Method and system for screening abnormal crystal grains on wafer, computer program product and device

By establishing a CDF probability map of wafer parameter data and setting DPAT specifications, the problem that the existing technology cannot effectively screen abnormal grains, and high-quality wafer shipment and intelligent machine control are achieved.

CN118940183AActive Publication Date: 2024-11-12GTA SEMICON CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202411027037.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-29
Publication Date
2024-11-12
Estimated Expiration
2044-07-29

AI Technical Summary

Technical Problem

The existing static and dynamic parameter average testing methods cannot effectively screen out abnormal grains of each wafer, resulting in difficult to guarantee electrical quality and the inability to effectively control the intelligent machine.

Method used

By obtaining the parameter data of the wafer, establishing a CDF probability map, setting the conversion point between the primary and secondary distributions of the data to the DPAT specification, filtering out grains outside the DPAT specification as abnormal grains, and realizing the abnormal grains Accurate separation.

Benefits of technology

It realizes accurate screening of the parameter data of each wafer, ensures zero defect shipment in electrical quality, improves wafer yield, and strengthens the intelligent control capabilities of the machine.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118940183A_ABST
    Figure CN118940183A_ABST
Patent Text Reader

Abstract

The invention provides a method and a system for screening abnormal crystal grains on a wafer, a computer program product and a device. The method comprises the following steps: acquiring parameter data of all crystal grains on a wafer to be detected; converting each piece of parameter data to establish a corresponding CDF probability graph, and setting a conversion point between data primary distribution and data secondary distribution of the CDF probability graph as a DPAT specification so as to separate normal crystal grains from abnormal crystal grains; and according to the set DPAT specification, screening out all crystal grains outside the DPAT specification on the wafer to be detected as abnormal crystal grains, and completing screening of the abnormal crystal grains on the wafer to be detected. According to the invention, grains with abnormal electrical properties can be screened out, and the purpose of zero-defect electrical shipment is achieved; the consistency of the quality and the performance of the product can be enhanced; parameter distribution from wafer test shipment to module terminal test may be tracked to monitor and prevent any anomalies on the manufacturing chain; and sustainable closed-loop analysis of yield data is realized, and intelligent management of machines and processes can be constructed.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of semiconductor manufacturing technology, and in particular to a method and system for screening abnormal grains on a wafer, a computer program product and a device. Background Art

[0002] Zero defects are the goal that all industries are constantly pursuing. In the automotive electronics industry, which has higher requirements for safety, the requirements for quality are more stringent. Measurement is an important means of process control in wafer fabs. By measuring the key parameters of the main production stages of the product, it is determined whether the product meets the design requirements to ensure that the final yield meets the requirements. At the same time, through measurement, we can understand the current problems and provide timely feedback to make process adjustments to achieve a high yield of the final product. When wafers are tested, a large amount of measurement data will be generated. Ideally, the measurement data (continuous numerical data) obtained by testing a certain measurement parameter of the same time period, the same machine, and the same batch / wafer should obey the normal distribution. However, in actual production situations, due to abnormalities such as the manufacturing process, the measurement data obtained from the test will have other forms of data distribution.

[0003] The AEC-Q001 specification proposes the Parametric Part Average Testing (PPAT) method. The PPAT method is a statistical method used to detect outliers (or scattered points) in semiconductor part parameters, and is used to eliminate abnormal parts from all parts tested. Figure 1 , which is a schematic diagram of the PAT limit and abnormal grains. Figure 1 As shown in the figure, by setting the PAT limit (Limit), including the upper Spec Limit (USL) and the lower Spec Limit (LSL), the scattered points (Outliers) whose parameter data exceeds the normal distribution (parameter anomaly) are screened out. The PPAT method mainly includes static PAT (Static PAT) and dynamic PAT (Dynamic PAT). However, the specification range of the limit setting for static PAT and dynamic PAT is calculated based on the standard deviation sigma value, and the specification range size is adjusted in stages to screen out abnormal grains, which cannot effectively screen out abnormal grains for each wafer. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a method and system for screening abnormal grains on a wafer, which can screen out abnormal grains whose parameter data of each wafer exceeds the normal distribution, ensure the electrical quality of the shipped grains, promote the improvement of the wafer yield and strengthen the intelligent control of the machine.

