Abnormality detection method and device, electronic device, and storage medium

Abstract

This disclosure relates to an anomaly detection method, apparatus, electronic device, and computer-readable storage medium, belonging to the field of semiconductor manufacturing testing technology, and can be applied to scenarios where the causes of anomalies occur during wafer testing. The method includes: acquiring test data of the test subject; determining at least one test indicator corresponding to the test data; the test indicator includes one or more of yield parameters, resistance parameters, and failure bit regions; determining the anomaly judgment result of the test subject under each test indicator; and determining the anomaly cause type of the test subject based on the determined anomaly judgment result. This disclosure, by analyzing wafer testing results from multiple dimensions, quickly analyzes the causes of test performance anomalies and provides timely warnings.

Description

Technical Field

[0001] This disclosure relates to the field of semiconductor manufacturing testing technology, and more specifically, to an anomaly detection method, an anomaly detection device, an electronic device, and a computer-readable storage medium. Background Technology

[0002] Semiconductor manufacturing processes include wafer fabrication and packaging / testing, each requiring separate wafer probing (CP) and final testing (FT). In testing equipment, testers are used to inspect chip functionality and performance, while probe stations and sorters connect the chip under test to the tester's functional modules. The wafer inspection stage requires both testers and probe stations.

[0003] If the testing equipment malfunctions, such as a probe card malfunction (e.g., pin malfunction) or a machine malfunction (e.g., a production card malfunction), it will cause product damage or measurement errors, further leading to a loss of yield.

[0004] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this disclosure is to provide an anomaly detection method, anomaly detection device, electronic device, and computer-readable storage medium, thereby overcoming, to at least a certain extent, the problem that existing anomaly detection methods require determining the cause of anomalies based on wafer test results and personal work experience, leading to the need for product retesting and a time-consuming anomaly finding process.

[0006] Other features and advantages of this disclosure will become apparent from the following detailed description, or may be learned in part by practice of the invention.

[0007] According to a first aspect of this disclosure, an anomaly detection method is provided, comprising: acquiring test data of a test subject; determining at least one test indicator corresponding to the test data; the test indicator including one or more of yield parameters, resistance parameters, and failure bit regions; determining anomaly judgment results of the test subject under each of the test indicators; and determining the anomaly cause type of the anomaly occurring in the test subject based on the determined anomaly judgment results.

[0008] In one exemplary embodiment of this disclosure, the method further includes: determining the test machine corresponding to the test subject, and obtaining the test log file of the test machine; performing structured processing on the test log file to generate initial test data; and performing data preprocessing on the initial test data to obtain the test data.

[0009] In one exemplary embodiment of this disclosure, the step of preprocessing the initial test data to obtain the test data includes: obtaining at least one group identifier; the group identifier being used to divide the initial test data; dividing the initial test data according to the at least one group identifier to obtain multiple test points; the test points including initial point test data; determining the data to be deleted for each of the initial point test data under different test parameters; filtering the data to be deleted from the multiple initial point test data to obtain multiple point test data; and generating the test data based on the multiple point test data.

[0010] In one exemplary embodiment of this disclosure, the test subject includes a wafer, the test index includes the yield parameter, and determining the anomaly judgment result of the test subject under each of the test indexes includes: obtaining at least one test failure parameter of the test subject in the subject frame dimension; the test subject includes multiple test points; determining the total number of wafers contained in each of the test points; determining the number of failed wafers corresponding to each of the test points under each of the test failure parameters; and determining the anomaly judgment result of the test subject under the yield parameter based on the total number of wafers and the number of failed wafers, as a first anomaly judgment result.

[0011] In one exemplary embodiment of this disclosure, determining the anomaly determination result of the test subject under the yield parameter as the first anomaly determination result based on the total number of wafers and the number of failed wafers includes: determining the proportion of failed wafers corresponding to the test subject based on the total number of wafers and the number of failed wafers; comparing the total number of wafers with a wafer number threshold to obtain a first comparison result; comparing the proportion of failed wafers with a proportion threshold to obtain a second comparison result; and determining the first anomaly determination result based on the first comparison result and the second comparison result.

[0012] In one exemplary embodiment of this disclosure, the test subject includes a wafer, the test index includes the resistance parameter, and determining the anomaly judgment result of the test subject under each of the test indexes includes: acquiring specified test data of the test subject under the resistance parameter within a specified time period; acquiring a specified group identifier, and grouping the specified test data according to the specified group identifier to obtain resistance group data; acquiring the point test results corresponding to each wafer under all test points, and determining the resistance threshold corresponding to the resistance group data according to the point test results; and determining the anomaly judgment result of the test subject under the resistance parameter according to the resistance threshold, as a second anomaly judgment result.

[0013] In one exemplary embodiment of this disclosure, determining the abnormality judgment result of the test subject under the resistance parameter based on the resistance threshold as the second abnormality judgment result includes: determining the total number of wafers in each test point and the number of abnormal wafers in each test point whose test results are greater than the resistance threshold; determining the abnormal wafer ratio based on the total number of wafers and the number of abnormal wafers; obtaining a parameter mean fitting line and determining the fitting line derivative corresponding to the parameter mean fitting line; and determining the second abnormality judgment result based on the total number of wafers, the abnormal wafer ratio, and the fitting line derivative.

[0014] In one exemplary embodiment of this disclosure, the test subject includes a wafer, the test index includes a failure bit region, and determining the anomaly judgment result of the test subject under each of the test indexes includes: determining the test parameters contained in each test point of the test subject; grouping the test points according to the test parameters to obtain grouped test points; determining the total number of wafers at each of the grouped test points; dividing the bit failure map corresponding to each of the grouped test points to obtain multiple matrix regions and generating a corresponding region list; determining the failure energy density of each of the matrix regions in the region list; and determining the anomaly judgment result of the test subject under the failure bit region based on the total number of wafers at the points and the multiple failure energy densities, as a third anomaly judgment result.

[0015] In one exemplary embodiment of this disclosure, determining the anomaly determination result of the test subject in the failure bit region as a third anomaly determination result based on the total number of wafers at the test site and multiple failure energy densities includes: defining the matrix region where the failure energy density is greater than the energy density threshold as a failure matrix region and saving the failure matrix region to a failure region list; deleting the secondary matrix regions contained in the failure matrix region from the region list; performing region superposition processing on different wafers in the failure region list to determine the number of failure bits in each test site after the region superposition processing; determining the wafer site average value based on the number of failure bits and the total number of wafers at the test site to obtain a superimposed failure bit map; and determining the third anomaly determination result based on the superimposed failure bit map.

[0016] In one exemplary embodiment of this disclosure, determining the third anomaly determination result based on the superimposed failure bitmap includes: generating a superimposed region list based on the superimposed failure bitmap; the superimposed region list includes multiple superimposed matrix regions; determining the number of superimposed failure bits and the superimposed failure energy density of each superimposed matrix region; determining the number of superimposed failure regions based on the number of superimposed failure bits and the superimposed failure energy density; comparing the number of superimposed failure regions with a pre-configured superimposed number threshold to determine the number of abnormal parameter combinations contained in each test point; and determining the third anomaly determination result based on the number of abnormal parameter combinations.

[0017] In one exemplary embodiment of this disclosure, generating a superimposed region list based on the superimposed failure bitmap includes: performing region division processing on the superimposed failure bitmap to obtain multiple superimposed matrix regions, and generating a corresponding initial superimposed region list; selecting superimposed matrix regions with superimposed failure energy densities greater than an energy density threshold as superimposed failure matrix regions, and saving the superimposed failure matrix regions to the superimposed failure region list; deleting secondary superimposed matrix regions contained in the superimposed failure matrix regions from the superimposed failure region list to obtain an updated superimposed failure region list.

[0018] In one exemplary embodiment of this disclosure, determining the type of anomaly cause for the test subject's anomaly based on the determined anomaly determination result includes: obtaining a pre-constructed anomaly type analysis model; obtaining the anomaly determination result of the test subject under each of the test indicators; the anomaly determination result includes one or more of a first anomaly determination result, a second anomaly determination result, and a third anomaly determination result; inputting multiple anomaly determination results into the anomaly type analysis model, the anomaly type analysis model including a calculated weight value corresponding to each anomaly determination result; and determining the anomaly type based on the multiple anomaly determination results and their respective calculated weight values.

[0019] In one exemplary embodiment of this disclosure, the calculated weight value corresponding to each anomaly determination result is calculated through the following steps: obtaining a training sample set and determining the total number of samples in the training sample set; determining the sample failure categories and sample parameter features included in the training sample set; the sample parameter features include sample parameter feature values; taking the number of training samples included in each sample failure category as the number of category samples; dividing the training sample set according to the sample parameter feature values ​​to obtain training sample subsets; determining the number of feature samples under each sample failure category in each training sample subset; and determining the calculated weight value based on the total number of samples, the number of category samples, and the number of feature samples.

[0020] According to a second aspect of this disclosure, an anomaly detection device is provided, comprising: a test data acquisition module for acquiring test data of a test subject; a test index determination module for determining at least one test index corresponding to the test data; the test index including one or more of yield parameters, resistance parameters, and failure bit regions; a judgment result determination module for determining an anomaly judgment result of the test subject under each of the test indexes; and an anomaly type determination module for determining the anomaly cause type of the anomaly occurring in the test subject based on the determined anomaly judgment result.

