CN101118423A - Serviceability selecting method and system of virtual measuring prediction model - Google Patents
Serviceability selecting method and system of virtual measuring prediction model Download PDFInfo
- Publication number
- CN101118423A CN101118423A CNA200610108224XA CN200610108224A CN101118423A CN 101118423 A CN101118423 A CN 101118423A CN A200610108224X A CNA200610108224X A CN A200610108224XA CN 200610108224 A CN200610108224 A CN 200610108224A CN 101118423 A CN101118423 A CN 101118423A
- Authority
- CN
- China
- Prior art keywords
- data
- virtual
- virtual measurement
- prediction model
- failure detection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 69
- 238000012423 maintenance Methods 0.000 claims abstract description 25
- 238000001514 detection method Methods 0.000 claims description 19
- 238000004519 manufacturing process Methods 0.000 claims description 16
- 238000003070 Statistical process control Methods 0.000 claims description 12
- 238000007726 management method Methods 0.000 claims description 12
- 238000012847 principal component analysis method Methods 0.000 claims description 9
- 238000000513 principal component analysis Methods 0.000 claims description 5
- 238000012544 monitoring process Methods 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 abstract description 4
- 238000003908 quality control method Methods 0.000 description 10
- 239000004065 semiconductor Substances 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000002872 Statistical quality control Methods 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 206010047571 Visual impairment Diseases 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000000491 multivariate analysis Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Images
Abstract
The present invention discloses an applicability choosing method of a dummy measure pre-estimation model. First, historical technical data of a machine table is obtained. The historical technical data is analyzed to get a data group after the machine table is maintained, and a plurality of dummy measure pre-estimation models is created according to the data group. The dummy measure pre-estimation model is added into a dummy measure model managing and choosing system in a dummy measure engine. The next maintenance of the machine table and the first technical data after a technology is performed for a wafer obtained, and an optimum dummy measure pre-estimation model is chosen according to the technical data to pre-estimate the wafer.
Description
Technical Field
The present invention relates to a quality control method for semiconductor manufacturing, and more particularly, to a method and system for selecting a virtual metrology prediction model for quality control in semiconductor manufacturing.
Background
Statistical Quality Control (SQC) is a technology for maintaining and improving product Quality, and Statistical Process Control (SPC) is one of the main tools, which focuses on the analysis of data during the manufacturing Process to determine the cause of product variation. Statistical quality management consists of two major components, statistical process control and sampling admission criteria. Statistical process control includes Quality Control (QC) processing and basic theories of probability and statistics and their applications. SPC is used to predictively monitor process variables or quality variables of products using process variables, and since there is a certain time lag between the process variables changing and the safety/quality issues, how to predict the quality variables in the shortest time is one of the important factors to be considered when evaluating the relative merits of SPC methods.
In the production process, several differences are allowed, and only these differences need to be properly controlled, and the degree of quality and quality needs to be within a certain range by means of the control process. The Quality Control (QC) is to extract samples during the manufacturing process, to measure the data of the samples, to perform statistical analysis and to draw a control chart to control whether the process is abnormal or not, or to extract several samples from a large batch of products to check the characteristics of the samples, to analyze the data to determine whether the products are qualified or not and whether the products need to be disposed.
In addition, the problem with using virtual metrology estimates in semiconductor processing in the past is that the tool conditions may change after a machine has undergone Maintenance (PM). For example, the original temperature is set at 250 ℃, which may become 260 ℃ or 270 ℃ after maintenance, resulting in the virtual measurement engine 260 not being able to operate or the estimated result being inaccurate due to incorrect settings.
Therefore, the invention provides a method for selecting the applicability of the virtual measurement estimation model applied to the quality control of semiconductor manufacturing, which can effectively control the state of the machine after maintenance so as to select the estimation model suitable for the state of the machine at that time.
