New qualified inspection methods and inspection systems for production tools

Abstract

A qualified inspection method and inspection system for newly added production tools, wherein the inspection method, after obtaining a number of new tool yield data and old tool yield data, performs data analysis on the new tool yield data and old tool yield data to determine whether the new tool yield data and old tool yield data belong to a high yield category or a relatively high yield category. If they do, the corresponding new tool yield data and old tool yield data are eliminated, and the remaining new tool yield data and old tool yield data are used as the screened new tool yield data and old tool yield data; based on the screened new tool yield data and old tool yield data, determine whether the new production tool is qualified. Through the aforementioned inspection method, the qualified inspection process of newly added production tools is standardized and streamlined, and the accuracy of the qualified inspection results of new tools is improved, and the efficiency of the qualified inspection process of new tools is improved.

Description

Technical Field

[0001] The present invention relates to the field of semiconductors, and in particular to a qualified inspection method and inspection system for newly added production tools. Background Art

[0002] An integrated circuit is a miniature electronic device or component. It is made by using semiconductor manufacturing processes such as oxidation, photolithography, diffusion, epitaxy, masking, sputtering, etc. to interconnect the transistors, resistors, capacitors, inductors and other components and wiring required in a circuit, and then make them on a small piece or several small pieces of semiconductor wafers or dielectric substrates, and then encapsulate them in a tube shell to become a miniature structure or chip with the required circuit function.

[0003] When manufacturing integrated circuits, each step of the semiconductor manufacturing process is carried out on corresponding semiconductor production tools (or production equipment), such as the oxidation process using corresponding furnace tube equipment and the photolithography process using corresponding photolithography equipment.

[0004] In order to increase production capacity, new production tools are usually added to the production line. Before the new production tools are officially put into production, the performance of the new production tools needs to be verified to determine whether the new production tools can be used for production or whether they are qualified. Currently, the qualification of the new production tools is usually determined by measuring the yield data of the wafers after the process is processed in the new production tools. There is no unified standard or process for the judgment process, and it is greatly affected by the subjective influence of the process or personnel. The accuracy of the inspection results needs to be improved. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a qualified inspection method and inspection system for newly added production tools, so as to standardize the inspection process and improve the accuracy of the inspection results.

[0006] The present invention provides a method for testing the quality of a newly added production tool, comprising:

[0007] Provide new production tools newly installed on the line and old production tools already on the line;

[0008] Providing a plurality of inspection wafers, wherein the plurality of inspection wafers are divided into a first portion of wafers and a second portion of wafers;

[0009] Performing corresponding process on the first part of wafers in the new production tool;

[0010] performing corresponding process on the second part of wafers in the old production tool;

[0011] Performing yield measurement on the first portion of wafers after being processed in the new production tool to obtain a number of new tool yield data;

[0012] Performing yield measurement on the second portion of wafers that have been processed in the old production tool to obtain a number of old tool yield data;

[0013] Performing data analysis on the new tool yield data and the old tool yield data to determine whether the new tool yield data and the old tool yield data belong to a high yield category or a relatively high yield category; if they do, excluding the corresponding new tool yield data and the old tool yield data, and using the remaining new tool yield data and the old tool yield data as the screened new tool yield data and the old tool yield data;

[0014] Based on the screened new tool yield data and old tool yield data, it is determined whether the new production tool is qualified.

[0015] Optionally, the several wafers to be inspected are several batches of wafers to be inspected, and the odd-numbered wafers in each batch are used as the first part of wafers, and the even-numbered wafers are used as the second part of wafers, or the even-numbered wafers in each batch are used as the first part of wafers, and the odd-numbered wafers are used as the second part of wafers.

[0016] Optionally, the method for performing data analysis on the plurality of new tool yield data and the old tool yield data adopts a data analysis method based on a fuzzy system model.

[0017] Optionally, a process of performing data analysis on the several new tool yield data and the old tool yield data to determine whether the several new tool yield data and the old tool yield data belong to a high yield category or a relatively high yield category includes: dividing the several old tool yield data into several clusters; constructing a fuzzy system model based on the several clusters, the fuzzy system model including category attribution and corresponding distribution function that conforms to the cluster characteristic distribution, the fuzzy system model is one of an α model, a β model or a γ model, the α model includes three category attribution and corresponding three distribution functions, the three category attribution are a low yield category, a medium yield category and a high yield category. Category, the β model includes two category attributions and two corresponding distribution functions, the two category attributions are a relatively low yield category and a relatively high yield category, the γ model includes one category attribution and a corresponding distribution function, the one category attribution is an overall yield category; a number of new tool yield data and old tool yield data are respectively projected into the fuzzy system model to obtain the category attribution corresponding to each new tool yield data and old tool yield data; according to the category attribution corresponding to each new tool yield data and old tool yield data, it is judged whether the several new tool yield data and old tool yield data belong to the high yield category or the relatively high yield category.

[0018] Optionally, the plurality of old tool yield data are divided into a plurality of clusters using a K-Means clustering algorithm.

[0019] Optionally, before dividing the plurality of old tool yield data into a plurality of clusters, the step of determining whether the number of the new tool yield data and the number of the old tool yield data are both greater than 10 is also included; if so, dividing the plurality of old tool yield data into a plurality of clusters is performed; if not, the inspection process is terminated.

[0020] Optionally, the process of dividing into several clusters, constructing a fuzzy system model, obtaining the category affiliation corresponding to each new tool yield data and old tool yield data, judging whether the several new tool yield data and old tool yield data belong to the high yield category or the relatively high yield category, and obtaining the screened new tool yield data and old tool yield data includes: when dividing the several old tool yield data into several clusters, pre-setting the K value in the K-Means clustering algorithm to be equal to 3, and then dividing the several old tool yield data into three clusters through the K-Means clustering algorithm; constructing a fuzzy system model according to the three clusters, the fuzzy system model is an α model; dividing the several new tool yield data into several clusters; The new tool yield data and the old tool yield data are projected into the α model respectively to obtain the category attribution corresponding to each new tool yield data and the old tool yield data; according to the category attribution corresponding to each new tool yield data and the old tool yield data, it is judged whether the several new tool yield data and the old tool yield data belong to the high yield category, if "belong", then the corresponding new tool yield data and the old tool yield data are eliminated, and the remaining new tool yield data and the old tool yield data are used as the screened new tool yield data and the old tool yield data; it is judged whether the number of the screened new tool yield data and the old tool yield data are both greater than 10, if "yes", then the screening based on the screened new tool yield data is performed. and the old tool yield data, to determine whether the new production tool is qualified, if "no", then reduce the K value by 1, and when the K value is equal to 2, divide the plurality of old tool yield data into two clusters by the K-Means clustering algorithm; according to the two clusters, construct a fuzzy system model, which is a β model; project the plurality of new tool yield data and the old tool yield data into the β model respectively, and obtain the category attribution corresponding to each new tool yield data and the old tool yield data; according to the category attribution corresponding to each new tool yield data and the old tool yield data, determine whether the plurality of new tool yield data and the old tool yield data belong to the high yield category, if "belonging" If "is equal to", the corresponding new tool yield data and old tool yield data are eliminated, and the remaining new tool yield data and old tool yield data are used as the screened new tool yield data and old tool yield data; continue to judge whether the number of the screened new tool yield data and old tool yield data are both greater than 10, if "yes", perform the step of judging whether the new production tool is qualified based on the screened new tool yield data and old tool yield data, if "no", reduce the K value by 1, and when the K value is equal to 1, divide the several old tool yield data into 1 cluster by the K-Means clustering algorithm; according to the 1 cluster, construct a fuzzy system model, and the fuzzy system model is a γ model;Projecting a number of new tool yield data and old tool yield data into the γ model respectively, obtaining the category attribution corresponding to each new tool yield data and old tool yield data, and directly using the new tool yield data and old tool yield data as the screened new tool yield data and old tool yield data. ;

[0021] Optionally, the process of judging whether the new production tool is qualified based on the screened new tool yield data and the old tool yield data includes: judging whether the screened new tool yield data and the old tool yield data meet condition 1, wherein condition 1 is that the mean of the screened new tool yield data ≥ the mean of the screened old tool yield data, and the standard deviation of the screened new tool yield data ≤ the standard deviation of the screened old tool yield data; if yes, the new production tool is judged to be qualified; if no, further judgment is performed.

