Method for diagnosing efficiency of machine
By detecting and analyzing the production indicators of machines in semiconductor manufacturing processes, and comparing the differences in production indicators with baselines, the system automatically identifies target machines with abnormal efficiency, solving the problem of high false alarm rates in existing technologies and achieving precision and real-time machine efficiency management.
Patent Information
- Application Number
- CN202411174462.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2024-08-13
- Filing Date
- 2024-08-26
- Publication Date
- 2026-02-13
Smart Images
Figure CN121523239A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to a method for diagnosing machine efficiency in a semiconductor manufacturing process. BACKGROUND
[0002] As the demand for semiconductor devices gradually increases, finding out inefficient machines and improving them immediately is one of the indispensable steps in a semiconductor manufacturing process. Generally speaking, this step monitors factors directly related to productivity, such as monitoring the time of cycle (TC) for generating a lot containing multiple wafers to monitor the productivity of the semiconductor manufacturing process. However, the current monitoring of TC is compared with the standard value set by the user, which not only has a high false alarm rate but also requires the user to set it based on experience. However, users' experiences are not the same, and machine abnormality reasons are constantly changing, making it difficult to improve the efficiency of the machine in management and statistics. On the other hand, users need to query and analyze data from different systems such as fault detection and classification (FDC) systems or statistical process control (SPC) systems to find out the cause of the problem. Therefore, those skilled in the art continue to improve the method of detecting inefficient machines. SUMMARY
[0003] The present application provides a method for diagnosing machine efficiency by detecting production indicators of machines and obtaining a first result of target machines that can show efficiency abnormalities by first analyzing production indicators of machines in the same group and second analyzing the obtained production indicators with historical data of each machine. This can significantly reduce the false alarm rate and immediately find target machines with efficiency abnormalities without relying on user experience.
[0004] One embodiment of the present application provides a method for diagnosing machine efficiency, comprising: detecting production indicators of each machine, wherein the production indicators include: a first indicator, which is the time of cycle (TC) for generating a lot containing multiple wafers; and a second indicator, which is the lot process time (LPT); first analyzing production indicators of machines in the same group; second analyzing the production indicators with historical data of each machine; and obtaining a first result of target machines that can show efficiency abnormalities according to the first analysis and / or the second analysis.
[0005] In some embodiments, the same group represents a group of machines with the same model and process parameters.
[0006] In some embodiments, the first analysis is a comparison of the production indicators of each machine with the production indicators of the machines in the same group having the best first indicator and / or the best second indicator, wherein the best first indicator is the time period (TC) having the lowest value, and the best second indicator is the lot good processing time (LPT) having the lowest value, to obtain a first difference value.
[0007] In some embodiments, a target machine showing efficiency abnormality in the first result is marked as abnormal when the first difference value exceeds the median and / or the mean ±10%.
[0008] In some embodiments, the historical data represents a baseline made based on the median and / or the mean of the production indicators of each machine during the first half of the year when detecting the production indicators of each machine.
[0009] In some embodiments, the second analysis is a comparison of the production indicators of each machine with the baseline to obtain a second difference value.
[0010] In some embodiments, a target machine showing efficiency abnormality in the first result is marked as abnormal when the second difference value exceeds the baseline ±10%.
[0011] In some embodiments, the method of diagnosing the efficiency of a machine further comprises: performing a third analysis on the target machine to obtain a second result capable of showing the cause of the efficiency abnormality, wherein the third analysis comprises a comparison of the residence time of each wafer in the target machine in the lot goods.
[0012] In some embodiments, the method of diagnosing the efficiency of a machine further comprises: performing a fourth analysis on the impact caused by the target machine to obtain a third result capable of showing the amount of machine loss and the amount of capacity loss.
[0013] In some embodiments, the method of diagnosing the efficiency of a machine further comprises: generating visual chart information according to the second result and the third result.
