A method, device and medium for detecting twin defects

Abstract

The disclosed embodiments provide a method, device and medium for detecting twin defects. The method may include: performing cluster analysis on historical crystal pulling parameters to obtain at least one cluster and the cluster center of each cluster; in the current crystal pulling process, collecting the current crystal pulling parameters in real time; obtaining the first distance between the current crystal pulling parameters and the cluster centers of each cluster; determining the cluster closest to the current crystal pulling parameters according to the first distance; when the current crystal pulling parameters are greater than the judgment threshold corresponding to the closest cluster, determining that the current crystal pulling parameters will cause twin defects in the crystal pulling process.

Description

Technical Field

[0001] The embodiments of the present disclosure relate to the field of semiconductor manufacturing technology, and in particular, to a method, device and medium for detecting twin defects. Background Art

[0002] The quality of semiconductor single crystal silicon substrates used in integrated circuits, such as defect size, determines whether the devices based on the substrate can work properly. For the current mainstream 12-inch large wafers, defect control is more difficult. Each processing technology of the wafer may introduce defects on the wafer, and each process can also be regarded as eliminating the damage caused by the previous processing technology. Therefore, the initial process of silicon wafer processing, that is, the crystal pulling process, is the key process section for defect control.

[0003] The crystal pulling process is a single crystal growth process with a very large length, temperature and time span. During this process, any slight parameter change may cause defects in the single crystal. Due to the large size and long cycle characteristics of the single crystal growth process, if defects occur, it will lead to serious yield loss and reduced production capacity.

[0004] At present, twin defects are more likely to occur at the end of the crystal pulling process. In the back-end process, twin defects may cause stress concentration at the twin boundary, resulting in fragmentation. It may also cause changes in conductivity, affecting the response of semiconductor devices made in the back-end process. Therefore, it is necessary to accurately detect twin defects and reduce the occurrence of twin defects. Summary of the invention

[0005] The embodiments of the present disclosure provide a method, device and medium for detecting twin defects, which can accurately detect the generation of twin defects to avoid the generation of twin defects during the crystal pulling process and improve the quality and yield of the crystal rod.

[0006] The technical solution of the embodiment of the present disclosure is implemented as follows:

[0007] In a first aspect, an embodiment of the present disclosure provides a method for detecting twin defects, the method comprising:

[0008] Performing cluster analysis on historical crystal pulling parameters to obtain at least one cluster and a cluster center of each cluster; wherein the historical crystal pulling parameters include crystal pulling parameters of a crystal rod that has been pulled and has twin defects during a crystal pulling process;

[0009] During the current crystal pulling process, the current crystal pulling parameters are collected in real time;

[0010] Acquire a first distance between the current crystal pulling parameter and a cluster center of each of the clusters;

[0011] Determine, according to the first distance, a cluster closest to the current crystal pulling parameter;

[0012] When the current crystal pulling parameter is greater than a determination threshold corresponding to the closest cluster, it is determined that the current crystal pulling parameter will cause twin defects to be generated during the crystal pulling process.

[0013] In some examples, before performing cluster analysis on the historical crystal pulling parameters, the method further includes:

[0014] Obtaining a wafer corresponding to any crystal ingot among the crystal ingots that have been pulled;

[0015] Measuring whether a crystal rod corresponding to the wafer has twin defects by X-ray diffraction morphology;

[0016] When it is determined that the crystal rod corresponding to the wafer has a twin defect, the wafer is meshed to obtain at least one mesh containing the twin defect;

[0017] The crystal pulling parameters of all grids containing twin defects are used as the historical crystal pulling parameters of the crystal rod with twin defects.

[0018] In some examples, for each of all grids containing twin defects, the crystal pulling parameters of the grid include: position parameters of the grid, pulling acceleration parameters, magnetic field strength parameters of the crystal pulling process, temperature parameters, silicon melt viscosity parameters during the crystal pulling process, silicon melt density parameters during the crystal pulling process, and silicon melt surface tension parameters during the crystal pulling process.

[0019] In some examples, obtaining a first distance between the current crystal pulling parameter and a cluster center of each of the clusters includes:

[0020] Determine the first distance by calculating the second norm between the current crystal pulling parameter and the cluster center of each of the clusters;

[0021] Alternatively, the first distance is determined by calculating the cosine similarity between the current crystal pulling parameter and the cluster center of each of the clusters;

[0022] Alternatively, the first distance is determined by calculating the Pearson correlation coefficient between the current crystal pulling parameter and the cluster center of each of the clusters.

[0023] In some examples, cluster analysis is performed on historical crystal pulling parameters to obtain at least one cluster and a cluster center of each cluster, including:

[0024] Set the number of clusters and set the initial cluster center for each cluster;

[0025] For each of all the historical crystal pulling parameters, calculating a second distance from an initial cluster center of each cluster;

[0026] Based on the minimum distance determination criterion and the second distance, determining the closest cluster for each historical crystal pulling parameter;

[0027] The cluster center of each cluster is determined according to the historical crystal pulling parameters in each cluster.

[0028] In some examples, for each of all historical crystal pulling parameters, calculating a second distance from an initial cluster center of each clustering cluster includes:

[0029] Determine the second distance by calculating the second norm of each historical crystal pulling parameter and the initial cluster center;

[0030] Alternatively, the second distance is determined by calculating the cosine similarity between each historical crystal pulling parameter and the initial cluster center;

[0031] Alternatively, the second distance is determined by calculating the Pearson correlation coefficient between each historical crystal pulling parameter and the initial cluster center.