[0021] In one exemplary embodiment of this disclosure, the anomaly detection device further includes a test data generation module, configured to determine the test machine corresponding to the test subject, obtain the test log file of the test machine; perform structured processing on the test log file to generate initial test data; and perform data preprocessing on the initial test data to obtain the test data.

[0022] In one exemplary embodiment of this disclosure, the test data generation module includes a test data generation unit, configured to acquire at least one group identifier; the group identifier is used to divide the initial test data; the initial test data is divided according to the at least one group identifier to obtain multiple test points; the test points include initial point test data; data to be deleted for each of the initial point test data under different test parameters is determined; the data to be deleted in the multiple initial point test data is filtered to obtain multiple point test data; and the test data is generated based on the multiple point test data.

[0023] In one exemplary embodiment of this disclosure, the test subject includes a wafer, the test index includes the yield parameter, and the judgment result determination module includes a first result determination unit, used to obtain at least one test failure parameter of the test subject in the subject frame dimension; the subject to be tested includes multiple test points; the total number of wafers contained in each of the test points is determined; the number of failed wafers corresponding to each of the test points under each of the test failure parameters is determined; and based on the total number of wafers and the number of failed wafers, the abnormal judgment result of the test subject under the yield parameter is determined as the first abnormal judgment result.

[0024] In one exemplary embodiment of this disclosure, the first result determination unit includes a first result determination subunit, configured to determine the proportion of failed wafers corresponding to the test subject based on the total number of wafers and the number of failed wafers; compare the total number of wafers with a wafer number threshold to obtain a first comparison result; compare the proportion of failed wafers with a proportion threshold to obtain a second comparison result; and determine the first anomaly determination result based on the first comparison result and the second comparison result.

[0025] In one exemplary embodiment of this disclosure, the test subject includes a wafer, the test index includes the resistance parameter, and the determination result module includes a second comparison result determination unit, used to acquire specified test data of the test subject under the resistance parameter within a specified time period; acquire a specified group identifier, group the specified test data according to the specified group identifier to obtain resistance group data; acquire the point test results corresponding to each wafer under all test points, determine the resistance threshold corresponding to the resistance group data according to the point test results; and determine the abnormal determination result of the test subject under the resistance parameter according to the resistance threshold, as a second abnormal determination result.

[0026] In one exemplary embodiment of this disclosure, the second comparison result determination unit includes a second comparison result determination subunit, configured to determine the total number of wafers at each test point and the number of abnormal wafers at each test point whose test results are greater than the resistance threshold; determine the abnormal wafer ratio based on the total number of wafers and the number of abnormal wafers; obtain a parameter mean fitting line and determine the derivative of the fitting line corresponding to the parameter mean fitting line; and determine the second abnormality determination result based on the total number of wafers, the abnormal wafer ratio, and the derivative of the fitting line.

[0027] In one exemplary embodiment of this disclosure, the test subject includes a wafer, the test index includes a failure bit region, and the determination result module includes a third result determination unit, used to determine the test parameters contained in each test point of the test subject; group the test points according to the test parameters to obtain grouped test points; determine the total number of point wafers in each grouped test point; divide the bit failure map corresponding to each grouped test point to obtain multiple matrix regions and generate a corresponding region list; determine the failure energy density of each matrix region in the region list; and determine the anomaly determination result of the test subject under the failure bit region according to the total number of point wafers and the multiple failure energy densities, as the third anomaly determination result.

[0028] In one exemplary embodiment of this disclosure, the third result determination unit includes a third result determination subunit, configured to: define the matrix region where the failure energy density is greater than the energy density threshold as a failure matrix region; save the failure matrix region to a failure region list; delete the secondary matrix regions contained in the failure matrix region from the region list; perform region superposition processing on different wafers in the failure region list to determine the number of failure bits in each test point after the region superposition processing; determine the wafer point average based on the number of failure bits and the total number of wafers at the point to obtain a superimposed failure bit map; and determine the third anomaly judgment result based on the superimposed failure bit map.

[0029] In one exemplary embodiment of this disclosure, the third result determination subunit is configured to perform: generating a list of superimposed regions based on the superimposed failure bitmap; the list of superimposed regions includes multiple superimposed matrix regions; determining the number of superimposed failure bits and the superimposed failure energy density of each superimposed matrix region; determining the number of superimposed failure regions based on the number of superimposed failure bits and the superimposed failure energy density; determining the number of superimposed failure regions based on a comparison with a pre-configured superimposed number threshold; and determining the number of abnormal parameter combinations contained in each test point based on the number of abnormal parameter combinations.

[0030] In one exemplary embodiment of this disclosure, the third result determination subunit includes an overlay list determination subunit, used to perform region division processing on the overlay failure bitmap to obtain multiple overlay matrix regions and generate a corresponding initial overlay region list; to take the overlay matrix regions with overlay failure energy density greater than the energy density threshold as overlay failure matrix regions and save the overlay failure matrix regions to the overlay failure region list; to delete the secondary overlay matrix regions contained in the overlay failure matrix regions from the overlay failure region list to obtain an updated overlay failure region list.

[0031] In one exemplary embodiment of this disclosure, the anomaly type determination module includes an anomaly type determination unit, configured to: acquire a pre-constructed anomaly type analysis model; acquire anomaly determination results of the test subject under each of the test indicators; the anomaly determination results include one or more of a first anomaly determination result, a second anomaly determination result, and a third anomaly determination result; input multiple anomaly determination results into the anomaly type analysis model, the anomaly type analysis model including calculated weight values ​​corresponding to each anomaly determination result; and determine the anomaly type based on the multiple anomaly determination results and their respective calculated weight values.

[0032] In one exemplary embodiment of this disclosure, the anomaly type determination unit includes a weight value determination subunit, configured to: acquire a training sample set; determine the total number of samples in the training sample set; determine the sample failure categories and sample parameter features included in the training sample set; the sample parameter features include sample parameter feature values; use the number of training samples included in each sample failure category as the category sample quantity; divide the training sample set according to the sample parameter feature values ​​to obtain training sample subsets; determine the number of feature samples under each sample failure category in each training sample subset; and determine the calculated weight value based on the total number of samples, the category sample quantity, and the feature sample quantity.

[0033] According to a third aspect of this disclosure, an electronic device is provided, comprising: a processor; and a memory storing computer-readable instructions that, when executed by the processor, implement the anomaly detection method according to any one of the preceding claims.

[0034] According to a fourth aspect of this disclosure, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the anomaly detection method according to any one of the preceding claims.

[0035] The technical solution provided in this disclosure may include the following beneficial effects:

[0036] The anomaly detection method in the exemplary embodiments of this disclosure, on the one hand, automatically analyzes and determines the anomaly type of the subject under test based on the anomaly judgment results under multiple test indicators, so that the anomaly cause determination process does not rely on human experience, thus improving the anomaly detection efficiency. On the other hand, by integrating the anomaly judgment results under multiple test indicators, the anomaly type of the subject under test can be quickly and effectively located, which can effectively ensure the accuracy of the anomaly detection results.

[0037] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0038] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure. It is obvious that the drawings described below are merely some embodiments of this disclosure, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort. In the drawings:

[0039] Figure 1 A flowchart illustrating an anomaly detection method according to an exemplary embodiment of the present disclosure is shown schematically;

[0040] Figure 2 A flowchart illustrating the determination of a first anomaly determination result according to an exemplary embodiment of the present disclosure is shown schematically;

[0041] Figure 3 A flowchart illustrating the determination of a second anomaly determination result according to an exemplary embodiment of the present disclosure is shown schematically;

[0042] Figure 4 A flowchart illustrating the determination of a third anomaly determination result according to an exemplary embodiment of the present disclosure is shown schematically;

[0043] Figure 5 A flowchart illustrating the determination of the anomaly type of the subject under test according to an exemplary embodiment of the present disclosure is shown.

[0044] Figure 6 A block diagram of an anomaly detection apparatus according to an exemplary embodiment of the present disclosure is shown schematically;

[0045] Figure 7 A block diagram of an electronic device according to an exemplary embodiment of the present disclosure is shown schematically;

[0046] Figure 8 The illustration shows a schematic diagram of a computer-readable storage medium according to an exemplary embodiment of the present disclosure. Detailed Implementation

[0047] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, they are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted.

[0048] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Numerous specific details are provided in the following description to give a thorough understanding of embodiments of this disclosure. However, those skilled in the art will recognize that the technical solutions of this disclosure can be practiced without one or more of the specific details described, or other methods, components, apparatuses, steps, etc., can be employed. In other instances, well-known structures, methods, apparatuses, implementations, materials, or operations are not shown or described in detail to avoid obscuring various aspects of this disclosure.

[0049] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, or in one or more software-hardened modules, or in different network and / or processor devices and / or microcontroller devices.

[0050] Wafer testing is an essential process on the production line. When anomalies occur, such as pin defects or equipment malfunctions like PE card defects, product damage or measurement errors can result, further leading to yield losses. Traditional methods require engineers with specialized backgrounds to observe wafer test results and, based on their experience, determine whether the anomalies are due to product process issues or testing equipment problems, a time-consuming process. Furthermore, since the stability of production line testing equipment is crucial for product yield, analyzing equipment malfunctions necessitates retesting, impacting production capacity to some extent.