[0005] In order to solve the above problems, the present invention provides a method for screening abnormal grains on a wafer, comprising: acquiring parameter data of all grains on the wafer to be tested; converting each of the parameter data to establish a corresponding CDF probability graph, and setting the transition point between the data main distribution and the data secondary distribution of the CDF probability graph as the DPAT specification to separate normal grains from abnormal grains; and, according to the set DPAT specification, screening out all grains on the wafer to be tested that are outside the DPAT specification as abnormal grains, thereby completing the screening of abnormal grains on the wafer to be tested.

[0006] In order to solve the above problems, the present invention also provides a screening system for abnormal grains on a wafer, comprising: an acquisition unit, used to acquire parameter data of all grains on the wafer to be tested; an establishment unit, used to convert each of the parameter data to establish a corresponding CDF probability graph, and set the transition point between the data main distribution and the data secondary distribution of the CDF probability graph as the DPAT specification to separate normal grains from abnormal grains; a screening unit, used to screen out all grains on the wafer to be tested that are outside the DPAT specification as abnormal grains according to the set DPAT specification, and complete the screening of abnormal grains on the wafer to be tested.

[0007] In order to solve the above problem, the present invention further provides a computer program product, including a computer program, and when the computer program is executed by a processor, the steps of the above method of the present invention are implemented.

[0008] In order to solve the above problem, the present invention further provides a computer device, including a memory for storing a computer program; a processor connected to the memory and used to execute the computer program to implement the steps of the above method of the present invention.

[0009] The above technical solution, by establishing a screening method by wafer, by parameter and by parameter data distribution, can screen out grains with abnormal electrical properties and achieve the goal of shipping with zero electrical defects; it can enhance the consistency of product quality and performance; it can track the parameter distribution from wafer test shipment to module terminal test to monitor and prevent any abnormalities in the manufacturing chain; sustainable closed-loop analysis of yield data can improve wafer yield, build intelligent management of machines and processes, and will have strict and effective benefits for intelligent control of production machine yield.

[0010] It should be understood that the above general description and the detailed description below are exemplary and explanatory only and cannot limit the present invention. The techniques, methods and devices known to ordinary technicians in the relevant field may not be discussed in detail, but where appropriate, the techniques, methods and devices should be considered as part of the authorization specification. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments of the present invention are briefly introduced below. 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 creative work.

[0012] Figure 1 Schematic diagram of PAT limit and abnormal grains;

[0013] Figure 2 A flow chart of a method for screening abnormal grains on a wafer provided by an embodiment of the present invention;

[0014] Figure 3 A CDF probability diagram of VDSON provided by an embodiment of the present invention;

[0015] Figure 4 A CDF probability diagram of ICES provided by an embodiment of the present invention;

[0016] Figure 5 A schematic diagram of a system for screening abnormal grains on a wafer provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0017] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part 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 ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] In order to screen abnormal grains on the wafer, static PAT or dynamic PAT is generally used.

[0019] Static PAT: Set fixed PAT limits, including the Upper Spec Limit (USL) and the Lower Spec Limit (LSL). All lots / wafers use the same PAT limits and are updated regularly. Standard deviation Sigma is calculated based on a large number of baseline lots. The upper and lower spec limits (USL) are usually set to ±3Sigma or as high as ±6Sigma. In the standard deviation Sigma calculation, all deviations from the mean are squared, so larger differences are weighted more heavily, and outliers can significantly affect it. In addition to the variables due to the lot, the Sigma value is increased by outliers, so that a wide specification range is set, and abnormal dies are often not intercepted, so abnormal and risky dies on the wafer are mistakenly placed.

[0020] Dynamic PAT: It uses dynamic limits (DPAT Limit) to allow each wafer to have its own limits. The parameter data is collected from the die on each wafer. The average value, sigma and dynamic limit of each parameter are calculated based on the parameter data on each wafer, so the DPAT specification will be within the SPAT range and more stringent than SPAT. Each die test is first screened by SPAT. After the wafer test is completed, the dynamic limit is calculated based on the parameter data on each wafer. Based on this dynamic limit, it is determined again whether the parameter value meets the requirements of good products. Strict DPAT specifications can certainly screen out more abnormal die, but the DPAT specification range is still calculated based on the standard deviation sigma value, and the use of ±6Sigma or ±3Sigma is a staged adjustment of the specification range to screen out abnormal die. It cannot effectively screen out abnormal die on each wafer.