[0032] In some examples, the determination threshold corresponding to the cluster includes: a threshold of at least one determination variable; wherein each of the determination variables corresponds to any parameter of the current crystal pulling parameters.

[0033] In some examples, after determining that the current crystal pulling parameters will cause twin defects in the crystal pulling process, the method further includes:

[0034] The current crystal pulling parameter is adjusted to be smaller than a determination threshold corresponding to the closest cluster.

[0035] In a second aspect, an embodiment of the present disclosure provides a device for detecting twin defects, the device comprising: a cluster analysis unit, a collection unit, an acquisition unit, a first determination unit, and a second determination unit; wherein,

[0036] The cluster analysis unit is configured to perform cluster analysis on historical crystal pulling parameters to obtain at least one cluster and a cluster center of each cluster; wherein the historical crystal pulling parameters include crystal pulling parameters of a crystal rod that has been pulled and has twin defects during a crystal pulling process;

[0037] The collecting unit is configured to collect the current crystal pulling parameters in real time during the current crystal pulling process;

[0038] The acquisition unit is configured to acquire a first distance between the current crystal pulling parameter and a cluster center of each of the clusters;

[0039] The first determining unit is configured to determine the cluster closest to the current crystal pulling parameter according to the first distance;

[0040] The second determination unit is configured to determine that the current crystal pulling parameter will cause twin defects in the crystal pulling process when the current crystal pulling parameter is greater than a determination threshold corresponding to the closest cluster.

[0041] In some examples, the acquisition unit is further configured to:

[0042] Obtaining a wafer corresponding to any crystal ingot among the crystal ingots that have been pulled;

[0043] Measuring whether a crystal rod corresponding to the wafer has twin defects by X-ray diffraction morphology;

[0044] When it is determined that the crystal rod corresponding to the wafer has a twin defect, the wafer is meshed to obtain at least one mesh containing the twin defect;

[0045] The crystal pulling parameters of all grids containing twin defects are taken as the historical crystal pulling parameters of the crystal rod with twin defects.

[0046] In some examples, for each of all grids containing twin defects, the crystal pulling parameters of the grid include: grid position parameters, pulling acceleration parameters, magnetic field strength parameters of the crystal pulling process, temperature parameters, silicon melt viscosity parameters during the crystal pulling process, silicon melt density parameters during the crystal pulling process, and silicon melt surface tension parameters during the crystal pulling process.

[0047] In some examples, the acquisition unit is configured to:

[0048] Determine the first distance by calculating the second norm between the current crystal pulling parameter and the cluster center of each of the clusters;

[0049] Alternatively, the first distance is determined by calculating the cosine similarity between the current crystal pulling parameter and the cluster center of each of the clusters;

[0050] Alternatively, the first distance is determined by calculating the Pearson correlation coefficient between the current crystal pulling parameter and the cluster center of each of the clusters.

[0051] In some examples, the cluster analysis unit is configured to:

[0052] Set the number of clusters and set the initial cluster center for each cluster;

[0053] For each of all the historical crystal pulling parameters, calculating a second distance from an initial cluster center of each cluster;

[0054] Based on the minimum distance criterion and the second distance, determining the closest cluster for each historical crystal pulling parameter;

[0055] The cluster center of each cluster is determined according to the historical crystal pulling parameters in each cluster.

[0056] In some examples, the cluster analysis unit is configured to:

[0057] Determine the second distance by calculating the second norm of each historical crystal pulling parameter and the initial cluster center;

[0058] Alternatively, the second distance is determined by calculating the cosine similarity between each historical crystal pulling parameter and the initial cluster center;

[0059] Alternatively, the second distance is determined by calculating the Pearson correlation coefficient between each historical crystal pulling parameter and the initial cluster center.

[0060] In some examples, the determination threshold corresponding to the cluster includes: a threshold of at least one determination variable; wherein each of the determination variables corresponds to any parameter of the current crystal pulling parameters.

[0061] In some examples, the apparatus includes an adjustment unit configured to:

[0062] After determining that the current crystal pulling parameter will cause twin defects in the crystal pulling process, the current crystal pulling parameter is adjusted to be smaller than a determination threshold corresponding to the closest cluster.

[0063] In a third aspect, an embodiment of the present disclosure provides a computing device, comprising: a processor and a memory; the processor is used to execute instructions stored in the memory to implement a method for detecting twin defects as described in the first aspect and any one of the examples thereof.

[0064] In a fourth aspect, an embodiment of the present disclosure provides a computer storage medium, wherein the storage medium stores at least one instruction, and the at least one instruction is used to be executed by a processor to implement a method for detecting twin defects as described in the first aspect and any one of the examples thereof.