[0051] Based on this, in this example embodiment, an anomaly detection method is first provided. The anomaly detection method of this disclosure can be implemented using a server or using a terminal device. The terminal described in this disclosure can include mobile terminals such as mobile phones, tablets, laptops, handheld computers, and personal digital assistants (PDAs), as well as fixed terminals such as desktop computers. Figure 1The illustration schematically shows a flow diagram of an anomaly detection method according to some embodiments of the present disclosure. Reference Figure 1 The anomaly detection method may include the following steps:

[0052] Step S110: Obtain test data of the test subject.

[0053] In some exemplary embodiments of this disclosure, the test subject can be the product to be tested during the manufacturing process. For example, in semiconductor manufacturing testing, the test subject can be a wafer, silicon wafer, etc. The test data can be relevant data generated by testing the test subject.

[0054] In semiconductor manufacturing, the testing process is a crucial step in the production line. This process generates test data related to the tested component. When analyzing the causes of anomalies, the test data generated during this process can be acquired first, analyzed, and ultimately the type of anomaly can be identified. For example, in wafer manufacturing, wafer testing is an essential step on the production line, and the data generated during wafer testing can be used as test data.

[0055] As will be readily understood by those skilled in the art, the test subject can also be other process products, such as silicon wafers, etc. This disclosure does not impose any special limitations on the specific type of test subject.

[0056] Step S120: Determine at least one test index corresponding to the test data; the test index includes one or more of the yield parameter, resistance parameter, and failure bit region.

[0057] In some exemplary embodiments of this disclosure, the test indicators can be test parameters under different dimensions. The yield parameter can be the pass rate parameter of the test subject, which is one of the product quality indicators, denoted as the BIN indicator. The resistance parameter can be the resistance parameter of the resistive element corresponding to the test subject, denoted as the contact resistance (CRES) indicator. The failure bit region can be the parameter indicator corresponding to the region where the failure bit is generated by the test subject, denoted as the Fail RegionLatch (FRL) indicator.

[0058] After obtaining the test data, multiple test indicators can be determined, such as yield parameters, resistance parameters, and failure bit regions, in order to determine the anomaly judgment results of the test subject under different data dimensions.

[0059] Step S130: Determine the anomaly judgment results of the test subject under each test indicator.

[0060] In some exemplary embodiments of this disclosure, the anomaly determination result can be the result obtained by performing anomaly analysis on the test subject under different test index dimensions. The test point can be the particle location corresponding to the test subject under various different test conditions. For example, in wafer testing, the test point can be a certain die location under the same test environment, test program, test probe, and test equipment.

[0061] After identifying multiple test metrics, corresponding algorithm models will be used to analyze anomalies in the test subject under those metrics, yielding corresponding anomaly determination results. Before analyzing the test subject under a specific test metric, the test data can be grouped to obtain multiple test points. Anomaly analysis will then be performed based on these test points to obtain anomaly determination results.

[0062] Step S140: Determine the type of abnormality cause for the test subject based on the determined abnormality judgment result.

[0063] In some exemplary embodiments of this disclosure, the anomaly type can be a specific type of cause that leads to an anomaly in the test subject. For example, the anomaly type in the semiconductor testing process can include product process anomalies or test equipment anomalies.

[0064] After determining the anomaly detection results for the test subject under different test indicators, the multiple anomaly detection results can be analyzed together. For example, the calculation weights used when calculating the anomaly detection results under different test indicators can be determined, and a comprehensive score can be determined based on the multiple anomaly detection values ​​and their corresponding calculation weights. The specific cause of the anomaly can then be determined based on the specific value of the comprehensive score, that is, the specific anomaly type can be identified.

[0065] According to the anomaly detection method in this example embodiment, on the one hand, based on the anomaly judgment results of the subject under test under multiple test indicators, the anomaly type of the subject under test is automatically analyzed and determined, so that the anomaly cause determination process does not rely on human experience, thus improving the anomaly detection efficiency. On the other hand, by integrating the anomaly judgment results under multiple test indicators, the anomaly type of the subject under test can be quickly and effectively located, which can effectively ensure the accuracy of the anomaly detection results.

[0066] The steps S110 to S140 of the anomaly detection method in this example embodiment will be further explained below.

[0067] In one exemplary embodiment of this disclosure, the test data of the test subject can be generated through the following steps: determining the test machine corresponding to the test subject and obtaining the test log file of the test machine; performing structured processing on the test log file to generate initial test data; and performing data preprocessing on the initial test data to obtain test data.

[0068] The test equipment refers to the instruments and equipment used to perform testing on the test subject. The test log file is a file composed of test logs generated during the testing process by the test equipment; test log files are typically unstructured data. Structured processing involves structuring the data in the unstructured test log file, resulting in data with clear, definable relationships between data points and containing a predefined model. Initial test data can be the data generated after structuring the test log file. Data preprocessing involves removing invalid, outlier, and duplicate data from the initial test data.

[0069] When determining the test data for the test subject, the test machine to be tested can be identified first, and test log files for a specified time period can be obtained from the test machine. Typically, the test log files obtained directly from the test machine are unstructured data. To facilitate subsequent data analysis, the test log files can be structured to generate corresponding initial test data. Since the initial test data is structured, different data points within it have specific relationships. After obtaining the initial test data, it can be stored in a big data platform. If the initial test data is needed, it can be obtained from the big data platform. For example, the obtained initial test data can be preprocessed, such as deleting invalid, outlier, and duplicate data, making the obtained test data more consistent and easier for subsequent calculations. In this embodiment, different test metrics can be used to analyze the test data and determine the anomaly detection results.

[0070] In one exemplary embodiment of this disclosure, the process of preprocessing initial test data to obtain test data can be performed through the following steps: obtaining at least one group identifier; the group identifier is used to divide the initial test data; the initial test data is divided according to the at least one group identifier to obtain multiple test points; the test points include initial point test data; determining the data to be deleted from each initial point test data under different test parameters; filtering the data to be deleted from the multiple initial point test data to obtain multiple point test data; and generating test data based on the multiple point test data.

[0071] The grouping identifier can be the identifier used when grouping the initial test data. The initial point test data can be multiple groups of data obtained after grouping the initial test data using the grouping identifier. Test parameters can be test parameters included under different test indicators. Data to be deleted can be invalid data, outliers, duplicate data, etc., existing in the initial point test data. Point test data can be the data obtained after deleting the data to be deleted from the initial point test data.

[0072] When generating test data based on initial test data, initial test data within a specified time period can be retrieved from a big data platform, such as test data from the past 4 hours. For the acquired initial test data, group identifiers can be determined for grouping. These group identifiers may include process step identifiers (STEP_ID), project identifiers (PROGRAM_ID), probe card identifiers (PROBE_CARD_ID), test equipment identifiers (TESTER_ID), and chip identifiers (CHIP_ID). Using these group identifiers as grouping criteria, the initial test data is divided, grouping data with the same step identifier, project identifier, probe card identifier, test equipment identifier, and chip identifier together. This group represents a specific die location within the same test environment, test program, test probe, and test equipment. The grouped data is collectively referred to as test points, denoted as Group. i Each test point contains the corresponding initial test data.

[0073] After obtaining the initial test data, we can determine the data to be deleted for each test point under different test parameters. For example, the data to be deleted may include Groups under the main bounding box dimension. i Invalid data for different test parameters. When the test parameter is a DC (Direct Current) dimension parameter related to contact resistance, parameters whose names begin with CRES are selected. Data to be deleted may also include data in Groupi corresponding to CRES parameters in the DC dimension where the test parameter result equals a preset threshold (e.g., 999). Data to be deleted may also include test data where the chip identifier (CHIP_ID) belongs to the wafer edge (frequent anomaly area, EE area).

[0074] After identifying the data to be deleted, it can be removed from the initial point test data to obtain multiple corresponding point test data. These multiple point test data can then be combined to form the test data, which is convenient for subsequent analysis.

[0075] In one exemplary embodiment of this disclosure, the test subject includes a wafer, and the test indicators include yield parameters. The anomaly determination result of the test subject under the yield parameter test indicators can be performed through the following steps: obtaining at least one test failure parameter of the test subject in the subject frame dimension; the test subject includes multiple test points; determining the total number of wafers contained in each test point of the test subject; determining the number of failed wafers corresponding to each test point under each test failure parameter; and determining the anomaly determination result of the test subject under the yield parameter based on the total number of wafers and the number of failed wafers, as the first anomaly determination result.

[0076] The subject bounding box dimension can be a data dimension that identifies different colored boxes. For example, one subject bounding box (bin) corresponds to one color, and different colors represent different bins. The test failure parameter can be the parameter that corresponds to the test result showing a failure after testing the subject. The number of failed wafers can be the number of wafers corresponding to the test failure parameter. The total number of wafers can be the number of wafers contained in a single test point. The first anomaly judgment result can be the judgment result obtained from anomaly analysis of the test subject under the yield parameter index.

[0077] refer to Figure 2 , Figure 2 The flowchart illustrating the determination of a first anomaly determination result according to an exemplary embodiment of the present disclosure is shown. For the acquired test data 210, in step S201, the test data 210 can be grouped using the above processing steps, i.e., using a grouping identifier to obtain multiple test points, so that the analysis unit of the test data is the grain location.

