Method and device for determining quality grade of wafer particles

Abstract

An embodiment of the present invention provides a method and device for determining the quality grade of wafer particles. The method obtains binning results of multiple wafer particles, determines at least two important bins based on the binning results, forms multiple bin combinations based on the values ​​of the important bins, determines the PPM of invalid wafer particles in each bin combination and the maximum value of the confidence interval of the PPM based on the binning results of the multiple wafer particles, and determines the quality grade of the wafer particles corresponding to each bin combination based on the maximum value of the confidence interval of the PPM of the multiple bin combinations. This method obtains bin combinations composed of important bins by performing statistical analysis on the binning results of a large number of wafer particles, and forms a corresponding relationship between the bin combination and the quality grade of the wafer particles. The quality grade of the wafer particles to be classified is subsequently predicted based on the corresponding relationship. This method has a higher efficiency and higher prediction accuracy for the quality grade of wafer particles.

Description

Technical Field

[0001] The present invention relates to the field of wafer testing, and in particular to a method and device for determining the quality grade of wafer particles. Background Art

[0002] Chip failures are inevitable during development, production, and use. After wafer production is complete, they undergo testing to verify wafer quality. This primarily involves circuit testing (CP test: chip probe test) and final testing (FT).

[0003] In the CP test, the position that needs to be repaired on each wafer particle is repaired, and the repair type of the position is classified. This classification operation is also called binning, and the types of bins include but are not limited to B, X, d, etc. The number of each repair type in each repaired particle (i.e., the value of each bin) is recorded. The result of binning can be expressed as: (chipID:0203 B:1, X:2, d:3, ...), where chipID represents the identifier of the wafer particle, and the numbers after B, X, and d represent the value of the bin. Wafers that pass the CP test will undergo FT, and will undergo aging (burn in, abbreviated as BI) in FT. The purpose is to provide a high temperature, high voltage, and high current environment for the wafer particles, so that the particles with poor quality will break down as quickly as possible in this environment, retaining the good particles, and then all particles that pass BI will be divided into quality grades. In the existing technology, relatively important bins are mainly obtained manually through experience, and the quality grade of the wafer particles is determined according to the values ​​of the relatively important bins.

[0004] The existing technology uses manual experience to classify wafer particles, which may lead to inaccurate quality grade classification and take a long time. Summary of the Invention

[0005] The present invention provides a method and device for determining the quality grade of wafer particles, which improves the prediction efficiency and prediction accuracy of the quality grade of wafer particles.

[0006] In a first aspect, the present invention provides a method for determining the quality grade of wafer particles, comprising:

[0007] Obtain binning results for multiple wafer particles, where each wafer particle's binning result includes the wafer particle's identifier, bin type, bin values, and data labels.

[0008] Determine at least two important bins according to the binning result, where the important bins have a greater impact on the quality of wafer particles;

[0009] Forming multiple bin combinations based on the values ​​of the important bins, each bin combination consists of the important bins, and at least one important bin has a different value in any two bin combinations;

[0010] Determine, based on the binning results of the plurality of wafer particles, a parts per million (PPM) of invalid wafer particles in each bin combination and a maximum confidence interval of the PPM;

[0011] Determining the wafer particle quality grade corresponding to each bin combination according to the maximum values ​​of the confidence intervals of the PPMs of the multiple bin combinations;

[0012] Each bin value of the wafer particles to be classified is matched with the multiple bin combinations, and the quality grade of the wafer particles corresponding to the successfully matched target bin combination is determined as the quality grade of the wafer particles to be classified.

[0013] Optionally, determining the wafer particle quality grade corresponding to each bin combination according to the maximum values ​​of the confidence intervals of the PPMs of the multiple bin combinations includes:

[0014] Sorting the multiple bin combinations according to the maximum values ​​of the confidence intervals of the multiple bin combinations;

[0015] Dividing the sorted multiple bin combinations into multiple models, each model corresponding to a wafer particle quality level;

[0016] Matching each bin value of the wafer particles to be classified with the multiple bin combinations, and determining the wafer particle quality grade corresponding to the successfully matched target bin combination as the quality grade of the wafer particles to be classified, includes:

[0017] Each bin value of the wafer particles to be classified is matched with the multiple bin combinations, and the quality grade of the wafer particles corresponding to the model to which the successfully matched target bin combination belongs is determined as the quality grade of the wafer particles to be classified.

[0018] Optionally, the number of the bin combinations is M, and the wafer particle quality levels include: valid and invalid;

[0019] The sorting of the multiple bin combinations according to the maximum values ​​of the confidence intervals of the multiple bin combinations includes:

[0020] Sort the multiple bin combinations according to the maximum values ​​of the confidence intervals of the multiple bin combinations from small to large;

[0021] The sorted bin combinations are divided into multiple models, each model corresponding to a wafer particle quality level, including:

[0022] The first M1 bins after sorting are combined to form a first model, and the remaining M-M1 bins are combined to form a second model. The wafer particle quality level corresponding to the first model is valid, and the wafer particle quality level corresponding to the second model is invalid.

[0023] Optionally, before determining at least two important bins according to the binning result, the method further includes:

[0024] For any bin, when the value type of the bin is less than or equal to a preset first threshold, determine not to cluster the bin;

[0025] When the number of values ​​in the bin is greater than the first threshold, the values ​​in the bin are clustered, and the number of values ​​in the bin after clustering is less than or equal to the first threshold.

[0026] Optionally, determining at least two important bins according to the binning result includes:

[0027] According to the data labels of the plurality of wafer particles and the value of each bin or the maximum value of each bin, a random forest algorithm is used to calculate the contribution of each bin, where the contribution is used to indicate the degree of influence of the bin on the quality of the wafer particles;

[0028] N bins with the largest contribution are selected as the important bins, where N is greater than or equal to 2.

[0029] Optionally, after calculating the contribution of each bin using a random forest algorithm according to the quality grades of the plurality of wafer particles and the value of each bin or the maximum value of each bin, the method further includes:

[0030] Calculate the PPM of invalid wafer particles in each bin;

[0031] Draw a correlation graph between the values ​​of each bin and the PPM of invalid wafer particles;

[0032] Based on the correlation diagram between the value of each bin and the PPM of invalid wafer particles, verify whether the contribution of the corresponding bin is valid;

[0033] The selecting N bins with the largest contribution as the important bins includes:

[0034] N bins with the largest contribution are selected from the bins with valid contribution as the important bins.