[0021] Since the calculation of standard deviation sigma is heavily weighted by outliers, Median Absolute Deviation (MAD) is proposed to be used in solving the weight of statistical discrete data. MAD is a robust statistic, Robust Sigma, which can handle outliers more strictly than standard deviation. Robust Sigma is smaller than standard deviation Sigma (Robust Sigma ≦ 0.675*Standard Deviation Sigma). Although a few outliers have no effect on MAD, MAD screening may still overlook abnormal grains with large sub-distribution or continuous distribution. In addition, the specification range of MAD increases or decreases in stages according to the robust sigma value, requiring ±6 Robust Sigma or ±3 Robust Sigma, which is either too loose or too tight, and becomes a decision that often has to be made in engineering. Although Robust Sigma is smaller than standard deviation Sigma and can screen out more outliers, it still cannot guarantee that abnormal grains on each wafer can be accurately removed.

[0022] To this end, the present invention proposes a method for screening abnormal grains on a wafer, by converting the parameter data to establish a corresponding CDF probability graph, and determining the DPAT specification by the slope conversion point of the data distribution on the CDF probability graph, it is possible to screen out abnormal grains whose parameter data of each wafer exceeds the normal distribution, accurately separate normal grains from abnormal grains, avoid the risk of misplacing abnormal grains within the data sub-distribution, ensure the electrical quality of the shipped grains, promote the improvement of the wafer yield and strengthen the intelligent control of the machine. A detailed explanation is given below.

[0023] See also Figure 2 , which is a flow chart of a method for screening abnormal grains on a wafer provided by an embodiment of the present invention. Figure 2 As shown, the method for screening abnormal grains on a wafer in this embodiment includes: step S1, obtaining parameter data of all grains on the wafer to be tested; step S2, converting each of the parameter data to establish a corresponding CDF probability graph, and setting the transition point between the data main distribution and the data secondary distribution of the CDF probability graph to the DPAT specification to separate normal grains from abnormal grains; and step S3, screening out all grains on the wafer to be tested that are outside the DPAT specification as abnormal grains according to the set DPAT specification, thereby completing the screening of abnormal grains on the wafer to be tested.

[0024] Regarding step S1, parameter data of all dies on the wafer to be tested is obtained. The parameter data is obtained by measuring or testing all dies distributed on a wafer or multiple wafers produced in the same batch; the dies refer to components or dies that can be measured or tested and are distributed on the wafer to be tested.

[0025] The wafer manufacturing process includes: wafer input, film deposition, mask, etching and diffusion of the wafers in turn. In order to monitor the process conditions of these processes, defect tests can be performed on the wafers after the film deposition operation, defect tests can be performed on the wafers after the mask processing, and tests can be performed on the wafers after etching, so as to obtain the parameter data of the front end of the process. It should be noted that the process technology detection in the front end of the process mainly tests the process performance parameters in the wafer manufacturing process. It is necessary to measure the defect type, film thickness, line width, key dimensions, etc. of the wafer, which belongs to the front-end test process, and the test results of the front-end test are usually presented in the form of wafer defect image types. The end of the process includes wafer acceptance test and wafer probe test. The WAT test link (corresponding to wafer acceptance test) and CP test link (corresponding to wafer probe test) at the end of the process are the process of testing the electrical performance and function of the wafer, and finally the functional level of the grain will be divided according to the test results. The test results are usually reflected in the form of electrical performance parameters and wafer product qualification rate. It should be noted that the parameter data on which some embodiments of the present invention are based may be parameter data at the front end of a process or parameter data at the end end of a process.