[0065] The disclosed embodiments provide a method, device and medium for detecting twin defects; cluster analysis is performed based on historical crystal pulling parameters that produce twin defects to obtain multiple clusters, and after obtaining the cluster closest to the current crystal pulling parameters, it is determined whether the current crystal pulling parameters will cause twin defects in the crystal pulling process according to the judgment threshold corresponding to the closest cluster. The generation of twin defects can be detected in real time and accurately to avoid the generation of twin defects in the crystal pulling process and improve the quality and yield of the crystal rod. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 A schematic diagram of a sliding surface provided in an embodiment of the present disclosure;

[0067] Figure 2A schematic diagram of dividing a crystal rod region provided in an embodiment of the present disclosure;

[0068] Figure 3 A schematic diagram of the relationship between silicon viscosity at the melting point and magnetic field strength provided in an embodiment of the present disclosure;

[0069] Figure 4 A schematic diagram of the relationship between the viscosity and temperature of silicon melt provided in an embodiment of the present disclosure;

[0070] Figure 5 A schematic diagram of the relationship between the density and temperature of silicon melt provided in an embodiment of the present disclosure;

[0071] Figure 6 A schematic diagram of the relationship between the surface tension and temperature of silicon melt provided in an embodiment of the present disclosure;

[0072] Figure 7 A schematic flow chart of a method for detecting twin defects provided in an embodiment of the present disclosure;

[0073] Figure 8 A schematic diagram of a twin defect detected according to an embodiment of the present disclosure;

[0074] Fig. 9 A schematic diagram of grid division provided in an embodiment of the present disclosure;

[0075] Fig.10 A schematic diagram of a device for detecting twin defects provided in an embodiment of the present disclosure;

[0076] Fig.11 A schematic diagram of the composition of a computing device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0077] The technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure.

[0078] Twin defects are generally considered to be formed by slip plane sliding caused by shear stress. In a single crystal silicon rod, the (111) crystal plane is a close-packed atomic plane and a slip plane. Figure 1 As shown in the figure, the shear stress on the slip surface can be calculated by the following formula:

[0079]

[0080] Where F represents the tensile force, S represents the cross-sectional area of ​​the crystal rod, λ represents the angle between the normal of the slip plane and the normal of the (001) crystal plane, Φ represents the angle between the radial direction of the slip plane and the normal of the (001) crystal plane, τ0 represents the stress on the cross-sectional area of ​​the crystal rod, and when τ0>τ 临界 When the sliding surface is formed, twins are generated at the slip plane.

[0081] In the crystal pulling process, pulling speed control and temperature control are the most important means of adjusting the crystal pulling process. In order to achieve the purpose of equal diameter control and defect type control, the pulling speed is always in a fine-tuning state, which means that the rate of change of the pulling speed, that is, the acceleration, is also changing all the time. As time goes by, the mass of the cooled solid crystal rod is getting larger and larger. The change in acceleration and mass causes the pulling force F to be a variable force. The change in F causes the shear stress on the (111) slip plane to be a variable force as well. According to the formation mechanism of deformation twins, it is necessary to ensure that this variable shear stress is always less than the critical shear stress τ 临界 state, only in this way can slip be avoided and the formation of twins be avoided.

[0082] During the crystal pulling process, the entire crystal rod can be divided into three areas. Figure 2 As shown, the area at and below the solid-liquid interface can be determined as the melt area, and the temperature of this area is usually greater than or equal to 1410°C. The area above the solid-liquid interface and with a temperature between 1100 and 1410°C can be determined as the plastic area, or it can be called the inelastic medium area. The area above the solid-liquid interface and with a temperature less than 1100°C can be determined as the elastic area, or it can be called the elastic medium area.

[0083] For the elastic medium region mentioned above, changes in the tension F usually do not produce twin defects.

[0084] For the above-mentioned inelastic medium region, the shear stress is related to the length of the crystal rod. Taking the length from the shoulder as the baseline to the solid-liquid interface as an example, the longer the length, the greater the mass of the crystal rod above the solid-liquid interface, resulting in a greater tensile force on the solid-liquid interface. In addition, the shear stress is also affected by the viscosity of the silicon melt, that is, the higher the viscosity of the silicon melt, the less the momentum transfer attenuation in the inelastic medium. The viscosity of the silicon melt is affected by the density of the silicon melt and the magnetic field strength of the crystal pulling process. For example, the greater the density of the silicon melt, the greater the viscosity; the viscosity of the silicon melt has a parabolic relationship with the magnetic field strength. In addition, the shear stress is also related to the surface tension of the silicon melt, because the size of the surface tension of the silicon melt has a significant effect on the atomic adsorption at the solid-liquid interface, and the surface tension of the silicon melt is related to the magnetic field strength. The above-mentioned silicon melt viscosity, silicon melt density and silicon melt surface tension, these crystal pulling parameters, will have abrupt properties near the melting point temperature, and the above-mentioned crystal pulling parameters also have a strong co-coupling effect. As a result, the calculation of the critical shear stress in the inelastic medium region is very complicated.

[0085] Based on the above analysis, it can be considered that the critical shear stress in the inelastic medium region (or plastic region) is a multivariate function of the position parameter X, the pulling acceleration parameter a, the magnetic field strength H during the crystal pulling process, the temperature T, the silicon melt viscosity η during the crystal pulling process, the silicon melt density ρ during the crystal pulling process, and the silicon melt surface tension σ during the crystal pulling process. For the silicon melt viscosity η, the silicon melt density ρ, and the silicon melt surface tension σ, they cannot be directly obtained by measurement in the actual crystal pulling process, but refer to Figures 3 to 6 From the relationship between silicon viscosity at the melting point and magnetic field strength, the relationship between silicon melt viscosity and temperature, the relationship between silicon melt density and temperature, and the relationship between silicon melt surface tension and temperature, it can be seen that these three parameters can be inferred based on the above relationships.

