JP2001521249A - Software system and method for extending classification and attributes in production analysis - Google Patents

Abstract

(57) [Summary] Software systems are used to extend classification attributes in the analysis of production data. The recipe editor (810) allows a user to create a recipe framework (820). The recipe is provided to the dataset framework (830) through the application module (840), and the page (860) is generated in a window format using the graphic user interface (GUI) framework (850). The recipe framework (820) is interfaced to a scheduler and event service (870) that automatically performs and issues production data analysis at specific times in the future. The software operates with greater flexibility in a Windows environment because it allows the user to add defect classifications by dragging and dropping with a computer mouse. When cluster classification is performed, the production map is filtered to remove non-defective data from the selected classification, and the attributes of the defect are easily made visible.

Description

DETAILED DESCRIPTION OF THE INVENTION

RELATED APPLICATIONS This application is hereby incorporated by reference into “Software System and Method for G
"Raphically Building Customized Recipe Flowcharts" 1997
No. 08 / 958,780, filed on Jan. 27, pending a pending P application.
Related to CT patent applications.

BACKGROUND OF THE INVENTIONField of the invention  The present invention relates to computer software for quality control application of a semiconductor production line.
Software that extends the namespace classification, especially in the analysis of production data.
About software packages.

[0003] 2.Description of the prior art  The software package allows the user to click the mouse or use the arrow keys
To interact with the program through the symbols on the computer screen.
So you can talk

[0004]

[Outside 1] It is increasingly written in a format. Such programs are typically Visual C
Written in a programming language such as ++ or Visual Basic, it allows programmers to call generalized programming objects to display selectable data entry fields. These programs are more user friendly and give greater flexibility because the user can select any of a number of program functions based on the displayed visual cues.

[0005] Object-oriented programming is a programmer that is modular in design and can operate on many different hardware systems.

[0006]

[Outside 2] To call specific functional subroutines.

[0007]

[Outside 3] One major advantage of is that many programming objects are included in the basic system. Examples of visible objects include classes for buttons, list boxes, dialog boxes, toolbars, scroll bars, menu bars, and title bars. Mouse functionality is also enabled.

[0008]

[Outside 4] Since the programmer working in the environment has access to all of these objects, the programmer's programming ability is greatly enhanced.

[0009] Quality control has historically relied on automated computer systems. Automated production lines replace workers, and automated quality control eliminates the need for inspectors. The semiconductor industry is one important user of quality control. Fine defects in a semiconductor wafer have a significant effect on the cost and yield of a semiconductor production process. Semiconductor technology is very sensitive to defects,
Eliminating the exact source of defects in the production process facilitates major advances in the semiconductor industry.

Currently, quality control software packages for the semiconductor industry are available,
Generally, the program is written specifically for the project. Although these programs are written to include a number of analytical utilities around the backbone system, the number of these utilities is essentially constant and the programs are difficult to change. Therefore, these packages are expensive and lack flexibility.

One method of measuring the number and location of defects in a semiconductor chip involves filtering or eliminating non-defective data from production data to leave only defective data. A map of actual production data may be a map of another set of actual production data, or
It can be compared to an ideal set of production data stored in computer memory. The defect data is then analyzed by number, density, location, type, and steps in the production process to modify the production process itself. Defect data can provide clues on how to increase production yields and reduce costs.

Various prior art data analysis systems and methods have been disclosed for industrial process control of semiconductor production. One of them is US Pat. No. 5,226,118 to Baker et al. The patent specification by Baker et al. Discloses a system for generating charts. The chart is displayed as a gallery of cells, where each cell is a unit of a two-dimensional array. Each cell in the chart is

[0013]

[Outside 5] , As well as a mathematical function of one or more other cells.

Other related patents are US Pat. Nos. 4,967,381, 4,951,190, 4,873,623, 4,843,538, and 4,805,089. And No. 4,679,137. These patent specifications disclose a complex multi-level set of data structures. While the control parameters are changed to some extent by the engineer, the set of parameters is predefined by the software package. Furthermore, not all parameters can be changed, and the extent to which an operator can change them depends on the level of authentication of the operator. These systems are somewhat primitive because they do not recognize the flexibility afforded by object-oriented visual programming.