[0026] In some embodiments, the parameter data includes at least one of the following data: the penetration current (ICES) between the collector and the emitter when the gate and the emitter are short-circuited, the source-drain voltage drop (VDSON) when the circuit is turned on, the leakage current (IGSS) between the gate and the source when the source and the drain are short-circuited, and the threshold voltage (VTH) at which the circuit or device starts to work or produces a specific effect. Among them, ICES is the penetration current between the collector and the emitter when the gate (similar to the base) and the emitter of the IGBT are short-circuited, or the leakage current. (ICES is formed by the front end P+ / N, Ndrift / PTub under reverse bias, and a few Ndrift / Gate leakage currents). ICES is usually determined by technology, manufacturing and design. When the device is in the off mode, it is subjected to a high voltage between the collector and the emitter. In addition to reducing the static loss of ICES, ICES is also an important parameter for testing the quality of the IGBT diffusion junction. ICES is related to BVCES, dopant profile, defects in junction depletion region, field line loops, etc. ICES of each die on a wafer is an important parameter that can be used to distinguish normal products from abnormal products under high voltage, eliminate die with voltage resistance risk, and ensure the normal quality and reliability of the die shipped.

[0027] Regarding step S2, each of the parameter data is converted to establish a corresponding CDF probability map, and the conversion point between the data main distribution and the data secondary distribution of the CDF probability map is set as the DPAT specification to separate normal grains from abnormal grains.

[0028] In this step, the ordinate of the CDF probability graph is a normal probability scale, the abscissa is parameter data, and the graph is a cumulative grain density function. When the parameter data of the grains on the wafer to be tested is converted into a CDF probability graph, the cumulative grain density distribution increases monotonically with the parameter data, the cumulative density of the grains on the data main distribution is linearly distributed with its parameter data and has a first slope, and the cumulative density of the grains on the data secondary distribution is linearly distributed with its parameter data and has a second slope or is distributed in a curve, the first slope is greater than the second slope, wherein the slope conversion point between the first slope and the second slope is set as the conversion point as the DPAT specification to separate normal grains from abnormal grains.

[0029] The parameter data obtained from the test of normal quality grains should obey the normal distribution. The parameter data falling within the normal distribution indicates that the tested grains have the same electrical characteristics. Parameter data beyond the normal distribution indicates that the grains do not meet the principle characteristics of the parameters, which is likely caused by manufacturing defects or abnormal mechanisms. Parameters such as leakage current are exponentially sensitive to voltage and temperature. The parameter data generated by the test of normal quality grains should obey the log-normal distribution. The static PAT or dynamic PAT method of judging whether the data obeys the normal distribution cannot effectively screen out abnormal grains on each wafer because the specification range of the limit value is calculated based on the standard deviation sigma value and the size of the specification range is adjusted in stages to screen out abnormal grains.

[0030] In order to determine whether the data obeys the normal distribution, this embodiment converts the parameter data of the grains into a CDF probability diagram. Specifically, CDF (Cumulative Density Function). When the parameter data of the wafer is converted into a CDF probability diagram, the cumulative grain density distribution increases monotonically with the parameter data. Among them, the cumulative density of normal grains is distributed in a straight line with their parameter data and has the steepest slope (main distribution); while defective or abnormal grains are obviously distributed on segments with different slopes (less than the slope of the straight line distribution segment on the main distribution) (secondary distribution), or on a curve (secondary distribution), or are discrete tailing abnormal grains (Outliers). By setting the slope conversion point of both the main distribution and the secondary distribution to the DPAT specification, normal grains can be accurately separated from abnormal grains. Since the DPAT specification is directly determined by the slope conversion point of the main distribution and the secondary distribution, it is better than the existing technology and avoids the risk of misplacing abnormal grains in the secondary distribution.

[0031] In some embodiments, the slope transition point between the primary distribution and the secondary distribution can be determined by second-order differentiation or Line-fit.