Disclosure of Invention
Based on the above objectives, the embodiment of the present invention discloses a system for selecting the applicability of a virtual measurement prediction model, which includes a control module, a process tool, a failure detection and classification system, a manufacturing execution system, and a virtual measurement engine, wherein the virtual measurement engine further includes a virtual measurement model management and selection system. The control module obtains historical process data of the machine, analyzes the historical process data, generates a plurality of data communities of the machine after maintenance, and establishes a plurality of virtual measurement estimation models according to the data communities. The processing machine performs a process on a wafer. The failure detection and classification system generates failure detection and classification data after the wafer has been processed. The manufacturing execution system sends a signal when the machine completes maintenance. The virtual metrology engine obtains the failure detection and classification data from the failure detection and classification system. The virtual measurement model management and selection system obtains the virtual measurement prediction models from the control module, obtains the failure detection and classification data from the failure detection and classification system when receiving the signal from the manufacturing execution system, and selects an optimal virtual measurement prediction model according to the falling points of the failure detection and classification data to predict the chip.
The embodiment of the invention further discloses a method for selecting the applicability of the virtual measurement prediction model. First, historical process data of a machine is obtained. And analyzing the historical process data, generating a data community after the machine maintenance, and establishing a plurality of virtual measurement estimation models according to the data community. Adding the virtual metrology pre-estimated models to a virtual metrology model management and selection system in a virtual metrology engine. The first process data after the next maintenance of the machine and the process of a wafer are obtained, and an optimal virtual measurement estimation model is selected according to the process data to estimate the wafer.
Drawings
FIG. 1 is a schematic diagram showing a virtual measurement engine.
Fig. 2 is a schematic diagram illustrating an architecture of a system for selecting a virtual measurement prediction model according to an embodiment of the invention.
Fig. 3 is a flowchart illustrating steps of a method for selecting a suitability of a virtual measurement prediction model according to an embodiment of the present invention.
Description of the figures
200-virtual engine pre-estimation system
210 manufacturing execution System
220-equipment control system and equipment automation program
230-process machine
240-measuring machine
250-failure detection and classification system
260-virtual measurement Engine
270 virtual measurement model management and selection system
280-data storage
290-on-line real-time statistical process control monitoring application technology module
Detailed Description
In order to make the objects, features and advantages of the present invention comprehensible, preferred embodiments accompanied with figures 1 to 3 are described in detail below. The present description provides various examples to illustrate the technical features of various embodiments of the present invention. The arrangement of the components in the embodiments is illustrative and not intended to limit the invention. And the repetition of the reference numbers in the embodiments is a simplified description and does not imply a relationship between the different embodiments.
The embodiment of the invention discloses a virtual measurement estimation method and a virtual measurement estimation system applied to quality control of semiconductor manufacturing.
To obtain detailed process tool performance data, a large number of high resolution system variables, i.e., a real-time process state Identification (SVID) value for each tool, such as temperature, pressure, gas flow rate, etc., are required. The method includes collecting the SVID of each tool (i.e. FDC data) by a Failure Detection and Classification (FDC) system, transmitting the obtained FDC data to a Virtual measurement Engine (Virtual measurement Engine) of the present invention, and performing an operation to generate a Virtual measurement result.
Virtual Metrology (VM) is a method of applying intelligent computing (computerized Intelligence) to estimate the quality of a project processed by a manufacturing tool in real time. If the estimated accuracy meets the requirement of the general advanced process/Equipment Control (APG/AEC), the virtual measurement can replace the traditional measurement machine. Therefore, the virtual measurement engine of the present invention is a system with built-in pre-estimation model (as shown in fig. 1), and when the FDC data is inputted into the virtual measurement engine, the virtual measurement result can be generated. The predictive model includes(which estimates film Thickness), CD = f (x) 1 ,x 2 ,...,x n ADI) (which is used to estimate the line width (CD)), and the like, wherein x i Expressed as SVID and ADI as displayPost-shadow process chip inspection (afterimage development inspection).
Then, as mentioned above, the condition of the tool may change after the tool is maintained. Referring to attachment 1, the dotted straight line represents the point in time when the machine is undergoing maintenance. After the machine is maintained, the Average value (Average) of the heater Power input (Stage _ Power) and the load position (RFLoad) parameter of the radio adapter will jump upwards greatly, then gradually become gentle, and the same condition will be repeated when the machine is maintained next time.