SUMMARY OF THE INVENTION The present invention is directed to a software system and method for extending cluster classification for production data analysis routines. The user selects a new category from the list box and drags and drops this new category with the computer mouse.
Drop it to add it to the table name space of the original classification. The new cluster classification is a namespace with attributes for performing the analysis routine of this cluster classification.

The set of classifications in the table name space is:

[0017]

[Outside 6] Executable to graphically display analysis results in an environment. A particular classification is a cluster of defect data in a production map. Properties identified by these cluster classifications are, for example, abrasive flaws, particles, pinholes, and blowouts. Performing a particular cluster classification filters the production data, leaving only a map of the particular defect in the cluster classification. The attributes of the classification cluster are
Analysis routines can be searched over the local network or the Internet, so that new cluster classifications from other production plants can be added.

An analysis routine of cluster classification using attributes for analyzing data from the production operation based on the production function generates a chart such as a Type Pareto or a bar graph. The software allows the user to color code the defects in each cluster classification for easy identification. The user
The filtered production map can be moved from one page to another, and the two pages can be compared side by side to look for patterns that indicate that a production defect has occurred. In this way, the pattern of defects on a given map can be more easily visualized.

[Detailed Description of Desirable Embodiment]Generate chart  Figure 1 shows one or more dialog boxes for specifying data analysis parameters.
FIG. 2 is a diagram showing a computer screen 100 having the software 101, 102, and 103. Do
The drop-down list boxes 101 and 102 are displayed on the X-axis and Y-axis of the bar graph.
Indicates whether the parameter should be placed. The parameters can be found in the drop-down menu.
Or a similar type of dialog box 101, 102
Selected using the mouse. The user sets the data analysis parameters on the X-axis and Y-axis.
Select any parameter in the menu of the dialog box 101 or 102 to make a selection.
Click the parameter.

Preferably, the dialog boxes 101, 102 are drop-down list boxes with a vertical scroll bar. The user can select the list boxes 101, 10
The list boxes 101 and 102 are dropped down by clicking the arrow boxes 111 and 112 on the side of 2. The user clicks on one of the set of text strings representing production variables and production functions in the column list in the list box rectangle. If the desired text string is found in the list boxes 101, 102, the user clicks on this text string,
As a result, this text string appears in the selected line of the list boxes 101 and 102.

One property of the present invention is to record the selected production variables and production functions in a macro property. The computer thus has the ability to learn the selection of variables and functions selected by the user and to replay specific variables and functions for a new data set. Thus, the analysis systems and methods are conventional in that new production parameters must be specified each time an analysis is performed, or where the user is restricted to select only from a specified set of pre-programmed analysis schemes. Different from technology. The macro property stores a specific set instruction for the production parameter that the user has previously selected from the dialog box and drop-down list boxes 101,102.

Parameters applicable for monitoring production defects may include numbers on the Y-axis and types on the X-axis. Then, a type / pareto chart is created and displayed.
For example, the number of defects in a semiconductor assembly line is selected as a parameter on the Y axis. The X-axis parameter is selected from a list containing the size of the defect or the date of production.

FIG. 2A is a diagram showing a type Pareto chart 200 of the number of defects for each defect class. At the top of the window is provided an appropriate title bar 210, where "
Type Pareto chart ". Tabs 221, 222, 223, and 224 indicating the number of pages directly accessible to the user are shown at the bottom of the screen. In FIG. 2A, the pages are “1 type Pareto”, “2 size histogram”,
Labeled "3 Wafer Map" and "4 Wafer Map Gallery". By clicking on these tabs at the bottom of the page, the user can move from page to page. The small box below the window 200 gives other information such as the total number of lots 230, the total number of wafers 231 and the total number of defects 232. The toolbar 240 above the window 200 shows another bar graph, FIG.
The toolbar 240 has buttons for further changing the magnification of the bar graph or the buttons for generating the calendar of FIG. 2B, with various buttons giving further selections such as accessing the wafer map of FIG. give.