[0032] In some embodiments, the slope transition point between the main distribution and the secondary distribution is determined by: (1) sorting the test values ​​of each grain on the wafer to be tested by the sub-test item; (2) calculating the distance value and slope of the test values ​​of each two adjacent grains; (3) sorting all the distance values ​​and slopes obtained by calculation, and finding the point where the deviation is greater than the preset threshold to confirm the slope transition point on the CDF probability map. After the sub-test item sorts the test values ​​of each grain on the wafer to be tested, the data that clearly does not meet the standard (data that exceeds the upper and lower limits of the specification) can be further eliminated. When calculating the distance value and slope of the test values ​​of each two adjacent grains a and b, the following convention can be used: distance formula l = ab, slope formula k = (Ya-Yb) / (Xa-Xb)). After sorting all the distance values ​​and slopes obtained by calculation, the point where the deviation is obvious can be found, and then combined with the CDF probability map, the slope transition point can be confirmed on the CDF probability map.

[0033] In some embodiments, first, the parameter data of a single wafer or a single batch of wafers are stored in a database or in a computer memory, and then a script or APP application software written in any scripting language (the script or application is used to implement the method of converting parameter data to establish a corresponding CDF probability map provided in some embodiments of the present invention) is used to convert each of the parameter data to establish a corresponding CDF probability map. It should be noted that any of the above scripting languages, including but not limited to Perl, Python, Java, etc.; any EDA software programming tool, etc., may also be used.

[0034] Regarding step S3, all the grains on the wafer to be tested that are outside the DPAT specification are screened out as abnormal grains according to the set DPAT specification, thereby completing the screening of abnormal grains on the wafer to be tested.

[0035] In this embodiment, the DPAT specification can screen out abnormal grains between the slope conversion point and the ± Sigma specification. In addition to the normal probability distribution on a linear or logarithmic scale for distinguishing outliers, the linear or logarithmic scale of the Poisson probability distribution, or other statistical probability distributions can also be used to judge parameter data. The advantage of this embodiment is that the conversion point of the slope of the CDF probability graph line segment is used to determine the boundary of the statistical distribution. As the DPAT specification, it can screen out abnormal grains between the slope conversion point and the ± Sigma specification, which is better than the prior art.

[0036] In this step, by setting the conversion point between the data main distribution and the data secondary distribution of the CDF probability map as the DPAT specification, it can be determined whether the distribution of the parameter data of the current grain in the corresponding CDF probability map is outside the DPAT specification according to the set DPAT specification; thus, it can be determined whether the current grain is an abnormal grain according to the distribution of the parameter data of the current grain in the corresponding CDF probability map; further traverse the distribution of all parameter data of all grains on the wafer in the corresponding CDF probability map, thereby completing the screening of abnormal grains on the wafer. If the current grain has a parameter data distribution abnormality, the current grain is determined to be an abnormal grain. The distribution of all parameter data of the abnormal grain in the corresponding CDF probability map can be further traversed to further confirm the abnormal mechanism of the abnormal grain.

[0037] In some embodiments, the parameter data of normal grains fall within the main data distribution of the CDF probability graph, and the parameter data of abnormal grains fall on a secondary data distribution outside the main data distribution of the CDF probability graph or are in a discrete state.

[0038] In some embodiments, the step of screening out all the grains on the wafer to be tested that are outside the DPAT specification as abnormal grains according to the set DPAT specification further includes: (1) traversing the distribution of all parameter data of the current grain on the wafer to be tested in the corresponding CDF probability diagram according to the set DPAT specification, if there is parameter data whose distribution in the corresponding CDF probability diagram is outside the DPAT specification, then determining that the current grain is an abnormal grain; (2) traversing the distribution of all parameter data of all grains on the wafer to be tested in the corresponding CDF probability diagram to complete the screening of abnormal grains on the wafer to be tested.