[0078] In step S202, parameters that failed the test are defined. Parameters under the BIN dimension whose names begin with BIN_COUNT and end with a special character or letter are selected as parameters that failed the test. In step S203, test point groups are statistically analyzed. i The total number of wafers tested within a group (i.e., within the group) is denoted as wft. i Then, calculate the Group for each test point separately. i The number of wafers that failed the test for the number of parameters within the group, i.e., the number of wafers that failed (pf). ij After determining each test point group i After determining the total number of wafers and the number of failed wafers, further analysis can be performed based on each test point group. iThe total number of wafers and the number of failed wafers determine the anomaly judgment result of the test subject under the yield parameter index, which is used as the first anomaly judgment result. Through the above processing steps, based on the test data of the test anomaly parameters, the frequency of test anomalies occurring at the test points can be analyzed, and whether there is a concentration of test anomalies can be analyzed to infer whether the test equipment is abnormal.

[0079] In one exemplary embodiment of this disclosure, determining the anomaly judgment result of the test subject under the yield parameter based on the total number of wafers and the number of failed wafers, as the first anomaly judgment result, includes: determining the proportion of failed wafers corresponding to the test subject based on the total number of wafers and the number of failed wafers; comparing the total number of wafers with a wafer number threshold to obtain a first comparison result; comparing the proportion of failed wafers with a proportion threshold to obtain a second comparison result; and determining the first anomaly judgment result based on the first comparison result and the second comparison result.

[0080] The failure rate can be defined as the proportion of failed wafers at each test point to the total number of wafers. The wafer quantity threshold is a value compared to the total number of wafers. The proportion threshold is a value compared to the failure rate. The first comparison result is obtained by comparing the total number of wafers with the wafer quantity threshold. The second comparison result is obtained by comparing the failure rate with the proportion threshold.

[0081] Continue to refer to Figure 2 In step S204, calculate the Group for each test point. i The percentage of test parameters that failed within a group, i.e. denoted as pr ij In step S205, each test point Group is traversed. i For parameters that fail the test, the total number of wafers is compared with a wafer quantity threshold, for example, the wafer quantity threshold can be set to 3; at the same time, the determined proportion of failed wafers is compared with a proportion threshold, for example, the proportion threshold can be set to 0.75.

[0082] If the total number of wafers wft is satisfied i Greater than 3 and there are unacceptable wafer ratios (PRs) ij If the value is greater than 0.75, then in step S206, Group can be considered... i If the BIN_TAG is F, it indicates that the testing equipment at that grain location is malfunctioning; if the Group i The percentage of failed wafers corresponding to multiple failed test parameters included. ij If all values ​​are less than 0.75, the next step can be determined. In step S207, if the total number of wafers wfti The sum of wafer proportions greater than 3 and not passing If the value is greater than 0.75, then in step S208, the Groupi's determination tag BIN_TAG can be considered to be F, indicating that the testing equipment at that grain location is malfunctioning. Otherwise, in step S209, Groupi can be considered to be... i If the BIN_TAG is P, it means that the testing equipment at that grain location is functioning normally.

[0083] In one exemplary embodiment of this disclosure, determining the anomaly judgment result of the test subject under the resistance parameter test index can be performed through the following steps: obtaining specified test data of the test subject under the resistance parameter within a specified time period in the past; obtaining a specified group identifier, and grouping the specified test data according to the specified group identifier to obtain resistance group data; obtaining the point test results corresponding to all test points of each wafer, and determining the resistance threshold corresponding to the resistance group data according to the point test results; determining the anomaly judgment result of the test subject under the resistance parameter according to the resistance threshold, as the second anomaly judgment result.

[0084] The specified test data can be the test data of the test subject under the specified resistance parameters within a specified time period in the past. The specified group identifier can be the identifier used to group the specified test data. The resistance group data can be the data obtained after grouping the specified test data according to the specified group identifier. The point test results can be the test results of each wafer at different test points. The resistance threshold can be a pre-configured value used for comparison with the point test results. The second anomaly judgment result can be the judgment result obtained by performing anomaly analysis on the test subject under the resistance parameter index.

[0085] refer to Figure 3 , Figure 3A flowchart illustrating the determination of a second anomaly determination result according to an exemplary embodiment of this disclosure is shown. In step S201, the test data is grouped, making the analysis unit the die location. In step S301, the test data is grouped in another way, making the analysis unit the test probe type. Specifically, test data of the test subject under resistance parameters within a specified time period can be obtained as the specified test data. For example, test data of wafer contact resistance parameters over the past 14 days can be obtained as the specified test data. After obtaining the specified test data, a predetermined specified group identifier can be obtained. For example, the specified group identifier may include a product identifier (PRODUCT_ID), a process identifier (STEP_ID), a project identifier (PROGRAM_ID), and a probe card type (PROBE_CARD_TYPE), etc. The specified test data is grouped according to the specified group identifier to obtain resistance group data, denoted as CRESRange. i .

[0086] In step S302, the RESRange data for each resistance value is determined. i The upper bound of the maximum tolerable fluctuation is used as the resistance threshold. For example, the CRESRange can be calculated from the 95th quantile (RES_P95) of the test results for all points on each wafer. i The upper bound of the maximum tolerable fluctuation is obtained, and the resistance threshold is shown in Formula 1.

[0087]

[0088] Among them, "AVG()" can be used to calculate the mean; "STDDEV()" can be used to return the standard deviation function, indicating the return of the sample standard deviation; "RES_P95i" can be used to represent the 95th quantile of the test data under the resistance parameter.

[0089] After obtaining the resistance threshold, the abnormal judgment result of the test subject under the resistance parameter can be determined based on the resistance threshold, and used as the second abnormal judgment result.

[0090] In one exemplary embodiment of this disclosure, the total number of wafers at each test point and the number of abnormal wafers at each test point whose test results are greater than the resistance threshold are determined; the proportion of abnormal wafers is determined based on the total number of wafers and the number of abnormal wafers; a parameter mean fitting line is obtained, and the derivative of the fitting line corresponding to the parameter mean fitting line is determined; a second abnormality determination result is determined based on the total number of wafers, the proportion of abnormal wafers, and the derivative of the fitting line.

[0091] The number of abnormal wafers can be the number of wafers whose test results exceed the resistance threshold at each test point. The abnormal wafer ratio can be the proportion of abnormal wafers to the total number of wafers. The parameter mean fitting line can be a fitting line drawn based on the mean value calculated from the test data within a specified time interval at preset time intervals. The derivative of the fitting line can be the value obtained by performing a first derivative calculation on the fitting line.

[0092] In step S303, the test point group is determined. i Total number of wafers tested in wft i And statistics on Groups i The total number of wafers whose test results exceed the resistance threshold upper_spec, i.e., the number of abnormal wafers pf. ij In step S304, Group is calculated. i The percentage of wafers whose parameter test results exceed the resistance threshold upper_spec within the group, i.e., the percentage of abnormal wafers. denoted as pr ij In step S305, the mean value of the wafer position parameter is calculated at preset time intervals. For example, if the specified time interval is set to 4 hours and the preset time interval is set to 10 minutes, the data is grouped at 10-minute intervals, and the mean parameter level within each 10-minute interval of the 4-hour period is calculated. In step S306, a linear fitting line is plotted based on the determined mean parameter values ​​to obtain the mean parameter fitting line. The derivative of the obtained mean parameter fitting line is calculated once, i.e., the slope value is calculated, to obtain the slope of the fitting line. i .

[0093] In step S307, the total number of wafers is compared with a wafer quantity threshold (first threshold), the abnormal wafer ratio is compared with an abnormal ratio threshold (second threshold), and it is determined whether the upper bound of the maximum resistance fluctuation is greater than 0. If the total number of wafers is wft i Greater than 3, and the proportion of abnormal wafers (pr) ij Greater than 0.75, and the fitted linear derivative Slope i If the value is greater than 0, then in step S308, CRES_TAG can be determined to be F, indicating that the test equipment at that die location is malfunctioning; otherwise, in step S309, CRES_TAG is determined to be P, indicating that the test equipment at that die location is normal.

[0094] Through the above processing steps, based on the rising contact resistance data of the test results, we can focus on analyzing the local trend of the parameter values ​​at the test points, and infer whether the test equipment is abnormal from whether the test results show a rising trend.

[0095] In one exemplary embodiment of this disclosure, the anomaly determination result of the test subject under the test index of the failure bit region can be performed through the following steps: determining the test parameters contained in each test point of the test subject; grouping the test points according to the test parameters to obtain grouped test points; determining the total number of wafers in each group of test points; dividing the bit failure map corresponding to each group of test points to obtain multiple matrix regions and generating a corresponding region list; determining the failure energy density of each matrix region in the region list; and determining the anomaly determination result of the test subject under the failure bit region based on the total number of wafers at the test points and the multiple failure energy densities, as the third anomaly determination result.

[0096] The test parameters can be parameters included in the test points. Grouped test points can be multiple different groups resulting from grouping test points according to the test parameters. The total number of wafers per test point can be the total number of wafers included in the grouped test points. The matrix region can be the region obtained after dividing the bit failure map of a single wafer in the grouped test points. The region list can be a list composed of relevant data from the matrix regions. The failure energy density can be the density of failure bits contained in each matrix region. The third anomaly judgment result can be the judgment result obtained from anomaly analysis of the test subject under the failure bit region index.

[0097] In step S201, the test data 210 is grouped, allowing the analysis unit of the test data to be the grain location. The test parameters contained in each test point are determined; for example, each test point contains K test parameters. (Reference) Figure 4 , Figure 4 The flowchart illustrating the determination of a third anomaly determination result according to an exemplary embodiment of the present disclosure is shown. In step S401, the bit failure map of a single wafer in the grouped test points is divided into regions to obtain multiple matrix regions.