[0014] Based on the above, in the above method of diagnosing the efficiency of a machine, by detecting the production indicators of the machines and obtaining a first result of target machines showing efficiency abnormality by performing a first analysis on the production indicators of the machines in the same group and performing a second analysis on the obtained production indicators and the historical data of each machine, the proportion of false positives can be significantly reduced, and the target machines showing efficiency abnormality can be found immediately without relying on the experience of the user. BRIEF DESCRIPTION OF DRAWINGS
[0015] Figure 1A schematic diagram of a method for diagnosing efficiency of a machine according to an embodiment of the present application.
[0016] SYMBOL DESCRIPTION
[0017] 100: machine diagnosis system
[0018] 110: big data database
[0019] 120: sensing device
[0020] 130: processor
[0021] AM: machine efficiency poor analysis module
[0022] DM: poor efficiency detection module
[0023] ID: poor efficiency machine production index difference
[0024] LE: machine capacity loss evaluation module
[0025] LPT: second production index difference detection
[0026] M1: first machine
[0027] M2: second machine
[0028] MA: machine availability
[0029] MC1, MC2: machine type classification
[0030] PC1, PC2: product type classification
[0031] PP: product type proportion
[0032] R1: first result
[0033] R2: second result
[0034] R3: third result
[0035] R4: fourth result
[0036] RT: wafer residence time
[0037] TC: first production index difference detection
[0038] W1: first wafer
[0039] W2: second wafer DETAILED DESCRIPTION
[0040] The present application can be more fully understood from the following description of various
[0041] It should be understood that when an element as a layer, region or plate is referred to as being "on" or "connected to" another element, it can be directly on or connected to the other element or intervening elements can also be present. In contrast, when an element is referred to as being "directly on" or "directly connected to" another element, there are no intervening elements present. As used herein, "connected" can mean physically and / or electrically connected, and "electrical connection" or "coupling" can be an electrical connection that can exist between two elements, with other elements present between the two elements. As used herein, "electrical connection" can include a physical connection (e.g., a wired connection) and a physical disconnection (e.g., a wireless connection).
[0042] As used herein, "about," "approximately," or "substantially" includes the exact stated value and average values within an acceptable range of deviation for the particular value in light of the measurement and the error associated with the measurement (i.e., the limitations of the measurement system). For example, "about" can mean within one or more standard deviations, or within ± 30%, ± 20%, ± 10%, ± 5% of the stated value. Further, as used herein, "about," "approximately," or "substantially" can select a more acceptable range of deviation or standard deviation for optical properties, etching properties, or other properties, and can not apply one standard deviation to all properties.
[0043] The terminology used herein is for the purpose of describing illustrative embodiments only and is not intended to be limiting of the disclosure. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.
[0044] Figure 1 A schematic diagram of a method of diagnosing efficiency of a machine according to an embodiment of the present application.
[0045] Referring to Figure 1 The method of diagnosing efficiency of a machine according to the present application can be performed, for example, by a machine diagnosing system 100. In some embodiments, the machine diagnosing system 100 can include a big data database 110, a sensing device 120, and a processor 130.
[0046] In some embodiments, the big data database 110 can be configured to store data for the underperformance detection module DM, the first result R1 obtained by the underperformance detection module DM, data for the machine underperformance analysis module AM, the second result R2 obtained by the machine underperformance analysis module AM, data for the machine capacity loss evaluation module LE, and the third result R3 obtained by the machine capacity loss evaluation module LE.
[0047] In some embodiments, the sensing device 120 can be configured to detect production metrics or wafer time histories for the first wafer W1 processed by the first machine M1 and the second wafer W2 processed by the second machine M2. Notably, Figure 1 The number of the sensing device 120, the first machine M1, the second machine M2, the first wafer W1, and the second wafer W2 shown in FIG. 1 is merely an example, and the number is not limited to Figure 1 the number shown in FIG. 1.