[0086] According to the above analysis, if the multivariate functional relationship between the critical shear stress in the plastic zone and the crystal pulling parameters such as position parameter X, pulling acceleration parameter a, magnetic field intensity H, temperature T, silicon melt viscosity η, silicon melt density ρ and silicon melt surface tension σ can be obtained, then not only can the occurrence of twin defects be accurately detected during the crystal pulling process, but also the occurrence of twin defects can be reduced by adjusting the crystal pulling parameters related to these factors.

[0087] Based on this, see Figure 7 , which shows a method for detecting twin defects provided by an embodiment of the present disclosure, which method may include steps S710 to S750.

[0088] In step S710, cluster analysis is performed on historical crystal pulling parameters to obtain at least one cluster and a cluster center of each cluster.

[0089] In the disclosed embodiment, twin defect detection can be performed on the crystal rods obtained in the historical crystal pulling process, so that the crystal rods with twin defects can be selected from all the crystal rods in the historical crystal pulling process.

[0090] In some examples, before clustering analysis is performed on historical crystal pulling parameters with twin defects, historical crystal pulling parameters need to be collected, and the process may include:

[0091] Obtaining a corresponding wafer based on any one of the pulled crystal rods;

[0092] Performing X-ray diffraction morphology measurement on the wafer, and determining whether a crystal rod corresponding to the wafer has a twin defect according to the measurement result;

[0093] When it is determined that the crystal rod corresponding to the wafer has a twin defect, the wafer is meshed to obtain at least one mesh containing the twin defect;

[0094] The crystal pulling parameters of all grids containing twin defects are collected as the historical crystal pulling parameters of the crystal rod with twin defects.

[0095] For the above example, in detail, first, for each crystal rod, the internal crystal defects and morphology of the wafer produced can be measured by X-ray Diffraction Topography (XRT) technology to determine whether there is a twin defect. Figure 8 For example, Figure 8 The right side of the diagram shows a wafer with twin defects. Figure 8 The enlarged image of the circled part shown on the left is the measurement result of the twin defect on the wafer. For XRT technology, its measurement results can usually only give a qualitative indication of whether twins exist, such as through Boolean values ​​(0 means no twins; 1 means twins).

[0096] Next, for the wafer with twins, it can be divided into grids. For example, each grid can be divided into 2mm*2mm to 8mm*8mm in size, and the following is obtained: Fig. 9 Grid shown.

[0097] Finally, for each network, the corresponding crystal pulling parameters can be determined, and the crystal pulling parameters may include the position parameter X, the pulling acceleration parameter a, the magnetic field strength H during the crystal pulling process, the temperature T, the silicon melt viscosity η during the crystal pulling process, the silicon melt density ρ during the crystal pulling process, and the silicon melt surface tension σ during the crystal pulling process. For each grid, the position parameter can be the coordinates (x, y, z) of the position of the grid centroid point in space with the shoulder as the baseline. The magnetic field strength H can be measured by a sensor. The temperature T can be calculated from the center temperature according to the law of heat conduction, and the center temperature is measured by an infrared thermometer of the crystal pulling furnace. Although η, ρ, and σ cannot be directly measured, they can be obtained from their respective temperature dependence curves (such as Figures 4 to 6 As shown in the calculation. Fig. 9 As shown, the crystal pulling parameters of the i-th grid can be marked as: i ,y i , z i , H i , T i , η i , ρ i , σ i and a i .

[0098] After the crystal pulling parameters of all grids (ie, historical crystal pulling parameters) are obtained, multiple clusters may be formed by clustering and a corresponding cluster center may be determined for each cluster.

[0099] In some examples, a K-means clustering algorithm may be used to form multiple clusters and determine a corresponding cluster center for each cluster. Each cluster corresponds to a cause of twin defects, and the current crystal pulling parameters collected subsequently may determine the closest cluster according to the first distance between the cluster centers, and determine whether the current crystal pulling parameters will cause twin defects in the crystal pulling process based on the judgment threshold corresponding to the closest cluster.

[0100] In step S720, during the current crystal pulling process, current crystal pulling parameters are collected in real time.

[0101] Similar to the aforementioned historical crystal pulling parameters, the current crystal pulling parameters may include the position parameter X of the current crystal pulling process, the pulling acceleration parameter a, the magnetic field strength H of the crystal pulling process, the temperature T, the silicon melt viscosity η during the crystal pulling process, the silicon melt density ρ during the crystal pulling process, and the silicon melt surface tension σ during the crystal pulling process. The method for obtaining each parameter is the same as the method for obtaining the aforementioned historical crystal pulling parameters, and will not be described in detail here.

[0102] In step S730, a first distance between the current crystal pulling parameter and the cluster center of each cluster is obtained.

[0103] In the embodiment of the present disclosure, the first distance can be calculated based on the second norm, cosine similarity or Pearson correlation coefficient. Specifically, the first distance can be determined by calculating the second norm of the current crystal pulling parameter and the cluster center of each cluster cluster. The first distance can also be determined by calculating the cosine similarity of the current crystal pulling parameter and the cluster center of each cluster cluster. The first distance can also be determined by calculating the Pearson correlation coefficient of the current crystal pulling parameter and the cluster center of each cluster cluster.