[0022] Optionally, the further judgment includes: judging whether the screened new tool yield data and the old tool yield data meet condition 2, wherein condition 2 is that the mean of the screened new tool yield data is less than the mean of the screened old tool yield data, and the standard deviation of the screened new tool yield data is greater than the standard deviation of the screened old tool yield data; if “yes”, the new production tool is judged to be unqualified; if “no”, a t-test is performed to judge whether there is a significant difference between the screened new tool yield data and the old tool yield data; if “no”, the new production tool is judged to be qualified; if so, judging whether the screened new tool yield data and the old tool yield data meet condition 3; wherein condition 3 is that the mean of the screened new tool yield data is ≥ the mean of the screened old tool yield data; if “yes”, the new production tool is judged to be qualified; if “no”, the new production tool is judged to be qualified or unqualified.

[0023] The present invention also provides a new production tool qualification inspection system, comprising:

[0024] A wafer providing unit, used for providing a plurality of test wafers, wherein the plurality of test wafers are divided into a first portion of wafers and a second portion of wafers;

[0025] A new production tool for performing corresponding process processing on the first part of wafers;

[0026] An old production tool, used for performing corresponding process treatment on the second portion of wafers;

[0027] A yield data measurement unit is used to measure the yield of the first part of the wafers after the process is processed in the new production tool to obtain a number of new tool yield data, and to measure the yield of the second part of the wafers after the process is processed in the old production tool to obtain a number of old tool yield data;

[0028] A data analysis unit is used to perform data analysis on the plurality of new tool yield data and the old tool yield data to determine whether the plurality of new tool yield data and the old tool yield data belong to a high yield category or a relatively high yield category. If they do, the corresponding new tool yield data and the old tool yield data are eliminated, and the remaining new tool yield data and the old tool yield data are used as the screened new tool yield data and the old tool yield data;

[0029] A judging unit is used to judge whether the new production tool is qualified based on the screened new tool yield data and the old tool yield data.

[0030] Optionally, the several wafers to be inspected provided by the wafer providing unit are several batches of wafers to be inspected, and the odd-numbered wafers in each batch are used as the first part of wafers, and the even-numbered wafers are used as the second part of wafers, or the even-numbered wafers in each batch are used as the first part of wafers, and the odd-numbered wafers are used as the second part of wafers.

[0031] Optionally, the method for the data analysis unit to perform data analysis on the plurality of new tool yield data and the old tool yield data adopts a data analysis method based on a fuzzy system model.

[0032] Optionally, the data analysis unit performs data analysis on the several new tool yield data and the old tool yield data, and the process of judging whether the several new tool yield data and the old tool yield data belong to the high yield category or the relatively high yield category includes: dividing the several old tool yield data into several clusters; constructing a fuzzy system model according to the several clusters, the fuzzy system model including category attribution and corresponding distribution function that conforms to the cluster characteristic distribution, the fuzzy system model is one of the α model, β model or γ model, the α model includes three category attribution and corresponding three distribution functions, the three category attribution are low yield category, medium yield category and high yield categories, the β model includes two category attributions and two corresponding distribution functions, the two category attributions are a relatively low yield category and a relatively high yield category, the γ model includes one category attribution and a corresponding distribution function, the one category attribution is an overall yield category; a number of new tool yield data and old tool yield data are respectively projected into the fuzzy system model to obtain the category attribution corresponding to each new tool yield data and old tool yield data; according to the category attribution corresponding to each new tool yield data and old tool yield data, it is judged whether the several new tool yield data and old tool yield data belong to the high yield category or the relatively high yield category.

[0033] Optionally, the plurality of old tool yield data are divided into a plurality of clusters using a K-Means clustering algorithm.

[0034] Optionally, it also includes a data sample number judgment unit, which is used to judge whether the number of new tool yield data and old tool yield data are both greater than 10 before the data analysis unit divides the old tool yield data into several clusters. If "yes", the step of dividing the old tool yield data into several clusters is performed; if "no", the inspection process is terminated.

[0035] Optionally, the data analysis unit performs the division into several clusters, constructs a fuzzy system model, obtains the category affiliation corresponding to each new tool yield data and old tool yield data, determines whether the several new tool yield data and old tool yield data belong to the high yield category or the relatively high yield category, and obtains the screened new tool yield data and old tool yield data. The process includes: when dividing the several old tool yield data into several clusters, presetting the K value in the K-Means clustering algorithm to be equal to 3, and then dividing the several old tool yield data into three clusters through the K-Means clustering algorithm; constructing a fuzzy system model according to the three clusters, wherein the fuzzy system model is an α model; A number of new tool yield data and old tool yield data are projected into the α model respectively to obtain the category attribution corresponding to each new tool yield data and old tool yield data; according to the category attribution corresponding to each new tool yield data and old tool yield data, it is judged whether the new tool yield data and old tool yield data belong to the high yield category, if "belong", the corresponding new tool yield data and old tool yield data are eliminated, and the remaining new tool yield data and old tool yield data are used as the screened new tool yield data and old tool yield data; it is judged whether the number of the screened new tool yield data and old tool yield data is greater than 10, if "yes", the new tool yield data based on the screened The step of judging whether the new production tool is qualified by using the yield data of the new tool and the yield data of the old tool, if "no", reducing the K value by 1, and when the K value is equal to 2, dividing the yield data of the old tools into 2 clusters by the K-Means clustering algorithm; constructing a fuzzy system model according to the 2 clusters, and the fuzzy system model is a β model; projecting the yield data of the new tool and the yield data of the old tool into the β model respectively, and obtaining the category attribution corresponding to each new tool yield data and the yield data of the old tool; judging whether the yield data of the new tool and the yield data of the old tool belong to the high yield category according to the category attribution corresponding to each new tool yield data and the yield data of the old tool, and if "belongs", then the corresponding new tool yield data and old tool yield data are eliminated, and the remaining new tool yield data and old tool yield data are used as the screened new tool yield data and old tool yield data; continue to judge whether the number of screened new tool yield data and old tool yield data are both greater than 10, if "yes", then perform the step of judging whether the new production tool is qualified based on the screened new tool yield data and old tool yield data, if "no", then reduce the K value by 1, and when the K value is equal to 1, the several old tool yield data are divided into 1 cluster by the K-Means clustering algorithm; according to the 1 cluster, construct a fuzzy system model, and the fuzzy system model is a γ model;Projecting a number of new tool yield data and old tool yield data into the γ model respectively, obtaining the category attribution corresponding to each new tool yield data and old tool yield data, and directly using the new tool yield data and old tool yield data as the screened new tool yield data and old tool yield data. ;

[0036] Optionally, the judgment unit judges whether the new production tool is qualified based on the screened new tool yield data and the old tool yield data, and the process includes: judging whether the screened new tool yield data and the old tool yield data meet condition 1, and condition 1 is that the mean of the screened new tool yield data ≥ the mean of the screened old tool yield data, and the standard deviation of the screened new tool yield data ≤ the standard deviation of the screened old tool yield data. If "yes", the new production tool is determined to be qualified, if "no", further judgment is performed.

[0037] Optionally, the further judgment includes: judging whether the screened new tool yield data and the old tool yield data meet condition 2, wherein condition 2 is that the mean of the screened new tool yield data is less than the mean of the screened old tool yield data, and the standard deviation of the screened new tool yield data is greater than the standard deviation of the screened old tool yield data; if “yes”, the new production tool is judged to be unqualified; if “no”, a t-test is performed to judge whether there is a significant difference between the screened new tool yield data and the old tool yield data; if “no”, the new production tool is judged to be qualified; if so, judging whether the screened new tool yield data and the old tool yield data meet condition 3; wherein condition 3 is that the mean of the screened new tool yield data is ≥ the mean of the screened old tool yield data; if “yes”, the new production tool is judged to be qualified; if “no”, the new production tool is judged to be qualified or unqualified.