The software package that creates the bar graph 200 of FIG. 2A includes a dialog box button 241 in the toolbar kit 240 to generate the chart of FIG. 2B. A dialog box 250 appears containing a string selection drop down list box 277 shown in FIG. 2B. Then, the user selects a method of inputting a date from the string selection box 277. If the user chooses to select a date using calendar 270, the month and year for calendar 270 are selected from additional drop-down list boxes 272,273. Alternatively, the customer may use the string selection box 277 to enter production variables in other ways, such as from a list box of text strings such as "last 10 days", "last 10 lots", or leave blank. You can enter a date range from the keyboard in the field box. FIG. 2B shows that calendar display 270 includes a plurality of push buttons 271 for each day of calendar 270. The user selects one or more push buttons 271 on calendar 270 to indicate one or more days for which production data should be displayed in a bar graph format. The time of a given calendar date can be selected by dialog boxes 274 and 275, or the calendar can be turned by month with arrow keys 276.

Referring now to FIG. 3, the software package allows the user to
Includes a toolbar kit feature that allows two or more bar graphs 310, 320 to be placed side by side for comparison by dividing 00. For example, the user may create a bar graph 310 of the number of defects per layer number for the first production lot.
Can be compared against a similar bar graph 320 for the second production lot. The bar graphs 310, 320 can be formed as embedded packaging objects that can be dragged and dropped into other windows, or the bar graphs 310, 320 can be stored as files. Packaged objects may be represented as icons or other graphic images. The various pages are color coded for easy viewing.

[0026] 2.Access to production map  Referring now to FIG. 4A, the software package of the present invention includes a display window.
A user for accessing and displaying the production wafer map 410 in the dove 400
Includes utilities. The user can select the plus or minus zoom button bar 430
Select to expand (zoom in) or shrink (zoom out) the production map
). Each particular defect is located on the right side of the map 410 to facilitate visual identification.
It is color-coded according to the side defect number scale 440. Side by side with wafer map
And type imported from windows 310 and 320 in FIG.
A chart 450 and a size histogram chart 460 are shown.

The actual production map is compared to the production map in the wafer map gallery to filter out all patterns except defects. Using a computer cursor, the user drags and drops the packaged image of the production map 410 onto one of the maps in the wafer map gallery. The actual and ideal production maps in the wafer map gallery represent what a particular product would look like if the product had no defects. The defect-free product map 480 can be accessed by selecting the tab 470 at the bottom of the window 400 to turn the page to the wafer map gallery window.

FIG. 4B shows the production map 440 after being filtered by the information in the wafer map gallery 470. Wafer Map Gallery 47
When the actual production map 410 is subtracted from the zero ideal production map, only the defect map 460 remains. Selection of production parameters and wafer map gallery 4
By combining with the filtering from 70, the user identifies the defect pattern in a particular lot number or layer number.

[0029] 3.Creating a flowchart recipe  Referring to FIG. 5, the present invention further provides a program for constructing a customized flowchart 510 according to a selected recipe of a conditional event.
Includes RAM function. In the context of this application, a conditional event is dependent on other values
Defined as an event that occurs or does not occur. This value is the raw
It may be the result of a calculation of one or several previous production functions of the production parameters.
Conditional events create one or more branches in the flowchart depending on this value.
Make Conditional event recipes can be tailored to the desired data analysis and response scheme
Thus, a series of linked conditional events created by the user.

The flowchart construction program function operates as a “macro” in the computer by recording the keystrokes in the casebook and reproducing the keystrokes in the casebook upon a command from the user. The "macro" type program function of the present invention memorizes a series of production parameters specified by the user through a dialog box.

The casebook macro is then converted into a flowchart 510 by software and displayed. The selected production parameters and production functions are each assigned to a geometric element in the flowchart, such as a box, a circle, a diamond, depending on the function. The geometric elements of the flowchart correspond to the standard definitions in the essential sense of the flowchart: circles are input / output functions, diamonds are branch conditions, and boxes are data calculations. As shown in FIG. 5, a user may create a flowchart by dragging and dropping node 520 from dialog box 530 and linking it to one of the previous nodes 540 in flowchart 510. The user selects a particular flowchart 510 from the current page by pressing a particular button 550 on the toolbar 560 when trying to perform a particular analytical scheme on the production data. By re-recording the selection of the conditional event, displaying the new flowchart again, and again executing the new flowchart more directly through the node dialog box 530, the recipe represented by the flowchart can be easily changed by the user.