[0035] Optionally, obtaining binning results of a plurality of wafer particles includes:

[0036] Obtaining circuit test results and final test results of the wafer from a database, wherein the circuit test results include binned data under different test conditions;

[0037] The test data of the same wafer particle is associated through the wafer identification and the wafer particle identification;

[0038] Screening out a plurality of usable wafer particles according to the final test results of the wafer particles;

[0039] According to the bin of the available wafer particles, data labels are added to the available wafer particles.

[0040] Optionally, before determining at least two important bins according to the binning result, the method further includes:

[0041] Remove wafer particles with abnormal bin values.

[0042] Optionally, determining the PPM of invalid wafer particles in each bin combination according to the binning results of the plurality of wafer particles includes:

[0043] According to the binning results of the plurality of wafer particles, counting the total amount of wafer particles and the number of invalid wafer particles in each bin combination;

[0044] The PPM of invalid wafer particles in each bin combination is obtained according to the ratio of the number of invalid wafer particles in each bin combination to the total number of wafer particles.

[0045] A second aspect of the present invention provides a device for determining the quality grade of wafer particles, comprising:

[0046] An acquisition module is used to obtain the binning results of multiple wafer particles. The binning result of each wafer particle includes the identification of the wafer particle, the type of bin, the value of each bin and the data label;

[0047] A first determining module is configured to determine at least two important bins according to the binning result, wherein the important bins are bins that have a greater impact on the quality of wafer particles;

[0048] a combining module, configured to form a plurality of bin combinations according to the values ​​of the important bins, wherein each bin combination is composed of the important bins, and at least one important bin value in any two bin combinations is different;

[0049] A second determining module is configured to determine, based on the binning results of the plurality of wafer particles, a parts per million (PPM) of invalid wafer particles in each bin combination and a maximum confidence interval of the PPM;

[0050] A third determination module is configured to determine the wafer particle quality grade corresponding to each bin combination according to the maximum values ​​of the confidence intervals of the PPMs of the multiple bin combinations;

[0051] The fourth determination module is used to match each bin value of the wafer particles to be classified with the multiple bin combinations, and determine the wafer particle quality level corresponding to the successfully matched target bin combination as the quality level of the wafer particles to be classified.

[0052] In a third aspect, the present invention provides an electronic device comprising: at least one processor and a memory;

[0053] The memory stores computer-executable instructions;

[0054] The at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor performs the method according to the first aspect of the present invention.

[0055] In a fourth aspect, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by a processor, they are used to implement the method described in the first aspect of the present invention.

[0056] In a fifth aspect, the present invention provides a computer program product, comprising a computer program, which, when executed by a processor, implements the method described in the first aspect of the present invention.

[0057] An embodiment of the present invention provides a method and device for determining the quality grade of wafer particles, obtaining binning results of multiple wafer particles, wherein the binning result of each wafer particle includes an identifier of the wafer particle, a bin type, values ​​of each bin, and a data label, determining at least two important bins based on the binning results, forming multiple bin combinations based on the values ​​of the important bins, determining the PPM of invalid wafer particles in each bin combination and the maximum value of the confidence interval of the PPM based on the binning results of the multiple wafer particles, determining the quality grade of the wafer particles corresponding to each bin combination based on the size of the maximum value of the confidence interval of the PPM of the multiple bin combinations, matching each bin value of the wafer particles to be classified with the multiple bin combinations, and determining the quality grade of the wafer particles corresponding to the successfully matched target bin combination as the quality grade of the wafer particles to be classified. This method performs statistical analysis on the binning results of a large number of wafer particles to obtain bin combinations consisting of important bins, and forms a corresponding relationship between the bin combinations and the quality grades of the wafer particles. Subsequently, based on this corresponding relationship and the bin values ​​of the wafer particles to be classified, the quality grades of the wafer particles to be classified can be predicted. This method has higher efficiency and higher prediction accuracy for the quality grades of wafer particles. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0059] Figure 1 This is a flow chart of a method for determining the quality grade of wafer particles provided in the first embodiment of the present invention;

[0060] Figure 2 Schematic diagram of the formation of bin combination;

[0061] Figure 3 Schematic diagram of the PPM of failed wafer particles and the proportion of wafer particles obtained by predicting wafer particle grades using the method of the present invention;

[0062] Figure 4 A flow chart of a method for determining the quality grade of wafer particles provided in the second embodiment of the present invention;

[0063] Figure 5 A schematic diagram of bin value clustering;

[0064] Figure 6 This is a flowchart of a method for determining important bins provided in Example 3 of the present invention;

[0065] Figure 7 This is a schematic diagram of the ranking of the contribution of each bin obtained using the random forest algorithm;

[0066] Figure 8 Schematic diagram of the correlation between the value of each bin and PPM;

[0067] Figure 9 Schematic diagram of the PPM and wafer particle ratio when screening different bins for important bins;

[0068] Figure 10 A schematic diagram of the structure of a device for determining the quality grade of wafer particles provided in a fourth embodiment of the present invention;

[0069] Figure 11 A structural diagram of an electronic device provided in Embodiment 5 of the present invention.

[0070] The above drawings illustrate specific embodiments of the present invention, which will be described in more detail below. These drawings and the accompanying description are not intended to limit the scope of the present invention in any way, but rather to illustrate the concept of the present invention to those skilled in the art by reference to specific embodiments. DETAILED DESCRIPTION

[0071] Exemplary embodiments will be described in detail herein, examples of which are illustrated in the accompanying drawings. In the following description, when referring to the drawings, like numbers in different figures represent like or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all possible embodiments consistent with the present invention. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present invention, as detailed in the appended claims.

[0072] A wafer is made of pure silicon (Si). It typically comes in sizes ranging from 6 inches, 8 inches, and 12 inches. A wafer is the silicon wafer used in the manufacture of silicon semiconductor integrated circuits. Because of its round shape, it's called a wafer.

[0073] After the wafer manufacturing process, it undergoes dicing, testing, and packaging. The final product, a complete and stable die, is also known as a wafer, wafer particle, or particle. A die is an unpackaged bare chip or single unit. Dies that pass testing are packaged to create chips. Dies that pass testing are considered qualified, while those that fail testing are considered scrap.

[0074] CP testing tests every die on an entire wafer. Probes are inserted into the chip pins to test the chip's performance and functionality while the chip is still in the wafer stage. This helps identify bad dies in the wafer, reducing subsequent packaging and testing costs. Bad dies are those that fail quality testing.