[0038] Compared with the prior art, the technical solution of the present invention has the following advantages:

[0039] The qualified inspection method of the newly added production tools of the present invention, after obtaining a number of new tool yield data and old tool yield data, performs data analysis on the new tool yield data and old tool yield data to determine whether the new tool yield data and old tool yield data belong to the high yield category or the relatively high yield category. If they "belong", the corresponding new tool yield data and old tool yield data are eliminated, and the remaining new tool yield data and old tool yield data are used as the screened new tool yield data and old tool yield data; based on the screened new tool yield data and old tool yield data, determine whether the new production tool is qualified. Through the above inspection method, the qualified inspection process of the newly added production tools is standardized and streamlined, and in the inspection process, a number of new tool yield data and old tool yield data are used as original data, and the new tool yield data and old tool yield data of the high yield category or the relatively high yield category that will have a greater impact on the accuracy and effectiveness of the new tool qualified inspection results are used to improve the accuracy of the new tool qualified inspection results and improve the efficiency of the new tool qualified inspection process.

[0040] The process of dividing the data into several clusters, constructing a fuzzy system model, obtaining the category attribution corresponding to each new tool yield data and old tool yield data, judging whether the several new tool yield data and old tool yield data belong to the high yield category or the relatively high yield category, and obtaining the screened new tool yield data and old tool yield data includes: when dividing the several old tool yield data into several clusters, presetting the K value in the K-Means clustering algorithm to be equal to 3, and then dividing the several old tool yield data into three clusters by using the K-Means clustering algorithm; constructing a fuzzy system model according to the three clusters, wherein the fuzzy system model is an α model; dividing the several new tool yield data into several clusters; and old tool yield data are projected into the α model respectively to obtain the category attribution corresponding to each new tool yield data and old tool yield data; according to the category attribution corresponding to each new tool yield data and old tool yield data, it is judged whether the several new tool yield data and old tool yield data belong to the high yield category, if "belong", then the corresponding new tool yield data and old tool yield data are eliminated, and the remaining new tool yield data and old tool yield data are used as the screened new tool yield data and old tool yield data; it is judged whether the number of the screened new tool yield data and old tool yield data is greater than 10, if "yes", then the screening based on the screened new tool yield data and old tool yield data is performed. The step of judging whether the new production tool is qualified is as follows: if “no”, the K value is reduced by 1. When the K value is equal to 2, the several old tool yield data are divided into two clusters by the K-Means clustering algorithm; according to the two clusters, a fuzzy system model is constructed, and the fuzzy system model is a β model; the several new tool yield data and the old tool yield data are respectively projected into the β model to obtain the category attribution corresponding to each new tool yield data and the old tool yield data; according to the category attribution corresponding to each new tool yield data and the old tool yield data, it is judged whether the several new tool yield data and the old tool yield data belong to the high yield category, if “belonging to ", then the corresponding new tool yield data and old tool yield data are eliminated, and the remaining new tool yield data and old tool yield data are used as the screened new tool yield data and old tool yield data; continue to judge whether the number of screened new tool yield data and old tool yield data are both greater than 10, if "yes", then perform the step of judging whether the new production tool is qualified based on the screened new tool yield data and old tool yield data, if "no", then reduce the K value by 1, and when the K value is equal to 1, the several old tool yield data are divided into 1 cluster by the K-Means clustering algorithm; according to the 1 cluster, construct a fuzzy system model, and the fuzzy system model is a γ model;Project a number of new tool yield data and old tool yield data into the γ model respectively, obtain the category attribution corresponding to each new tool yield data and old tool yield data, and directly use the new tool yield data and old tool yield data as the screened new tool yield data and old tool yield data. Further improve the accuracy of the obtained category attribution corresponding to each new tool yield data and old tool yield data, so as to further improve the accuracy of the new tool qualification inspection result. ; BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1-Figure 4 A schematic diagram of a flow chart of a qualified inspection method for a newly added production tool according to an embodiment of the present invention;

[0042] Figure 5-Figure 8 This is a structural schematic diagram of a qualified inspection process for a newly added production tool in an embodiment of the present invention;

[0043] Fig. 9 This is a schematic diagram of the structure of a qualified inspection system for newly added production tools in an embodiment of the present invention. DETAILED DESCRIPTION

[0044] As mentioned in the background technology, there is no unified standard or process for judging whether new production tools are qualified, and the process is greatly affected by the subjective influence of the process or personnel, and the accuracy of the inspection results needs to be improved.

[0045] To this end, the present invention provides a qualified inspection method and inspection system for newly added production tools. The inspection method, after obtaining a number of new tool yield data and old tool yield data, performs data analysis on the new tool yield data and old tool yield data to determine whether the new tool yield data and old tool yield data belong to a high yield category or a relatively high yield category. If they "belong", the corresponding new tool yield data and old tool yield data are eliminated, and the remaining new tool yield data and old tool yield data are used as the screened new tool yield data and old tool yield data; based on the screened new tool yield data and old tool yield data, it is determined whether the new production tool is qualified. Through the above inspection method, the qualified inspection process of newly added production tools is standardized and streamlined, and in the inspection process, a number of new tool yield data and old tool yield data are used as original data, and the new tool yield data and old tool yield data of the high yield category or the relatively high yield category that will have a greater impact on the accuracy and effectiveness of the new tool qualified inspection results are used to improve the accuracy of the new tool qualified inspection results and improve the efficiency of the new tool qualified inspection process.

[0046] In order to make the above-mentioned objects, features and advantages of the present invention more clearly understood, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. When describing the embodiments of the present invention in detail, for the sake of convenience, the schematic diagrams will not be partially enlarged according to the general proportion, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0047] refer to Figure 1 An embodiment of the present invention provides a semiconductor product grading method, comprising the steps of:

[0048] Step S20, providing new production tools newly installed on the line and old production tools already on the line;

[0049] Step S21, providing a plurality of test wafers, wherein the plurality of test wafers are divided into a first portion of wafers and a second portion of wafers;

[0050] Step S22, performing corresponding process treatment on the first part of wafers in the new production tool;

[0051] Step S23, performing corresponding process on the second part of wafers in the old production tool;

[0052] Step S24, performing yield measurement on the first portion of wafers after being processed in the new production tool to obtain some new tool yield data;

[0053] Step S25, performing yield measurement on the second portion of wafers processed in the old production tool to obtain some old tool yield data;

[0054] Step S26, determining whether the number of the new tool yield data and the old tool yield data are both greater than 10, if "yes", proceeding to step S27, if "no", proceeding to step S29, and ending the inspection process.

[0055] Step S27, analyzing the new tool yield data and the old tool yield data to determine whether the new tool yield data and the old tool yield data belong to a high yield category or a relatively high yield category. If they do, the corresponding new tool yield data and the old tool yield data are removed, and the remaining new tool yield data and the old tool yield data are used as the screened new tool yield data and the old tool yield data;

[0056] Step S28, judging whether the new production tool is qualified based on the screened new tool yield data and the old tool yield data.

[0057] The above process is described in detail below with reference to the accompanying drawings.

[0058] Step S20 is performed to provide new production tools newly installed on the line and old production tools already on the line.

[0059] The old production tools and the newly installed new production tools are tools or equipment used to perform corresponding process processing on wafers on the production line (Fab). The old production tools are already in use on the production line, and their various performance and yield rates meet the process requirements. The newly installed new production tools are equipment that needs to be inspected and needs to be judged whether they are qualified, and have not yet been officially used in production.

[0060] The new production tools newly installed on the line and the old production tools already on the line are the same tools or the same type of tools. In a specific embodiment, the new production tools and the old production tools are production tools or production equipment that can perform (semiconductor) process treatments such as oxidation, lithography, deposition, ion implantation, epitaxy, grinding, etching or sputtering. Specifically, the new production tools and the old production tools can both be one of furnace tube equipment, lithography equipment, deposition equipment, ion implantation equipment, grinding equipment, etching equipment or sputtering equipment. In other embodiments, the new production tools and the old production tools can also be production tools or production equipment used for other process treatments on the production line.

