Chip yield monitoring method and device, electronic device, and storage medium

Abstract

The embodiment of the present invention discloses a chip yield monitoring method and device, electronic device, and storage medium, which relate to the field of semiconductor processing technology and can effectively improve the accuracy and timeliness of chip yield monitoring. The method includes: determining the yield information of each monitoring object according to the test results of the chip in a preset test, wherein the yield information includes chip yield and / or yield loss, and each of the monitoring objects includes any one of the following: chips from the same batch, chips on the same wafer, and chips in the same preset area on the same wafer; according to the yield information, each of the monitoring objects is divided into at least two object sets; a target set is searched from the at least two object sets to adjust the chip process according to the process information corresponding to each monitoring object in the target set, wherein the target set carries early warning information of fluctuations in the yield information. The present invention can be used for yield control in semiconductor processing.

Description

Technical Field

[0001] The present invention relates to the field of semiconductor processing technology, and in particular to a chip yield monitoring method and device, electronic equipment, and storage medium. Background Art

[0002] In semiconductor processing technology, the chip production process is very complex. The entire tape-out process involves dozens or even more steps, including photolithography, etching, material deposition, and ion implantation. Each step may introduce manufacturing defects or cause fluctuations in process characteristic parameters such as resistance and capacitance. Therefore, after manufacturing is completed, some non-compliant dies may exist on the wafer. These dies need to be screened through CP (Chip Probe) testing. The proportion of qualified products to the total number is the chip yield.

[0003] Chip yield plays a crucial role in the entire semiconductor manufacturing process, representing the stability of the fab's manufacturing process. In actual production, process fluctuations can occur between batches, wafers, and even within different regions of the same wafer, causing some chips to have yields below normal. Furthermore, testing issues such as probe wear, poor contact, and machine malfunctions can also lead to additional yield losses.

[0004] To promptly identify these yield issues and mitigate significant yield losses, in related technologies, after CP testing is completed, yield engineers typically use specialized software to extract test data from test logs and then analyze the yield data using data analysis software such as EXCEL and JMP. Because large-scale tape-outs generate massive amounts of yield data, yield engineers typically analyze the data by using data analysis software to draw wafer maps and other methods to detect abnormal yields. However, this typically requires yield engineers to make judgments based on actual conditions and experience, which can result in missed or misjudgment decisions and can also delay yield analysis, reducing timeliness. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a chip yield monitoring method and device, electronic device, and storage medium, which can effectively improve the accuracy and timeliness of chip yield monitoring.

[0006] In a first aspect, an embodiment of the present invention provides a chip yield monitoring method, the method comprising: determining the yield information of each monitoring object based on the test results of the chip in a preset test, wherein the yield information includes chip yield and / or yield loss, and each of the monitoring objects includes any one of the following: chips from the same batch, chips on the same wafer, and chips within the same preset area on the same wafer; dividing each of the monitoring objects into at least two object sets based on the yield information; searching for a target set from the at least two object sets to adjust the chip process based on the process information corresponding to each monitoring object in the target set, wherein the target set carries early warning information of fluctuations in the yield information.

[0007] Optionally, determining the yield information of each monitored object separately based on the test results of the chip in the preset test includes: obtaining the number of failed chips that fall into each failure category in each monitored object from the test log, the number including the total failure number of failed chips in each failure category and / or the number of each type of failure of failed chips in each failure category; determining the yield information of each monitored object separately based on the number of failed chips that fall into each failure category in each monitored object and the total number of chips in the monitored object, to obtain first yield information, wherein the first yield information includes the yield corresponding to the total failure number and / or the yield loss of each type corresponding to the number of each type of failure.

[0008] Optionally, determining the yield information of each monitored object based on the test results of the chip in the preset test includes: obtaining the number of failed chips falling into each failure category in each monitored object from the test log, and obtaining the number of failures in each category; determining the failure groups corresponding to each failure category, wherein the failure groups are divided according to the causes of chip failure, each failure group corresponds to at least one failure category, and each failure category corresponds to one failure group; determining the number of failed chips falling into each failure group in each monitored object based on the number of failures in each category and the failure groups corresponding to each failure category, and obtaining the number of failures in each group; determining the yield loss of each failure group in each monitored object based on the number of failures in each group in each monitored object and the total number of chips in the same monitored object, and obtaining the second yield information.

[0009] Optionally, the failure group includes at least one of the following: a defect-caused failure group, a device characteristic-caused failure group, a test-caused failure group, and an undetermined failure group.

[0010] Optionally, dividing each of the monitored objects into at least two object sets according to the yield information includes: dividing each of the monitored objects into at least two object sets in each of the failure groups according to the second yield information; searching for a target set from the at least two object sets includes: searching for a corresponding alternative set from the at least two object sets corresponding to each of the failure groups, wherein the alternative set carries a first warning information of the yield information fluctuation; inputting the first warning information into a preset model to obtain the target set, wherein the preset model includes an artificial intelligence model and / or a script model, and the target set carries the second warning information of the yield information fluctuation.

[0011] Optionally, dividing each of the monitored objects into at least two object sets in each of the failure groups based on the second yield information includes: clustering the yield loss of each of the monitored objects in each of the failure groups according to a preset clustering algorithm, so as to divide each of the monitored objects into at least two object sets in each of the failure groups; and / or classifying the yield loss of each of the monitored objects in each of the failure groups according to preset rules, so as to divide each of the monitored objects into at least two object sets in each of the failure groups.

[0012] Optionally, dividing each of the monitored objects into at least two object sets based on the yield information includes at least one of the following: clustering the chip yield or the yield loss of each of the monitored objects according to a preset clustering algorithm to divide each of the monitored objects into at least two object sets; classifying the chip yield or the yield loss of each of the monitored objects according to preset rules to divide each of the monitored objects into at least two object sets.

[0013] Optionally, the method further includes: in response to the object set obtained according to the preset clustering algorithm being inconsistent with the object set obtained according to the preset rule, selecting one of the object sets as the division result based on a preset strategy.

[0014] Optionally, searching for a target set from the at least two object sets includes at least one of the following: searching for a target set from the at least two object sets based on the degree of difference in the yield information between each of the object sets; searching for a target set from the at least two object sets based on the number of the monitoring objects contained in each of the object sets; searching for a target set from the at least two object sets based on the difference between each of the object sets and a preset set rule.

[0015] Optionally, after searching the target set from the at least two object sets, the method further includes: displaying the process information corresponding to each of the monitoring objects in the target set in a visual manner, the process information including at least one of the following: chip yield, yield loss, processing information, and test information.

[0016] In the second aspect, an embodiment of the present invention also provides a chip yield monitoring device, comprising: a determination unit, for determining the yield information of each monitoring object according to the test results of the chip in a preset test, wherein the yield information includes chip yield and / or yield loss, and each of the monitoring objects includes any one of the following: chips from the same batch, chips on the same wafer, and chips in the same preset area on the same wafer; a division unit, for dividing each of the monitoring objects into at least two object sets according to the yield information; a search unit, for searching a target set from the at least two object sets, so as to adjust the chip process according to the process information corresponding to each monitoring object in the target set, wherein the target set carries early warning information of fluctuations in the yield information.