A user may wish to, for example, regularly execute a flowchart 510 that produces a standard deviation of defects in a second layer of a product for the last 10 days. This particular flowchart 510 has been previously created and recorded by the user. Each page in this casebook is represented by a corresponding graphical node symbol 541, 542, 543, 544, 545, displayed in window 500 upon selection.
And 546. After being created, the flowchart 510 is displayed in a linked window in the display window 500 shown in FIG. The user then generates one or a series of bar graphs as shown in FIG.
Other functions are performed by selecting this particular flowchart 510 therein.

The analytical subroutine represented by the flowchart can be executed automatically by the user on a regular basis. Rather than specifying a certain date,
A comprehensive grouping of production data, such as "0 days" or "last 10 lots" is specified. This analytical subroutine is then executed, for example, every week or every other week without operator intervention. Then, at these times, one or more analytical scheme reports are automatically generated. The report is generated automatically at the operator's computer where the report is stored as a data file or at a printing location. The operator specifies the number of copies of these reports to be generated, and the location where the reports are generated is selected via a network printer.

A plurality of these reports are automatically issued according to pre-established rules without having to spend operator time each time. Copies are sent to the selected person responsible for overseeing production. Alternatively, these reports may be sent regularly in electronic form through a network server instead of printing on a network printer. The network can be a local area network or an Internet location on the World Wide Web. Electronic publishing has the advantage of saving paper and reducing costs.

It is also possible to automatically use the results of previous data analysis subroutines to conditionally change the data analysis scheme and subsequent report issuance. If a condition is detected where the result of a particular analysis exceeds a preset value, another predetermined analysis may be performed to more accurately identify the source of the defect. In practice, a feedback loop is established to maintain tight control between the test results and the analysis scheme. If the result of the analysis exceeds a predetermined control level, a notification may be sent to the engineer via an e-mail or pager system to take the specified repair action. The report can be returned to the production line to automatically initiate process changes over the Internet, for example, to launch a Java applet. Thus, the analytical tool of the present invention is not a mere data collection mechanism because process control can be automatically performed through a feedback loop.

One application of the analysis routine of the present invention is to determine when a defect is an “addition,” ie, a defect that does not exist in a previous layer. This added defect is subtracted from the defect that existed on the previous layer, and then the automatic defect control (
ADC) program.

[0037] 4.System architecture  As shown in FIG. 6A, the software system of the present invention has a three-layer architecture.
Use a key. The top layer 610 is the presentation layer, which is visible to the user.
Object finger consisting of menus, buttons, dialog boxes, and charts
Direction

[0038]

[Outside 7] Environment. The middle layer 620 distributes objects through a server and the Internet. The CORBA (Common Object Request Architecture) IIOP (Inte
r-Internet Object Protocol) middleware layer. The lowest layer 630, the storage layer, is a database containing crude production data that is connected to other servers and presentation layers by the CORBA layer.

FIG. 6B illustrates how the CORBA middleware layer interfaces with the storage layer 630 and the namespace service 640 to change the cluster classification. CORBA is a mechanism for distributing objects through a microcomputer and an operating system. The CORBA middleware layer allows the server to access objects that may exist anywhere on the network so that the new cluster classification is represented from the namespace service 640 to the dataset service 650. Some Irish clients may develop a new method of clustering 660 or custom compiling 670, for example, at that location. This method may be relevant to the needs of US clients. CORBA may obtain a namespace service 640 from an Irish client and apply it to a dataset service 650 from the United States through a server without having to access a production database located in the United States to perform the analysis.

FIG. 6C shows how CORBA is a client personal computer 6.
51 through the server to another workstation and database server 652
It is a figure which shows whether interface connection is carried out. The application server 653 stores and distributes all various calculation modules or objects. Application server 653 may be identical to database server 652, and may be client workstation 651 or any other location on the network. The recipe is contained within the application server 653, and once the recipe is executed, the results may be published on a printer 654, a web (WEB) server 655, or a client personal computer 656.

FIG. 7 is a diagram showing details of the software architecture of the presentation layer written in the Visual C ++ programming language. Recipe editor 71
0 allows the user to create a recipe framework 720. The recipe is stored in the data set framework 7 through the application module 740.
20 and using a graphic user interface (GUI) framework 750,

[0042]

[Outside 8] A page 760 is generated in the format. The recipe framework 730 is also interfaced with the scheduler and event service 770 to provide for the automatic execution and publication of production data analysis at a specified time in the future.