[0061] In step S21 , a plurality of inspection wafers are provided, wherein the plurality of inspection wafers are divided into a first portion of wafers and a second portion of wafers.

[0062] The test wafers are wafers that need to undergo relevant process processing on a new production tool or an old production tool and then undergo yield testing. The test wafers are divided into a first part of wafers and a second part of wafers. The first part of wafers are subsequently subjected to corresponding process processing in the new production tool, and the second part of wafers are subsequently subjected to corresponding process processing in the old production tool.

[0063] In one embodiment, the plurality of wafers to be inspected are batches of wafers to be inspected, and the odd-numbered wafers in each batch are used as the first part of wafers and the even-numbered wafers are used as the second part of wafers, or the even-numbered wafers in each batch are used as the first part of wafers and the odd-numbered wafers are used as the second part of wafers. When performing subsequent inspections, the yield of the wafer itself has the least effect on the inspection results, thereby improving the accuracy of the inspection results.

[0064] In one embodiment, the number of the first part of wafers and the number of the second part of wafers are both greater than 10, thereby increasing the number of valid samples of yield data subsequently acquired.

[0065] Perform step S22 to perform corresponding process treatment on the first part of wafers in the new production tool; perform step S23 to perform corresponding process treatment on the second part of wafers in the old production tool.

[0066] Specifically, the first part of the wafers are sequentially subjected to corresponding process treatments in the new production tool, and the second part of the wafers are sequentially subjected to corresponding process treatments in the old production tool. The process treatment is one of oxidation, photolithography, deposition, ion implantation, epitaxy, grinding, etching or sputtering.

[0067] Perform step S24 to measure the yield of the first portion of wafers after process processing in the new production tool to obtain some new tool yield data; perform step S25 to measure the yield of the second portion of wafers after process processing in the old production tool to obtain some old tool yield data.

[0068] The process of measuring the first portion of wafers and the process of measuring the second portion of wafers are performed on the same measuring equipment, and the measuring procedures and parameters to be measured are the same.

[0069] Each wafer has a yield data after measurement, that is, each wafer in the first part of wafers has a corresponding new tool yield data, and each wafer in the second part of wafers has a corresponding old tool yield data after measurement.

[0070] In one embodiment, the new tool yield data and the new tool yield data are both average values ​​of several yields obtained by measurements on the corresponding wafer.

[0071] In one embodiment, step S26 is also required to determine whether the number of the new tool yield data and the old tool yield data are both greater than 10. If "yes", step S27 is performed, if "no", step S29 is performed to end the inspection process.

[0072] The purpose of performing step S26 is to ensure that there are enough samples when performing data analysis in the subsequent step S27, thereby improving the accuracy of the data analysis. In other embodiments, step S26 may not be performed, and step S27 may be performed directly.

[0073] Perform step S27 to analyze the new tool yield data and the old tool yield data to determine whether the new tool yield data and the old tool yield data belong to the high yield category or the relatively high yield category. If they do, discard the corresponding new tool yield data and the old tool yield data, and use the remaining new tool yield data and the old tool yield data as the screened new tool yield data and the old tool yield data.

[0074] The study found that in the process of judging whether a new tool is qualified, the yield data of new tools in the high-yield category or the relatively high-yield category and the yield data of old tools will have a greater impact on the accuracy and effectiveness of the new tool qualification inspection results. Therefore, in the process of judging whether a new tool is qualified, it is necessary to eliminate the yield data of new tools in the high-yield category or the relatively high-yield category and the yield data of old tools in the high-yield category to improve the accuracy of the new tool qualification inspection results and improve the new tool qualification inspection process.

[0075] The method for performing data analysis on the yield data of the new tools and the yield data of the old tools adopts a data analysis method based on fuzzy system models (DA-FSMs).

[0076] In one embodiment, please refer to Figure 2 , the specific process of step S27 may include: step S270, dividing the plurality of old tool yield data into a plurality of clusters; step S271, constructing a fuzzy system model according to the plurality of clusters, the fuzzy system model including category attribution and corresponding distribution function that conforms to the cluster characteristic distribution, the fuzzy system model is one of an α model, a β model or a γ model, the α model includes three category attributions and corresponding three distribution functions, the three category attributions are a low yield category, a medium yield category and a high yield category, the β model includes two category attributions and corresponding two distribution functions, the two category attributions are a relatively low yield category and a relatively high yield category, the γ model includes a category attribution and a corresponding distribution function number, and the one category is assigned to the overall yield category; step S272, projecting a number of new tool yield data and old tool yield data into the fuzzy system model respectively to obtain the category assignment corresponding to each new tool yield data and old tool yield data; step S273, judging whether the new tool yield data and old tool yield data belong to the high yield category or the relatively high yield category according to the category assignment corresponding to each new tool yield data and old tool yield data; step S274, if they "belong to" the high yield category or the relatively high yield category, then eliminating the corresponding new tool yield data and old tool yield data, and the remaining new tool yield data and old tool yield data are used as the screened new tool yield data and old tool yield data.

[0077] Specifically, in step S270, the plurality of old tool yield data may be divided into a plurality of clusters by using a K-Means clustering algorithm or other grouping or clustering algorithms.

[0078] In one embodiment, a K-MEANS algorithm is used to divide the yield data of the plurality of old tools into a plurality of clusters as an example for illustration, including the steps of:

[0079] (1) The yield data of the old tools are set as a point set S, which needs to be divided into N categories or clusters, where N is set as needed;

[0080] (2) Set K equal to N and randomly select N points as the initial center points;

[0081] (3) Calculate the distance between each point and the N center points, select the nearest center point, and divide it into the group centered on the center point;

[0082] (4) Recalculate the center points of N new clusters;

[0083] (5) If the center point remains unchanged, the K-Means process ends. Otherwise, repeat steps (3) and (4).

[0084] In the present application, the plurality of old tool yield data are divided into at most 3 clusters, which may be 3 clusters, 2 clusters, or 1 cluster, so that the efficiency of the subsequent construction of the fuzzy system model can be improved, and the constructed fuzzy system model can more easily and accurately reflect the category of the new tool and old tool yield data. In other embodiments, the plurality of old tool yield data may be divided into more clusters.

[0085] In one embodiment, the K value is taken as 3 as an example for explanation. Figure 5 , Figure 5 The top graph is a distribution curve of yield data of several old tools, where the horizontal axis represents yield and the vertical axis represents quantity. Figure 5 The middle figure is the distribution diagram of the three clusters after K-Means, where the horizontal axis represents the yield and the vertical axis represents the quantity. The three clusters correspond to three center points, namely C1, C2 and C3.

[0086] Continue to refer Figure 2, proceed to step S271, and construct a fuzzy system model according to the plurality of clusters, wherein the fuzzy system model includes category attribution and corresponding distribution function that conform to the cluster characteristic distribution. When constructing the fuzzy system model, the number of the category attribution and the corresponding distribution function is determined according to the number of clusters. For example, when divided into 3 clusters, there are 3 category attributions and corresponding 3 distribution functions. In one embodiment, the fuzzy system model is one of an α model, a β model or a γ model, and the α model includes three category attributions and corresponding three distribution functions, and the three category attributions are a low yield category, a medium yield category and a high yield category, and the β model includes two category attributions and corresponding two distribution functions, and the two category attributions are a relatively low yield category and a relatively high yield category, and the γ model includes one category attribution and a corresponding one distribution function, and the one category attribution is an overall yield category. Specifically, when the several old tool yield data are divided into 3 clusters in step S270, the fuzzy system model constructed is an α model, when the several old tool yield data are divided into 2 clusters in step S270, the fuzzy system model constructed is a β model, and when the several old tool yield data are divided into 1 cluster in step S270, the fuzzy system model constructed is a γ model.