[0048] In some embodiments, the processor 130 can be configured to execute the underperformance detection module DM to generate the first result R1 (i.e., output underperformance machines). In other embodiments, the processor 130 is configured to execute the machine underperformance analysis module AM to generate the second result R2 (i.e., output underperformance reasons). In yet other embodiments, the processor 130 is configured to execute the machine capacity loss evaluation module LE to generate the third result R3 (i.e., output machine loss degrees). In some alternative embodiments, the processor 130 is configured to generate a fourth result R4 (i.e., output visualization chart information) according to the second result R2 and the third result R3.
[0049] In the present embodiment, the method of diagnosing machine efficiency can include the following steps.
[0050] First, production metrics for each machine are detected, wherein the production metrics include a first metric, which is a time of cycle (TC) for outputting a lot containing a plurality of wafers, and a second metric, which is a lot process time (LPT). In some embodiments, the TC and the LPT for each machine can be detected by the sensing device 120.
[0051] Then, the detected production indicators can be analyzed by the inefficient detection module DM to obtain the first result R1 of target machines showing efficiency abnormality. In the inefficient detection module DM, machine type classification MC1 and product classification PC1 are first performed, then the production indicators of machines in the same group are analyzed first, and the detected production indicators are analyzed secondly with the historical data of each machine, and then the first result R1 of target machines showing efficiency abnormality can be obtained according to the first analysis and / or the second analysis, so that the proportion of false positives can be significantly reduced and the target machines showing efficiency abnormality can be found immediately without relying on user experience.
[0052] In the present embodiment, the first analysis can be used for the difference detection TC of the first production indicator and the difference detection LPT of the second production indicator. For example, the machines in the same group can be classified by machine type classification MC1 and product classification PC1, and the TC and LPT of the obtained machines in the same group can be detected. In some embodiments, the same group represents a group of machines with the same model and process parameters. In some embodiments, the first analysis is to compare the production indicators (such as TC and LPT) of each machine with the production indicators of the machines in the same group with the best first indicator (i.e. TC) and / or the best second indicator (i.e. LPT), to obtain a first difference value, wherein the best first indicator refers to the time period (TC) with the lowest value, and the best second indicator refers to the batch goods handling time (LPT) with the lowest value. In some embodiments, more than ±10% of the median and / or average of the first difference value is marked as abnormal and shown as the target machine showing efficiency abnormality in the first result R1.
[0053] In the present embodiment, the second analysis can be used for the difference detection TC of the first production indicator and the difference detection LPT of the second production indicator. For example, the historical data of the production indicators of each machine can be a baseline made based on the median and / or average of the production indicators of each machine during the first half of the year when detecting the production indicators. The historical data can be stored in the big data database 110, and when performing the second analysis, the production indicators (TC or LPT) of each machine are compared with the baseline to obtain a second difference value. In some embodiments, more than ±10% of the second difference value is marked as abnormal and shown as the target machine showing efficiency abnormality in the first result R1.
[0054] In some embodiments, the first result R1 can be the target machine tool determined as efficiency abnormal in the first analysis and / or the target machine tool determined as efficiency abnormal in the second analysis. In some embodiments, the first result R1 can be the target machine tool determined as efficiency abnormal in the difference detection TC of the first production index and / or the target machine tool determined as efficiency abnormal in the difference detection LPT of the second production index.
[0055] In some embodiments, the method of diagnosing the efficiency of the machine tool can include performing further analysis on the first result R1 by the machine tool efficiency poor analysis module AM to obtain a second result R2 capable of showing the cause of the efficiency abnormal. In the efficiency poor analysis module AM, machine tool type classification MC2 and product classification PC2 are first performed, and then a third analysis is performed on the target machine tool shown as efficiency abnormal in the first result R1 to obtain the second result R2 capable of showing the cause of the efficiency abnormal, wherein the third analysis includes comparing the residence time of each wafer in the target machine tool in the batch goods (i.e. Figure 1 the abnormal wafer residence time), so that the action, time, position, etc. of the wafer in the target machine tool can be further analyzed by detecting the abnormal wafer residence time to obtain the second result R2 capable of showing the cause of the efficiency abnormal. In some embodiments, the further analysis can include machine tool transfer step clustering and machine tool transfer position analysis.