Once the analytical routine has been executed, the results are displayed in a graphical representation such as a bar chart, graph, or filtered production map.
Each cluster classification is displayed in different colors to make it easier to see. For example, a filtered production map is generated by color-coding friction flaws in red, particles in blue, and pinholes in green. Or, for comparison, bar graphs were generated and placed side by side on a computer screen, one bar graph showing the standard deviation of abrasion in lot 1 for the last 10 days in red and a second bar graph showing the most recent The standard deviation of pinholes in lot 1 for 10 days is shown in green.

[0044] 5.Classification extensions  Optical patterns to allow the computer to classify some types of defects
There is an on-line recognition system. For example, a defect such as a friction scratch is
It has some similarity, ie the defect is long and thin. Characteristic of this class of defects
By separating similarities, the computer recognizes based on the separated features.
Classify. However, various patterns that can be identified by prior art pattern recognition systems.
Class flaws are typically fixed by a specific software program.
I have. Therefore, new defect categories can be added as they become identifiable.
It is desirable that this is possible.

Defects often appear as groups or clusters because the factors that cause defects are repeated in general areas on the wafer. These defect groups are called cluster classification. An optical pattern recognition system working with a neural network computer can identify these cluster classifications in the production map. Each cluster classification represents a filter that makes only the defects of that cluster classification visible. Multiple defect filters may be applied sequentially, so that only the combination of defects applied may be manipulated by a particular data analysis routine.

The present invention includes software procedures for integrating and extending the attributes of a new cluster classification into a namespace for analysis routines. In the context of the present application, a namespace is similar to a file name, but includes additional executable attributes that tell the operating system to perform a given operation by the extension of the namespace.
Namespaces with attributes have, for example, a “.doc” extension

[0047]

[Outside 9] File. The “.doc” extension signal indicates that the user

[0048]

[Outside 10] Signals the operating system to do something when you directly click the file name to open the document. Otherwise, the user first

[0049]

[Outside 11] You have to start the word processing program, then click on the "Open File" list box on the title bar, then select the correct drive, and finally use the scroll bar to find and click on the correct file. "
. The "doc" extension is an attribute that tells the computer to perform these tasks when a file name is selected, greatly simplifying the execution. A namespace is a collection of files represented by file names that include an associated executable or other film type, which are executed when a file name is selected.

FIG. 8A is a diagram showing an essential format of a name space data file. The cluster classification 800 has a name space 801 having a file name 820 and an attribute 803. The attribute 803 has a definition 804 consisting of a number of executable files and other types of files 805. In this case, file 805 may include other files 850 for the operating system to perform optical pattern recognition and search for procedures to filter and display data analysis in the chart.
To search for.

In the present invention, the attributes of the cluster classification namespace include an associated executable utility for semiconductor defect analysis. Attributes included by this namespace include, for example, (1) continuous instructions associated with the execution of the optical pattern recognition program executed by the neural network, and (2) data collected from production lots, layers, dates, etc. A series of instructions associated with performing the analysis scheme, and (3) a data analysis scheme stored at various locations in the network (eg, from a production facility of a foreign subsidiary over the Internet, or to another user's network over an intranet). It may include a sequence of instructions associated with importing and executing (from a hard drive).

Referring to FIG. 8B, the extension is applied to cluster classification of semiconductor defects. The program includes a table name space 810 for the defect cluster classification 811. The attribute of the name space 810 indicates to the computer the cluster classification name space 811
Contains one or more executable files or other types of files that tell what to do when selected by clicking with a computer mouse. These cluster classifications 811 may include semiconductor defects 812, such as abrasions, particles, blowouts, pinholes, etc., or the defects may be classified by size. When the user selects the abrasion cluster classification namespace 812 with the cursor, the attributes of the abrasion namespace 812 are executed, thereby initiating the neural network pattern identification procedure and retrieving the data analysis parameters of the data analysis scheme. This data analysis scheme is performed, and any programs or data files are imported over the network as desired.