[0087] In one embodiment, please refer to Figure 5 , Figure 5 The bottom figure in the figure is a distribution line graph of several old tool yield data obtained by building a fuzzy system model, where the horizontal axis represents the yield and the vertical axis represents the probability. Based on this distribution line graph and the three center points C1, C2, and C3, the category attribution and corresponding distribution function that conforms to the cluster characteristic distribution in the fuzzy system model can be obtained. For details, please refer to Figure 6 , Figure 6 is a schematic diagram of the structure of the α model. The α model is a fuzzy system model constructed when the yield data of the old tools are divided into three clusters. The α model includes three categories and three corresponding distribution functions. The three categories are low yield category f1, medium yield category f2 and high yield category f3. The three categories are related to the three distribution functions f1(x j ), f2(x j ), f3(x j ), C1, C2, and C3 represent the yield values ​​corresponding to the three center points, and x j represents the yield variable.

[0088] In another embodiment, please refer to Figure 7 , Figure 7The structural diagram of the β model is shown in FIG. β model is a fuzzy system model constructed when the yield data of the old tools are divided into two clusters. The β model includes two categories and two corresponding distribution functions. The two categories are low yield category f4 and high yield category f5. The two categories are related to the two distribution functions f4(x j ), f5(x j ), C1 and C2 represent the yield values ​​corresponding to the two center points, and x j represents the yield variable.

[0089] In another embodiment, please refer to Figure 8 , Figure 8 The schematic diagram of the structure of the γ model is a fuzzy system model constructed when the yield data of the old tools are divided into one cluster. The γ model includes: a category attribution and a corresponding distribution function, the category attribution is the overall yield category f6, the category attribution and a distribution function f6(x j ) corresponds to x j represents the yield variable.

[0090] In step S272, a number of new tool yield data and old tool yield data are respectively projected into the fuzzy system model to obtain the corresponding category of each new tool yield data and old tool yield data. Specifically, a number of new tool yield data and old tool yield data are respectively projected into one of the α model, β model or γ model to obtain the corresponding category of each new tool yield data and old tool yield data. The corresponding category is the category corresponding to the distribution function when the maximum probability is obtained by calculating a certain distribution function. For example, when a number of new tool yield data and old tool yield data are respectively projected into the α model, a number of new tool yield data and old tool yield data are used as variables x j Project to Figure 6 The distribution function f1(x j ), f2(x j ) and f3(x j ), we can obtain the corresponding probability if the distribution function f1(x j ) is the largest probability, then the category corresponding to the new tool yield data or the old tool yield data is classified as the “low yield category”. If the distribution function f2(x j ) is the largest probability, then the category corresponding to the new tool yield data or the old tool yield data belongs to the “medium yield category”. If the distribution function f3(x j) is the largest, then the category corresponding to the new tool yield data or the old tool yield data is classified as a "high yield category". The process of projecting the new tool yield data and the old tool yield data into the β model or the γ model to obtain the category attribution corresponding to each new tool yield data and the old tool yield data is similar to the process of projecting them into the α model to obtain the category attribution corresponding to each new tool yield data and the old tool yield data.

[0091] Step S273, judging whether the new tool yield data and old tool yield data belong to the high yield category or the relatively high yield category according to the category affiliation corresponding to each of the new tool yield data and the old tool yield data; Step S274, if they "belong to" the high yield category or the relatively high yield category, then eliminating the corresponding new tool yield data and old tool yield data, and the remaining new tool yield data and old tool yield data are used as the screened new tool yield data and old tool yield data. By eliminating the new tool yield data and old tool yield data of the high yield category or the relatively high yield category, the accuracy and efficiency of the judgment can be improved when performing step S28 later.

[0092] In one embodiment, in order to further improve the accuracy of the classification of each new tool yield data and the old tool yield data, so as to further improve the accuracy of the new tool qualification test result, reference Figure 3, when performing step S270 to divide the plurality of old tool yield data into a plurality of clusters, the K value in the K-Means clustering algorithm is pre-set to be equal to 3, and then the plurality of old tool yield data are divided into three clusters by the K-Means clustering algorithm; when performing step S271, a fuzzy system model is constructed according to the three clusters, and the fuzzy system model is an α model, and the fuzzy system model includes a category attribution and a corresponding distribution function that conforms to the cluster feature distribution, and performing step S272 to project a plurality of new tool yield data and old tool yield data into the α model respectively, and obtain the category attribution corresponding to each new tool yield data and the old tool yield data; performing step S273 according to each of the new tool yield data When judging whether the several new tool yield data and the old tool yield data belong to the high yield category or the relatively high yield category, it is only necessary to judge whether the several new tool yield data and the old tool yield data belong to the high yield category according to the category corresponding to each of the new tool yield data and the old tool yield data; perform step S274. If they "belong to" the high yield category or the relatively high yield category, the corresponding new tool yield data and the old tool yield data are eliminated, and the remaining new tool yield data and the old tool yield data are used as the screened new tool yield data and the old tool yield data, and it is only necessary to perform step S275. If they "belong to" the high yield category, the corresponding new tool yield data and the old tool yield data are eliminated, and the remaining new tool yield data and the old tool yield data are used as the screened new tool yield data and the old tool yield data. The yield data of the new tool and the yield data of the old tool are used as the screened yield data of the new tool and the yield data of the old tool; after obtaining the screened yield data of the new tool and the yield data of the old tool, step S275 needs to be performed to determine whether the number of the screened yield data of the new tool and the yield data of the old tool are both greater than 10. If "yes", step S28 is performed to determine whether the new production tool is qualified based on the screened yield data of the new tool and the yield data of the old tool. If "no", step S276 is performed to reduce the K value by 1, and then when the K value is equal to 2, step S270 is continued to divide the several old tool yield data into 2 clusters by the K-Means clustering algorithm; then step S271 is performed to classify the two clusters according to the two clusters. , construct a fuzzy system model, the fuzzy system model is a β model; then proceed to step S272, project a number of new tool yield data and old tool yield data into the β model respectively, and obtain the category attribution corresponding to each new tool yield data and old tool yield data; then proceed to step S273, according to the category attribution corresponding to each new tool yield data and old tool yield data, determine whether the several new tool yield data and old tool yield data belong to the high yield category, and then proceed to step S274, if they "belong" to the high yield category, then eliminate the corresponding new tool yield data and old tool yield data, and the remaining new tool yield data and old tool yield data are used as the screened new tool yield data and old tool yield data;Then, step S275 is performed to continue to determine whether the number of the new tool yield data and the old tool yield data after screening is greater than 10. If "yes", step S28 is performed to determine whether the new production tool is qualified based on the new tool yield data and the old tool yield data after screening. If "no", step S276 is performed to reduce the K value by 1, and then step S270 is performed to divide the old tool yield data into 1 cluster by using the K-Means clustering algorithm when the K value is equal to 1; step S271 is performed to construct a fuzzy system model according to the 1 cluster, and the fuzzy system model is a γ model; step S272 is performed to project the new tool yield data and the old tool yield data into the γ model respectively, obtain the category attribution corresponding to each new tool yield data and the old tool yield data, and directly use the new tool yield data and the old tool yield data as the new tool yield data and the old tool yield data after screening. ;

[0093] Continue to refer Figure 1 , proceed to step S28, based on the screened new tool yield data and the old tool yield data, determine whether the new production tool is qualified.

[0094] In one embodiment, reference Figure 4 , the process of judging whether the new production tool is qualified based on the screened new tool yield data and the old tool yield data includes: performing step S280, judging whether the screened new tool yield data and the old tool yield data meet condition 1, and condition 1 is that the mean of the screened new tool yield data ≥ the mean of the screened old tool yield data, and the standard deviation of the screened new tool yield data ≤ the standard deviation of the screened old tool yield data, if "yes", perform step S285, then judge that the new production tool is qualified, if "no", perform further judgment. The purpose of the next step of judgment is to further verify the data that does not meet condition 1, so as to further judge whether the new production tool is qualified, prevent qualified production tools from being judged as unqualified, and further improve the accuracy of the new tool qualification inspection results. In other embodiments, no further judgment step may be performed.