[0017] Optionally, the determination unit includes: a first acquisition module, used to obtain from the test log the number of failed chips in each monitoring object that fall into each failure category, the number including the total failure number of failed chips in each failure category and / or the number of each type of failure of failed chips in each failure category; a first determination module, used to determine the yield information of each monitoring object according to the number of failed chips in each monitoring object that fall into each failure category and the total number of chips in the monitoring object, to obtain first yield information, wherein the first yield information includes the yield corresponding to the total failure number and / or the yield loss of each type corresponding to the number of each type of failure.

[0018] Optionally, the determination unit includes: a second acquisition module, used to obtain the number of failed chips falling into each failure category in each of the monitored objects from the test log, and obtain the number of failures in each category; a second determination module, used to determine the failure groups corresponding to each of the failure categories, wherein the failure groups are divided according to the causes of chip failure, each of the failure groups corresponds to at least one of the failure categories, and each of the failure categories corresponds to one failure group; a third determination module, used to determine the number of failed chips falling into each failure group in each of the monitored objects based on the number of failures in each category and the failure groups corresponding to each failure category, and obtain the number of failures in each group; a fourth determination module, used to determine the yield loss of each failure group in each of the monitored objects based on the number of failures in each group in each of the monitored objects and the total number of chips in the same monitored object, and obtain second yield information.

[0019] Optionally, the failure group includes at least one of the following: a defect-caused failure group, a device characteristic-caused failure group, a test-caused failure group, and an undetermined failure group.

[0020] Optionally, the division unit is specifically used to divide each monitored object into at least two object sets in each failure group according to the second yield information; the search unit includes: a search module, used to search for a corresponding alternative set from the at least two object sets corresponding to each failure group, wherein the alternative set carries a first warning information of yield information fluctuation; an input module, used to input the first warning information into a preset model to obtain the target set, wherein the preset model includes an artificial intelligence model and / or a script model, and the target set carries a second warning information of yield information fluctuation.

[0021] Optionally, the division unit includes: a first clustering module, used to cluster the yield loss of each monitored object in each failure group according to a preset clustering algorithm, so as to divide each monitored object into at least two object sets in each failure group; and / or a first classification module, used to classify the yield loss of each monitored object in each failure group according to a preset rule, so as to divide each monitored object into at least two object sets in each failure group.

[0022] Optionally, the division unit includes: a second clustering module, used to cluster the chip yield or the yield loss of each monitored object according to a preset clustering algorithm, so as to divide each monitored object into at least two object sets; and / or a second classification module, used to classify the chip yield or the yield loss of each monitored object according to preset rules, so as to divide each monitored object into at least two object sets.

[0023] Optionally, the device further includes: a selection unit configured to select one of the object sets as a division result based on a preset strategy in response to the object set obtained according to the second clustering module being inconsistent with the object set obtained according to the second classification module.

[0024] Optionally, the search unit is specifically used for at least one of the following: searching for a target set from the at least two object sets based on the degree of difference in the yield information between each of the object sets; searching for a target set from the at least two object sets based on the number of the monitored objects contained in each of the object sets; searching for a target set from the at least two object sets based on the difference between each of the object sets and a preset set rule.

[0025] Optionally, the device also includes: a display unit, which is used to visually display the process information corresponding to each monitoring object in the target set after searching the target set from the at least two object sets, and the process information includes at least one of the following: chip yield, yield loss, processing information, and test information.

[0026] In a third aspect, an embodiment of the present invention further provides an electronic device, comprising: a housing, a processor, a memory, a circuit board, and a power supply circuit, wherein the circuit board is placed inside the space enclosed by the housing, and the processor and the memory are arranged on the circuit board; a power supply circuit for supplying power to various circuits or devices of the above-mentioned electronic device; the memory is used to store executable program code; the processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, and is used to execute the chip yield monitoring method provided by any embodiment of the present invention.

[0027] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, which stores one or more programs, and the one or more programs can be executed by one or more processors to implement the chip yield monitoring method provided by any embodiment of the present invention.

[0028] The chip yield monitoring method and apparatus, electronic device, and storage medium provided by the embodiments of the present invention can determine the yield information of each monitoring object, such as chips from the same batch, chips on the same wafer, or chips within the same preset area on the same wafer, based on the test results of the chips in a preset test. Based on the yield information, each monitoring object is divided into at least two object sets, and a target set is searched from the at least two object sets, wherein the target set carries warning information of yield information fluctuation. In this way, by dividing the monitoring objects into different object sets based on the yield information and determining whether each object set carries warning information of yield information fluctuation, the target set can be found from each object set. Based on the process information corresponding to each monitoring object in the target set, the cause of the yield information fluctuation warning can be analyzed to make corresponding process adjustments. The entire process does not require the use of dedicated software for plotting and analysis, nor does it require manual judgment of yield anomalies based on actual conditions and experience, thereby effectively improving the accuracy and timeliness of chip yield monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0030] Figure 1 A flow chart of a chip yield monitoring method provided by an embodiment of the present invention;

[0031] Figure 2 A schematic diagram of using the K-means algorithm to partition an object set in an embodiment of the present invention;

[0032] Figure 3 A schematic diagram of dividing an object set using preset rules in an embodiment of the present invention;

[0033] Figure 4 Schematic diagram of object set division of monitoring object LotA-19 wafer in each failure group in an embodiment of the present invention;

[0034] Figure 5 A detailed flow chart of a chip yield monitoring method provided by an embodiment of the present invention;

[0035] Figure 6 for Figure 5 A structural diagram of a system based on a chip yield monitoring method;

[0036] Figure 7 for Figure 5 A schematic diagram of data feature extraction in the embodiment shown;

[0037] Figure 8 A schematic diagram of artificial intelligence model training in an embodiment of the present invention;

[0038] Figure 9 In the embodiments of the present invention, Figure 8 A schematic diagram of chip yield monitoring using an artificial intelligence model trained in the embodiment shown;

[0039] Figure 10 A schematic structural diagram of a chip yield monitoring device provided by an embodiment of the present invention;

[0040] Figure 11 A schematic structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0041] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0042] It should be understood that the embodiments described are only a portion of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative work are within the scope of protection of the present invention.

[0043] In a first aspect, an embodiment of the present invention provides a chip yield monitoring method, which can effectively improve the accuracy and efficiency of chip yield monitoring.

[0044] like Figure 1 As shown, an embodiment of the present invention provides a chip yield monitoring method, comprising:

[0045] S11, determining yield information of each monitoring object based on the test results of the chip in a preset test, wherein the yield information includes chip yield and / or yield loss, and each monitoring object includes any of the following: chips from the same batch, chips on the same wafer, and chips within the same preset area on the same wafer;

[0046] In semiconductor processing, the manufacture of chips often requires a series of process flows. To improve productivity, a batch of wafers generally undergoes these processes together, and the chips on this batch of wafers can be referred to as chips from the same batch. In each wafer of the same batch, tens of thousands of dies can be set on each wafer. Each dies can be separated from the wafer by dicing and packaged to obtain the corresponding chip product. The chip in the embodiments of the present invention can generally refer to dies that have not yet been packaged and / or chip products obtained after the dies are packaged.