[0056] In some embodiments, the method of diagnosing the efficiency of the machine tool can include performing further analysis on the first result R1 by the machine tool capacity loss evaluation module LE to obtain a third result R3 capable of showing the machine tool loss amount and the capacity loss amount. In the machine tool capacity loss evaluation module LE, machine tool utilization rate MA and product proportion PP are first evaluated, and the evaluation results can be stored in the big data database 110. Then, a fourth analysis is performed on the impact caused by the target machine tool to obtain the third result R3 capable of showing the machine tool loss amount and the capacity loss amount. The fourth analysis includes calculating the machine tool loss amount and the capacity loss amount based on the efficiency poor machine tool production index difference ID, respectively. For example, the machine tool loss amount and the capacity loss amount can be obtained by the following formula 1 and formula 2, respectively, wherein the time in formula 1 can be 24 hours (i.e. the average daily loss amount) or 168 hours (i.e. the average weekly loss amount), but is not limited thereto. And the best production index in formula 2 can refer to the best TC or LPT.
[0057] [Formula 1]
[0058] Machine tool loss amount = (production index difference x product proportion x output) / (machine tool utilization rate x time)
[0059] [Formula 2]
[0060] Capacity loss amount = (production index difference x product type ratio x production) / optimal production index
[0061] In some embodiments, the method of diagnosing machine efficiency can include generating visualized chart information (i.e., fourth result R4) from the second result R2 and the third result R3.
[0062] In summary, in the method of diagnosing machine efficiency in the above embodiments, the first result of the target machine capable of showing efficiency abnormality can be obtained by detecting the production index of the machine, and by performing the first analysis on the production index of the same group of machines and the second analysis on the obtained production index and the historical data of each machine, so that the proportion of false positives can be significantly reduced and the target machine with efficiency abnormality can be found immediately without relying on the experience of the user.
Claims
1. A method for diagnosing machine efficiency, comprising: detecting production indicators of each machine, wherein the production indicators comprise: a first indicator, which is a time of cycle (TC) of producing a lot containing a plurality of wafers; and a second indicator, which is a lot process time (LPT); performing a first analysis on the production indicators of the same group of machines; performing a second analysis on the production indicators with historical data of each machine; and obtaining a first result of target machines showing efficiency abnormality according to the first analysis and / or the second analysis.
2. The method of claim 1, wherein the same group represents a group of machines having the same model and process parameters.
3. The method of claim 1, wherein the first analysis is comparing the production indicators of each machine with the production indicators of the machines having the best first indicator and / or the best second indicator in the same group, to obtain a first difference value, wherein the best first indicator refers to the time of cycle (TC) having the lowest value, and the best second indicator refers to the lot process time (LPT) having the lowest value.
4. The method of claim 3, wherein the target machines showing efficiency abnormality in the first result are indicated as abnormal if the first difference value exceeds the median and / or the mean of ±10%.
5. The method of claim 1, wherein the historical data represents a baseline of each machine based on the median and / or the mean of the production indicators of each machine during the first half of the year when the production indicators are detected.
6. The method of claim 5, wherein the second analysis is comparing the production indicators of each machine with the baseline, to obtain a second difference value.
7. The method of claim 6, wherein the target machines showing efficiency abnormality in the first result are indicated as abnormal if the second difference value exceeds the baseline of ±10%.
8. The method of claim 1, further comprising: performing a third analysis on the target machines, to obtain a second result showing the cause of efficiency abnormality, wherein the third analysis comprises comparing the residence time of each wafer in the lot in the target machines.
9. The method of claim 8, further comprising: performing a fourth analysis on the impact caused by the target machines, to obtain a third result showing the amount of machine loss and the amount of capacity loss.
10. The method of claim 9, further comprising: generating visualized chart information according to the second result and the third result.