FIG. 8B further shows a list box 820 for adding a cluster classification to the table name space 810. The list box 820 is a drop-down list box having a plurality of expected cluster classifications 821. The user may select one of these cluster classifications 821 by using the computer cursor 8.
22 and click on the drop-down list box 820.
21 is added to the left column 811 of the table name space 810. The new cluster classification 821 is preferably created by dragging and dropping the new cluster classification 821 using the cursor 822 to a drop-down list box 820.
Is moved to the table name space 810. Thereby, the user has more flexibility in inserting new attributes 803 into the namespace 810 than is possible with a static software package. When performing the cluster classification 821, the user filters the production map so that only the friction data remains. Then, the analysis of the friction data can be easily performed.

FIG. 8C is a block diagram showing functional parts of the cluster class extension system of the present invention. Each client A, B, and C (831, 832, 833) is connected to a presentation layer (834, 835, 836) for a respective computer. Clients A and B are in the United States and Client C is in Ireland. Each client A, B, and C has a namespace 83 containing a file name 838 and an attribute 839 in a respective relational database 840, 841.
7 and the clients A and B have the cluster classification file name 838 and the attribute 83
Share a relational database 840 containing nine definitions 842. Middleware layer 834 connects client C to clients A and B over the Internet. Client C, located in Ireland, has its own relational database 841 containing namespace 837 and definitions 842, while client C also has namespace 844 containing file names 845 with definitions 847 and attributes 846. Client C has a namespace 844
Has been determined to be a particularly good cluster classification for the analysis of semiconductor data, and US clients A and B seek to obtain this. Client C thereby sends namespace 844 to clients A and B.

FIG. 8D shows that after clients A and B have initiated a request for namespace 844 (step 850) and when client C sends namespace 844 to clients A and B (step 851). 5 is a flowchart showing what happens. Client A drags a new cluster classification 821 representing namespace 844 into Client A's table namespace 810.
Drop (step 852). Client A executes namespace 844 by selecting cluster classification 821 (step 853). Name space 844 has attributes 846. Execution of the namespace 844 with the attribute 846 causes the client A computer to request transmission of the namespace definition 847 over the Internet via the middleware layer 843 (step 85).
4). The computer of the client C transmits the definition 847 in the relational database 841 of the client C to the client A via the middleware layer 843 (step 855). At the same time, definition 847 is placed in relational database 840 shared by clients A and B (step 856). Attribute 846 also identifies the presence of a particular cluster classification 821 by the optical pattern recognizer and neural network computer (step 857) and performs a particular filtering routine on the production data (step 858), thereby analyzing the data. Generate the chart (step 859) including the extension.

Although the present invention has been described with reference to preferred embodiments, it should be understood that modifications, substitutions, and alterations may be readily made by those skilled in the art without departing from the spirit of the invention. is there. Accordingly, the invention is not limited by what is claimed.

[Brief description of the drawings]

FIG. 1 is a diagram showing a computer screen for designating data analysis parameters.

FIG. 2A is a diagram showing a histogram of a bar graph generated by software, and 2B is a diagram showing a calendar generated by software.

FIG. 3 is a diagram showing a display window in which two or more bar graphs are arranged side by side for comparison.

FIG. 4A is a diagram showing zoom characteristics on a software package for viewing production data, and 4B is a diagram showing a production map after filtering.

FIG. 5 is a diagram showing a display window of a flowchart created by a software package.

FIG. 6A is a diagram showing the entire system architecture, and 6B and 6C are diagrams showing interfaces of the middle layer architecture.

FIG. 7 illustrates the operation of higher level software.

8A is a diagram showing a format for a name space, 8B is a diagram showing a list of cluster classifications for analyzing defect data and a list box for adding a cluster classification, and 8C is a diagram showing the Internet. FIG. 8D is a block diagram illustrating attributes that operate through the Internet, and 8D is a flowchart illustrating attributes that operate through the Internet.