[0075] FT is a test for packaged chips. Only the die that passes the CP test (i.e., passes) will be packaged and undergo FT.

[0076] During CP testing, each wafer chip's repaired location is repaired and the repair type is classified. This classification operation is also called binning. Bin types include but are not limited to B, X, and d. The number of each repair type in each repaired chip (i.e., the value of each bin) is recorded. The binning result can be expressed as: (chipID:0203 B:1, X:2, d:3, ...), where chipID represents the wafer chip's identifier, and the numbers after B, X, and d represent the bin values.

[0077] Wafer particles that pass the CP test will undergo FT, and then BI will be performed during FT. The purpose is to provide the wafer particles with a high temperature, high voltage, and high current environment, so that the particles of poor quality will be damaged as quickly as possible in this environment, while the good particles will be retained. All particles that pass BI will then be classified into quality grades. For example, they can be divided into 10 categories, 1, 2, 3, ..., among which wafer particles of grades 1, 2, 3, 4, 5, and 6 are all good dies, and wafer particles of other grades are all bad dies. Good dies are dies of qualified quality, and bad dies are dies of unqualified quality. This is just an example, and the classification of wafer particle quality grades is not limited to the 10 categories mentioned above, and different classification methods can be used.

[0078] Currently, CP testing is primarily performed at three sites: PRE_BI, PRE_HT, and PRE_LT. Each site uses different test conditions and bins the results based on the corresponding test conditions. Specifically, each site performs repairs on wafer particles and bins the repair types. PRE_BI performs burn-in testing, PRE_HT performs high-temperature (HT) testing, and PRE_LT performs low-temperature (LT) testing.

[0079] Currently, engineers primarily rely on experience to determine relatively important bins (for example, PRE_BI: F, I, j, B, X, d; PRE_HT: G, S, K, B06, p; PRE_LT: q, W). Important bins are those that have a significant impact on wafer particle quality. Bins F, I, j, B, X, d are important bins in the PRE_BI test results; G, S, K, B06, p are important bins in the PRE_HT test results; and q, W are important bins in the PRE_LT test results. As can be seen, different test conditions produce different bin types in the binning results.

[0080] Then, the values ​​of each bin are divided empirically, and some wafer particles with relatively good quality are obtained based on the division results. However, the number of wafer particles obtained through division is very small. Currently, the fail ppm of wafer particles obtained through empirical division is 1.9K, and the total proportion of wafer particles is only about 15%. The total proportion of wafer particles refers to the ratio of wafer particles with qualified quality to the total wafer particles obtained after the bin value division. PPM refers to parts per million, and fail ppm refers to the number of failed wafer particles in 1 million wafer particles. A fail ppm of 1.9K means that the number of failed wafer particles in 1 million wafer particles is 1.9K, and 1.9K is 1900. Failed wafer particles are also called failed wafer particles, and failed wafer particles refer to wafer particles with unqualified quality.

[0081] The present invention mainly establishes a statistical model through a large amount of existing data, scientifically and conveniently divides the bin values, and retains the most wafer particles of qualified quality while minimizing the fail ppm.

[0082] Figure 1 This is a flow chart of a method for determining the quality grade of wafer particles provided in the first embodiment of the present invention, such as Figure 1 As shown, the method provided in this embodiment includes the following steps.

[0083] S101 , obtaining binning results of a plurality of wafer particles, wherein the binning result of each wafer particle includes an identifier of the wafer particle, a bin type, values ​​of each bin, and a data label.

[0084] For example, the wafer's CP test results and FT results are obtained from a database. The CP test results include binned data under different test conditions, such as the binned results under the test conditions corresponding to the PRE_BI, PRE_HT, and PRE_LT sites. The test data for the same wafer particle is associated using the wafer ID and the wafer particle ID. Based on the wafer particle's FT results, multiple available wafer particles are screened, and data tags are added to the available wafer particles based on their bins.

[0085] A single wafer may have multiple locations requiring repair, each belonging to a different bin. Binning can be understood as counting the number of locations within a wafer belonging to each bin type. For example, we can count the number of locations belonging to X, D, F, and I within a wafer. B, X, d, F, and I are all bin types, and the number of locations belonging to each bin is the bin value.

[0086] Assuming there are 10 million wafers on a wafer, each wafer is tested for CP, generating binned data for each wafer. The wafer CP test results include binned data for all 10 million wafers under various test conditions. This binned data includes the wafer identifier, bin type, and the values ​​for each bin.

[0087] Wafers that pass the CP test undergo FT, which applies BI to each wafer to quickly destroy low-quality particles under these conditions while preserving the good ones. The same wafer uses the same wafer identifier and wafer ID in both the CP and FT tests. Therefore, the CP and FT test data for the same wafer can be linked using the wafer ID and wafer identifier to form the test data for that wafer.

[0088] The data label of a wafer particle is used to indicate whether the wafer particle passes the test. Optionally, the wafer particles in bins (1-6) in the FTBI are filtered out, and the data label "pass" is added to the wafer particles in bin 6, and the data label "fail" is added to the remaining wafer particles. Pass means the test passed, and fail means the test failed.

[0089] S102 : Determine at least two important bins based on the binning results of the plurality of wafer particles, where the important bins are bins that have a greater impact on the quality of the wafer particles.

[0090] Data sets contain hundreds or even thousands of features, and it's often necessary to select those features that have the greatest impact on the results for further modeling. Related methods include principal component analysis, lasso, and random forests. This example analyzes the binning results of multiple wafer particles and selects the bins that have the greatest impact on wafer particle quality.

[0091] Taking the random forest algorithm as an example, one of its output variables is feature importance, or important features. Important features are those that are closely related to the dependent variable and have a significant impact on changes in the dependent variable. The idea behind using random forests to assess feature importance is relatively simple. It primarily measures the contribution of each feature in each tree in the random forest, then takes the average and compares the contributions of different features. Contribution metrics include the Gini index and the out-of-bag (OOB) error rate.

[0092] In this embodiment, the data label of the wafer particle is used as the dependent variable, and the value of each bin or the maximum value is used as the independent variable. The contribution of each bin is calculated through the random forest model, and the bins are further sorted from large to small according to the contribution. The higher the contribution of the bin, the greater the contribution. Therefore, the top bins are taken as important bins.