[0047] As mentioned in the background art, since there may be inconsistent process conditions in the process flow of chip manufacturing, not every chip obtained is a qualified product, resulting in chip yield loss. In an embodiment of the present invention, the chip yield is monitored, that is, it is necessary to monitor the chip yield of each monitored object, the yield loss, whether it is within the normal range, or whether there is an early warning of chip yield or yield loss. Among them, the chip yield of each monitored object is the ratio of the number of qualified chips in the monitored object to the total number of chips contained in the monitored object. The yield loss of each monitored object is the ratio of the number of unqualified chips in the monitored object to the total number of chips contained in the monitored object, and yield loss = 1-chip yield. Chip yield and yield loss can be collectively referred to as yield information.

[0048] In order to understand the fluctuation of chip yield or yield loss at different scales so as to make process improvements, in an embodiment of the present invention, chip yield monitoring can be performed on monitoring objects of different scales. For example, if it is necessary to understand the chip yield fluctuation between different batches, the chips of the same batch can be used as one monitoring object, and the chips of different batches can be different monitoring objects; if it is necessary to understand the chip yield fluctuation between different wafers in the same batch, the chips on the same wafer of the same batch can be used as one monitoring object, and the chips of different wafers of the same batch can be different monitoring objects; if it is necessary to understand the chip yield fluctuation between different areas in the same wafer, the chips in the same preset area on the same wafer can be used as one monitoring object, and the chips in different preset areas on the same wafer can be different monitoring objects.

[0049] In order to realize chip yield monitoring, in one embodiment of the present invention, the specific numerical value of the yield information of each monitoring object can be determined by various means. For example, in this step, the chip yield and / or yield loss of each monitoring object can be determined separately based on the test results of the chip in the preset test. Among them, the preset test can be any test that can determine whether the chip is a qualified product, such as CP test. The test results can be either historical data of the chip test or real-time data currently being tested, and the embodiments of the present invention are not limited to this.

[0050] S12, dividing each of the monitoring objects into at least two object sets according to the yield information;

[0051] After obtaining the yield information of each monitoring object, in this step, each monitoring object can be divided into two or more object sets based on the difference in yield information. For example, monitoring objects with the same or similar chip yield (or yield loss) can be divided into the same object set, while monitoring objects with significantly different chip yields (or yield losses) can be divided into different object sets. The specific division criteria can be set and adjusted based on the distribution characteristics of chip yield, yield loss, and the specific process used by the chip.

[0052] S13, searching for a target set from the at least two object sets to adjust the chip process according to the process information corresponding to each monitoring object in the target set, wherein the target set carries warning information of the yield information fluctuation.

[0053] Based on the various object sets obtained in step S12, in this step, the target set can be searched in these object sets. Since each object set is divided according to the yield information (chip yield and / or yield loss) of the monitored object, different object sets have correspondingly different chip yield characteristics and carry different chip yield and / or yield loss fluctuation information. The chip yield and / or yield loss fluctuations of some object sets may be normal fluctuations, while the chip yield and / or yield loss fluctuations of some object sets may be abnormal (for example, the fluctuation range may be too large, exceeding the preset threshold). These abnormal fluctuations of chip yield and / or yield loss are warning information of yield information fluctuations. When selecting the target set, the object set carrying the warning information of yield information fluctuations can be used as the target set.

[0054] The chip yield monitoring method provided by an embodiment of the present invention can determine the yield information of each monitoring object, such as chips from the same batch, chips on the same wafer, or chips within the same preset area on the same wafer, based on the test results of the chips in a preset test. Based on the yield information, each monitoring object is divided into at least two object sets, and a target set is searched from the at least two object sets, wherein the target set carries warning information of yield information fluctuation. In this way, by dividing the monitoring objects into different object sets based on the yield information and determining whether each object set carries warning information of yield information fluctuation, the target set can be found from each object set. Based on the process information corresponding to each monitoring object in the target set, the cause of the yield information fluctuation warning can be analyzed to make corresponding process adjustments. The entire process does not require the use of dedicated software for plotting and analysis, nor does it require manual judgment of yield anomalies based on actual conditions and experience, thereby effectively improving the accuracy and timeliness of chip yield monitoring.

[0055] Specifically, the basis of yield monitoring is first to determine yield information such as chip yield and yield loss for each monitored object. To this end, in embodiments of the present invention, yield information can be determined using various test data, including test data generated by real-time testing and historical test data. For example, in one embodiment of the present invention, in step S11, determining the yield information of each monitored object based on the test results of the chip in a preset test may include: obtaining the number of failed chips falling into each failure category within each monitored object from the test log, wherein the number may include the total number of failed chips in each failure category and / or the number of each type of failure for failed chips in each failure category; and determining the yield information of each monitored object based on the number of failed chips falling into each failure category within each monitored object and the total number of chips within the monitored object, thereby obtaining first yield information, wherein the first yield information includes the yield corresponding to the total number of failures and / or the yield loss for each type of failure corresponding to the number of each type of failure.

[0056] In a specific implementation, during chip testing, unqualified products obtained through testing are generally classified to obtain the failure category of the chip. Taking CP testing as an example, in CP testing, it is often necessary to test the chip (crystal) under multiple conditions and multiple items. If all test items are passed, the CP test of the chip is passed and it is a qualified product; if any of the test items are not passed, the CP test of the chip is not passed, the chip is a failed crystal, an unqualified product, and the chip is classified into a corresponding failure category, such as the failbin1 category. In this way, there will be a corresponding number of chips under various failure categories. In an embodiment of the present invention, the number of failed chips in each failure category under each monitoring object can be counted, and the yield loss and yield of the monitoring object can be calculated accordingly, wherein the yield loss of the monitoring object is equal to the ratio of the sum of the number of failed chips in each failure category of a monitoring object to the total number of chips in the monitoring object.

[0057] Optionally, in one embodiment of the present invention, if it is necessary to focus on the total chip yield of the monitoring object without distinguishing each failure category, the total number of failed chips in each failure category within each monitoring object can be obtained from the test log to calculate the yield corresponding to the total number of failures. For example, the chips on the same wafer are a monitoring object, where the total number of failed chips of various categories on wafer W1 is 50, and there are 5,000 chips on wafer W1, then the yield of wafer W1 is (5,000-50) / 5,000=99%; the total number of failed chips of various categories on wafer W2 is 100, and there are 5,000 chips on wafer W2, then the yield of wafer W2 is (5,000-100) / 5,000=98%. Through this yield calculation method, the chip yield fluctuation of the monitoring object can be grasped as a whole, so as to preliminarily but quickly discover whether there is a problem with the yield of the monitoring object.