[0095] In one embodiment, please continue to refer to Figure 4, the further judgment includes: performing step S281 to judge whether the screened new tool yield data and the old tool yield data meet condition 2, the condition 2 is that the mean of the screened new tool yield data is less than the mean of the screened old tool yield data, and the standard deviation of the screened new tool yield data is greater than the standard deviation of the screened old tool yield data. If "yes", perform step S285 to determine that the new production tool is unqualified. If "no", perform step S282 to perform a t-test to determine the yield of the screened new tool. Is there a significant difference between the data and the yield data of the old tool? If "No", proceed to step S284 to determine whether the new production tool is qualified. If so, proceed to step S283 to determine whether the screened new tool yield data and the old tool yield data meet condition 3, and condition 3 is that the mean of the screened new tool yield data ≥ the mean of the screened old tool yield data. If "Yes", proceed to step S284 to determine whether the new production tool is qualified. If "No", proceed to step S285 to determine whether the new production tool is qualified or not.

[0096] In one embodiment, when performing step S282, the t-test (Student's t test) adopts a two-sided test, the statistical significance level α=0.05, and two hypothesis tests: H0: there is a significant difference between the new tool yield data and the old tool yield data, H1: there is no significant difference between the new tool yield data and the old tool yield data, the t-test will obtain one of the results (support H0 and reject H1) or (support H1 and reject H0). If H0 is supported and H1 is rejected, it means that our first hypothesis H0 (there is a significant difference) is proven to be correct, that is, there is a significant difference between the new tool yield data and the old tool yield data. On the contrary, if the hypothesis H1 is supported, it means that there is no significant difference between the new tool yield data and the old tool yield data.

[0097] The embodiment of the present invention also provides a new production tool qualification inspection system, referring to Fig. 9 ,include:

[0098] The wafer providing unit 301 is used to provide a plurality of test wafers, wherein the plurality of test wafers are divided into a first portion of wafers and a second portion of wafers;

[0099] A new production tool 302, used for performing corresponding process treatment on the first part of wafers;

[0100] An old production tool 303, used for performing corresponding process treatment on the second portion of wafers;

[0101] The yield data measurement unit 304 is used to measure the yield of the first part of the wafers after the process is processed in the new production tool to obtain a number of new tool yield data, and to measure the yield of the second part of the wafers after the process is processed in the old production tool to obtain a number of old tool yield data;

[0102] The data analysis unit 305 is used to perform data analysis on the new tool yield data and the old tool yield data to determine whether the new tool yield data and the old tool yield data belong to the high yield category or the relatively high yield category. If they do, the corresponding new tool yield data and the old tool yield data are removed, and the remaining new tool yield data and the old tool yield data are used as the screened new tool yield data and the old tool yield data;

[0103] The judging unit 306 is used to judge whether the new production tool is qualified based on the screened new tool yield data and old tool yield data.

[0104] In one embodiment, the plurality of wafers to be inspected provided by the wafer providing unit 301 are plurality of batches of wafers to be inspected, and the odd-numbered wafers in each batch are used as the first part of wafers and the even-numbered wafers are used as the second part of wafers, or the even-numbered wafers in each batch are used as the first part of wafers and the odd-numbered wafers are used as the second part of wafers.

[0105] The method for the data analysis unit 305 to analyze the plurality of new tool yield data and the old tool yield data adopts a data analysis method based on a fuzzy system model.

[0106] In one embodiment, the data analysis unit 305 performs data analysis on the new tool yield data and the old tool yield data, and the process of determining whether the new tool yield data and the old tool yield data belong to a high yield category or a relatively high yield category includes: dividing the old tool yield data into a plurality of clusters; constructing a fuzzy system model according to the plurality of clusters, wherein the fuzzy system model includes a category attribution and a corresponding distribution function that conforms to the cluster characteristic distribution, wherein the fuzzy system model is one of an α model, a β model or a γ model, wherein the α model includes three category attributions and corresponding three distribution functions, wherein the three category attributions are a low yield category, a medium yield category, a low yield category, a medium yield category, a high ... The beta model includes two categories and two corresponding distribution functions, the two categories are the low yield category and the high yield category, the gamma model includes one category and one corresponding distribution function, the one category is the overall yield category; a number of new tool yield data and old tool yield data are respectively projected into the fuzzy system model to obtain the category attribution corresponding to each new tool yield data and old tool yield data; according to the category attribution corresponding to each new tool yield data and old tool yield data, it is judged whether the several new tool yield data and old tool yield data belong to the high yield category or the high yield category.

[0107] In one embodiment, the plurality of old tool yield data are divided into a plurality of clusters using a K-Means clustering algorithm or other grouping or clustering algorithms.

[0108] In one embodiment, a data sample number judgment unit (not shown in the figure) is also included, which is used to judge whether the number of new tool yield data and old tool yield data are both greater than 10 before the data analysis unit divides the old tool yield data into several clusters. If "yes", the step of dividing the old tool yield data into several clusters is performed, and if "no", the inspection process is terminated.

[0109] In one embodiment, the data analysis unit 305 performs the division into several clusters, constructs a fuzzy system model, obtains the category attribution corresponding to each new tool yield data and old tool yield data, determines whether the several new tool yield data and old tool yield data belong to the high yield category or the relatively high yield category, and obtains the screened new tool yield data and old tool yield data. The process includes: when dividing the several old tool yield data into several clusters, presetting the K value in the K-Means clustering algorithm to be equal to 3, and then dividing the several old tool yield data into three clusters by using the K-Means clustering algorithm; constructing a fuzzy system model according to the three clusters, and the fuzzy system model is α Model; project several new tool yield data and old tool yield data into the α model respectively to obtain the category belonging to each new tool yield data and old tool yield data; according to the category belonging to each new tool yield data and old tool yield data, determine whether the several new tool yield data and old tool yield data belong to the high yield category, if "belong", remove the corresponding new tool yield data and old tool yield data, and use the remaining new tool yield data and old tool yield data as the screened new tool yield data and old tool yield data; determine whether the number of the screened new tool yield data and old tool yield data is greater than 10, if "yes", perform the screening based on the screened The step of judging whether the new production tool is qualified is performed by using the new tool yield data and the old tool yield data, if "no", reducing the K value by 1, and when the K value is equal to 2, dividing the plurality of old tool yield data into two clusters by using the K-Means clustering algorithm; constructing a fuzzy system model according to the two clusters, and the fuzzy system model is a β model; projecting the plurality of new tool yield data and the old tool yield data into the β model respectively, and obtaining the category attribution corresponding to each new tool yield data and the old tool yield data; judging whether the plurality of new tool yield data and the old tool yield data belong to the high yield category according to the category attribution corresponding to each new tool yield data and the old tool yield data , if "belongs", then the corresponding new tool yield data and old tool yield data are eliminated, and the remaining new tool yield data and old tool yield data are used as the screened new tool yield data and old tool yield data; continue to judge whether the number of screened new tool yield data and old tool yield data are both greater than 10, if "yes", then perform the step of judging whether the new production tool is qualified based on the screened new tool yield data and old tool yield data, if "no", then reduce the K value by 1, and when the K value is equal to 1, the several old tool yield data are divided into 1 cluster by the K-Means clustering algorithm; according to the 1 cluster, construct a fuzzy system model, and the fuzzy system model is a γ model;Projecting a number of new tool yield data and old tool yield data into the γ model respectively, obtaining the category attribution corresponding to each new tool yield data and old tool yield data, and directly using the new tool yield data and old tool yield data as the screened new tool yield data and old tool yield data. ;

[0110] In one embodiment, the judgment unit 306 judges whether the new production tool is qualified based on the screened new tool yield data and the old tool yield data, and the process includes: judging whether the screened new tool yield data and the old tool yield data meet condition 1, and condition 1 is that the mean of the screened new tool yield data ≥ the mean of the screened old tool yield data, and the standard deviation of the screened new tool yield data ≤ the standard deviation of the screened old tool yield data. If "yes", the new production tool is determined to be qualified, and if "no", further judgment is performed.