[0093] Compared with the prior art method of using empirical values ​​to determine important bins, the important bins obtained by the statistical analysis method of this embodiment are more accurate.

[0094] S103. Form multiple bin combinations according to the values ​​of the important bins. Each bin combination consists of important bins, and at least one important bin in any two bin combinations has a different value.

[0095] A bin combination can consist of all or some important bins. The maximum number of bin combinations formed by important bins is linked to the types of bin values ​​and is equal to the product of the types of values ​​of the important bins that make up the bin combination. For example, if there are two important bins, bin1 has two types of values, and bin2 has three types of values, then the number of bin combinations formed by all important bins is 2*3=6. If there are three important bins, bin1 has two types of values, bin2 has three types of values, and bin3 has five types of values, then the number of bin combinations formed by all important bins is 2*3*5=30. Some bin combinations formed by this method do not exist in reality, so these bin combinations are removed.

[0096] It is understandable that only some important bins can be used to form a bin combination. For example, if there are 3 important bins, only two important bins can be used to form a bin combination, or if there are 5 important bins, only three important bins can be used to form a bin combination.

[0097] In this embodiment, among the multiple bin combinations formed, at least one important bin value in any two bin combinations is different.

[0098] Figure 2 This is a schematic diagram of the formation of bin combination, such as Figure 2 As shown in Table a, the binning results of 20 wafer particles are shown. The first column is the identification of the wafer particle, the second column is the value of bin1, the third column is the value of bin2, and the fourth column is the label. The value of the label is 0 and 1. 0 means pass, that is, the wafer particle is valid, and 1 means fail, that is, the wafer particle is invalid.

[0099] In Table a, bin1 has two possible values: 1 and 2, and bin2 has three possible values: 3, 4, and 5. Replace bin1 value 1 with A1 and bin1 value 2 with A2. Replace bin2 value 3 with B1, bin2 value 4 with B2, and bin3 value with B5 to obtain Table b. It should be understood that A1, A2, A3, B1, B2, and B3 are just examples; the values ​​of bin1 and bin2 can also be replaced with other letters or numbers.

[0100] Counting the various combinations in Table b, there are five bin combinations, as shown in Table c. Any combination of the two values ​​of bin 1 and the three values ​​of bin 2 can form six bin combinations: A1B1, A1B2, A1B3, A2B1, A2B2, and A3B3. However, Table c only has five bin combinations, omitting the bin combination A1B3 because it doesn't exist in Table a.

[0101] In another optional implementation, the bin values ​​in Table a may not be replaced, and the original bin values ​​may be directly used to form bin combinations.

[0102] In another optional implementation, various possible bin combinations are formed directly according to the bin value types in table a. For example, any combination of the two values ​​of bin1 and the three values ​​of bin2 can form 6 bin combinations. There is no need to consider whether each bin combination actually exists. If a formed bin combination does not exist in table a, then the statistical values ​​of the bin combination are all 0, and the statistical values ​​of other bin combinations are actual values.

[0103] S104 : Determine the parts per million (PPM) of invalid wafer particles in each bin combination and the maximum value of the confidence interval of the PPM based on the binning results of the plurality of wafer particles.

[0104] After forming the bin combination, the total number of wafer particles and the number of invalid wafer particles in each bin combination are counted based on the binning results of multiple wafer particles. The PPM of invalid wafer particles in each bin combination is obtained based on the ratio of the number of invalid wafer particles in each bin combination to the total number of wafer particles. The maximum confidence interval of the PPM can be obtained based on the PPM of the invalid wafer particles.

[0105] The maximum value of the PPM confidence interval is used to indicate the reliability of the PPM. The smaller the maximum value of the PPM confidence interval, the greater the reliability of the PPM. For example, if one wafer die fails in 100 wafers, the PPM of the former is larger than that of the latter in 1000 wafers, but the reliability of the latter is better. The confidence interval of the PPM can be a Bernoulli confidence interval.

[0106] Still Figure 2 For example, the total number of wafer particles with bin combination A1B2 in 20 wafer particles is 8, and the number of failed wafer particles in A1B2 combination is 1. The PPM of failed wafer particles in A1B2 combination = (ratio of number of failed wafer particles / total number of wafer particles) * 1000000 = (1 / 8) * 1000000 = 125000. The same method is used to obtain the statistical data of other bin combinations in turn to form Figure 2 The bin combination list shown in Table c.

[0107] Figure 2 The bin combination list shown in Table c includes the total number of wafer particles, the number of failed wafer particles, the PPM of failed wafer particles, and PPMUL, where PPMUL is the maximum confidence interval value of PPM.

[0108] S105 : Determine the wafer particle quality grade corresponding to each bin combination according to the maximum values ​​of the confidence intervals of the PPMs of the multiple bin combinations.

[0109] In this embodiment, the quality grades of wafer particles are divided according to the maximum value of the confidence interval of PPM, forming a correspondence between bin combinations and quality grades. This correspondence can reflect the intrinsic connection between the bin value and the quality grade. The correspondence between all bin combinations and quality grades can also be understood as a statistical model. Subsequently, the quality grade of wafer particles can be predicted based on this statistical model and the bin value of the wafer particles.

[0110] Compared to categorizing quality levels based on PPM, categorizing quality levels based on the confidence interval of the PPM of a bin combination is more accurate. After categorizing quality levels for multiple bin combinations, multiple bin combinations may correspond to the same quality level. For example, if there are 10 bin combinations, bin combinations 1-3 correspond to quality level 1, bin combinations 4-5 correspond to quality level 2, bin combinations 6-7 correspond to quality level 3, and bin combinations 8-10 correspond to quality level 4.

[0111] S106 , matching each bin value of the wafer particles to be classified with multiple bin combinations, and determining the wafer particle quality grade corresponding to the successfully matched target bin combination as the quality grade of the wafer particles to be classified.

[0112] Specifically, based on the binning results of the wafer particles to be classified, it is determined whether the values ​​of each bin of the wafer particles to be classified are the same as the values ​​of the bins of a certain bin combination. If they are the same, it is considered that the wafer particles to be classified match the bin combination, and accordingly, the quality levels of the two are also the same, so the quality level corresponding to the matching bin combination is used as the quality level of the wafer particles to be classified.