[0104] In step S740, a cluster closest to the current crystal pulling parameter is determined according to the first distance;

[0105] In the disclosed embodiment, after the first distance between the current crystal pulling parameter and the cluster center of each cluster, the cluster with the shortest first distance is used as the cluster closest to the current crystal pulling parameter. Based on the fact that each cluster corresponds to a cause of twin defects, if the current crystal pulling parameter causes twin defects in the crystal pulling process, it is most likely caused by the cause of twin defects corresponding to the closest cluster.

[0106] In step S750, when the current crystal pulling parameter is greater than the determination threshold corresponding to the closest cluster, it is determined that the current crystal pulling parameter will cause twin defects in the crystal pulling process.

[0107] In the disclosed embodiment, the judgment threshold corresponding to the cluster includes: a threshold of at least one judgment variable; wherein each judgment variable corresponds to any parameter of the current crystal pulling parameter. In detail, among the aforementioned seven crystal pulling parameters that can affect the critical shear stress of the plastic zone, η, ρ and σ are all related to temperature. The position parameter is positively correlated with the quality of the crystal rod, and the quality of the crystal rod is positively correlated with the pulling force. Therefore, the fundamental factors for the generation of twins are the three crystal pulling parameters of pulling acceleration a, temperature T and magnetic field H. For these three crystal pulling parameters, considering all possible combinations of causes for the generation of twins, the number of clusters can be set to Each cluster corresponds to at least one cause of twinning, i.e., a crystal pulling parameter. Based on this, each cluster can set a judgment threshold for its corresponding crystal pulling parameter. Combined with the aforementioned closest cluster, if the crystal pulling parameter corresponding to the closest cluster in the current crystal pulling parameter is greater than the judgment threshold, it means that the current crystal pulling parameter will cause twin defects in the crystal pulling process. Otherwise, it means that the current crystal pulling parameter will not cause twin defects in the crystal pulling process.

[0108] For the above 7 clusters, specifically, if the numerical changes of the pulling acceleration a, temperature T and magnetic field H in the historical crystal pulling parameters within a certain cluster are relatively significant, for example, the extreme difference of the pulling acceleration a, temperature T and magnetic field H in the historical crystal pulling parameters within the cluster is less than the pre-set threshold for characterizing significant changes, then the generation of twins in the cluster is caused by the three parameters of pulling acceleration a, temperature T and magnetic field H. Set judgment thresholds for pulling acceleration a, temperature T and magnetic field H respectively, such as the minimum values ​​of the three parameters of pulling acceleration a, temperature T and magnetic field H. If the pulling acceleration a, temperature T and magnetic field H of the current crystal pulling parameters are all greater than the corresponding judgment thresholds of pulling acceleration a, temperature T and magnetic field H, then it can be determined that the current crystal pulling parameters will cause twin defects in the crystal pulling process.

[0109] If the numerical changes of the two parameters of the pulling acceleration a, temperature T and magnetic field H in the historical crystal pulling parameters in a certain cluster are relatively significant, then the twinning of this cluster is caused by the two parameters with significant changes. Set the corresponding judgment thresholds for these two parameter factors respectively. If the two parameter values ​​of the current crystal pulling parameters are greater than the corresponding judgment thresholds, it can be determined that the current crystal pulling parameters will cause twin defects in the crystal pulling process. For example, in the historical crystal pulling parameters of a cluster, the change of temperature T is not obvious, and the range ΔT<5 to 20K. In addition, compared with the other 6 clusters, the temperature T is not the maximum value, and the mean Mean(T) is not the largest among the 7 clusters. Then it can be determined that the twinning defects of this cluster are caused by the combined influence of the fluctuation of the pulling acceleration a and the fluctuation of the magnetic field intensity H. Therefore, the minimum value of the pulling acceleration a and the minimum value of the magnetic field intensity H in this cluster are used as the judgment threshold in the "pulling acceleration-magnetic field intensity" dual variable model.

[0110] If the numerical changes of only one parameter of the pulling acceleration a, temperature T and magnetic field H in the historical pulling parameters of a cluster are relatively significant, then the twinning of the cluster is caused by the parameter with significant changes. Set the corresponding judgment threshold for this parameter factor. If the parameter value of the current pulling parameter is greater than the corresponding judgment threshold, it can be determined that the current pulling parameter will cause twin defects in the pulling process. For example, in the historical pulling parameters of a cluster, the numerical changes of temperature T and magnetic field intensity H are not obvious, the extreme difference ΔT<5~20K, ΔH<0.3~3T. In addition, compared with the other 6 clusters, T and H are not the maximum values, and Mean(T) and Mean(H) are not the largest among the 7 clusters. Then it can be determined that the twinning defects of the cluster are completely caused by the fluctuation of the pulling acceleration a. Therefore, the minimum value of a in the cluster is used as the threshold of the pulling acceleration in the pulling acceleration univariate model.

[0111] Through the above technical solution, cluster analysis is performed based on the historical crystal pulling parameters that produce twin defects to obtain multiple clusters. After obtaining the cluster closest to the current crystal pulling parameters, the judgment threshold corresponding to the closest cluster is used to determine whether the current crystal pulling parameters will cause twin defects in the crystal pulling process. The generation of twin defects can be detected in real time and accurately to avoid the generation of twin defects in the crystal pulling process and improve the quality and yield of the crystal rod.