[0039] See also Figure 3 , which is a CDF probability diagram of VDSON provided by an embodiment of the present invention. Figure 3 As shown, the parameter data VDSON shows a normal main distribution and the existence of a secondary distribution. The abnormal pattern is obviously caused by the wafer edge grains. Figure 3 Part (a) shows the parameter data of the grains using different specifications: (i) standard USL / LSL, (ii) mean ±3Sigma, (iii) median ±6MAD, (iv) median ±3MAD and (v) distribution slope conversion point. Figure 3 Part (b) uses various specifications to generate the corresponding wafer yield map: (i) standard USL / LSL, (ii) median ±3MAD, (iii) median ±6MAD, (iv) mean ±3Sigma. Figure 3It can be seen that using the standard USL / LSL, the screening shows that the wafer yield is high, but in fact, the problems at the edge of the wafer are not screened out, nor are many abnormal grains with abnormal VDSON. (i)-(iv) can be used to quantitatively compare the effectiveness of various specifications in screening out abnormal grains. Some of the definitions of terms are as follows: CP (Chip Probing, wafer test) PASS: CP test passed. CP VCESAT (Voltage Collector-Emitter Saturation): CP source-drain saturation voltage drop. CP Other Fail: CP other failure; refers to other failure items (over-specification) that appeared in the wafer test in the case, including ICES, IGES, VTH, etc. mentioned above. DPAT INK: Items excluded by the dynamic part average test point; ink is a dot on the ink point, which is a way to deal with failed grains, and here it refers to waste. Among them, VCESAT and VDSON are both source-drain saturation voltage drops when the circuit is turned on. The device measured in the case in the figure is an IGBT (Insulated Gate Bipolar Transistor), which is composed of two sub-devices, a MOSFET (Metal-Oxide-Semiconductor Field-Effect Transistor) and a BJT (Bipolar Junction Transistor). When expressed using the electrodes of the MOSFET, it can be called VDSON (source-drain saturation voltage drop), and when expressed using the electrodes of the BJT, it can be called VCESAT (emitter-collector saturation voltage drop), which are interchangeable general terms in the industry.

[0040] See also Figure 4 , which is a CDF probability diagram of ICES provided by an embodiment of the present invention. Among them, Figure 4 Part (a) shows the relationship between the CDF probability diagram and the penetration current in the linear ICES. Figure 4Part (b) shows the relationship between the CDF probability graph and the penetration current in the logarithmic ICES. Among them, it is not easy to find the ICES slope conversion point for linear ICES. ICES is usually determined by technology, manufacturing and design. When the device is in the off mode and is subjected to high voltage between the collector and the emitter, in addition to reducing the static loss of ICES, ICES is also an important parameter for testing the quality of the IGBT diffusion junction (junction). Since the penetration current (ICES) and Vjunction are exponential functions in principle, when Vjunction is normally distributed, ICES~exp±(qVj / KT) is a logarithmic normal distribution. Let Vjunction be a standard normal variable, μ and σ be two real numbers, and the distribution of the random variable ICES=exp(±μ±σV) is a logarithmic normal distribution with parameters μ and σ. Normal grain ICES shows a main distribution close to a straight line segment. Abnormal grain ICES tail divergence can be easily screened out in the CDF probability plot of logarithmic ICES by setting the DPAT specification from each line slope transition point (marked with a five-pointed star).

[0041] In the CDF probability diagram, the line segments with different slopes of the grain and parameter data indicate different physical mechanisms. Therefore, by detecting whether there is a secondary distribution outside the main distribution on each wafer (and whether there is a single-mode distribution outside the main mode), if there is a secondary distribution, it means that there are other abnormal mechanisms; further by analyzing the physical mechanism of the abnormal distribution, find out its root cause to correct the problem and prevent it from happening again; this is of great help in improving the uniformity and yield stability of the wafer. In addition, the average value (or center value) of each wafer should also follow the normal distribution (central limit theory); if a non-normal distribution is formed, it can guide the inspection of the differences between wafers caused by the mismatch of the machine or chamber.

[0042] In addition to screening out abnormal grains to ensure high-quality wafer shipments, the present invention can also be applied to equipment control and optimization of wafer factories, and can be specifically applied to at least one of the following aspects: (1) Analyze the distribution of each parameter data on each wafer in the corresponding CDF probability map, use different specification cards to screen out the corresponding abnormal grains, and obtain the abnormal grain distribution (Wafer Bin Map) on the corresponding wafer, which is an important measure to detect abnormal distribution and improve the uniformity of wafer performance. And / or, (2) By detecting whether there is a secondary distribution other than the main distribution on each wafer, if there is a secondary distribution, further analyze the physical mechanism of the abnormal distribution, wherein the line segments with different slopes of the grain and parameter data in the CDF probability map indicate different physical mechanisms. And / or, (3) In the CDF probability map, tighten the normal distribution of the parameter data by increasing the line segment slope of the parameter data to improve the consistency of the grain quality and performance. And / or, (4) analyzing whether the average value or center value of each wafer follows a normal distribution to screen and compare wafers, so as to further check whether the differences between wafers are caused by machine or chamber mismatch, so as to achieve the goal of machine equipment matching and control.