[0113] In steps S101-S105, a large amount of data analysis can be used to obtain the correspondence between each bin combination and the quality grade. Subsequently, prediction is performed based on the correspondence, and the prediction efficiency is very high. In addition, this embodiment divides the quality grade according to the maximum value of the confidence interval of the PPM of each bin combination, and the correspondence between the bin combination and the quality grade is more accurate, thereby improving the prediction accuracy of the wafer particle quality grade.

[0114] Figure 3 Schematic diagram of the PPM of failed wafer particles and the proportion of wafer particles obtained by using the method of the present invention to predict the wafer particle grade, as shown in FIG. Figure 3 As shown, the larger the PPM of failed wafer particles, the larger the proportion of wafer particles. When the PPM of failed wafer particles is controlled to 1.9K, the proportion of wafer particles screened out by the method of the present invention is about 80%, while the proportion of wafer particles screened out by engineers based on experience can only reach about 15%.

[0115] In this embodiment, the binning results of multiple wafer particles are obtained, and the binning results of each wafer particle include the identification of the wafer particle, the type of bin, the value of each bin, and the data label. At least two important bins are determined based on the binning results. Multiple bin combinations are formed based on the values ​​of the important bins. Based on the binning results of the multiple wafer particles, the PPM of the invalid wafer particles of each bin combination and the maximum value of the confidence interval of the PPM are determined. According to the maximum value of the confidence interval of the PPM of the multiple bin combinations, the quality grade of the wafer particles corresponding to each bin combination is determined. The bin values ​​of the wafer particles to be classified are matched with the multiple bin combinations, and the quality grade of the wafer particles corresponding to the target bin combination that is successfully matched is determined as the quality grade of the wafer particles to be classified. This method obtains a bin combination composed of important bins by statistically analyzing the binning results of a large number of wafer particles, and forms a corresponding relationship between the bin combination and the quality grade of the wafer particles. Subsequently, based on the corresponding relationship and the bin value of the wafer particles to be classified, the quality grade of the wafer particles to be classified can be predicted. This method has a higher prediction efficiency and a higher prediction accuracy for the quality grade of the wafer particles.

[0116] Based on the first embodiment, the second embodiment of the present invention provides a method for determining the quality grade of wafer particles. The implementation of the same method steps refers to the description of the first embodiment and will not be repeated in this embodiment. Figure 4 This is a flow chart of a method for determining the quality grade of wafer particles provided in the second embodiment of the present invention, as shown in FIG. Figure 4 As shown, the method provided in this embodiment includes the following steps.

[0117] S201 , obtaining binning results of a plurality of wafer particles, wherein the binning result of each wafer particle includes an identifier of the wafer particle, a bin type, values ​​of each bin, and a data label.

[0118] S202. For any bin, when the value types of the bin are less than or equal to a preset first threshold, determine not to cluster the bin; when the value types of the bin are greater than the first threshold, cluster the bin values, and the value types of the bin after clustering are less than or equal to the first threshold.

[0119] Optionally, before determining the important bins, this embodiment determines whether the bins of wafer particles need to be clustered. For any bin, when the number of bin value types is less than or equal to a preset first threshold, it is determined not to cluster the bin. When the number of bin value types is greater than the first threshold, the bin values ​​are clustered, and the number of bin value types after clustering is less than or equal to the first threshold.

[0120] Among 10 million wafer particles, some bins have very few value types. For example, bin 1 has only two values: 0 and 1. Some bins may have many value types, such as 15, 20, 30 or more. Figure 5 A schematic diagram of bin value clustering, such as Figure 5 As shown, bin1 has a total of 8 values, which vary from wafer to wafer. Some wafer particles have similar values ​​for bin1, so they can be clustered. For example, the values ​​1, 2, and 2 of wafer particles 1-3 are clustered into one category, the values ​​10, 9, and 8 of wafer particles 4-6 are clustered into one category, and the values ​​24, 25, and 26 of wafer particles 7-9 are clustered into one category. Figure 5 This is just an example. In actual applications, the number of wafer particles is in the millions or tens of millions, and the bin values ​​of all wafer particles may be much larger, so clustering is needed.

[0121] The first threshold is 10, for example. When the number of value types in a bin is less than or equal to 10, clustering is not required. When the number of value types in a bin is greater than 10, the values ​​of the bin are clustered, and clustering makes the number of value types in the bin less than or equal to 10.

[0122] Any existing clustering method may be used for clustering, for example, K-mean clustering may be used for clustering, and this embodiment does not limit this.

[0123] Optionally, before clustering, wafer particles with abnormal bin values ​​are removed. For example, for a certain bin, most of its values ​​are between 100-200. When a value of 10000 appears, it is considered an abnormal value, and the wafer particle corresponding to the abnormal value is removed.

[0124] S203 : Determine at least two important bins according to the binning result, where the important bins have a greater impact on the quality of wafer particles.

[0125] S204. Form multiple bin combinations according to the values ​​of the important bins. Each bin combination consists of important bins, and at least one important bin in any two bin combinations has a different value.

[0126] S205 : Determine the PPM of invalid wafer particles in each bin combination and the maximum value of the confidence interval of the PPM according to the binning results of the plurality of wafer particles.

[0127] S206 , sorting the multiple bin combinations from small to large according to the maximum values ​​of their confidence intervals.

[0128] The smaller the maximum value of the PPM confidence interval is, the better the quality of the corresponding wafer particles is.

[0129] S207 , dividing the sorted multiple bin combinations into multiple models, each model corresponding to a wafer particle quality level.

[0130] In this embodiment, when determining the wafer particle quality grade corresponding to each bin combination based on the maximum value of the confidence interval of the PPM of multiple bin combinations, the method of steps S206 and S207 is adopted. Among them, among the multiple bin combinations after sorting, the higher the bin combination is, the better the quality of the wafer particles corresponding to it. The multiple bin combinations are divided into multiple models according to the sorting. The sorting of the bin combinations in the same model is continuous. For example, there are 10 bin combinations in total. Then the bin combination ranked first can be divided into model 1, the bin combinations ranked 2-4 can be divided into model 2, the bin combinations ranked 5-8 can be divided into model 3, and the bin combinations ranked 9-10 can be divided into model 4. Define model 1 to correspond to wafer particle quality grade 1, model 2 to wafer particle quality grade 2, model 3 to wafer particle quality grade 3, and model 4 to wafer particle quality grade 4. Among them, the higher the model is in the sorting, the better the quality of the wafer particles corresponding to it. Optionally, wafer particles with quality grades 1 and 2 are valid, and wafer particles with quality grades 3 and 4 are invalid.