To solve such problems, parameters related to the equipment and process can be sent to the chart1, the virtual measurement engine generates a virtual measurement model after operation. As shown in appendix 2, the machine and process related data in a certain time interval (e.g., the time period between run number 5744 and 6747, such as the left double-arrow section) is trained (Training) with a virtual metrology function to generate a virtual metrology model, i.e., THK = f (Stage _ Power, RFLoad) =0.15 × Stage _ Power 2 +5 + RFDOad +52, and then the variation of the machine condition is solved by using the model. However, the model does not effectively solve the variation of the machine status in the right double-headed segment, mainly because the model is not established by using the relevant data of the machines and processes in the segment. It should be noted that the above description is only exemplary, and not intended to limit the present invention.
In view of the above, the present invention establishes respective virtual measurement models for the machine status after each maintenance of the machine, and matches with a Principal Component Analysis (PCA) method to effectively solve the above problems.
The principal component analysis method of the invention mainly reduces the Dimension (Dimension) of the process variable displayed on the control chart, and removes parameters which have no influence on the process, so as to effectively distinguish which process parameters are factors which really influence the change of the machine condition. Referring to the accessory 3, when the process parameters (e.g., temperature, pressure, gas flow) are represented in three-dimensional state (X, Y, Z axes are S1, S2, S3, respectively (S is represented as the state variable identifier (StatusVariableID) of the tool), the analysis result may be incorrect due to the overlapping of the control points of the process parameters, so all the control points must be projected onto a two-dimensional plane, and the original three axes of S1, S2, S3 are converted into the X and Y axes of Com1 and Com2 (Com is represented as the tool Component (Component)) by using the principal Component analysis method, where Com1 = a 1 s 1 +a 2 s 2 +...+a n s n And Com2= b 1 s 1 +b 2 s 2 +...+b n s n Where a is a load parameter and s is a state variable identification code (StatusVariab)leID). Therefore, each control point can be clearly displayed, so that the principal component analysis result can be more accurate.
Fig. 2 is a schematic diagram illustrating an architecture of a system for selecting a suitability of a virtual measurement prediction model according to an embodiment of the invention.
The Virtual measurement pre-estimation system 200 of the embodiment of the invention includes a manufacturing execution system 210, a machine control system and a machine equipment automation program 220, a process machine 230, a measurement machine 240, an FDC system 250, a Virtual measurement engine 260, a data warehouse 280, and an online Real-Time (RT-Time) statistical process control monitoring application technology module (hereinafter, referred to as RT-SPC module) 290, where the Virtual measurement engine 260 further includes a Virtual measurement Model management and selection system 270, where a bold solid line represents a control operation related to a wafer and a dotted line represents a control operation related to a Virtual measurement Model (Virtual measurement Model).
First, the RT-SPC module 390 obtains the historical process data of the tool, i.e. the first process data after the tool is maintained, and analyzes and generates a data community (Group) after the tool is maintained by using the principal component analysis method (as shown in attachment 4). Then, for each data community, a virtual measurement pre-estimation model is respectively established according to the process data of the machine in each maintenance cycle (as shown in the attached component 5), and all the established virtual measurement pre-estimation models are added to the virtual measurement model management and selection system 270 in the virtual measurement engine 260.
Next, when the machine completes maintenance, a signal is sent to the virtual measurement engine 260. When the next wafer is processed in the processing tool 230, the tool control system and the tool automation program 220 are used to sample the wafer, and the FDC system 250 sends the FDC data to the virtual metrology engine 260 when the wafer is finished. When receiving the FDC data, the virtual metrology model management and selection system 270 determines which virtual metrology prediction model area the process data falls in based on the obtained FDC data and by a principal component analysis method, and then selects an optimal virtual metrology prediction model (as shown in the attachment 6) according to the falling point to predict the wafer, for example, the film thickness or the line width.
Fig. 3 is a flowchart illustrating steps of a method for selecting a suitability of a virtual measurement prediction model according to an embodiment of the present invention.
First, historical process data of a tool, i.e., first process data after maintenance of the tool, is obtained, and a data community (Group) after maintenance of the tool is analyzed and generated by a principal component analysis method (step S1). As shown in the attachment 4, the first process data after each maintenance of the machine is obtained, and after the analysis by the principal component analysis method, four data communities shown in the diagram of the attachment 4 can be obtained.
