic diagram of a reference crystal rod image with an identified crystal neck region and crystal crown region according to one embodiment of the present invention. Figure 4B is a schematic diagram of the mechanism for obtaining the actual diameter of the crystal crown region according to Figure 4A.

Claims

1. A method for determining the stability of crystal growth, comprising: Obtain multiple reference crystal rod images associated with a specific crystal rod, and identify a crystal neck region and a crystal crown region of the specific crystal rod in each of the reference crystal rod images; estimate an actual diameter of the crystal crown region of each of the reference crystal rod images based on the crystal neck region and the crystal crown region of each of the reference crystal rod images; and evaluate a growth stability of the specific crystal rod based on the actual diameter of the crystal crown region of each of the reference crystal rod images.

2. The method as described in claim 1, wherein the reference ingot images are multiple images continuously captured of the particular ingot during growth.

3. The method of claim 1, wherein the step of estimating the actual diameter of the crown region of each reference crystal ingot image based on the neck region and the crown region of each reference crystal ingot image includes: Estimate a specific ratio between the diameter of the neck region and the diameter of the crown region in each of the reference crystal rod images; The actual diameter of the crown region of each reference crystal rod image is estimated based on the specific scale corresponding to each reference crystal rod image and the actual diameter of the crystal neck region.

4. The method of claim 1, wherein the step of evaluating the growth stability of a particular crystal rod based on the actual diameter of the crown region of each of the reference crystal rod images includes: The degree of change in the crown diameter of a particular crystal rod is determined based on the actual diameter of the crown region corresponding to each reference crystal rod image; The specific crystal rod is determined to be in stable growth if the degree of change in the crown diameter of the specific crystal rod is within a specified range; and the specific crystal rod is determined to be in unstable growth if the degree of change in the crown diameter of the specific crystal rod is not within the specified range.

5. The method of claim 1, wherein the step of identifying the neck region and the crown region of the particular crystal ingot in each of the reference crystal ingot images includes: Each reference crystal ingot image is input into a recognition model, wherein the recognition model identifies the corresponding crystal neck region and crystal crown region in each reference crystal ingot image in accordance with each reference crystal ingot image.

6. The method as described in claim 5, further comprising: Multiple first crystal rod images corresponding to a crystal crown growth stage are acquired, wherein each first crystal rod image is marked with a reference crystal neck region and a reference crystal crown region; the first crystal rod images are used as multiple first training data for training the recognition model; and the recognition model is trained to learn the features of the reference crystal neck region and the reference crystal crown region in each first crystal rod image by feeding the first training data into the recognition model.

7. The method of claim 6, wherein the identification model comprises a masked region convolutional neural network model.

8. The method as described in claim 6, wherein the reference crystal neck region is marked as a crystal neck boundary and the reference crystal crown region is marked as a crystal crown boundary.

9. A device for determining the stability of crystal growth, comprising: A storage circuit storing program code; a processor coupled to the storage circuit and accessing the program code to execute: acquiring a plurality of reference ingot images associated with a particular ingot, and identifying a neck region and a crown region of the particular ingot in each of the reference ingot images; estimating an actual diameter of the crown region of each of the reference ingot images based on the neck region and the crown region of each of the reference ingot images; and evaluating a growth stability of the particular ingot based on the actual diameter of the crown region of each of the reference ingot images.

Images