[0131] S208 , matching each bin value of the wafer particles to be classified with multiple bin combinations, and determining the wafer particle quality grade corresponding to the model to which the successfully matched target bin combination belongs as the quality grade of the wafer particles to be classified.

[0132] Compare the bin values ​​of the wafer particles to be classified with the values ​​of the multiple bin combinations. If the bin values ​​of the wafer particles to be classified are equal to the value of a certain bin combination, the quality level of the wafer particles corresponding to the model to which the bin combination belongs is determined as the quality level of the wafer particles to be classified.

[0133] Assume that the number of bin combinations is M, and the wafer particle quality levels include: valid and invalid. Accordingly, after sorting the multiple bin combinations from small to large according to the maximum value of the confidence interval of the multiple bin combinations, the first M1 bin combinations after sorting are taken to form the first model, and the remaining M-M1 bin combinations form the second model. The wafer particle quality level corresponding to the first model is valid, and the wafer particle quality level corresponding to the second model is invalid.

[0134] It can be understood that in other embodiments of the present invention, multiple bin combinations can also be sorted from large to small according to the maximum value of the confidence interval of multiple bin combinations. Accordingly, among the sorted multiple bin combinations, the quality of the wafer particles corresponding to the later bin combinations is better.

[0135] In this embodiment, before determining the important bins, whether to cluster the bins is determined based on the number of value types in each bin. Bins with many value types are clustered, and bins with few value types are not clustered, thereby reducing the amount of data calculations.

[0136] Based on the first and second embodiments, the third embodiment of the present invention provides a method for determining an important bin. Figure 6 This is a flow chart of a method for determining an important bin provided in the third embodiment of the present invention, as shown in FIG. Figure 6 As shown, the method provided in this embodiment includes the following steps.

[0137] S301 , using a random forest algorithm to calculate the contribution of each bin based on data labels of multiple wafer particles and the value of each bin or the maximum value of each bin.

[0138] S302 , sorting the bins from large to small according to their contribution.

[0139] Optionally, the contribution of each bin is sorted, and the contribution sorting is the sorting of the importance of the bin. Figure 7 This is a schematic diagram of the ranking of the contribution of each bin obtained using the random forest algorithm. Figure 7 As shown, the contribution of bin i, j, d, G, F, S, B, K, X, q, p, w, and y gradually decreases. Figure 7 The contribution of the bin shown is obtained through a random forest model with the data label of the wafer particle (i.e., the fail or pass label of the wafer particle) as the dependent variable and the value of each bin as the independent variable.

[0140] Optionally, according to the sorting diagram, the first 4 bins or the first 5 bins can be selected as important bins.

[0141] S303: Calculate the PPM of invalid wafer particles in each bin.

[0142] S304 , draw a correlation graph between the value of each bin and the PPM of invalid wafer particles.

[0143] Figure 8 This is a schematic diagram of the correlation between the values ​​of each bin and PPM. Figure 8, which shows the correlation diagrams of bin G, X, I, d, q and PPM in sequence, wherein the value of each bin is positively correlated with PPM, and in actual process, the value of the bin may also be negatively correlated with PPM.

[0144] S305 : Verify whether the contribution of the corresponding bin is valid based on a correlation diagram between the value of each bin and the PPM of invalid wafer particles.

[0145] The correlation graph between the bin value and PPM is used to verify whether the bin contribution is valid. If the bin contribution is large (i.e., the ranking is relatively high) and the bin value is related to PPM, then the bin contribution is determined to be valid. If the bin contribution is large, but the bin value is not related to PPM, then the bin contribution is determined to be invalid.

[0146] S306 : Select N bins with the largest contribution from the bins with valid contribution as important bins.

[0147] Figure 9 A diagram showing the ratio of PPM to wafer particles when screening different bins for important bins. Figure 9 The figure shows the relationship between PPM and wafer particle ratio when all bins, FiiBXdG, FiiBXdGS, FiiBXdGSK or FiiBXdS are selected as important bins. Among them, when FiiBXdGS is used as the important bin, the wafer particle ratio is the best. When the PPM is 1K, the wafer particle ratio is 55%.

[0148] In this embodiment, a random forest algorithm is used to select important bins from multiple bins, and a correlation diagram between the bin values ​​and PPM is used to further verify whether the order of the bin contributions is correct or effective, thereby ensuring the accuracy of the selected important bins.

[0149] Figure 10 This is a schematic diagram of the structure of the device for determining the quality grade of wafer particles provided in the fourth embodiment of the present invention, as shown in FIG. Figure 10 As shown, the device 100 provided in this embodiment includes.

[0150] An acquisition module 11 is used to obtain binning results of multiple wafer particles, where the binning result of each wafer particle includes the identification of the wafer particle, the bin type, the value of each bin, and a data label;

[0151] A first determining module 12 is configured to determine at least two important bins according to the binning result, wherein the important bins are bins that have a greater impact on the quality of wafer particles;

[0152] a combining module 13, configured to form a plurality of bin combinations according to the values ​​of the important bins, wherein each bin combination is composed of the important bins, and at least one important bin value in any two bin combinations is different;

[0153] A second determining module 14 is configured to determine the parts per million (PPM) of invalid wafer particles in each bin combination and a maximum confidence interval of the PPM according to the binning results of the plurality of wafer particles;

[0154] A third determining module 15 is configured to determine the wafer particle quality grade corresponding to each bin combination according to the maximum values ​​of the confidence intervals of the PPMs of the multiple bin combinations;

[0155] The fourth determination module 16 is used to match each bin value of the wafer particles to be classified with the multiple bin combinations, and to match the wafer particles corresponding to the successfully matched target bin combination.

[0156] Optionally, the third determining module 15 is specifically configured to:

[0157] Sorting the multiple bin combinations according to the maximum values ​​of the confidence intervals of the multiple bin combinations;

[0158] Dividing the sorted multiple bin combinations into multiple models, each model corresponding to a wafer particle quality level;

[0159] The fourth determination module 16 is specifically used to match each bin value of the wafer particles to be classified with the multiple bin combinations, and determine the wafer particle quality grade corresponding to the model to which the successfully matched target bin combination belongs as the quality grade of the wafer particles to be classified.