[Procedure amendment]

[Submission date] May 29, 2000 (2000.5.29)

[Procedure amendment 1]

[Document name to be amended] Statement

[Correction target item name] Claim 1

[Correction method] Change

[Correction contents]

[Procedure amendment 2]

[Document name to be amended] Statement

[Correction target item name] Claim 9

[Correction method] Change

[Correction contents]

[Procedure amendment 3]

[Document name to be amended] Statement

[Correction target item name] Claim 16

[Correction method] Change

[Correction contents]

[Procedure amendment 4]

[Document name to be amended] Drawing

[Correction target item name] Fig. 6

[Correction method] Change

[Correction contents]

FIG. 6

[Procedure amendment 5]

[Document name to be amended] Drawing

[Correction target item name] Fig. 7

[Correction method] Change

[Correction contents]

FIG. 7

──────────────────────────────────────────────────続 き Continuation of front page (81) Designated country EP (AT, BE, CH, CY, DE, DK, ES, FI, FR, GB, GR, IE, IT, LU, MC, NL, PT, SE ), JP (72) Inventor Zhou, Steve United States of America 95014 Cupertino Glenora Drive 21073 (72) Inventor Siflet, Richard United States of America 95125 San Jose Willow Glen Way 866 (72) Inventor Kurcarni, Ashok United States of America California 95120 San Jose Valley Quayle Circle 1149 F-term (reference) 4M106 AA01 BA20 CA27 DH01 DJ20 DJ21 DJ23 5B046 AA08 CA06 DA01 GA01 HA05 KA05

Claims (24)

[Claims] Establishing a table name space for an original semiconductor cluster classification having a first attribute; selecting a new semiconductor cluster classification having a second attribute from a list box; Adding the selected new classification; selecting the new classification from the table name space; and performing an analysis routine on data from the new classification with the second attribute. How to analyze semiconductor production data. 2. The method of claim 1, wherein the performing step includes: importing a namespace including the second attribute for the new classification; performing a pattern recognition function for the new classification; Generating a graphical representation of the result of the analysis routine for the new classification with the attributes, further comprising at least one of the following steps:
The method of claim 1. 3. The method of claim 2, wherein said importing step further comprises the step of importing said namespace via the Internet or a local network. 4. The method of claim 1, further comprising displaying a plurality of graphical representations of the original classification and the new classification side by side. 5. The method of claim 1, wherein said step of performing an analysis routine includes generating information regarding clustering of defects. 6. The method of claim 5, wherein the step of generating information relating to cluster classification of defects further comprises the step of generating information relating to flaws, particles, pinholes, and blowouts. 7. The method of claim 1, further comprising importing the new classification via a server. 8. The method of claim 1, further comprising the step of importing the new classification via the Internet. 9. A table name space of an original semiconductor cluster classification having a first attribute, a list box including a plurality of new semiconductor cluster classifications having a second attribute, the first attribute or the second attribute. An analysis routine to be performed on data from either the original class or the new class with attributes. 10. The analysis routine includes: importing a namespace having the second attribute for the new classification; a pattern recognition function for the new classification; 10. The system of claim 9, further comprising at least one of: a graphical representation of a result of the analysis routine. 11. The system according to claim 10, wherein said namespace is imported via the Internet or a local network. 12. The system of claim 9, wherein the analysis routine is one of a specific lot identifier, a layer number, or a number, average, or standard deviation of defects over a time period. 13. The system of claim 9, further comprising a plurality of graphical representations of the original classification and the new classification displayed side by side. 14. The system of claim 9, wherein the step of performing an analysis routine generates information regarding cluster classification of defects. 15. The system of claim 14, wherein the cluster classification of defects includes abrasions, particles, pinholes, and blowouts. 16. A table name space means for storing and displaying an original semiconductor cluster classification having a first attribute; a list box means for storing and displaying a plurality of new semiconductor cluster classifications having a second attribute; An analysis routine to be performed on data from either the original class or the new class with the first attribute or the second attribute. 17. The analysis routine includes: means for importing a namespace including the second attribute for the new classification; means for performing a pattern recognition function for the new classification; 17. The system of claim 16, further comprising at least one of: and means for graphically representing the results of an analysis routine for the new classification. 18. The system according to claim 17, wherein said namespace is imported via the Internet or a local network. 19. The system of claim 16, wherein said analysis routine means performs a calculation of the number, average, or standard deviation of defects over a particular lot identifier, layer number, or time period. 20. The system of claim 16, further comprising means for displaying a plurality of said original classifications and said new classifications side by side. 21. The system of claim 16, wherein said analyzing routine step generates information regarding cluster classification of defects. 22. The system of claim 20, wherein the cluster classification of the defects includes abrasions, particles, pinholes, and blowouts. 23. The system of claim 16, further comprising means for importing the new classification via a server. 24. The system of claim 16, further comprising means for importing the new classification via the Internet.