[0058] Optionally, in another embodiment of the present invention, if it is necessary to pay attention to the yield loss of each type under each failure category of the monitored object, the number of failures of each type of failed chips in each failure category within each of the monitored objects can be obtained from the test log, so as to calculate the yield loss of each type corresponding to the number of failures of each type. For example, the chips on the same wafer are a monitored object, wherein the number of failed chips in the failbin7 category on wafer W3 is 200, and the number of failed chips in the failbin19 category is 50. There are 5000 chips on wafer W3. Then, the yield loss of each type of failbin7 category in wafer W3 is 200 / 5000=4%, and the yield loss of each type of failbin19 category is 50 / 5000=1%. Through this yield calculation method, the yield information fluctuation of the monitored object can be grasped at a more detailed level, so as to find out the specific yield problem of the monitored object.

[0059] The above embodiment uses chips on the same wafer as one monitoring object, but the embodiments of the present invention are not limited thereto. Optionally, in other embodiments of the present invention, if chips from the same batch or chips from the same preset area on the same wafer are used as one monitoring object, the corresponding chip yield calculation method is similar, except that the number of failed chips required to calculate the chip yield is specifically the number of failed chips from the same batch, or the number of failed chips from the same preset area on the same wafer. Accordingly, the total number of chips within the monitoring object is specifically the total number of chips from the same batch, or the total number of chips from the same preset area on the same wafer.

[0060] Furthermore, in the above embodiment, although the yield loss of the monitored object can be calculated separately from the perspective of each failure category, in some cases, the division of failure categories in the test data may be too detailed, resulting in a large number of failure categories and chip yield problems that are too dispersed, making it difficult to discover and summarize yield problems. To this end, in another embodiment of the present invention, step S11 determines the yield information of each monitoring object according to the test results of the chip in the preset test, and specifically may include: obtaining the number of failed chips falling into each failure category in each monitoring object from the test log, and obtaining the number of failures in each category; determining the failure groups corresponding to each failure category, wherein the failure groups are divided according to the causes of chip failure, each failure group corresponds to at least one failure category, and each failure category corresponds to one failure group; determining the number of failed chips falling into each failure group in each monitoring object according to the number of failures in each category and the failure groups corresponding to each failure category, and obtaining the number of failures in each group; determining the yield loss of each failure group in each monitoring object according to the number of failures in each group and the total number of chips in the same monitoring object, and obtaining the second yield information.

[0061] Specifically, in this embodiment, failure groups are provided in addition to failure categories. This allows one or more failure categories to be grouped together into the same failure group based on corresponding rules, and the yield loss corresponding to each failure group of the monitored object is calculated to obtain second yield information. For example, in one embodiment of the present invention, although failbin20 and failbin3 are different failure categories, they are both failures caused by process defect-related issues. Therefore, failbin20 and failbin3 can be grouped into the same failure group, such as a defect-caused failure group. When determining yield information, if the number of failed chips falling into failure group GP1 on the monitored wafer W4 is 40, i.e., the number of failures per group corresponding to failure group GP1 is 40, and the total number of chips on wafer W4 is 10,000, then the yield loss corresponding to failure group GP1 is 40 / 10,000 = 0.4%. In this way, by grouping failure categories into corresponding failure groups, yield information can be calculated using various failure groups as statistical calibers, depending on specific needs and circumstances. This allows the calculated yield information to more prominently reflect the process issues of concern and provide improved operability.

[0062] Optionally, in one embodiment of the present invention, the failure group may include one or more of the following: a defect-caused failure group, a device characteristic-caused failure group, a test-caused failure group, and an undetermined failure group. The defect-caused failure group may include chip failures caused by process defects, such as metal wire breakage that prevents the corresponding chip function from being realized; the device characteristic-caused failure group may include manufactured chips that do not meet device performance requirements, such as excessive signal delay; the test-caused failure group may include chip failures caused by improper test operations, such as poor contact during testing; and the undetermined failure group may include chip failures caused by other reasons that cannot be categorized for the time being.

[0063] It should be noted that although in this embodiment, failure groups are divided based on the causes of chip failures, the embodiments of the present invention are not limited thereto. In other embodiments of the present invention, the rules for dividing failure groups can be determined based on specific needs, thereby greatly enhancing the flexibility of chip yield monitoring. For example, in one embodiment of the present invention, if one is particularly concerned about how the synergistic effects of failbin4, failbin5, and failbin6 affect chip yield, then failbin4, failbin5, and failbin6 can be divided into the same failure group, while all other failure categories can be divided into another failure group.

[0064] In the aforementioned embodiments, the yield information determination methods using failure categories, failure groups, and the monitoring object as a whole as the statistical caliber of the yield information are respectively introduced. Regardless of which of the above yield information determination methods is adopted, after obtaining the yield information, each of the monitoring objects can be divided into at least two object sets based on the yield information. Specifically, dividing each of the monitoring objects into at least two object sets based on the yield information can include at least one of the following: clustering the chip yield or yield loss of each of the monitoring objects according to a preset clustering algorithm to divide each of the monitoring objects into at least two object sets; classifying the chip yield or yield loss of each of the monitoring objects according to preset rules to divide each of the monitoring objects into at least two object sets.

[0065] For example, in one embodiment of the present invention, the chip yield or yield loss of each monitoring object can be clustered according to a preset clustering algorithm (such as the K-means algorithm) to divide each monitoring object into at least two object sets. For example, the process and classification effect of the K-means algorithm can be as follows: Figure 2 shown.

[0066] Optionally, in another embodiment of the present invention, the chip yield of each monitoring object or the yield loss corresponding to the chip yield can be classified according to a preset rule (for example, with a preset yield as a set dividing line) to divide each monitoring object into at least two object sets. Exemplarily, the process and effect of group division according to the preset rule can be as follows: Figure 3 shown.

[0067] Furthermore, in one or more embodiments of the present invention, the two aforementioned set partitioning methods may be used simultaneously. Optionally, when the two aforementioned set partitioning methods are used simultaneously, if the object set obtained according to the preset clustering algorithm is inconsistent with the object set obtained according to the preset rule, one of the object sets may be selected as the partitioning result based on a preset strategy.

[0068] For example, in one embodiment of the present invention, when the results of K-means and rule judgment are consistent, any one of the results can be adopted; when K-means determines that the chip yield of the monitored object belongs to the normal set, and the preset rule determines that the chip yield of the monitored object belongs to the abnormal set, the K-means result can be selected and the preset rule can be checked and corrected; when K-means determines that the chip yield of the monitored object belongs to the abnormal set, and the preset rule determines that the chip yield of the monitored object belongs to the normal set, the preset rule judgment result can be adopted, that is, it is considered that the chip yield of the monitored object belongs to the normal set.

[0069] The K-means algorithm randomly selects k initial cluster centers and analyzes the distance between data points and the cluster centers. After repeated iterations, the iterative analysis stops when no significant difference is found. The pre-defined rule determination process can refer to the normal distribution of yield loss and the data dispersion diagram to pre-define the rules for dividing the object set.