[0112] For the aforementioned Figure 7 In some possible implementations of the technical solution shown, cluster analysis is performed on historical crystal pulling parameters of a crystal rod with twin defects to obtain at least one cluster and a cluster center of each cluster, including:

[0113] Set the number of clusters and set the initial cluster center for each cluster;

[0114] For each of all the historical crystal pulling parameters, calculating a second distance from an initial cluster center of each cluster;

[0115] Based on the minimum distance determination criterion and the second distance, determining the closest cluster for each historical crystal pulling parameter;

[0116] The cluster center of each cluster is determined according to the historical crystal pulling parameters in each cluster.

[0117] For the above implementation, in some examples, calculating the second distance between each of all historical crystal pulling parameters and the initial cluster center of each cluster includes:

[0118] Determine the second distance by calculating the second norm of each historical crystal pulling parameter and the initial cluster center;

[0119] Alternatively, the second distance is determined by calculating the cosine similarity between each historical crystal pulling parameter and the initial cluster center;

[0120] Alternatively, the second distance is determined by calculating the Pearson correlation coefficient between each historical crystal pulling parameter and the initial cluster center.

[0121] For the above implementation and examples, in detail, the number of clusters is set to n, and an initial cluster center is randomly set for each cluster.

[0122] First, the distance between each historical crystal pulling parameter and each of the n cluster centers is calculated, and the cluster corresponding to the cluster center with the smallest distance is taken as the cluster to which the historical crystal pulling parameter belongs.

[0123] Next, after calculating the clusters to which all historical crystal pulling parameters belong, for each cluster, the average distance between all historical crystal pulling parameters to which it belongs is calculated as a new cluster center of each cluster.

[0124] Then, for the new n cluster centers, the cluster to which each crystal pulling history data belongs is calculated again according to the distance.

[0125] The above three steps are iterated multiple times until convergence, and the final clusters and the cluster center of each cluster are recorded.

[0126] From the detailed implementation process of the above clustering analysis, we can know that the K-means clustering algorithm can be implemented by only specifying the number of clusters and the calculation method of the distance. The other steps are the process of iteratively calculating the mean value to obtain the cluster center by utilizing the connection between the data itself, and no more details need to be specified.

[0127] For the aforementioned Figure 7 In some possible implementations of the technical solution shown, after determining that the current crystal pulling parameters will cause twin defects in the crystal pulling process, the method further includes:

[0128] The current crystal pulling parameter is adjusted to be smaller than a determination threshold corresponding to the closest cluster.

[0129] For the above-mentioned implementation method, in the embodiment of the present disclosure, since each cluster corresponds to at least one parameter factor, and a judgment threshold is set corresponding to each parameter factor, when adjusting, the parameter value corresponding to the parameter factor in the current crystal pulling parameter can be reduced so that it is less than the corresponding judgment threshold, thereby avoiding the generation of twin defects during the crystal pulling process.

[0130] Based on the same inventive concept as the above technical solution, see Fig.10 , which shows a device 100 for detecting twin defects provided by an embodiment of the present disclosure, the device 100 comprises: a cluster analysis unit 1001, a collection unit 1002, an acquisition unit 1003, a first determination unit 1004 and a second determination unit 1005; wherein,

[0131] The cluster analysis unit 1001 is configured to perform cluster analysis on historical crystal pulling parameters to obtain at least one cluster and a cluster center of each cluster; wherein the historical crystal pulling parameters include crystal pulling parameters of a crystal rod that has been pulled and has twin defects during a crystal pulling process;

[0132] The collecting unit 1002 is configured to collect the current crystal pulling parameters in real time during the current crystal pulling process;

[0133] An acquisition unit 1003 is configured to acquire a first distance between the current crystal pulling parameter and the cluster center of each cluster;

[0134] The first determining unit 1004 is configured to determine a cluster with the closest current crystal pulling parameters when any of the first distances is less than a distance threshold;

[0135] The second determination unit 1005 is configured to determine that the current crystal pulling parameter will cause twin defects in the crystal pulling process when the current crystal pulling parameter is greater than a determination threshold corresponding to the closest cluster.

[0136] In some examples, the collection unit 1002 is further configured to:

[0137] Obtaining a wafer corresponding to any crystal ingot among the crystal ingots that have been pulled;

[0138] Measuring whether a crystal rod corresponding to the wafer has twin defects by X-ray diffraction morphology;

[0139] When it is determined that the crystal rod corresponding to the wafer has a twin defect, the wafer is meshed to obtain at least one mesh containing the twin defect;

[0140] The crystal pulling parameters of all grids containing twin defects are taken as the historical crystal pulling parameters of the crystal rod with twin defects.

[0141] In some examples, for each of all grids containing twin defects, the crystal pulling parameters of the grid include: grid position parameters, pulling acceleration parameters, magnetic field strength parameters of the crystal pulling process, temperature parameters, silicon melt viscosity parameters during the crystal pulling process, silicon melt density parameters during the crystal pulling process, and silicon melt surface tension parameters during the crystal pulling process.