[0098] The specific processing steps include: statistical analysis of each test point Group. i The test points contain K test parameters. After dividing the test points according to the test parameters, multiple grouped test points can be obtained, denoted as Group. ik After identifying multiple test sites for each group, the statistics for each group of test sites are calculated. ik The total number of points on the wafers is denoted as wft. ik Then, group ik The bit failure map of a single wafer is divided into multiple matrix regions from large to small according to the axis of symmetry, and a region list AREA_LIST is generated based on the obtained multiple matrix regions.

[0099] In step S402, the bit failure energy density corresponding to each matrix region in the region list AREA_LIST is calculated in a progressive manner from largest to smallest. For example, a certain matrix region is an 8x8 matrix, and the matrix element is denoted as A. ij Therefore, A ij The value can be 0 or 1. After determining the value corresponding to each matrix element in the matrix region, the failure energy density of each matrix region can be determined, denoted as S, where,

[0100] After determining multiple failure energy densities, the anomaly judgment result of the test subject in the failure bit region can be determined based on the total number of wafers at the test site and the multiple failure energy densities, and this result serves as the third anomaly judgment result.

[0101] In one exemplary embodiment of this disclosure, a matrix region with a failure energy density greater than an energy density threshold is designated as a failure matrix region, and the failure matrix region is saved to a failure region list; secondary matrix regions contained in the failure matrix region are deleted from the region list; different wafers are superimposed on the failure region list to determine the number of failure bits in each test point after the region superposition process; the average value of the wafer points is determined based on the number of failure bits and the total number of wafers at the points to obtain a superimposed failure bit map; and a third anomaly determination result is determined based on the superimposed failure bit map.

[0102] The failure matrix region can be a matrix region where the failure energy density is greater than the energy density threshold. The failure region list can be a list composed of relevant data from the failure matrix regions, denoted as PTN_LIST. Secondary matrix regions can be smaller regions contained within the matrix region. Region overlay processing can be the process of overlaying regions from different wafers in the failure region list. The number of failure bits can be the sum of the number of failure bits at each test point after region overlay processing. The wafer point average can be the average number of wafers in each group of test points. The overlay failure bit map can be the bit failure map obtained after region overlay processing.

[0103] Continue to refer to Figure 4After determining the failure energy density corresponding to each group test point, in step S403, each failure energy density is compared with an energy density threshold to determine whether each failure energy density is greater than the energy density threshold. In step S404, if the failure energy density is not greater than the energy density threshold, the matrix region corresponding to that failure energy density is skipped. In step S405, if the failure energy density is greater than the energy density threshold, the failure matrix region is saved to the failure region list PTN_LIST. Furthermore, it is determined whether the failure matrix region contains a secondary matrix region; where a secondary matrix region can be a smaller matrix region than the failure matrix region. If the failure matrix region contains a secondary matrix region, the secondary matrix region contained in the failure matrix region can be deleted from the region list AREA_LIST.

[0104] For example, assuming there are j regions in the region list AREA_LIST, and the energy density threshold can be set to 0.75, determine the S of each region in the region list AREA_LIST. j Is it greater than 0.75? If it is, treat the matrix region as a failure matrix region and save it to the failure region list PTN_LIST; in addition, delete the smaller region contained in the failure matrix region from the region list AREA_LIST; if it is not, skip the region.

[0105] In step S406, the failure region lists of different wafers are superimposed, and the average value of each point is calculated. Different wafers undergo region superposition processing in the failure region list PTN_LIST to determine the number of failure bits in each test point after region superposition processing. The average value of each wafer point is determined based on the number of failure bits and the total number of wafers at that point, and a superimposed failure bit map is generated based on the obtained average wafer point value. Furthermore, the third anomaly determination result is determined based on the superimposed failure bit map.

[0106] For example, different wafers are superimposed in the failure region list PTN_LIST, and the values ​​of each group of test points are summed to obtain the number of failed bits for each group of test points. ikl Based on the obtained number of invalid bits, sum ikl Divide by the total number of points of wafers, denoted as wft ik The average value (avg) of each group of test points is obtained. ikl This forms a superimposed failure bitmap.

[0107] In one exemplary embodiment of this disclosure, the process of determining the third anomaly determination result based on the superimposed failure bitmap can be performed through the following steps: generating a superimposed region list based on the superimposed failure bitmap; the superimposed region list includes multiple superimposed matrix regions; determining the number of superimposed failure bits and the superimposed failure energy density of each superimposed matrix region; determining the number of superimposed failure regions based on the number of superimposed failure bits and the superimposed failure energy density; determining the number of abnormal parameter combinations contained in each test point by comparing the number of superimposed failure regions with a pre-configured superimposed number threshold; and determining the third anomaly determination result based on the number of abnormal parameter combinations.

[0108] The overlay region list can be a list of related data from multiple overlay matrix regions. Each overlay matrix region can be a different matrix region obtained by dividing a bit failure map into regions according to its axis of symmetry from largest to smallest after overlay processing. The number of overlay failure bits can be the number of failure bits contained in each overlay matrix region. The overlay failure energy density can be the density of failure bits contained in each overlay matrix region. The number of overlay failure regions can be a value determined based on the number of failure bits and the overlay failure energy density in each overlay failure region. The overlay number threshold can be a pre-set value used for comparison with the number of overlay failure regions. The number of abnormal parameter combinations can be the specific number of abnormal parameter combinations.

[0109] Continue to refer to Figure 4 In step S407, the superimposed bit failure map is divided into regions to obtain a list of superimposed regions. In an exemplary embodiment of this disclosure, the generation of the superimposed region list can be performed through the following steps: performing region division processing on the superimposed failure bit map to obtain multiple superimposed matrix regions, and generating a corresponding initial list of superimposed regions; taking the superimposed matrix regions with superimposed failure energy density greater than the energy density threshold as superimposed failure matrix regions, and saving the superimposed failure matrix regions to the list of superimposed failure regions; deleting the secondary superimposed matrix regions contained in the superimposed failure matrix regions from the list of superimposed failure regions to obtain an updated list of superimposed failure regions.

[0110] The initial stacking region list can be a list of regions generated from the relevant data of multiple stacking matrix regions obtained after directly dividing the stacking failure bitmap into regions. The stacking failure matrix region can be a stacking matrix region where the stacking failure energy density is greater than an energy density threshold. The secondary stacking matrix region can be a smaller region contained within the stacking failure matrix region.

[0111] In step S408, the superimposed bit failure map is divided into regions to obtain a list of superimposed regions. Specifically, the superimposed bit failure map is divided into different matrix regions from large to small according to the axis of symmetry, resulting in multiple superimposed matrix regions. An initial list of superimposed regions, denoted as N_AREA_LIST, is generated based on the relevant data of the multiple superimposed matrix regions.

[0112] In step S409, the inter-cell failure energy density within the region is calculated progressively. After generating the initial stacking region list, the stacking failure energy density corresponding to different stacking matrix regions in the initial stacking region list N_AREA_LIST is calculated progressively from largest to smallest, and denoted as SS.

[0113] In step S410, it is determined whether the superposition failure energy density is greater than the energy density threshold. Assuming there are m superposition matrix regions in the initial superposition region list N_AREA_LIST, the SS of each region in the initial superposition region list N_AREA_LIST is determined. m Is it greater than the energy density threshold of 0.75? In step S411, if the superimposed failure energy density is not greater than the energy density threshold, then skip the superimposed matrix region corresponding to that failure energy density.

[0114] In step S412, if the failure energy density is greater than the energy density threshold, the superposition matrix region is saved to the superposition failure region list STARK_PTN_LIST. Furthermore, it is determined whether the superposition matrix region contains a secondary superposition matrix region; the secondary superposition matrix region can be a smaller matrix region than the superposition matrix region. If the superposition matrix region contains a secondary superposition matrix region, the secondary superposition matrix region contained within it can be deleted from the initial superposition region list N_AREA_LIST. Through the above processing steps, an updated superposition failure region list is finally generated.

[0115] After generating the list of superimposed failure regions, the subsequent steps are executed. In step S413, the number of superimposed failure regions is determined. Specifically, the number of failure bits in each superimposed matrix region in the list of superimposed failure regions is summed to determine the total number of superimposed failure bits, and the superimposed failure energy density of each superimposed matrix region is determined. The sum of the products between the number of superimposed failure bits and the superimposed failure energy density is taken as the number of superimposed failure regions.

[0116] After determining the number of stacked failure regions, in step S414, the number of stacked failure regions is compared with a stacking quantity threshold. For example, the stacking quantity threshold can be set to 64, and it is determined whether the number of stacked failure regions is greater than 64. In step S415, if the number of stacked failure regions is greater than the stacking quantity threshold, Groupik is determined to be an abnormal parameter combination, that is, the group (grain position, i.e., test BIN) is abnormal. In step S416, if the number of stacked failure regions does not exceed the stacking quantity threshold, Groupik is determined to be a normal parameter combination, that is, the group is normal.

[0117] In step S417, it is determined whether the number of abnormal groups is greater than or equal to 1. That is, it is determined whether at least one abnormal parameter combination exists among the multiple group test points. In step S418, if at least one abnormal parameter combination exists, FRL_TAG is determined to be F, indicating that the test equipment at that die location is abnormal. In step S419, if no abnormal parameter combination exists, FRL_TAG is determined to be P, indicating that the test equipment at that die location is normal.

[0118] Through the above-mentioned process of solving the third anomaly determination result, based on the Fail RegionCount data, we can mainly analyze the concentration of failure bit regions in the test points, and infer whether the test equipment is abnormal from whether there is a block failure in the test results.