[0070] As can be seen from the above embodiments, the second yield information is obtained by grouping and counting the yield loss of the monitored object according to the failure group. Each failure group in any monitored object corresponds to its own yield loss. Therefore, in step S12, each monitored object can also be divided into object sets according to the second yield information. For example, in one embodiment of the present invention, in step S12, dividing each of the monitored objects into at least two object sets according to the yield information can specifically include: dividing each of the monitored objects into at least two object sets in each of the failure groups according to the second yield information. Optionally, when dividing each of the monitored objects into at least two object sets in each of the failure groups according to the second yield information, a variety of set division methods can be adopted. For example, in one embodiment of the present invention, the yield loss of each of the monitored objects can be clustered in each of the failure groups according to a preset clustering algorithm, so as to divide each of the monitored objects into at least two object sets in each of the failure groups. In another embodiment of the present invention, the yield loss of each of the monitored objects can also be classified in each of the failure groups according to preset rules, so as to divide each of the monitored objects into at least two object sets in each of the failure groups. The specific set division principle is similar to that of the aforementioned embodiment. Furthermore, in one or more embodiments of the present invention, the above-mentioned two set division methods can also be used simultaneously. When the above-mentioned two set division methods are used simultaneously, in response to the object set obtained according to the preset clustering algorithm being inconsistent with the object set obtained according to the preset rules, one of the object sets can be selected as the division result based on the preset strategy.

[0071] Optionally, the preset strategy may be, for example, as follows: if one set partitioning method R partitions the monitored object M into the object set P, then regardless of which other object set Q the monitored object M is partitioned into by another set partitioning method T, it is considered that the monitored object M should be partitioned into the object set P according to the partitioning result of the set partitioning method R. The object set P can be set as needed.

[0072] For example, if Figure 4 As shown, in one embodiment of the present invention, taking the monitoring object LotA-19 (wafer No. 19 of batch A) as an example, in the defect-related (ie defect-caused failure group, see Figure 4 (a)), device characteristics related (i.e., device characteristics-induced failure group, see Figure 4 (b)) and test-related yield loss (i.e., test-induced failure group, see Figure 4(c)) The results of K-means and preset rules are consistent. The yield loss related to defects is judged to belong to the abnormal set (set Q), and the yield loss related to device characteristics and test-related yield loss are judged to belong to the normal set (set P). Figure 4 (d)), the results of K-means and preset rules are inconsistent. K-means determines that it belongs to the abnormal set (set Q), while the preset rules determine that it belongs to the normal set (set P). Then the yield loss of the undetermined failure group can be determined as belonging to the normal set (set P) according to the preset rules.

[0073] After dividing each of the monitoring objects into at least two object sets, a target set can be searched from the at least two object sets in step S13. Since the target set carries warning information about yield information fluctuations, that is, the chip yield or yield loss of each monitoring object in the target set requires vigilance and attention. Therefore, finding the target set is equivalent to finding each monitoring object with a chip yield or yield loss that requires attention, so that the process can be adjusted and optimized based on the process parameters of these monitoring objects.

[0074] Optionally, in one embodiment of the present invention, searching for a target set from at least two object sets may specifically include one or more of the following: searching for a target set from the at least two object sets based on the degree of difference in yield information between each of the object sets; searching for a target set from the at least two object sets based on the number of the monitored objects contained in each of the object sets; searching for a target set from the at least two object sets based on the difference between each of the object sets and a preset set rule.

[0075] For example, in one embodiment of the present invention, if the chip yield of each monitored object in the object set ST1 is between 85% and 90%, the chip yield of each monitored object in the object set ST2 is between 75% and 84%, and the chip yield of each monitored object in the object set ST3 is between 40% and 50%, then since the chip yield of ST3 is too different from the chip yields of ST1 and ST2, ST3 can be selected as the target set.

[0076] Optionally, in another embodiment of the present invention, if there are 10,000 monitoring objects in the object set ST4, 2,000 monitoring objects in the object set ST5, and 900 monitoring objects in the object set ST6, then since the number of chips in ST4 is too different from the number of chips in ST5 and ST6, ST4 can be selected as the target set.

[0077] Optionally, in another embodiment of the present invention, if the preset set rule is that the number of monitored objects in the set should be greater than 5, and there are 3 monitored objects in the object set ST7, then since ST7 is different from the preset set rule, ST7 can be selected as the target set.

[0078] When selecting a target set from each object set, if each object set is divided according to the second yield information, since the second yield information uses the failure group as the statistical caliber, factors related to the failure group can also be taken into consideration when selecting the target set. For example, in one embodiment of the present invention, searching for a target set from the at least two object sets in step S13 can specifically include: searching for a corresponding alternative set from the at least two object sets corresponding to each failure group, wherein the alternative set carries first warning information of yield information fluctuations; inputting the first warning information into a preset model to obtain the target set, wherein the preset model includes an artificial intelligence model and / or a script model, and the target set carries second warning information of yield information fluctuations.

[0079] For example, in one embodiment of the present invention, chips from the same batch can be identified as one monitoring target, while chips from different batches can be identified as different monitoring targets. For three different monitoring targets (lots A, B, and C), the yield losses for failures due to defects are: 15% for lot A, 40% for lot B, and 20% for lot C. The yield losses for failures due to device characteristics are: 25% for lot A, 27% for lot B, and 10% for lot C. The yield losses for failures due to testing are: 10% for lot A, 4% for lot B, and 2% for lot C. Based on the yield loss of these monitored objects, the monitored objects in each failure group can be divided into at least two object sets. For example, in the defect-caused failure group, the monitored objects can be divided into two sets, such as batch lotA and batch lotC are classified into one object set set1, and lotB is classified into another object set set2; in the device characteristic-caused failure group, the monitored objects can be divided into two sets, such as batch lotA and batch lotB are classified into one object set set3, and lotC is classified into another object set set4; in the test-caused failure group, the monitored objects can be divided into two sets, such as batch lotB and batch lotC are classified into one object set set5, and lotA is classified into another object set set6.

[0080] After obtaining the object set, we can search for alternative sets within each failure group. Continuing with the previous example, based on yield information fluctuations, assuming that set1 has normal fluctuations, set2 has abnormal fluctuations, set3 has abnormal fluctuations, set4 has normal fluctuations, and both set5 and set6 have normal fluctuations, we can use set2 and set3, the object sets with abnormal yield fluctuations, as alternative sets. Set2 carries the first warning information for failures caused by defects, and set3 carries the first warning information for failures caused by device characteristics.

[0081] After selecting the alternative set, the fluctuation of at least part of the yield information of the monitored object can be known. For example, through the first warning information of failure caused by device characteristics carried by set3, it can be known that the chips in batches lotA and lotB have abnormal yield loss fluctuation caused by device characteristics. Through the first warning information of failure caused by defects carried by set2, it can be known that the chips in batch lotB have abnormal yield loss fluctuation caused by defects.