Then, for each data community, a virtual measurement estimation model is respectively established according to the process data of the machine in each maintenance period (step S2). As shown in the appendix 5, a first virtual measurement prediction Model (Model 1), a second virtual measurement prediction Model (Model 2), a third virtual measurement prediction Model (Model 3) and a fourth virtual measurement prediction Model (Model 4) are respectively established for the four data communities.
Next, add all the created virtual metrology pre-estimated models to a virtual metrology model management and selection system in a virtual metrology engine (step S3). Then, the first process data after the next machine maintenance and process execution is obtained, and the main component analysis method is used to determine which virtual measurement estimation model is used for estimation after the next maintenance (step S4). As shown in the appendix 6, when the projected point of the process data falls in the area of the fourth virtual measurement prediction Model (Model 4), the fourth virtual measurement prediction Model is selected for prediction.
The invention relates to a method for selecting the applicability of a virtual measurement prediction model applied to quality control of semiconductor manufacturing, which effectively controls the condition of a machine after maintenance by using a principal component analysis method in statistical multivariate analysis so as to select the prediction model suitable for the condition of the machine at the time, and solves the problem of inaccurate virtual measurement model caused by the change of the condition of the machine after maintenance by matching with system integration.
While the foregoing is directed to embodiments of the present invention, other and further embodiments of the invention may be devised without departing from the basic scope thereof, and the scope thereof is determined by the claims that follow.
Claims (10)
1. A system for selecting applicability of a virtual measurement prediction model, comprising:
the control module is used for acquiring historical process data of the machine, analyzing the historical process data, generating a plurality of data communities of the machine after maintenance, and establishing a plurality of virtual measurement estimation models according to the data communities;
a process tool for performing a process on a wafer;
a failure detection and classification system that generates failure detection and classification data after the wafer has been processed;
a manufacturing execution system, which sends a signal when the machine completes maintenance; and
a virtual measurement engine that obtains the failure detection and classification data from the failure detection and classification system, and further comprising:
and a virtual measurement model management and selection system, which obtains the virtual measurement pre-estimation models from the control module, obtains the failure detection and classification data from the failure detection and classification system when receiving the signal from the manufacturing execution system, and selects an optimal virtual measurement pre-estimation model according to the falling point of the failure detection and classification data to pre-estimate the wafer.
2. The system of claim 1, wherein the control module analyzes and generates the data populations by a principal component analysis.
3. The system of claim 2, wherein the control module establishes a virtual metrology prediction model for each data population based on process data of the tool during each maintenance cycle.
4. The system for selecting a virtual metrology prediction model of claim 1 wherein the virtual metrology model management and selection system determines within which region of the virtual metrology prediction model the failure detection and classification data falls by a principal component analysis method.
5. The system of claim 1, wherein the control module is an online real-time statistical process control and monitoring application module.
6. A method for selecting the applicability of a virtual measurement prediction model comprises the following steps:
obtaining historical process data of a machine;
analyzing the historical process data and generating a data community after the machine maintenance;
establishing a plurality of virtual measurement estimation models according to the data communities;
adding the virtual measurement pre-estimation models to a virtual measurement model management and selection system in a virtual measurement engine;
obtaining first process data after the machine station is maintained for the next time and a process is executed on a wafer;
selecting an optimal virtual metrology prediction model according to the process data to predict the wafer.
7. The method of claim 6, further comprising analyzing and generating the data populations by a principal component analysis method.
8. The method according to claim 7, further comprising establishing a virtual metrology prediction model for each data population based on process data of the tool during each repair/maintenance cycle.
9. The method of claim 6, further comprising determining which region of the virtual metrology prediction model the failure detection and classification data falls within by a principal component analysis.