[0142] In some examples, the acquisition unit 1003 is configured to:

[0143] Determine the first distance by calculating the second norm between the current crystal pulling parameter and the cluster center of each of the clusters;

[0144] Alternatively, the first distance is determined by calculating the cosine similarity between the current crystal pulling parameter and the cluster center of each of the clusters;

[0145] Alternatively, the first distance is determined by calculating the Pearson correlation coefficient between the current crystal pulling parameter and the cluster center of each of the clusters.

[0146] In some examples, the cluster analysis unit 1001 is configured to:

[0147] Set the number of clusters and set the initial cluster center for each cluster;

[0148] For each of all the historical crystal pulling parameters, calculating a second distance from an initial cluster center of each cluster;

[0149] Based on the minimum distance criterion and the second distance, determining the closest cluster for each historical crystal pulling parameter;

[0150] The cluster center of each cluster is determined according to the historical crystal pulling parameters in each cluster.

[0151] In some examples, the cluster analysis unit 1001 is configured to:

[0152] Determine the second distance by calculating the second norm of each historical crystal pulling parameter and the initial cluster center;

[0153] Alternatively, the second distance is determined by calculating the cosine similarity between each historical crystal pulling parameter and the initial cluster center;

[0154] Alternatively, the second distance is determined by calculating the Pearson correlation coefficient between each historical crystal pulling parameter and the initial cluster center.

[0155] In some examples, the determination threshold corresponding to the cluster includes: a threshold of at least one determination variable; wherein each of the determination variables corresponds to any parameter of the current crystal pulling parameters.

[0156] In some examples, the apparatus further includes an adjustment unit configured to:

[0157] After determining that the current crystal pulling parameter will cause twin defects in the crystal pulling process, the current crystal pulling parameter is adjusted to be smaller than a determination threshold corresponding to the closest cluster.

[0158] Please refer to Fig.11 , which shows a structural block diagram of a computing device provided by an exemplary embodiment of the present application. In some examples, the computing device 110 may be at least one of a smart phone, a smart watch, a desktop computer, a laptop, a virtual reality terminal, an augmented reality terminal, a wireless terminal, and a laptop portable computer. The computing device 110 has a communication function and can access a wired network or a wireless network. The computing device 110 may generally refer to one of a plurality of terminals, and those skilled in the art may know that the number of the above terminals may be more or less. In some examples, the computing device 110 may receive data based on the wired network or wireless network to which it is connected. It can be understood that the computing device 110 undertakes the calculation and processing work of the technical solution of the present invention, and the embodiments of the present disclosure are not limited to this.

[0159] like Fig.11 As shown, the computing device in the present application may include one or more of the following components: a processor 1110 and a memory 1120 .

[0160] Optionally, the processor 1110 uses various interfaces and lines to connect various parts of the entire computing device, and executes various functions of the computing device and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 1120, and calling data stored in the memory 1120. Optionally, the processor 1110 can be implemented in at least one hardware form of digital signal processing (DSP), field-programmable gate array (FPGA), and programmable logic array (PLA). The processor 1110 can integrate one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), a neural network processor (NPU), and a baseband chip. Among them, the CPU mainly processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing the content that needs to be displayed on the touch display; the NPU is used to implement artificial intelligence (AI) functions; and the baseband chip is used to process wireless communications. It is understandable that the above-mentioned baseband chip may not be integrated into the processor 1110, but may be implemented by a separate chip.

[0161] The memory 1120 may include a random access memory (RAM) or a read-only memory (ROM). Optionally, the memory 1120 includes a non-transitory computer-readable storage medium. The memory 1120 may be used to store instructions, programs, codes, code sets, or instruction sets. The memory 1120 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above various method embodiments, etc.; the data storage area may store data created according to the use of the computing device, etc.

[0162] In addition, those skilled in the art will appreciate that the structure of the computing device shown in the above figures does not constitute a limitation on the computing device, and the computing device may include more or fewer components than shown, or combine certain components, or arrange the components differently. For example, the computing device also includes a display screen, a camera assembly, a microphone, a speaker, a radio frequency circuit, an input unit, a sensor (such as an acceleration sensor, an angular velocity sensor, a light sensor, etc.), an audio circuit, a WiFi module, a power supply, a Bluetooth module, and other components, which will not be described in detail here.

[0163] An embodiment of the present application further provides a computer-readable storage medium storing at least one instruction, wherein the at least one instruction is used to be executed by a processor to implement the method for detecting twin defects as described in the above embodiments.

[0164] An embodiment of the present application also provides a computer program product, which includes computer instructions, which are stored in a computer-readable storage medium; a processor of a computing device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computing device executes to implement the method for detecting twin defects described in the above-mentioned embodiments.

[0165] Those skilled in the art should be aware that in one or more of the above examples, the functions described in the embodiments of the present application can be implemented with hardware, software, firmware, or any combination thereof. When implemented using software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein the communication media include any media that facilitates the transmission of a computer program from one place to another. The storage medium can be any available medium that a general or special-purpose computer can access.

[0166] It should be noted that the technical solutions described in the embodiments of the present disclosure can be combined arbitrarily without conflict.