[0082] However, due to the large number of monitoring objects and the complicated classification and failure grouping, in order to more quickly clarify whether the yield information fluctuation of each monitoring object is abnormal and the cause of the abnormality, in one embodiment of the present invention, the first warning information carried by each alternative set can be input into a preset model, and the target set can be obtained through the preset model, and the target set carries the second warning information of the yield information fluctuation. For example, the first warning information of the defect-induced failure carried by set2 and the first warning information of the device characteristic-induced failure carried by set3 can be input into the preset model, and the corresponding output results can be obtained. According to the output results of the preset model, it can be known that batch lotA has abnormal chip yield fluctuations caused by device characteristics, and batch lotB has abnormal chip yield fluctuations caused by defects and abnormal chip yield fluctuations caused by device characteristics, so that it can be simply, clearly and clearly pointed out whether the yield information fluctuation of each monitoring object is abnormal and the cause of the abnormality.

[0083] In addition, in an embodiment of the present invention, the preset model can be constructed in a variety of ways. For example, it can be an artificial intelligence model obtained through training based on artificial intelligence and big data, or it can be a script model written according to a preset programming language and algorithm. The embodiment of the present invention does not limit this.

[0084] Since the target set carries early warning information of yield information fluctuations, that is to say, the chip yield or yield loss of each monitoring object in the target set needs to be alerted and paid attention to. Therefore, finding the target set is equivalent to finding each monitoring object whose chip yield or yield loss needs to be paid attention to, so that the process can be adjusted and optimized according to the process parameters of these monitoring objects. In order to make the yield monitoring effect more intuitive and effectively improve the efficiency of finding process problems and user experience, further, in one embodiment of the present invention, after searching the target set from the at least two object sets, the chip yield monitoring method provided by the embodiment of the present invention can also include: displaying the process information corresponding to each of the monitoring objects in the target set in a visual manner, and the process information includes at least one of the following: chip yield, yield loss, processing information, and test information.

[0085] The chip yield monitoring method provided by the embodiment of the present invention is described in detail below through specific examples.

[0086] Figure 5 A flowchart of a chip yield monitoring method provided by an embodiment of the present invention, Figure 6 for Figure 5 Schematic diagram of the system structure on which the monitoring method is based. Figure 6 As shown, test data can be input into the real-time data processing module and the offline data processing module. The real-time data processing module processes new yield data; the offline data processing layer processes historical yield data for training machine learning models; and the data analysis and storage layer uses the model trained with historical yield data to test new yield data and finally stores the results in a database. The index database is used to store large amounts of data, while the cache database is used to temporarily store suspicious data or data requiring further manual processing.

[0087] Combine Figure 5 and Figure 6 The chip yield monitoring method provided by the embodiment of the present invention may include:

[0088] S201, preprocessing the test data of the monitored object;

[0089] For example, sensitive information contained in CP test data is removed and non-compliant CP test data is deleted based on the byte size of the test data.

[0090] S202, extracting data features from the preprocessed test data;

[0091] Optionally, the data features may mainly include relevant information in the CP test that can reflect whether the chip test is qualified, which test failed, what failure category it belongs to, etc., and may also include chip positioning information, etc.

[0092] Optionally, various failed chips can be further classified into corresponding failure groups based on the extracted data features. For example, according to the causes of chip failure, various failure categories can be divided into defect-related failures (defect-caused failure group), device feature-related failures (device feature-caused failure group), test-related failures (test-caused failure group), and failures that cannot be classified (undetermined cause failure group). The failed chips can be classified into corresponding failure groups based on the failure categories corresponding to the failed chips and the failure groups corresponding to the failure categories. For example, a data feature extraction process can be as follows: Figure 7 shown.

[0093] S203, determining the yield information of each monitored object based on the extracted data features;

[0094] Optionally, first yield information may be calculated for the entire monitored object or for each failure category in the monitored object. The first yield information may include both the yield and the yield loss corresponding to each failure category. In the case where each failure category corresponds to a different failure group, second yield information may also be calculated for each failure group. The second yield information may include the yield loss corresponding to each failure group in each monitored object.

[0095] S204. Divide the monitored objects in each failure group into at least two object sets based on the second yield information;

[0096] S205. Searching for a corresponding candidate set from the at least two object sets corresponding to each failure group, wherein the candidate set carries first warning information of yield information fluctuation;

[0097] S206. Input the first warning information into a preset model to obtain the target set, wherein the preset model includes an artificial intelligence model and / or a script model, and the target set carries the second warning information of yield information fluctuation.

[0098] Optionally, the AI ​​model can be trained in the following ways: Figure 8As shown, the labeled feature data is input into the machine learning training module and trained according to the training logic. Taking LotC-17 as an example, the neural network judgment process is as follows: its defect feature abnormality activates the "Defect or device characteristic abnormality" neuron 3 containing "defect abnormality." This activates neurons 1 through 5 in the first hidden layer. Then, neurons 1 through 3 in the first hidden layer combine to activate neuron 1 in the second hidden layer. Neurons 1, 3, 4, and 5 in the first hidden layer combine to activate neuron 2 in the second hidden layer. Finally, neurons 2 through 4 in the first hidden layer combine to activate neuron 3 in the second hidden layer. Finally, the results from neurons 1 through 3 in the second hidden layer are combined and similar results are used to determine that LotC-17 is a defect.

[0099] After the model training is completed, the first warning information can be input into the preset model, for example, Figure 9 As shown, the LotA-19 results, marked by the yield data processing module, are input into the yield inspection unit. The processed LotA-19 data has the highest similarity to the "Defect feature abnormality, device-related, test-related, and other features normal" data from the inspection unit, and is therefore identified as a defect. The LotA-19 yield data and the judgment results are stored in the abnormal yield database.

[0100] S207 . Display the process information corresponding to each monitoring object in the target set in a visual manner, where the process information includes at least one of the following: chip yield, yield loss, processing information, and test information.

[0101] In a second aspect, an embodiment of the present invention further provides a chip yield monitoring device, which can effectively improve the accuracy and efficiency of chip yield monitoring.

[0102] like Figure 10 As shown, the chip yield monitoring device provided by the embodiment of the present invention may include:

[0103] A determination unit 31 is configured to determine yield information of each monitoring object based on the test results of the chip in a preset test, wherein the yield information includes chip yield and / or yield loss, and each monitoring object includes any of the following: chips from the same batch, chips on the same wafer, and chips within the same preset area on the same wafer;

[0104] a dividing unit 32, configured to divide each of the monitoring objects into at least two object sets according to the yield information;

[0105] The search unit 33 is used to search for a target set from the at least two object sets to adjust the chip process according to the process information corresponding to each monitoring object in the target set, wherein the target set carries the warning information of the yield information fluctuation.

[0106] The chip yield monitoring device provided by an embodiment of the present invention can determine the yield information of each monitoring object, such as chips from the same batch, chips on the same wafer, or chips within the same preset area on the same wafer, based on the test results of the chips in a preset test. Based on the yield information, each monitoring object is divided into at least two object sets, and a target set is searched from the at least two object sets, wherein the target set carries warning information of yield information fluctuation. In this way, by dividing the monitoring objects into different object sets based on the yield information and determining whether each object set carries warning information of yield information fluctuation, the target set can be found from each object set. Based on the process information corresponding to each monitoring object in the target set, the cause of the yield information fluctuation warning can be analyzed to make corresponding process adjustments. The entire process does not require the use of dedicated software for graphing and analysis, nor does it require manual judgment of yield anomalies based on actual conditions and experience, thereby effectively improving the accuracy and timeliness of chip yield monitoring.