[0043] The method for screening abnormal grains on a wafer provided by the above-mentioned embodiment of the present invention can screen out grains with abnormal electrical properties by wafer, parameter and distribution, so as to achieve the goal of shipping with zero electrical defects; it can enhance the consistency of product quality and performance; it can track the parameter distribution from wafer testing and shipment to module terminal testing to monitor and prevent any abnormalities in the manufacturing chain; sustainable closed-loop analysis of yield data can improve wafer yield, build intelligent management of machines and processes, and will have strict and effective benefits for the intelligent control of production machine yield.

[0044] Based on the same inventive concept, the present invention also provides a system for screening abnormal grains on a wafer, which is used to execute the method for screening abnormal grains on a wafer described in the present invention.

[0045] Figure 5 A schematic diagram of a system for screening abnormal grains on a wafer provided by an embodiment of the present invention. Figure 5 As shown, the screening system for abnormal grains on the wafer includes: an acquisition unit 51, an establishment unit 52, and a screening unit 53.

[0046] Specifically, the acquisition unit 51 is used to acquire parameter data of all grains on the wafer to be tested; the establishment unit 52 is used to convert each parameter data to establish a corresponding CDF probability graph, and set the transition point between the data main distribution and the data secondary distribution of the CDF probability graph as the DPAT specification to separate normal grains from abnormal grains; the screening unit 53 is used to screen out all grains on the wafer to be tested that are outside the DPAT specification as abnormal grains according to the set DPAT specification, thereby completing the screening of abnormal grains on the wafer to be tested.

[0047] Based on the same inventive concept, the present invention further provides a computer program product, including a computer program, which implements the steps of the aforementioned method of the present invention when executed by a processor.

[0048] Based on the same inventive concept, the present invention also provides a computer device, comprising: a memory for storing a computer program; a processor connected to the memory and for executing the computer program to implement the steps of the aforementioned method of the present invention.

[0049] Based on the same inventive concept, the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the program is executed by a processor, the steps of the aforementioned method of the present invention are implemented.

[0050] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synch Link) DRAM (SLDRAM), memory bus (Ram Bus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0051] It should be noted that the terms "including" and "having" and their variations involved in the document of the present invention are intended to cover non-exclusive inclusions. The terms "first", "second", etc. are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence, unless the context clearly indicates, and it should be understood that the data used in this way can be interchanged under appropriate circumstances. The term "one or more" depends at least in part on the context, and can be used to describe features, structures or characteristics in a singular sense, or can be used to describe features, structures or combinations of features in a plural sense. The term "based on" can be understood as not necessarily intended to express a set of exclusive factors, but can alternatively, also at least in part depending on the context, allow the presence of other factors that are not necessarily explicitly described. In addition, the embodiments of the present invention and the features in the embodiments can be combined with each other without conflict. In addition, in the above description, the description of well-known components and technologies is omitted to avoid unnecessary confusion of the concept of the present invention. In the above embodiments, each embodiment focuses on the differences from other embodiments, and the same / similar parts between the embodiments can be referred to each other.

[0052] The above is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. It should be noted that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention, and these improvements and modifications should also be regarded as the protection scope of the present invention.

Claims

1. A method for screening abnormal grains on a wafer, characterized in that: include: Obtain parameter data of all dies on the wafer to be tested; Convert each of the parameter data to establish a corresponding CDF probability graph, and set the conversion point between the data main distribution and the data secondary distribution of the CDF probability graph as a DPAT specification to separate normal grains from abnormal grains; as well as According to the set DPAT specification, all the grains on the wafer to be tested that are outside the DPAT specification are screened out as abnormal grains, thereby completing the screening of abnormal grains on the wafer to be tested.