[0167] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for detecting twin defects, characterized in that: The method comprises: Performing cluster analysis on historical crystal pulling parameters to obtain at least one cluster and a cluster center of each cluster; wherein the historical crystal pulling parameters include crystal pulling parameters of a crystal rod that has been pulled and has twin defects during a crystal pulling process, and each cluster corresponds to a cause of twin defects; During the current crystal pulling process, the current crystal pulling parameters are collected in real time; Acquire a first distance between the current crystal pulling parameter and a cluster center of each of the clusters; Determine, according to the first distance, a cluster closest to the current crystal pulling parameter; When the current crystal pulling parameter is greater than the judgment threshold corresponding to its closest cluster, it is determined that the current crystal pulling parameter will cause twin defects in the crystal pulling process; wherein each cluster corresponds to at least one parameter factor, and each parameter factor is correspondingly provided with a judgment threshold.

2. The method according to claim 1, characterized in that Before performing cluster analysis on the historical crystal pulling parameters, the method further includes: Obtaining a wafer corresponding to any crystal ingot among the crystal ingots that have been pulled; Measuring whether a crystal rod corresponding to the wafer has twin defects by X-ray diffraction morphology; When it is determined that the crystal rod corresponding to the wafer has a twin defect, the wafer is meshed to obtain at least one mesh containing the twin defect; The crystal pulling parameters of all grids containing twin defects are used as the historical crystal pulling parameters of the crystal rod with twin defects.

3. The method according to claim 2, characterized in that For each of all grids containing twin defects, the crystal pulling parameters of the grid include: position parameters of the grid, pulling acceleration parameters, magnetic field strength parameters of the crystal pulling process, temperature parameters, silicon melt viscosity parameters during the crystal pulling process, silicon melt density parameters during the crystal pulling process, and silicon melt surface tension parameters during the crystal pulling process.

4. The method according to claim 1, characterized in that: The obtaining of a first distance between the current crystal pulling parameter and a cluster center of each of the clusters comprises: Determine the first distance by calculating the second norm between the current crystal pulling parameter and the cluster center of each of the clusters; Alternatively, the first distance is determined by calculating the cosine similarity between the current crystal pulling parameter and the cluster center of each of the clusters; Alternatively, the first distance is determined by calculating the Pearson correlation coefficient between the current crystal pulling parameter and the cluster center of each of the clusters.

5. The method according to claim 1, characterized in that The cluster analysis of the historical crystal pulling parameters is performed to obtain at least one cluster and the cluster center of each cluster, including: Set the number of clusters and set the initial cluster center for each cluster; For each of all the historical crystal pulling parameters, calculating a second distance from an initial cluster center of each cluster; Based on the minimum distance determination criterion and the second distance, determining the closest cluster for each historical crystal pulling parameter; The cluster center of each cluster is determined according to the historical crystal pulling parameters in each cluster.

6. The method according to claim 5, characterized in that The step of calculating, for each of all the historical crystal pulling parameters, a second distance from an initial cluster center of each clustering cluster comprises: Determine the second distance by calculating the second norm of each historical crystal pulling parameter and the initial cluster center; Alternatively, the second distance is determined by calculating the cosine similarity between each historical crystal pulling parameter and the initial cluster center; Alternatively, the second distance is determined by calculating the Pearson correlation coefficient between each historical crystal pulling parameter and the initial cluster center.

7. The method according to claim 1, characterized in that The determination threshold corresponding to the cluster includes: a threshold of at least one determination variable; wherein each determination variable corresponds to any parameter of the current crystal pulling parameters.

8. The method according to any one of claims 1 to 7, characterized in that: After determining that the current crystal pulling parameters will cause twin defects in the crystal pulling process, the method further includes: The current crystal pulling parameter is adjusted to be smaller than a determination threshold corresponding to the closest cluster.

9. A device for detecting twin defects, characterized in that: The device comprises: a cluster analysis unit, a collection unit, an acquisition unit, a first determination unit and a second determination unit; wherein, The cluster analysis unit is configured to perform cluster analysis on historical crystal pulling parameters to obtain at least one cluster and a cluster center of each cluster; wherein the historical crystal pulling parameters include crystal pulling parameters of a crystal rod that has been pulled and has twin defects during a crystal pulling process, and each cluster corresponds to a cause of twin defects; The collecting unit is configured to collect the current crystal pulling parameters in real time during the current crystal pulling process; The acquisition unit is configured to acquire a first distance between the current crystal pulling parameter and a cluster center of each of the clusters; The first determining unit is configured to determine the cluster closest to the current crystal pulling parameter according to the first distance; The second determination unit is configured to determine that the current crystal pulling parameter will cause twin defects in the crystal pulling process when the current crystal pulling parameter is greater than the judgment threshold corresponding to the closest cluster; wherein each cluster corresponds to at least one parameter factor, and each parameter factor is correspondingly provided with a judgment threshold.

10. The device according to claim 9, characterized in that The device further comprises an adjustment unit configured to: After determining that the current crystal pulling parameter will cause twin defects in the crystal pulling process, the current crystal pulling parameter is adjusted to be smaller than a determination threshold corresponding to the closest cluster.

11. A computing device, characterized in that: The computing device comprises: a processor and a memory; the processor is used to execute instructions stored in the memory to implement the method for detecting twin defects as claimed in any one of claims 1 to 8.

12. A computer storage medium, characterized in that: The storage medium stores at least one instruction, and the at least one instruction is used to be executed by a processor to implement the method for detecting twin defects according to any one of claims 1 to 8.

Images