[0160] Optionally, the number of the bin combinations is M, and the wafer particle quality levels include: valid and invalid;

[0161] The sorting of the multiple bin combinations according to the maximum values ​​of the confidence intervals of the multiple bin combinations includes:

[0162] Sort the multiple bin combinations according to the maximum values ​​of the confidence intervals of the multiple bin combinations from small to large;

[0163] The sorted bin combinations are divided into multiple models, each model corresponding to a wafer particle quality level, including:

[0164] The first M1 bins after sorting are combined to form a first model, and the remaining M-M1 bins are combined to form a second model. The wafer particle quality level corresponding to the first model is valid, and the wafer particle quality level corresponding to the second model is invalid.

[0165] Optionally, a clustering module is also included, which is used to: for any bin, when the value type of the bin is less than or equal to a preset first threshold, determine not to cluster the bin; when the value type of the bin is greater than the first threshold, cluster the values ​​of the bin, and the value type of the bin after clustering is less than or equal to the first threshold.

[0166] Optionally, the first determining module 12 is specifically configured to:

[0167] According to the data labels of the plurality of wafer particles and the value of each bin or the maximum value of each bin, a random forest algorithm is used to calculate the contribution of each bin, where the contribution is used to indicate the degree of influence of the bin on the quality of the wafer particles;

[0168] N bins with the largest contribution are selected as the important bins, where N is greater than or equal to 2.

[0169] Optionally, the first determining module 12 is further configured to:

[0170] Calculate the PPM of invalid wafer particles in each bin;

[0171] Draw a correlation graph between the values ​​of each bin and the PPM of invalid wafer particles;

[0172] Based on the correlation diagram between the value of each bin and the PPM of invalid wafer particles, verify whether the contribution of the corresponding bin is valid;

[0173] N bins with the largest contribution are selected from the bins with valid contribution as the important bins.

[0174] Optionally, the acquisition module 11 is specifically configured to:

[0175] Obtaining circuit test results and final test results of the wafer from a database, wherein the circuit test results include binned data under different test conditions;

[0176] The test data of the same wafer particle is associated through the wafer identification and the wafer particle identification;

[0177] Screening out a plurality of usable wafer particles according to the final test results of the wafer particles;

[0178] According to the bin of the available wafer particles, data labels are added to the available wafer particles.

[0179] Optionally, an abnormality removal module is also included to remove wafer particles with abnormal bin values.

[0180] Optionally, the second determining module 14 is specifically configured to:

[0181] According to the binning results of the plurality of wafer particles, counting the total amount of wafer particles and the number of invalid wafer particles in each bin combination;

[0182] The PPM of invalid wafer particles in each bin combination is obtained according to the ratio of the number of invalid wafer particles in each bin combination to the total number of wafer particles.

[0183] The device of this embodiment can be used to execute the method described in any one of the above embodiments 1 to 3. The specific implementation method and technical effects are similar and will not be described again here.

[0184] It should be noted that, when the device provided in the above embodiment executes the data analysis and processing method, the division of the above functional modules is only used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0185] Figure 11 A structural diagram of an electronic device provided in Embodiment 5 of the present invention.

[0186] Device 200 may include one or more of the following components: a processing component 202 , a memory 204 , a power component 206 , a multimedia component 208 , an audio component 210 , an input / output (I / O) interface 212 , a sensor component 214 , and a communication component 216 .

[0187] The processing component 202 generally controls the overall operation of the device 200, such as operations associated with display, phone calls, data communications, camera operation, and recording operations. The processing component 202 may include one or more processors 220 to execute instructions to perform all or part of the steps of the above-described method. In addition, the processing component 202 may include one or more modules to facilitate interaction between the processing component 202 and other components. For example, the processing component 202 may include a multimedia module to facilitate interaction between the multimedia component 208 and the processing component 202.

[0188] The memory 204 is configured to store various types of data to support operations on the device 200. Examples of such data include instructions for any application or method operating on the device 200, contact data, phone book data, messages, pictures, videos, etc. The memory 204 can be implemented by any type of volatile or non-volatile storage device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk.

[0189] The power supply component 206 provides power to the various components of the device 200. The power supply component 206 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the device 200.

[0190] The multimedia component 208 includes a screen that provides an output interface between the device 200 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen can be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor can not only sense the boundaries of the touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 208 includes a front camera and / or a rear camera. When the device 200 is in an operating mode, such as a shooting mode or a video mode, the front camera and / or the rear camera can receive external multimedia data. Each front camera and rear camera can be a fixed optical lens system or have a focal length and optical zoom capability.

[0191] The audio component 210 is configured to output and / or input audio signals. For example, the audio component 210 includes a microphone (MIC) that is configured to receive external audio signals when the device 200 is in an operating mode, such as a call mode, a recording mode, and a voice recognition mode. The received audio signal may be further stored in the memory 204 or transmitted via the communication component 216. In some embodiments, the audio component 210 further includes a speaker for outputting audio signals.

[0192] I / O interface 212 provides an interface between processing component 202 and peripheral interface modules, such as a keyboard, click wheel, buttons, etc. These buttons may include but are not limited to: a home button, volume buttons, a start button, and a lock button.

[0193] The sensor assembly 214 includes one or more sensors for providing various aspects of the status assessment of the device 200. For example, the sensor assembly 214 can detect the open / closed state of the device 200, the relative positioning of components, such as the display and keypad of the device 200. The sensor assembly 214 can also detect changes in the position of the device 200 or a component of the device 200, the presence or absence of user contact with the device 200, the orientation or acceleration / deceleration of the device 200, and temperature changes of the device 200. The sensor assembly 214 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. The sensor assembly 214 may also include an optical sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor assembly 214 may also include an accelerometer, a gyroscope, a magnetic sensor, a pressure sensor, or a temperature sensor.

[0194] The communication component 216 is configured to facilitate wired or wireless communication between the device 200 and other devices. The device 200 can access a wireless network based on a communication standard, such as WiFi, 2G or 3G, or a combination thereof. In an exemplary embodiment, the communication component 216 receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 216 also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology and other technologies.

[0195] In an exemplary embodiment, the device 200 can be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the method described in any of the above embodiments 1 to 3.

[0196] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as the memory 204 including instructions. The instructions can be executed by the processor 220 of the device 200 to perform the above method. For example, the non-transitory computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc. When the instructions in the storage medium are executed by the processor of the electronic device, the electronic device can perform the method described in any one of the above embodiments 1 to 3.

[0197] An embodiment of the present invention further provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements the method described in the above embodiment 1. The specific implementation method and technical effects are similar and will not be repeated here.