[0111] In one embodiment, the further judgment includes: judging whether the screened new tool yield data and the old tool yield data meet condition 2, wherein condition 2 is that the mean of the screened new tool yield data is less than the mean of the screened old tool yield data, and the standard deviation of the screened new tool yield data is greater than the standard deviation of the screened old tool yield data; if "yes", the new production tool is judged to be unqualified; if "no", a t-test is performed to judge whether there is a significant difference between the screened new tool yield data and the old tool yield data; if "no", the new production tool is judged to be qualified; if so, judging whether the screened new tool yield data and the old tool yield data meet condition 3; wherein condition 3 is that the mean of the screened new tool yield data is ≥ the mean of the screened old tool yield data; if "yes", the new production tool is judged to be qualified; if "no", the new production tool is judged to be qualified or unqualified.

[0112] It should be noted that the limitations or descriptions of the same or similar parts in this embodiment (inspection system) as those in the aforementioned embodiment (inspection system) will not be repeated in this embodiment. Please refer to the limitations or descriptions of the corresponding parts in the aforementioned embodiment.

[0113] Although the present invention has been disclosed as above in the form of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art may make possible changes and modifications to the technical solution of the present invention by using the methods and technical contents disclosed above without departing from the spirit and scope of the present invention. Therefore, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the protection scope of the technical solution of the present invention.

Claims

1. A method for testing the quality of a newly added production tool, characterized in that: include: Provide new production tools newly installed on the line and old production tools already on the line; Providing a plurality of inspection wafers, wherein the plurality of inspection wafers are divided into a first portion of wafers and a second portion of wafers; Performing corresponding process on the first part of wafers in the new production tool; performing corresponding process on the second portion of wafers in the old production tool; Performing yield measurement on a first portion of wafers after being processed in the new production tool to obtain a number of new tool yield data; Performing yield measurement on the second portion of wafers processed in the old production tool to obtain a number of old tool yield data; Performing data analysis on the new tool yield data and the old tool yield data to determine whether the new tool yield data and the old tool yield data belong to a high yield category or a relatively high yield category; if they do, excluding the corresponding new tool yield data and the old tool yield data, and using the remaining new tool yield data and the old tool yield data as the screened new tool yield data and the old tool yield data; Based on the screened new tool yield data and old tool yield data, it is determined whether the new production tool is qualified.

2. The qualified inspection method for newly added production tools according to claim 1, characterized in that: The several inspection wafers are several batches of wafers to be inspected, and the odd-numbered wafers in each batch are correspondingly used as the first part of wafers and the even-numbered wafers are used as the second part of wafers, or the even-numbered wafers in each batch are used as the first part of wafers and the odd-numbered wafers are used as the second part of wafers.

3. The qualified inspection method for newly added production tools according to claim 1, characterized in that: The method for analyzing the yield data of the new tools and the yield data of the old tools adopts a data analysis method based on a fuzzy system model.

4. The qualified inspection method for newly added production tools as claimed in claim 2, characterized in that: The process of performing data analysis on the several new tool yield data and the old tool yield data to determine whether the several new tool yield data and the old tool yield data belong to a high yield category or a relatively high yield category includes: dividing the several old tool yield data into several clusters; constructing a fuzzy system model according to the several clusters, wherein the fuzzy system model includes category attribution and corresponding distribution function that conform to the cluster characteristic distribution, the fuzzy system model is one of an α model, a β model or a γ model, the α model includes three category attribution and corresponding three distribution functions, and the three category attributions are a low yield category, a medium yield category and a high yield category The β model includes two categories and two corresponding distribution functions, the two categories are a relatively low yield category and a relatively high yield category, and the γ model includes one category and one corresponding distribution function, the one category is an overall yield category; a number of new tool yield data and old tool yield data are respectively projected into the fuzzy system model to obtain the category attribution corresponding to each new tool yield data and old tool yield data; according to the category attribution corresponding to each new tool yield data and old tool yield data, it is judged whether the several new tool yield data and old tool yield data belong to the high yield category or the relatively high yield category.

5. The qualified inspection method for newly added production tools as claimed in claim 3, characterized in that: The yield data of the plurality of old tools are divided into a plurality of clusters using a K-Means clustering algorithm.

6. The qualified inspection method for newly added production tools as claimed in claim 5, characterized in that: Before dividing the said plurality of old tool yield data into a plurality of clusters, the step also includes: determining whether the number of the said new tool yield data and the old tool yield data are both greater than 10; if "yes", dividing the said plurality of old tool yield data into a plurality of clusters is performed; if "no", the inspection process is terminated.

7. The qualified inspection method for newly added production tools as claimed in claim 6, characterized in that: The process of dividing the data into several clusters, constructing a fuzzy system model, obtaining the category attribution corresponding to each new tool yield data and old tool yield data, judging whether the several new tool yield data and old tool yield data belong to the high yield category or the relatively high yield category, and obtaining the screened new tool yield data and old tool yield data includes: when dividing the several old tool yield data into several clusters, presetting the K value in the K-Means clustering algorithm to be equal to 3, and then dividing the several old tool yield data into three clusters by using the K-Means clustering algorithm; constructing a fuzzy system model according to the three clusters, wherein the fuzzy system model is an α model; dividing the several new tool yield data and the old tool yield data into several clusters; The tool yield data are projected into the α model respectively to obtain the category attribution corresponding to each new tool yield data and the old tool yield data; according to the category attribution corresponding to each new tool yield data and the old tool yield data, it is judged whether the several new tool yield data and the old tool yield data belong to the high yield category, if "belong", the corresponding new tool yield data and the old tool yield data are eliminated, and the remaining new tool yield data and the old tool yield data are used as the screened new tool yield data and the old tool yield data; it is judged whether the number of the screened new tool yield data and the old tool yield data are both greater than 10, if "yes", the screening based on the screened new tool yield data and the old tool yield data is performed. Tool yield data, a step of judging whether the new production tool is qualified, if "no", then subtract 1 from the K value, when the K value is equal to 2, the several old tool yield data are divided into two clusters by the K-Means clustering algorithm; according to the two clusters, a fuzzy system model is constructed, and the fuzzy system model is a β model; the several new tool yield data and the old tool yield data are respectively projected into the β model to obtain the category attribution corresponding to each new tool yield data and the old tool yield data; according to the category attribution corresponding to each new tool yield data and the old tool yield data, it is judged whether the several new tool yield data and the old tool yield data belong to the high yield category, if "belongs ", then the corresponding new tool yield data and old tool yield data are eliminated, and the remaining new tool yield data and old tool yield data are used as the screened new tool yield data and old tool yield data; continue to judge whether the number of screened new tool yield data and old tool yield data are both greater than 10, if "yes", then perform the step of judging whether the new production tool is qualified based on the screened new tool yield data and old tool yield data, if "no", then reduce the K value by 1, and when the K value is equal to 1, the several old tool yield data are divided into 1 cluster by the K-Means clustering algorithm; according to the 1 cluster, construct a fuzzy system model, and the fuzzy system model is a γ model;Projecting a number of new tool yield data and old tool yield data into the γ model respectively, obtaining the category attribution corresponding to each new tool yield data and old tool yield data, and directly using the new tool yield data and old tool yield data as the screened new tool yield data and old tool yield data. ; 8. The qualified inspection method for newly added production tools as claimed in claim 3, characterized in that: The process of judging whether the new production tool is qualified based on the screened new tool yield data and the old tool yield data includes: judging whether the screened new tool yield data and the old tool yield data meet condition 1, wherein condition 1 is that the mean of the screened new tool yield data ≥ the mean of the screened old tool yield data, and the standard deviation of the screened new tool yield data ≤ the standard deviation of the screened old tool yield data; if "yes", the new production tool is judged to be qualified; if "no", further judgment is performed.