[0107] Optionally, the determination unit 31 includes: a first acquisition module, used to obtain from the test log the number of failed chips in each monitoring object that fall into each failure category, the number including the total failure number of failed chips in each failure category and / or the number of each type of failure of failed chips in each failure category; a first determination module, used to determine the yield information of each monitoring object according to the number of failed chips in each monitoring object that fall into each failure category and the total number of chips in the monitoring object, to obtain first yield information, wherein the first yield information includes the yield corresponding to the total failure number and / or the yield loss of each type corresponding to the number of each type of failure.

[0108] Optionally, the determination unit 31 includes: a second acquisition module, used to obtain the number of failed chips falling into each failure category in each of the monitored objects from the test log, and obtain the number of failures in each category; a second determination module, used to determine the failure groups corresponding to each of the failure categories, wherein the failure groups are divided according to the causes of chip failure, each of the failure groups corresponds to at least one of the failure categories, and each of the failure categories corresponds to one failure group; a third determination module, used to determine the number of failed chips falling into each of the failure groups in each of the monitored objects based on the number of failures in each category and the failure groups corresponding to the failure categories, and obtain the number of failures in each group; a fourth determination module, used to determine the yield loss of each failure group in each of the monitored objects based on the number of failures in each group in each of the monitored objects and the total number of chips in the same monitored object, and obtain second yield information.

[0109] Optionally, the failure group includes at least one of the following: a defect-caused failure group, a device characteristic-caused failure group, a test-caused failure group, and an undetermined failure group.

[0110] Optionally, the division unit 32 is specifically used to divide each of the monitored objects into at least two object sets in each of the failure groups according to the second yield information; the search unit includes: a search module, used to search for a corresponding alternative set from the at least two object sets corresponding to each of the failure groups, wherein the alternative set carries a first warning information of yield information fluctuations; an input module, used to input the first warning information into a preset model to obtain the target set, wherein the preset model includes an artificial intelligence model and / or a script model, and the target set carries a second warning information of yield information fluctuations.

[0111] Optionally, the division unit 32 includes: a first clustering module, used to cluster the yield loss of each monitored object in each failure group according to a preset clustering algorithm, so as to divide each monitored object into at least two object sets in each failure group; and / or a first classification module, used to classify the yield loss of each monitored object in each failure group according to a preset rule, so as to divide each monitored object into at least two object sets in each failure group.

[0112] Optionally, the division unit 32 includes: a second clustering module, used to cluster the chip yield or the yield loss of each monitored object according to a preset clustering algorithm, so as to divide each monitored object into at least two object sets; and / or a second classification module, used to classify the chip yield or the yield loss of each monitored object according to preset rules, so as to divide each monitored object into at least two object sets.

[0113] Optionally, the device further includes: a selection unit configured to select one of the object sets as a division result based on a preset strategy in response to the object set obtained according to the second clustering module being inconsistent with the object set obtained according to the second classification module.

[0114] Optionally, the search unit 33 is specifically used for at least one of the following: searching for a target set from the at least two object sets based on the degree of difference in the yield information between each of the object sets; searching for a target set from the at least two object sets based on the number of the monitored objects contained in each of the object sets; searching for a target set from the at least two object sets based on the difference between each of the object sets and the preset set rules.

[0115] Optionally, the device also includes: a display unit, which is used to visually display the process information corresponding to each monitoring object in the target set after searching the target set from the at least two object sets, and the process information includes at least one of the following: chip yield, yield loss, processing information, and test information.

[0116] In a third aspect, an embodiment of the present invention further provides an electronic device capable of improving chip output yield.

[0117] like Figure 11 As shown, the electronic device provided by an embodiment of the present invention may include: a shell 51, a processor 52, a memory 53, a circuit board 54 and a power supply circuit 55, wherein the circuit board 54 is placed inside the space enclosed by the shell 51, and the processor 52 and the memory 53 are arranged on the circuit board 54; the power supply circuit 55 is used to supply power to various circuits or devices of the above-mentioned electronic device; the memory 53 is used to store executable program code; the processor 52 runs the program corresponding to the executable program code by reading the executable program code stored in the memory 53, so as to execute the chip yield monitoring method provided by any of the aforementioned embodiments.

[0118] The specific execution process of the above steps by the processor 52 and the steps further executed by the processor 52 by running the executable program code can be found in the description of the above embodiment and will not be repeated here.

[0119] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, which stores one or more programs, and the one or more programs can be executed by one or more processors to implement any chip yield monitoring method provided in the aforementioned embodiments, thereby also achieving the corresponding technical effects, which have been described in detail above and will not be repeated here.

[0120] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply the existence of any such actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

[0121] Each embodiment in this specification is described in a related manner. The same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

[0122] In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0123] For the convenience of description, the above device is described as being divided into various units / modules based on their functions. Of course, when implementing the present invention, the functions of each unit / module can be implemented in the same or multiple software and / or hardware.

[0124] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing related hardware through a computer program. The program can be stored in a computer-readable storage medium, and when executed, the program can include the processes in the above-described method embodiments. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0125] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A chip yield monitoring method, characterized in that: include: Determine yield information for each monitored object based on the test results of the chip in a preset test, wherein the yield information includes chip yield and / or yield loss, and each monitored object includes any of the following: chips from the same batch, chips on the same wafer, and chips within the same preset area on the same wafer; Dividing each of the monitoring objects into at least two object sets according to the yield information; Searching a target set from the at least two object sets to adjust the chip process according to the process information corresponding to each monitoring object in the target set, wherein the target set carries the early warning information of the yield information fluctuation; The step of searching for a target set from the at least two object sets includes at least one of the following: searching for a target set from the at least two object sets according to the degree of difference in the yield information between the object sets; searching for a target set from the at least two object sets according to the number of the monitored objects included in each of the object sets; According to the difference between each of the object sets and the preset set rule, a target set is searched from the at least two object sets.

2. The method according to claim 1, characterized in that Determining the yield information of each monitoring object according to the test results of the chip in the preset test includes: Obtaining from the test log the number of failed chips in each failure category within each monitoring object, the number including the total number of failed chips in each failure category and / or the number of each failure category of failed chips in each failure category; Based on the number of failed chips falling into each failure category in each monitoring object and the total number of chips in the monitoring object, the yield information of each monitoring object is determined respectively to obtain first yield information, wherein the first yield information includes the yield corresponding to the total number of failures and / or the yield loss of each category corresponding to the number of each category of failures.

3. The method according to claim 1, characterized in that Determining the yield information of each monitoring object according to the test results of the chip in the preset test includes: Obtaining from the test log the number of failed chips that fall into each failure category in each monitoring object to obtain the number of failures in each category; Determining a failure group corresponding to each failure category, wherein the failure groups are divided according to causes of chip failure, each failure group corresponds to at least one failure category, and each failure category corresponds to one failure group; Determine the number of failed chips in each monitoring object that fall into each failure group based on the number of failures of each type in each monitoring object and the failure groups corresponding to each failure type, to obtain the number of failures in each group; According to the number of failures in each group in each monitoring object and the total number of chips in the same monitoring object, the yield loss of each failure group in each monitoring object is determined respectively to obtain second yield information.