[0198] Other embodiments of the present invention will readily occur to those skilled in the art after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, with the true scope and spirit of the invention being indicated by the following claims.

[0199] It should be understood that the present invention is not limited to the exact construction described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof, which is limited only by the appended claims.

Claims

1. A method for determining the quality grade of wafer particles, characterized in that: include: Obtain binning results for multiple wafer particles, where each wafer particle's binning result includes the wafer particle's identifier, bin type, bin values, and data labels. Determine at least two important bins according to the binning result, where the important bins have a greater impact on the quality of wafer particles; Forming multiple bin combinations based on the values ​​of the important bins, each bin combination consists of the important bins, and at least one important bin has a different value in any two bin combinations; Determine, based on the binning results of the plurality of wafer particles, a parts per million (PPM) of invalid wafer particles in each bin combination and a maximum confidence interval of the PPM; Determining the wafer particle quality grade corresponding to each bin combination according to the maximum values ​​of the confidence intervals of the PPMs of the multiple bin combinations; Each bin value of the wafer particles to be classified is matched with the multiple bin combinations, and the quality grade of the wafer particles corresponding to the successfully matched target bin combination is determined as the quality grade of the wafer particles to be classified.

2. The method according to claim 1, characterized in that The step of determining the wafer particle quality grade corresponding to each bin combination according to the maximum values ​​of the confidence intervals of the PPMs of the multiple bin combinations includes: Sorting the multiple bin combinations according to the maximum values ​​of the confidence intervals of the multiple bin combinations; Dividing the sorted multiple bin combinations into multiple models, each model corresponding to a wafer particle quality level; Matching each bin value of the wafer particles to be classified with the multiple bin combinations, and determining the wafer particle quality grade corresponding to the successfully matched target bin combination as the quality grade of the wafer particles to be classified, includes: Each bin value of the wafer particles to be classified is matched with the multiple bin combinations, and the quality grade of the wafer particles corresponding to the model to which the successfully matched target bin combination belongs is determined as the quality grade of the wafer particles to be classified.

3. The method according to claim 2, characterized in that The number of bin combinations is M, and the wafer particle quality levels include: valid and invalid; The sorting of the multiple bin combinations according to the maximum values ​​of the confidence intervals of the multiple bin combinations includes: Sort the multiple bin combinations according to the maximum values ​​of the confidence intervals of the multiple bin combinations from small to large; The sorted bin combinations are divided into multiple models, each model corresponding to a wafer particle quality level, including: The first M1 bins after sorting are combined to form a first model, and the remaining M-M1 bins are combined to form a second model. The wafer particle quality level corresponding to the first model is valid, and the wafer particle quality level corresponding to the second model is invalid.

4. The method according to any one of claims 1 to 3, characterized in that Before determining at least two important bins according to the binning result, the method further includes: For any bin, when the value type of the bin is less than or equal to a preset first threshold, determine not to cluster the bin; When the number of values ​​in the bin is greater than the first threshold, the values ​​in the bin are clustered, and the number of values ​​in the bin after clustering is less than or equal to the first threshold.

5. The method according to claim 4, characterized in that The determining of at least two important bins according to the binning result includes: According to the data labels of the plurality of wafer particles and the value of each bin or the maximum value of each bin, a random forest algorithm is used to calculate the contribution of each bin, where the contribution is used to indicate the degree of influence of the bin on the quality of the wafer particles; N bins with the largest contribution are selected as the important bins, where N is greater than or equal to 2.

6. The method according to claim 5, characterized in that After calculating the contribution of each bin using a random forest algorithm based on the quality grades of the plurality of wafer particles and the value of each bin or the maximum value of each bin, the method further includes: Calculate the PPM of invalid wafer particles in each bin; Draw a correlation graph between the values ​​of each bin and the PPM of invalid wafer particles; Based on the correlation diagram between the value of each bin and the PPM of invalid wafer particles, verify whether the contribution of the corresponding bin is valid; The selecting N bins with the largest contribution as the important bins includes: N bins with the largest contribution are selected from the bins with valid contribution as the important bins.

7. The method according to any one of claims 1 to 3, characterized in that The obtaining of binning results of a plurality of wafer particles includes: Obtaining circuit test results and final test results of the wafer from a database, wherein the circuit test results include binned data under different test conditions; The test data of the same wafer particle is associated through the wafer identification and the wafer particle identification; Screening out a plurality of usable wafer particles according to the final test results of the wafer particles; According to the bin of the available wafer particles, data labels are added to the available wafer particles.

8. The method according to claim 7, characterized in that Before determining at least two important bins according to the binning result, the method further includes: Remove wafer particles with abnormal bin values.

9. The method according to any one of claims 1 to 3, characterized in that The step of determining the PPM of invalid wafer particles in each bin combination according to the binning results of the plurality of wafer particles includes: According to the binning results of the plurality of wafer particles, counting the total amount of wafer particles and the number of invalid wafer particles in each bin combination; The PPM of invalid wafer particles in each bin combination is obtained according to the ratio of the number of invalid wafer particles in each bin combination to the total number of wafer particles.

10. A device for determining the quality grade of wafer particles, characterized in that: include: An acquisition module is used to obtain the binning results of multiple wafer particles. The binning result of each wafer particle includes the identification of the wafer particle, the type of bin, the value of each bin and the data label; A first determining module is configured to determine at least two important bins according to the binning result, wherein the important bins are bins that have a greater impact on the quality of wafer particles; a combining module, configured to form a plurality of bin combinations according to the values ​​of the important bins, wherein each bin combination is composed of the important bins, and at least one important bin value in any two bin combinations is different; A second determining module is configured to determine, based on the binning results of the plurality of wafer particles, a parts per million (PPM) of invalid wafer particles in each bin combination and a maximum confidence interval of the PPM; A third determination module is configured to determine the wafer particle quality grade corresponding to each bin combination according to the maximum values ​​of the confidence intervals of the PPMs of the multiple bin combinations; The fourth determination module is used to match each bin value of the wafer particles to be classified with the multiple bin combinations, and determine the wafer particle quality level corresponding to the successfully matched target bin combination as the quality level of the wafer particles to be classified.

11. An electronic device, characterized in that: include: at least one processor and memory; The memory stores computer-executable instructions; The at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor performs the method according to any one of claims 1 to 9.

12. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 9 when executed by a processor.

13. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 9 is implemented.

Images