[0119] In one exemplary embodiment of this disclosure, a pre-built anomaly type analysis model is obtained; the anomaly determination results of the test subject under various test indicators are obtained; the anomaly determination results include one or more of a first anomaly determination result, a second anomaly determination result, and a third anomaly determination result; multiple anomaly determination results are input into the anomaly type analysis model, which includes a calculated weight value corresponding to each anomaly determination result; and the anomaly type is determined based on the multiple anomaly determination results and their corresponding calculated weight values.

[0120] The anomaly type analysis model is a computational model used to determine the anomaly type of the test subject. The calculated weight values ​​can be the weight values ​​used for different anomaly determination results when calculating the comprehensive score.

[0121] refer to Figure 5 , Figure 5The flowchart illustrating the process of determining the anomaly type of a subject under test according to an exemplary embodiment of the present disclosure is shown. When determining the anomaly type of the subject under test, a pre-built anomaly type analysis model can be used, which calculates the anomaly judgment results under different data dimensions. Specifically, different anomaly judgment results can be determined as follows: First test data 511 is acquired; in step S510, the first test data 511 is grouped under yield parameter indicators to realize that the analysis unit of the subject under test is the die location; in step S520, the anomaly judgment result of each die location under the yield parameter dimension is determined, i.e., the first anomaly judgment result, corresponding to the first result score 512 (BIN_TAG).

[0122] Similarly, for the second test data 521, it is grouped under the electrical resistance parameter index to make the analysis unit of the subject under test the grain location. Then, through step S530, the anomaly judgment result of each grain location in the electrical resistance dimension is determined as the second anomaly judgment result, corresponding to the second result score 522 (CRES_TAG). For the second test data 531, it is grouped under the failure bit region parameter index to make the analysis unit of the subject under test the grain location. Then, through step S540, the anomaly judgment result of each grain location in the failure bit region is determined as the third anomaly judgment result, corresponding to the second result score 532 (FRL_TAG).

[0123] After determining the three result scores, these three different result scores can be converted into corresponding contribution values ​​in the anomaly type analysis model calculation process, denoted as the first contribution value 513 (BIN_SCORE), the second contribution value 523 (CRES_SCORE), and the third contribution value 533 (FRL_SCORE), respectively. Specifically, if the value of BIN_TAG / CRES_TAG / FRL_TAG is P, then the corresponding BIN_SCORE / CRES_SCORE / FRL_SCORE is assigned a value of 0; if the value of BIN_TAG / CRES_TAG / FRL_TAG is F, then the corresponding BIN_SCORE / CRES_SCORE / FRL_SCORE is assigned a value of 1.

[0124] During the calculation, Groupi's BIN_SCORE, CRES_SCORE, and FRL_SCORE are used as feature inputs to the anomaly type analysis model. The calculation weight value corresponding to each anomaly judgment result is obtained, and the comprehensive contribution value 530 under the three data dimensions is calculated based on their respective calculation weight values.

[0125] After determining the overall contribution value 530, in step S550, it is determined whether the overall contribution value is greater than the contribution value threshold, thereby determining the anomaly cause type Target. For example, after calculating each Group i After calculating the contribution value SCORE, the comprehensive contribution value is SUM(BIN_FRAC*BIN_SCORE+CRES_FRAC*CRES_SCORE+FRL_FRAC*FRL_SCORE); where BIN_FRAC, CRES_FRAC and FRL_FRAC are the calculated weight values ​​in the BIN dimension, CRES dimension and FRL dimension, respectively.

[0126] In step S560, if the overall contribution value is greater than the contribution value threshold, the anomaly type of the test subject is considered to be a test equipment problem, and the corresponding Target = 1. In step S570, if the overall contribution value is less than or equal to the contribution value threshold, the anomaly type of the test subject is considered to be a product problem, and the corresponding Target = 0, thus finally determining the anomaly cause type of the test subject's anomaly.

[0127] In one exemplary embodiment of this disclosure, a training sample set is obtained, and the total number of samples in the training sample set is determined; the sample failure categories and sample parameter features included in the training sample set are determined; the sample parameter features include sample parameter feature values; the number of training samples included in each sample failure category is taken as the number of category samples; the training sample set is divided according to the sample parameter feature values ​​to obtain training sample subsets; the number of feature samples under each sample failure category in each training sample subset is determined; and the weight values ​​are calculated based on the total number of samples, the number of category samples, and the number of feature samples.

[0128] The training sample set can be a dataset used to train the anomaly type analysis model. The total number of samples can be the number of samples included in the training sample set. The sample failure category can be the anomaly failure category that the test subject may have. For example, a sample failure category can represent a specific category of die location anomalies; for wafer failures, the sample failure category can include whether the die location anomaly is a product problem or a test equipment problem. Sample parameter features can be multiple different data dimensions corresponding to the training sample set. Sample parameter feature values ​​can be the anomaly judgment results of the training sample set under multiple different data dimensions. The number of category samples can be the number of samples included in the sample failure category. The training sample subset can be multiple sample subsets obtained after partitioning the training sample set according to the sample parameter feature values. The number of feature samples can be the number of samples included in each training sample subset.

[0129] Before calculating the overall contribution value, pre-calculated weight values ​​can be obtained. Specifically, the weight values ​​for the first contribution value 513, the second contribution value 523, and the third contribution value 533 are BIN_FRAC, CRES_FRAC, and FRL_FRAC, respectively. The calculation process for these weight values ​​is as follows:

[0130] Assuming a pre-configured training sample set D is provided, the total number of samples in the training sample set D can be determined, denoted as |D|. For the data in the training sample set, the sample failure categories and sample parameter features included in the training sample set are determined; where the sample failure category can be represented by k, and the sample parameter feature can be represented by m. For example, in this embodiment, the sample failure category k can be 2, and the sample parameter feature m can be 3; where the sample failure category 2 (k = 1, 2) can represent two types of anomalies, respectively indicating whether the grain position anomaly is a product problem or a test equipment problem; the three sample parameter features (m = 1, 2, 3) represent the anomaly determination results in the three dimensions of BIN / CRES / FRL.

[0131] Determine the number of training samples included in each failure category, as the number of category samples, i.e., |C|. k | This can indicate that it belongs to sample failure category C. k The number of samples; among which, After determining the number of samples in each failure category, the training sample set can be partitioned based on the sample parameter feature values ​​to obtain corresponding training sample subsets. Assuming feature A has two distinct values, D can be divided into two subsets, namely training sample subsets D1 and D2; where |Di| represents the number of samples in Di. Let subset Di belong to class C. k The sample set is D ik ,|D ik | can represent D ik The number of samples in the dataset refers to the number of feature samples. After determining the above quantities, the weight values ​​can be calculated based on the total number of samples, the number of category samples, and the number of feature samples, as shown in Formulas 2-4.

[0132]

[0133] Where H(D) represents the information entropy contained in the training sample set; |D| represents the total number of samples in the training sample set; |C k | This can indicate that it belongs to sample failure category C. k The number of samples.

[0134]

[0135] Where H(D|A) can represent the information entropy contained in feature A in the training sample set; |D| can represent the total number of samples in the training sample set; and |Di| can represent the number of class samples contained in the training sample subset Di.

[0136]

[0137] Wherein, FRAC(A) can represent the calculated weight value corresponding to feature A (such as the BIN index); H(D) can represent the information entropy contained in the training sample set; H(D|A) can represent the information entropy contained in feature A in the training sample set.

[0138] Similarly, the weight values ​​for CRES and FRL features can also be calculated using the above steps and applied to the subsequent calculation of the overall contribution value.

[0139] After calculating the anomaly types and other anomaly data of the test subject through the above processing methods, one or more of the following graphs can be used to present the statistics of the number of anomalies of all test equipment under a specific condition: line graph, histogram, bar chart, etc. The changing trend of the number of anomalies of test equipment can be presented in the form of line graph, box plot, etc., to assist users in performing equipment health checks and preventing problems before they occur.

[0140] In summary, the anomaly detection method disclosed herein acquires test data of the test subject; determines at least one test indicator corresponding to the test data; the test indicator includes one or more of yield parameters, resistance parameters, and failure bit regions; determines the anomaly judgment result of the test subject under each test indicator; and determines the anomaly cause type of the test subject's anomaly based on the determined anomaly judgment result. On the one hand, based on the anomaly judgment results of the test subject under multiple test indicators, the anomaly type of the test subject is automatically analyzed and determined, eliminating reliance on human experience in the anomaly cause determination process, improving anomaly detection efficiency, and preventing human error (Mis-Operation). On the other hand, by integrating the anomaly judgment results under multiple test indicators, the anomaly type of the test subject can be quickly and effectively located, effectively ensuring the accuracy of the anomaly detection results. Furthermore, by plotting a trend chart of the frequency of anomalies occurring on the same device, the performance stability monitoring of the test equipment can be provided. Finally, by automatically generating trend charts of the number of test equipment anomalies in different dimensions, users can be reminded to pay attention to the health status of the equipment and intervene in a timely manner.

[0141] It should be noted that although the steps of the method in this invention are described in a specific order in the accompanying drawings, this does not require or imply that the steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0142] Furthermore, in this example embodiment, an anomaly detection device is also provided. (See reference...) Figure 6 The anomaly detection device 600 may include: a test data acquisition module 610, a test index determination module 620, a judgment result determination module 630, and an anomaly type determination module 640.