4. The method according to claim 3, characterized in that The failure group includes at least one of the following: a defect-caused failure group, a device characteristic-caused failure group, a test-caused failure group, and an undetermined failure group.

5. The method according to claim 3, characterized in that The dividing each of the monitoring objects into at least two object sets according to the yield information includes: According to the second yield information, in each of the failure groups, each of the monitoring objects is divided into at least two object sets; The searching for a target set from the at least two object sets includes: Searching for a corresponding candidate set from the at least two object sets corresponding to each failure group, wherein the candidate set carries the first warning information of the yield information fluctuation; The first warning information is input into a preset model to obtain the target set, wherein the preset model includes an artificial intelligence model and / or a script model, and the target set carries the second warning information of the yield information fluctuation.

6. The method according to claim 5, characterized in that The dividing, in each of the failure groups, of the monitoring objects into at least two object sets according to the second yield information includes: Clustering the yield loss of each of the monitored objects in each of the failure groups according to a preset clustering algorithm, so as to divide each of the monitored objects into at least two object sets in each of the failure groups; and / or According to a preset rule, the yield loss of each monitoring object is classified in each failure group, so that each monitoring object is divided into at least two object sets in each failure group.

7. The method according to claim 1, characterized in that The dividing each of the monitoring objects into at least two object sets according to the yield information includes at least one of the following: Clustering the chip yield or the yield loss of each monitoring object according to a preset clustering algorithm to divide each monitoring object into at least two object sets; According to preset rules, the chip yield or the yield loss of each monitoring object is classified to divide each monitoring object into at least two object sets.

8. The method according to claim 7, characterized in that The method further comprises: In response to the object set obtained according to the preset clustering algorithm being inconsistent with the object set obtained according to the preset rule, one of the object sets is selected as the division result based on a preset strategy.

9. The method according to any one of claims 1 to 8, characterized in that After searching the target set from the at least two object sets, the method further includes: The process information corresponding to each monitoring object in the target set is displayed in a visual manner, where the process information includes at least one of the following: chip yield, yield loss, processing information, and test information.

10. A chip yield monitoring device, characterized in that: include: a determination unit, configured to determine yield information of each monitoring object based on test results of the chip in a preset test, wherein the yield information includes chip yield and / or yield loss, and each monitoring object includes any of the following: chips from the same batch, chips on the same wafer, and chips within the same preset area on the same wafer; a dividing unit, configured to divide each of the monitoring objects into at least two object sets according to the yield information; a search unit, configured to search a target set from the at least two object sets to adjust the chip process according to the process information corresponding to each monitoring object in the target set, wherein the target set carries warning information of the yield information fluctuation; The search unit is specifically configured to perform at least one of the following: searching for a target set from the at least two object sets according to the degree of difference in the yield information between the object sets; searching for a target set from the at least two object sets according to the number of the monitored objects included in each of the object sets; According to the difference between each of the object sets and the preset set rule, a target set is searched from the at least two object sets.

11. The device according to claim 10, characterized in that The determining unit includes: A first acquisition module is configured to acquire, from the test log, the number of failed chips in each failure category within each monitoring object, the number including the total number of failed chips in each failure category and / or the number of each failure category of failed chips in each failure category; The first determination module is used to determine the yield information of each monitored object according to the number of failed chips falling into each failure category in each monitored object and the total number of chips in the monitored object, so as to obtain first yield information, wherein the first yield information includes the yield corresponding to the total number of failures and / or the yield loss of each category corresponding to the number of each category of failures.

12. The device according to claim 10, characterized in that The determining unit includes: A second acquisition module is configured to acquire, from the test log, the number of failed chips in each failure category in each monitoring object, to obtain the number of failures in each category; A second determination module is configured to determine a failure group corresponding to each failure category, wherein the failure groups are divided according to causes of chip failure, each failure group corresponds to at least one failure category, and each failure category corresponds to one failure group; a third determining module, configured to determine the number of failed chips in each monitoring object that fall into each failure group based on the number of failures of each type in each monitoring object and the failure groups corresponding to each failure type, to obtain the number of failures in each group; The fourth determination module is used to determine the yield loss of each failure group in each monitoring object according to the number of failures in each group in each monitoring object and the total number of chips in the same monitoring object, so as to obtain second yield information.

13. The device according to claim 12, characterized in that The failure group includes at least one of the following: a defect-caused failure group, a device characteristic-caused failure group, a test-caused failure group, and an undetermined failure group.

14. The device according to claim 12, characterized in that The division unit is specifically configured to divide each of the monitoring objects into at least two object sets in each of the failure groups according to the second yield information; The search unit includes: a search module, configured to search for a corresponding candidate set from the at least two object sets corresponding to each failure group, wherein the candidate set carries first warning information of yield information fluctuation; An input module is used to input the first warning information into a preset model to obtain the target set, wherein the preset model includes an artificial intelligence model and / or a script model, and the target set carries the second warning information of yield information fluctuations.

15. The device according to claim 14, characterized in that The division unit includes: a first clustering module, configured to cluster the yield loss of each of the monitored objects in each of the failure groups according to a preset clustering algorithm, so as to divide each of the monitored objects into at least two object sets in each of the failure groups; and / or The first classification module is configured to classify the yield loss of each monitoring object in each failure group according to a preset rule, so as to divide each monitoring object into at least two object sets in each failure group.

16. The device according to claim 10, characterized in that The division unit includes: A second clustering module is configured to cluster the chip yield or the yield loss of each monitoring object according to a preset clustering algorithm, so as to divide each monitoring object into at least two object sets; and / or The second classification module is configured to classify the chip yield or the yield loss of each monitoring object according to a preset rule, so as to divide each monitoring object into at least two object sets.

17. The device according to claim 16, characterized in that The device further comprises: A selection unit is configured to select one of the object sets as a division result based on a preset strategy in response to the object set obtained according to the second clustering module being inconsistent with the object set obtained according to the second classification module.

18. The device according to any one of claims 10 to 17, characterized in that Also includes: A display unit is used to visually display the process information corresponding to each monitoring object in the target set after searching the target set from the at least two object sets, and the process information includes at least one of the following: chip yield, yield loss, processing information, and test information.

19. An electronic device, characterized in that: The electronic device includes: a housing, a processor, a memory, a circuit board and a power supply circuit, wherein the circuit board is placed inside the space enclosed by the housing, and the processor and the memory are arranged on the circuit board; the power supply circuit is used to supply power to various circuits or devices of the above-mentioned electronic device; the memory is used to store executable program code; the processor runs the program corresponding to the executable program code by reading the executable program code stored in the memory, and is used to execute the chip yield monitoring method described in any one of claims 1 to 9.

20. A computer-readable storage medium, characterized in that The computer-readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the chip yield monitoring method according to any one of claims 1 to 9.

Images