9. The qualified inspection method for newly added production tools as claimed in claim 8, characterized in that: The further judgment includes: judging whether the screened new tool yield data and the old tool yield data meet condition 2, wherein condition 2 is that the mean of the screened new tool yield data is less than the mean of the screened old tool yield data, and the standard deviation of the screened new tool yield data is greater than the standard deviation of the screened old tool yield data; if "yes", the new production tool is judged to be unqualified; if "no", a t-test is performed to judge whether there is a significant difference between the screened new tool yield data and the old tool yield data; if "no", the new production tool is judged to be qualified; if so, judging whether the screened new tool yield data and the old tool yield data meet condition 3; wherein condition 3 is that the mean of the screened new tool yield data is ≥ the mean of the screened old tool yield data; if "yes", the new production tool is judged to be qualified; if "no", the new production tool is judged to be qualified or unqualified.

10. A new production tool qualification inspection system, characterized in that: include: A wafer providing unit, used for providing a plurality of test wafers, wherein the plurality of test wafers are divided into a first portion of wafers and a second portion of wafers; A new production tool for performing corresponding process processing on the first part of wafers; An old production tool, used for performing corresponding process treatment on the second portion of wafers; A yield data measurement unit is used to measure the yield of a first portion of wafers after being processed in the new production tool to obtain a number of new tool yield data, and to measure the yield of a second portion of wafers after being processed in the old production tool to obtain a number of old tool yield data; a data analysis unit, for performing data analysis on the plurality of new tool yield data and the plurality of old tool yield data, and determining whether the plurality of new tool yield data and the plurality of old tool yield data belong to a high yield category or a relatively high yield category; if they do, excluding the corresponding new tool yield data and the plurality of old tool yield data, and using the remaining new tool yield data and the plurality of old tool yield data as the screened new tool yield data and the plurality of old tool yield data; A judging unit is used to judge whether the new production tool is qualified based on the screened new tool yield data and the old tool yield data.

11. The newly added production tool qualification inspection system as claimed in claim 10, characterized in that: The several inspection wafers provided by the wafer providing unit are several batches of wafers to be inspected, and correspondingly the odd-numbered wafers in each batch are used as the first part of wafers, and the even-numbered wafers are used as the second part of wafers, or correspondingly the even-numbered wafers in each batch are used as the first part of wafers, and the odd-numbered wafers are used as the second part of wafers.

12. The newly added production tool qualification inspection system as claimed in claim 10, characterized in that: The method for the data analysis unit to perform data analysis on the plurality of new tool yield data and the old tool yield data adopts a data analysis method based on a fuzzy system model.

13. The newly added production tool qualification inspection system as claimed in claim 12, characterized in that: The data analysis unit performs data analysis on the several new tool yield data and the old tool yield data to determine whether the several new tool yield data and the old tool yield data belong to the high yield category or the relatively high yield category, including: dividing the several old tool yield data into several clusters; constructing a fuzzy system model according to the several clusters, the fuzzy system model including category attribution and corresponding distribution function that conform to the cluster characteristic distribution, the fuzzy system model is one of the α model, the β model or the γ model, the α model includes three category attribution and corresponding three distribution functions, the three category attribution are low yield category, medium yield category and high yield category. Yield category, the β model includes two category attributions and two corresponding distribution functions, the two category attributions are a relatively low yield category and a relatively high yield category, the γ model includes one category attribution and a corresponding distribution function, the one category attribution is an overall yield category; a number of new tool yield data and old tool yield data are respectively projected into the fuzzy system model to obtain the category attribution corresponding to each new tool yield data and old tool yield data; according to the category attribution corresponding to each new tool yield data and old tool yield data, it is judged whether the several new tool yield data and old tool yield data belong to a high yield category or a relatively high yield category.

14. The newly added production tool qualification inspection system as claimed in claim 13, characterized in that: The yield data of the plurality of old tools are divided into a plurality of clusters using a K-Means clustering algorithm.

15. The newly added production tool qualification inspection system as claimed in claim 14, characterized in that: It also includes a data sample number judgment unit, which is used to judge whether the number of new tool yield data and old tool yield data are both greater than 10 before the data analysis unit divides the old tool yield data into several clusters. If "yes", the step of dividing the old tool yield data into several clusters is performed; if "no", the inspection process is terminated.

16. The newly added production tool qualification inspection system as claimed in claim 15, characterized in that: The data analysis unit divides the data into several clusters, constructs a fuzzy system model, obtains the category attribution corresponding to each new tool yield data and old tool yield data, determines whether the several new tool yield data and old tool yield data belong to the high yield category or the relatively high yield category, and obtains the screened new tool yield data and old tool yield data. The process includes: when dividing the several old tool yield data into several clusters, presetting the K value in the K-Means clustering algorithm to be equal to 3, and then dividing the several old tool yield data into three clusters by using the K-Means clustering algorithm; constructing a fuzzy system model according to the three clusters, wherein the fuzzy system model is an α model; dividing several new tool yield data into several clusters; The tool yield data and the old tool yield data are projected into the α model respectively to obtain the category attribution corresponding to each new tool yield data and the old tool yield data; according to the category attribution corresponding to each new tool yield data and the old tool yield data, it is judged whether the several new tool yield data and the old tool yield data belong to the high yield category, if "belong", the corresponding new tool yield data and the old tool yield data are eliminated, and the remaining new tool yield data and the old tool yield data are used as the screened new tool yield data and the old tool yield data; it is judged whether the number of the screened new tool yield data and the old tool yield data are both greater than 10, if "yes", the screening based on the screened new tool yield data is performed. The step of judging whether the new production tool is qualified is as follows: if "no", the K value is reduced by 1, and when the K value is equal to 2, the plurality of old tool yield data are divided into two clusters by the K-Means clustering algorithm; according to the two clusters, a fuzzy system model is constructed, and the fuzzy system model is a β model; the plurality of new tool yield data and the old tool yield data are respectively projected into the β model to obtain the category attribution corresponding to each new tool yield data and the old tool yield data; according to the category attribution corresponding to each new tool yield data and the old tool yield data, it is judged whether the plurality of new tool yield data and the old tool yield data belong to the high yield category, and if " belongs to", then the corresponding new tool yield data and old tool yield data are eliminated, and the remaining new tool yield data and old tool yield data are used as the screened new tool yield data and old tool yield data; continue to judge whether the number of screened new tool yield data and old tool yield data are both greater than 10, if "yes", then perform the step of judging whether the new production tool is qualified based on the screened new tool yield data and old tool yield data, if "no", then reduce the K value by 1, and when the K value is equal to 1, the several old tool yield data are divided into 1 cluster by the K-Means clustering algorithm; according to the 1 cluster, construct a fuzzy system model, and the fuzzy system model is a γ model;Projecting a number of new tool yield data and old tool yield data into the γ model respectively, obtaining the category attribution corresponding to each new tool yield data and old tool yield data, and directly using the new tool yield data and old tool yield data as the screened new tool yield data and old tool yield data. ; 17. The newly added production tool qualification inspection system as claimed in claim 10, characterized in that: The process of the judgment unit judging whether the new production tool is qualified based on the screened new tool yield data and the old tool yield data includes: judging whether the screened new tool yield data and the old tool yield data meet condition 1, wherein condition 1 is that the mean of the screened new tool yield data ≥ the mean of the screened old tool yield data, and the standard deviation of the screened new tool yield data ≤ the standard deviation of the screened old tool yield data; if "yes", the new production tool is judged to be qualified; if "no", further judgment is performed.

18. The newly added production tool qualification inspection system as claimed in claim 17, characterized in that: The further judgment includes: judging whether the screened new tool yield data and the old tool yield data meet condition 2, wherein condition 2 is that the mean of the screened new tool yield data is less than the mean of the screened old tool yield data, and the standard deviation of the screened new tool yield data is greater than the standard deviation of the screened old tool yield data; if "yes", the new production tool is judged to be unqualified; if "no", a t-test is performed to judge whether there is a significant difference between the screened new tool yield data and the old tool yield data; if "no", the new production tool is judged to be qualified; if so, judging whether the screened new tool yield data and the old tool yield data meet condition 3; wherein condition 3 is that the mean of the screened new tool yield data is ≥ the mean of the screened old tool yield data; if "yes", the new production tool is judged to be qualified; if "no", the new production tool is judged to be qualified or unqualified.

Images