2. The method according to claim 1, characterized in that The parameter data includes at least one of the following data: the penetration current between the collector and the emitter when the gate and the emitter are short-circuited, the source-drain voltage drop when the circuit is turned on, the leakage current appearing between the gate and the source when the source and the drain are short-circuited, and the threshold voltage at which the circuit or device starts to work or produces a specific effect.

3. The method according to claim 1, characterized in that The ordinate of the CDF probability graph is a normal probability scale, the abscissa is parameter data, and the graph is a cumulative grain density function; When the parameter data of the grains on the wafer to be tested is converted into a CDF probability diagram, the cumulative grain density distribution increases monotonically with the parameter data, the cumulative density of the grains on the data main distribution is distributed in a straight line with its parameter data and has a first slope, the cumulative density of the grains on the data secondary distribution is distributed in a straight line with its parameter data and has a second slope or is distributed in a curve, the first slope is greater than the second slope, wherein the slope conversion point between the first slope and the second slope is set as the conversion point as the DPAT specification to separate normal grains from abnormal grains.

4. The method according to claim 3, characterized in that The slope switching point is determined by second-order differentiation or Line-fit.

5. The method according to claim 3, characterized in that: The slope switching point is determined by: The sub-test items sort the test values ​​of each die on the wafer to be tested; Calculate the distance and slope between every two adjacent grain test values; All calculated distance values ​​and slopes are sorted, and points with deviations greater than a preset threshold are found to confirm the slope conversion point on the CDF probability map.

6. The method according to claim 1, characterized in that The parameter data of normal grains fall within the data main distribution of the CDF probability diagram, and the parameter data of abnormal grains fall on the data secondary distribution outside the data main distribution of the CDF probability diagram or are in a discrete state.

7. The method according to claim 1, characterized in that The step of screening out all the grains on the wafer to be tested that are outside the DPAT specification as abnormal grains according to the set DPAT specification further includes: Traversing the distribution of all parameter data of the current grain on the wafer to be tested in the corresponding CDF probability map according to the set DPAT specification, if there is a parameter data whose distribution in the corresponding CDF probability map is outside the DPAT specification, then determining that the current grain is an abnormal grain; The distribution of all parameter data of all grains on the wafer to be tested in the corresponding CDF probability graph is traversed to complete the screening of abnormal grains on the wafer to be tested.

8. The method according to claim 1, characterized in that The DPAT specification can screen out abnormal grains between the slope transition point and the ± Sigma specification.

9. The method according to claim 1, characterized in that: The method further comprises: Analyze the distribution of each parameter data on each wafer in the corresponding CDF probability map, use different specification gates to screen out corresponding abnormal grains, and obtain the distribution of abnormal grains on the corresponding wafer; and / or By detecting whether there is a secondary distribution other than the main distribution on each wafer, if there is a secondary distribution, further analyzing the physical mechanism of the abnormal distribution, wherein the line segments with different slopes of the grain and parameter data in the CDF probability diagram indicate different physical mechanisms; and / or In the CDF probability graph, tightening the normal distribution of parameter data by increasing the line slope of the parameter data to improve the consistency of grain quality and performance; and / or Analyze whether the average or center value of each wafer follows a normal distribution to screen and compare wafers to further check whether the differences between wafers are caused by tool or chamber mismatch.

10. A screening system for abnormal grains on a wafer, characterized in that: include: An acquisition unit, used for acquiring parameter data of all grains on the wafer to be tested; An establishing unit, used for converting each of the parameter data to establish a corresponding CDF probability graph, and setting a conversion point between a data main distribution and a data secondary distribution of the CDF probability graph as a DPAT specification, so as to separate normal grains from abnormal grains; The screening unit is used to screen out all the grains on the wafer to be tested that are outside the DPAT specification as abnormal grains according to the set DPAT specification, so as to complete the screening of abnormal grains on the wafer to be tested.

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

12. A computer device, characterized in that: include: Memory for storing computer programs; A processor, connected to the memory, is used to execute the computer program to implement the steps of the method according to any one of claims 1 to 9.

Citation Information

Patent Citations

  • NOR Flash product wafer testing and screening method

    CN115579044A

  • Yield rate determination method for wafer and method for multiple variable detection of wafer acceptance test

    TW201620057A