[0143] Specifically, the test data acquisition module 610 is used to acquire test data of the test subject; the test index determination module 620 is used to determine at least one test index corresponding to the test data; the test index includes one or more of yield parameters, resistance parameters, and failure bit regions; the judgment result determination module 630 is used to determine the abnormal judgment result of the test subject under each test index; and the abnormality type determination module 640 is used to determine the abnormality cause type of the abnormality of the test subject based on the determined abnormality judgment result.

[0144] In one exemplary embodiment of this disclosure, the anomaly detection device 600 further includes a test data generation module, used to determine the test machine corresponding to the test subject, obtain the test log file of the test machine; perform structured processing on the test log file to generate initial test data; and perform data preprocessing on the initial test data to obtain test data.

[0145] In one exemplary embodiment of this disclosure, the test data generation module includes a test data generation unit, configured to acquire at least one group identifier; the group identifier is used to divide the initial test data; the initial test data is divided according to the at least one group identifier to obtain multiple test points; the test points include initial point test data; the data to be deleted for each initial point test data under different test parameters is determined; the data to be deleted in the multiple initial point test data is filtered to obtain multiple point test data; and test data is generated based on the multiple point test data.

[0146] In one exemplary embodiment of this disclosure, the test subject includes a wafer, the test indicators include yield parameters, and the judgment result determination module 630 includes a first result determination unit, used to obtain at least one test failure parameter of the test subject in the subject frame dimension; the test subject includes multiple test points; the total number of wafers contained in each test point of the test subject is determined; the number of failed wafers corresponding to each test point under each test failure parameter is determined; based on the total number of wafers and the number of failed wafers, the abnormal judgment result of the test subject under the yield parameter is determined as the first abnormal judgment result.

[0147] In one exemplary embodiment of this disclosure, the first result determination unit includes a first result determination subunit, configured to determine the proportion of failed wafers corresponding to the test subject based on the total number of wafers and the number of failed wafers; compare the total number of wafers with a wafer number threshold to obtain a first comparison result; compare the proportion of failed wafers with a proportion threshold to obtain a second comparison result; and determine a first anomaly judgment result based on the first comparison result and the second comparison result.

[0148] In one exemplary embodiment of this disclosure, the test subject includes a wafer, the test index includes resistance parameters, and the judgment result determination module 630 includes a second comparison result determination unit, used to acquire specified test data of the test subject under the resistance parameters within a specified time period in the past; acquire a specified group identifier, and group the specified test data according to the specified group identifier to obtain resistance group data; acquire the point test results corresponding to each wafer under all test points, and determine the resistance threshold corresponding to the resistance group data according to the point test results; and determine the abnormal judgment result of the test subject under the resistance parameters according to the resistance threshold, as the second abnormal judgment result.

[0149] In one exemplary embodiment of this disclosure, the second comparison result determination unit includes a second comparison result determination subunit, used to determine the total number of wafers at each test point and the number of abnormal wafers at each test point whose test results are greater than the resistance threshold; determine the proportion of abnormal wafers based on the total number of wafers and the number of abnormal wafers; obtain the parameter mean fitting line and determine the derivative of the fitting line corresponding to the parameter mean fitting line; and determine the second abnormality judgment result based on the total number of wafers, the proportion of abnormal wafers, and the derivative of the fitting line.

[0150] In one exemplary embodiment of this disclosure, the test subject includes a wafer, the test index includes a failure bit region, and the judgment result determination module 630 includes a third result determination unit, used to determine the test parameters contained in each test point of the test subject; group the test points according to the test parameters to obtain grouped test points; determine the total number of point wafers in each grouped test point; divide the bit failure map corresponding to each grouped test point to obtain multiple matrix regions and generate a corresponding region list; determine the failure energy density of each matrix region in the region list; and determine the anomaly judgment result of the test subject under the failure bit region according to the total number of point wafers and the multiple failure energy densities, as the third anomaly judgment result.

[0151] In one exemplary embodiment of this disclosure, the third result determination unit includes a third result determination subunit, configured to: define the matrix region with failure energy density greater than the energy density threshold as the failure matrix region; save the failure matrix region to the failure region list; delete the secondary matrix regions contained in the failure matrix region from the region list; perform region superposition processing on different wafers in the failure region list to determine the number of failure bits in each test point after region superposition processing; determine the average value of wafer points based on the number of failure bits and the total number of wafers at the points to obtain a superimposed failure bit map; and determine the third anomaly judgment result based on the superimposed failure bit map.

[0152] In one exemplary embodiment of this disclosure, the third result determination subunit is configured to perform: generating a list of superimposed regions based on the superimposed failure bitmap; the list of superimposed regions includes multiple superimposed matrix regions; determining the number of superimposed failure bits and the superimposed failure energy density of each superimposed matrix region; determining the number of superimposed failure regions based on the number of superimposed failure bits and the superimposed failure energy density; determining the number of superimposed failure regions based on a comparison with a pre-configured superimposed number threshold; and determining the number of abnormal parameter combinations contained in each test point based on the number of abnormal parameter combinations.

[0153] In one exemplary embodiment of this disclosure, the third result determination subunit includes a superposition list determination subunit, which is used to perform region division processing on the superposition failure bitmap to obtain multiple superposition matrix regions and generate a corresponding initial superposition region list; superposition matrix regions with superposition failure energy density greater than the energy density threshold are taken as superposition failure matrix regions and saved to the superposition failure region list; secondary superposition matrix regions contained in the superposition failure matrix regions are deleted from the superposition failure region list to obtain an updated superposition failure region list.

[0154] In one exemplary embodiment of this disclosure, the anomaly type determination module 640 includes an anomaly type determination unit, used to obtain a pre-built anomaly type analysis model; obtain anomaly determination results of the test subject under various test indicators; the anomaly determination results include one or more of a first anomaly determination result, a second anomaly determination result, and a third anomaly determination result; input multiple anomaly determination results into the anomaly type analysis model, the anomaly type analysis model including the calculated weight values ​​corresponding to each anomaly determination result; and determine the anomaly type based on the multiple anomaly determination results and their respective calculated weight values.

[0155] In one exemplary embodiment of this disclosure, the anomaly type determination unit includes a weight value determination subunit, used to obtain a training sample set, determine the total number of samples in the training sample set; determine the sample failure categories and sample parameter features contained in the training sample set; the sample parameter features include sample parameter feature values; take the number of training samples contained in each sample failure category as the number of category samples; divide the training sample set according to the sample parameter feature values ​​to obtain training sample subsets; determine the number of feature samples under each sample failure category in each training sample subset; and determine the weight value calculation based on the total number of samples, the number of category samples, and the number of feature samples.

[0156] The specific details of the virtual modules of each anomaly detection device mentioned above have been described in detail in the corresponding anomaly detection methods, so they will not be repeated here.

[0157] It should be noted that although several modules or units of the anomaly detection device have been mentioned in the detailed description above, this division is not mandatory. In fact, according to embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0158] Furthermore, in an exemplary embodiment of this disclosure, an electronic device capable of implementing the above-described method is also provided.

[0159] Those skilled in the art will understand that various aspects of the present invention can be implemented as systems, methods, or program products. Therefore, various aspects of the present invention can be specifically implemented as entirely hardware embodiments, entirely software embodiments (including firmware, microcode, etc.), or embodiments combining hardware and software aspects, collectively referred to herein as “circuit,” “module,” or “system.”

[0160] The following is for reference. Figure 7 To describe an electronic device 700 according to such an embodiment of the present disclosure. Figure 7The electronic device 700 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments disclosed herein.

[0161] like Figure 7 As shown, the electronic device 700 is manifested in the form of a general-purpose computing device. The components of the electronic device 700 may include, but are not limited to: at least one processing unit 710, at least one storage unit 720, a bus 730 connecting different system components (including storage unit 720 and processing unit 710), and a display unit 740.

[0162] The storage unit stores program code that can be executed by the processing unit 710, causing the processing unit 710 to perform the steps described in the "Exemplary Methods" section above, according to various exemplary embodiments of this disclosure.

[0163] Storage unit 720 may include a readable medium in the form of a volatile storage unit, such as random access memory (RAM) 721 and / or cache memory 722, and may further include a read-only memory (ROM) 723.

[0164] Storage unit 720 may include a program / utility 724 having a set (at least one) program module 725, such program module 725 including but not limited to: operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.

[0165] Bus 730 can represent one or more of several types of bus structures, including memory cell bus or memory cell controller, peripheral bus, graphics acceleration port, processing unit, or local bus using any of the multiple bus structures.

[0166] Electronic device 700 can also communicate with one or more external devices 770 (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable a user to interact with electronic device 700, and / or with any device that enables electronic device 700 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 750. Furthermore, electronic device 700 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 760. As shown, network adapter 760 communicates with other modules of electronic device 700 via bus 730. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 700, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0167] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, terminal device, or network device, etc.) to execute the methods according to the embodiments of this disclosure.

[0168] In exemplary embodiments of this disclosure, a computer-readable storage medium is also provided, on which a program product capable of implementing the methods described above is stored. In some possible embodiments, various aspects of the invention may also be implemented as a program product comprising program code that, when the program product is run on a terminal device, causes the terminal device to perform the steps of the various exemplary embodiments of the invention described in the "Exemplary Methods" section above.