10. The method of claim 6, wherein the control module is an online real-time statistical process control and monitoring application module.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNA200610108224XA CN101118423A (en) | 2006-08-01 | 2006-08-01 | Serviceability selecting method and system of virtual measuring prediction model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNA200610108224XA CN101118423A (en) | 2006-08-01 | 2006-08-01 | Serviceability selecting method and system of virtual measuring prediction model |
Publications (1)
Publication Number | Publication Date |
---|---|
CN101118423A true CN101118423A (en) | 2008-02-06 |
Family
ID=39054568
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CNA200610108224XA Pending CN101118423A (en) | 2006-08-01 | 2006-08-01 | Serviceability selecting method and system of virtual measuring prediction model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN101118423A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101976045A (en) * | 2010-08-25 | 2011-02-16 | 江苏大学 | Panel quality virtual measurement method and system for TFT-LCD etching process |
CN102456083A (en) * | 2010-10-20 | 2012-05-16 | 北京北方微电子基地设备工艺研究中心有限责任公司 | Process data analyzing method and system |
TWI580912B (en) * | 2016-03-21 | 2017-05-01 | China Steel Corp | A Method for Predicting the Maintenance Time of Trolley Air Conditioning |
CN114841378A (en) * | 2022-07-04 | 2022-08-02 | 埃克斯工业(广东)有限公司 | Wafer characteristic parameter prediction method and device, electronic equipment and readable storage medium |
-
2006
- 2006-08-01 CN CNA200610108224XA patent/CN101118423A/en active Pending
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101976045A (en) * | 2010-08-25 | 2011-02-16 | 江苏大学 | Panel quality virtual measurement method and system for TFT-LCD etching process |
CN102456083A (en) * | 2010-10-20 | 2012-05-16 | 北京北方微电子基地设备工艺研究中心有限责任公司 | Process data analyzing method and system |
CN102456083B (en) * | 2010-10-20 | 2013-10-30 | 北京北方微电子基地设备工艺研究中心有限责任公司 | Process data analyzing method and system |
TWI580912B (en) * | 2016-03-21 | 2017-05-01 | China Steel Corp | A Method for Predicting the Maintenance Time of Trolley Air Conditioning |
CN114841378A (en) * | 2022-07-04 | 2022-08-02 | 埃克斯工业(广东)有限公司 | Wafer characteristic parameter prediction method and device, electronic equipment and readable storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
TWI625615B (en) | Prediction model building method and associated predicting method and computer software product | |
US8452441B2 (en) | Process quality predicting system and method thereof | |
US8437870B2 (en) | System and method for implementing a virtual metrology advanced process control platform | |
US8046193B2 (en) | Determining process condition in substrate processing module | |
Wells et al. | Statistical process monitoring approach for high-density point clouds | |
KR20180061769A (en) | Semiconductor Manufacturing Yield Prediction System and Method based on Machine Learning | |
JP4911055B2 (en) | Batch process data analysis apparatus and abnormality detection / quality estimation apparatus using the same | |
US10901398B2 (en) | Controller, control program, control system, and control method | |
CN101118422A (en) | Virtual measurement prediction generated by semi-conductor, method for establishing prediction model and system | |
US11940357B2 (en) | System for predicting anomalies of machining | |
CN110096036A (en) | A kind of determination method, device and equipment of equipment state | |
KR101998972B1 (en) | Method of analyzing and visualizing the cause of process failure by deriving the defect occurrence index by variable sections | |
CN101118423A (en) | Serviceability selecting method and system of virtual measuring prediction model | |
CN109816191A (en) | The qualitative forecasting method and its system of Multi-workstation System | |
CN116339145A (en) | Machine parameter regulation and control method, equipment and storage medium | |
CN103278714B (en) | A kind of virtual measurement method and system mixing processing procedure | |
CN116613086A (en) | Method for predicting waiting time in semiconductor factory | |
Tang et al. | An IoT inspired semiconductor Reliability test system integrated with data-mining applications | |
US20220414555A1 (en) | Prediction system, information processing apparatus, and information processing program | |
US20130030760A1 (en) | Architecture for analysis and prediction of integrated tool-related and material-related data and methods therefor | |
TWM618987U (en) | Tendency chart table analyzing platform using artificial intelligence | |
Wenzel et al. | Improving the accuracy of cycle time estimation for simulation in volatile manufacturing execution environments | |
TWI830193B (en) | Forecasting systems, information processing devices and information processing programs | |
TWI815123B (en) | Massive data analysis platform using artificial intelligence and its operation method | |
Arellano et al. | Process Monitoring in a Flexible Manufacturing Cell through Data Analytics and a Visualisation Tool |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
WD01 | Invention patent application deemed withdrawn after publication |