[0169] refer to Figure 8 As shown, a program product 800 for implementing the above-described method according to an embodiment of the present invention is described. It may employ a portable compact disc read-only memory (CD-ROM) and include program code, and may run on a terminal device, such as a personal computer. However, the program product of the present invention is not limited thereto. In this document, the readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, apparatus, or device.

[0170] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0171] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of sending, propagating, or transmitting programs for use by or in conjunction with an instruction execution system, apparatus, or device.

[0172] The program code contained on the readable medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0173] Program code for performing the operations of this invention can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0174] Furthermore, the above figures are merely illustrative of the processes included in the method according to exemplary embodiments of the present invention, and are not intended to be limiting. It is readily understood that the processes shown in the above figures do not indicate or limit the temporal order of these processes. Additionally, it is readily understood that these processes may be executed synchronously or asynchronously, for example, in multiple modules.

[0175] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the claims.

[0176] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.

Claims

1. An anomaly detection method, characterized in that, include: Obtain test data from the test subject; Determine multiple test indicators corresponding to the test data; the test indicators include yield parameters, resistance parameters, and failure bit regions. Determine the anomaly judgment result of the test subject under each of the test indicators; The type of abnormality cause for the test subject is determined based on the determined abnormality judgment result; The step of determining the type of abnormality cause for the test subject's abnormality based on the determined abnormality judgment result includes: Obtain a pre-built anomaly type analysis model; Obtain the anomaly determination results of the test subject under each of the test indicators; the anomaly determination results include a first anomaly determination result, a second anomaly determination result, and a third anomaly determination result. Multiple anomaly determination results are input into the anomaly type analysis model, which includes a calculated weight value corresponding to each anomaly determination result. The anomaly type is determined based on multiple anomaly determination results and their corresponding calculated weight values; The weight values ​​corresponding to each of the aforementioned anomaly determination results are calculated through the following steps: Obtain the training sample set and determine the total number of samples in the training sample set; Determine the failure categories and parameter characteristics of the samples included in the training sample set; The sample parameter features include sample parameter feature values; The number of training samples contained in each of the aforementioned sample failure categories is taken as the number of category samples; The training sample set is divided according to the feature values ​​of the sample parameters to obtain a training sample subset; Determine the number of feature samples under each failure category in each of the training sample subsets.

2. The method according to claim 1, characterized in that, The method further includes: Identify the test machine corresponding to the test subject and obtain the test log file of the test machine; The test log file is processed in a structured manner to generate initial test data; The initial test data is preprocessed to obtain the test data.

3. The method according to claim 2, characterized in that, The step of preprocessing the initial test data to obtain the test data includes: Obtain at least one group identifier; the group identifier is used to divide the initial test data. The initial test data is divided according to the at least one group identifier to obtain multiple test points; the test points include the initial test data. Determine the data to be deleted from each of the initial test points under different test parameters; The data to be deleted in the multiple initial point test data are filtered to obtain multiple point test data; The test data is generated based on test data from multiple of the aforementioned locations.

4. The method according to claim 1, characterized in that, The test subject includes a wafer, the test indicators include the yield parameters, and determining the anomaly judgment result of the test subject under each of the test indicators includes: Obtain at least one failed test parameter of the test subject in the subject frame dimension; the test subject includes multiple test points; Determine the total number of wafers contained in the test body at each of the test points; Determine the number of failed wafers corresponding to each of the aforementioned test points under each of the aforementioned test failure parameters; Based on the total number of wafers and the number of wafers that failed the test, the anomaly determination result of the test subject under the yield parameter is determined as the first anomaly determination result.

5. The method according to claim 4, characterized in that, The step of determining the anomaly judgment result of the test subject under the yield parameter based on the total number of wafers and the number of failed wafers, as the first anomaly judgment result, includes: Based on the total number of wafers and the number of failed wafers, determine the proportion of failed wafers corresponding to the test subject; The total number of wafers is compared with the wafer number threshold to obtain the first comparison result; The percentage of wafers that failed the test was compared with the percentage threshold to obtain a second comparison result; The first anomaly determination result is determined based on the first comparison result and the second comparison result.

6. The method according to claim 1, characterized in that, The test subject includes a wafer, the test index includes the resistance parameter, and determining the anomaly judgment result of the test subject under each of the test indexes includes: Obtain specified test data of the test subject under the specified resistance parameter within a specified time period in the past; Obtain a specified group identifier, and group the specified test data according to the specified group identifier to obtain resistance value group data; Obtain the test results of each wafer at all test points, and determine the resistance threshold corresponding to the resistance grouping data based on the test results. The abnormality determination result of the test subject under the resistance parameter is determined based on the resistance threshold, and is used as the second abnormality determination result.

7. The method according to claim 6, characterized in that, The step of determining the anomaly judgment result of the test subject under the resistance parameter based on the resistance threshold, as the second anomaly judgment result, includes: Determine the total number of wafers at each test point, and the number of abnormal wafers at each test point whose test results are greater than the resistance threshold. The proportion of abnormal wafers is determined based on the total number of wafers and the number of abnormal wafers; Obtain the parameter mean fitting line and determine the fitting line derivative corresponding to the parameter mean fitting line; The second anomaly determination result is determined based on the total number of wafers, the proportion of abnormal wafers, and the fitted linear derivative.

8. The method according to claim 1, characterized in that, The test subject includes a wafer, the test indicators include failure bit regions, and determining the anomaly judgment result of the test subject under each of the test indicators includes: Determine the test parameters contained in the test subject at each test point; The test points are grouped according to the test parameters to obtain grouped test points. Determine the total number of wafers at each of the grouped test sites; The bit failure map corresponding to each of the grouped test points is divided into multiple matrix regions, and a corresponding region list is generated. Determine the failure energy density of each matrix region in the region list; Based on the total number of wafers at the stated points and the multiple stated failure energy densities, the anomaly determination result of the test subject in the failed bit region is determined as the third anomaly determination result.

9. The method according to claim 8, characterized in that, The step of determining the anomaly judgment result of the test subject in the failure bit region based on the total number of wafers at the points and multiple failure energy densities, as the third anomaly judgment result, includes: The matrix region with a failure energy density greater than the energy density threshold is designated as the failure matrix region, and the failure matrix region is saved to the failure region list. Remove the secondary matrix regions contained in the failure matrix region from the region list; Different wafers are superimposed in the list of failure regions to determine the number of failure bits in each test point after the region superposition process. The average value of wafer points is determined based on the number of failed bits and the total number of wafer points to obtain a superimposed failure bit map; The third anomaly determination result is determined based on the superimposed failure bitmap.

10. The method according to claim 9, characterized in that, The step of determining the third anomaly determination result based on the superimposed failure bitmap includes: A list of superimposed regions is generated based on the superimposed failure bitmap; the list of superimposed regions includes multiple superimposed matrix regions. Determine the number of superposition failure bits and the superposition failure energy density in each superposition matrix region; The number of stacked failure regions is determined based on the number of stacked failure bits and the stacked failure energy density. The number of abnormal parameter combinations contained in each test point is determined by comparing the number of superimposed failure areas with a pre-configured superimposed number threshold. The third anomaly determination result is determined based on the number of combinations of the aforementioned abnormal parameters.

11. The method according to claim 10, characterized in that, The step of generating a list of superimposed regions based on the superimposed failure bitmap includes: The superimposed failure bitmap is divided into regions to obtain multiple superimposed matrix regions, and a corresponding initial superimposed region list is generated. The superimposed matrix region where the superimposed failure energy density is greater than the energy density threshold is taken as the superimposed failure matrix region, and the superimposed failure matrix region is saved to the superimposed failure region list. The secondary superimposed matrix regions contained in the superimposed failure matrix region are deleted from the superimposed failure region list to obtain an updated superimposed failure region list.

12. An anomaly detection device, characterized in that, include: The test data acquisition module is used to acquire test data of the test subject; The test index determination module is used to determine multiple test indices corresponding to the test data. The test indicators include yield parameters, resistance parameters, and failure bit regions; The judgment result determination module is used to determine the abnormal judgment result of the test subject under each of the test indicators; An anomaly type determination module is used to determine the anomaly cause type of the test subject's anomaly based on the determined anomaly judgment result; The step of determining the type of abnormality cause for the test subject's abnormality based on the determined abnormality judgment result includes: Obtain a pre-built anomaly type analysis model; Obtain the anomaly determination results of the test subject under each of the test indicators; the anomaly determination results include a first anomaly determination result, a second anomaly determination result, and a third anomaly determination result. Multiple anomaly determination results are input into the anomaly type analysis model, which includes a calculated weight value corresponding to each anomaly determination result. The anomaly type is determined based on multiple anomaly determination results and their corresponding calculated weight values; The weight values ​​corresponding to each of the aforementioned anomaly determination results are calculated through the following steps: Obtain the training sample set and determine the total number of samples in the training sample set; Determine the failure categories and parameter characteristics of the samples included in the training sample set; The sample parameter features include sample parameter feature values; The number of training samples contained in each of the aforementioned sample failure categories is taken as the number of category samples; The training sample set is divided according to the feature values ​​of the sample parameters to obtain a training sample subset; Determine the number of feature samples under each failure category in each of the training sample subsets.

13. An electronic device, characterized in that, include: processor; as well as A memory storing computer-readable instructions that, when executed by the processor, implement the anomaly detection method according to any one of claims 1 to 11.

14. A computer-readable storage medium having a computer program stored thereon, the computer program, when executed by a processor, implementing the anomaly detection method according to any one of claims 1 to 11.

Images