CN101504543B - Method and system for extracting key process parameters - Google Patents

Abstract

A system, method, and computer readable medium are provided for extracting key process parameters associated with a selected device parameter. In one embodiment, a gene map analysis is used to determine key process parameters. The genetic map analysis includes grouping highly correlated process parameters and determining a correlation of the groups to selected device parameters. In one embodiment, the group having the greatest correlation with the selected device parameter is displayed in a correlation matrix and/or a gene map.

Description

Method and system for extracting key process parameters

related application

Pursuant to 35 U.S.C. §119, this application claims priority to U.S. Provisional Patent No. 60/916,194, filed May 4, 2007, entitled "Method and Apparatus to Enable Accurate Wafer Prediction" rights, the entire contents of which are hereby incorporated by reference.

technical field

The invention relates to a method for extracting key process parameters, in particular to a method for extracting key process parameters related to selected device parameters.

Background technique

Integrated circuits are produced through a number of processes in a wafer fabrication facility. These processes and associated fabrication tools can include thermal oxidation, diffusion, ion implantation, rapid thermal processing (RTP), chemical vapor deposition (CVD), physical vapor deposition (PVD), epitaxial formation/growth processes, etching processes, lithography processes and/or other manufacturing processes and tools known in the art. In addition, the manufacturing process includes multiple metrology processes for monitoring and controlling the yield, quality and reliability of IC manufacturing.

During the production of integrated circuits, the manufacturing process generates large amounts of data. However, the process and equipment engineers responsible for the integrated circuit manufacturing process and equipment can only determine the manufacturing process parameters and integrated circuit performance through various experiments (such as DOE) or production experience (such as for semiconductor substrates or wafers) The relationship between the measurements made by integrated circuits). Accumulating such data and knowledge is resource-intensive. Furthermore, experiments can only be performed on active parameters (such as gas flow rate) and not on passive parameters (such as reflected power). In addition, it remains difficult to determine the effect of each parameter selected in an experiment (eg, in a batch operation) on integrated circuit performance.

Therefore, there is a need for a method that can strengthen the correlation between a judgment parameter (such as a process parameter) and the performance of an integrated circuit device.

Contents of the invention

A method is provided, including selecting plant parameters and collecting process data. The process data includes time series process data and values related to multiple process parameters. Summarizes timing process data. Perform correlation analysis. Correlation analysis identifies critical parameters, which are included in multiple process parameters. The key process parameters are related to the selected device parameters. Genetic maps of selected device parameters were generated.

A computer readable medium is also provided. A computer readable medium includes instructions for determining process parameters. These instructions include receiving process data and corresponding device performance data. The process data is related to the first, second, third, and fourth process parameters. Summarize process data. The first, second, third, and fourth process parameters are grouped into a first group and a second group. The correlation among the first, second, third, and fourth process parameters is used for grouping. A correlation matrix for the first population is generated. Generate gene maps for device parameters. The gene map includes the relative correlation of the first population to the device parameter, and the relative correlation of the second population to the device parameter.

A system is further provided. The system is operable to collect manufacturing data. Manufacturing data includes timing data. The system summarizes this time series data. A correlation between the first manufacturing parameter and the second manufacturing parameter is determined. A correlation between the group comprising the first fabrication parameter and the second fabrication parameter and integrated circuit performance is also determined. A gene map showing the relative correlation between cohorts and integrated circuit performance is generated.

Description of drawings

Aspects of the invention are best understood from the following detailed description when read in conjunction with the accompanying drawings. It is emphasized that, in accordance with the standards practiced in the industry, various features are not drawn to scale. In fact, the dimensions of the various features may be arbitrarily expanded or reduced for clarity of discussion.

Fig. 1 is a flowchart illustrating an embodiment of a method for extracting key process parameters;

Figure 2 is a block diagram illustrating an embodiment of a computer system;

Fig. 3 is a flowchart showing an embodiment of a correlation analysis method;

Figure 4 is a diagram illustrating an embodiment of a one-level binary tree used in the method of Figure 3;

Figure 5 is a document/screen shot showing an embodiment of a dependency matrix; and

Figure 6 is a file/screen image showing an example of a gene map.

Detailed ways

It is to be understood that specific embodiments are provided herein by way of example to teach the broader inventive concepts, and that those skilled in the art will readily apply the teachings of the invention to other methods or devices. In addition, it is to be understood that the methods and apparatuses discussed herein involve some known structures and/or processes. Since these structures and processes are well known in the art, they will be described in general terms only. Some intermediate structures and/or processes will be omitted in this description, and their contents are only design choices. In addition, for the purpose of convenience and example, component symbols are repeated in all the drawings, and the repetition does not mean that any features or steps must be combined in all the drawings.

Referring to FIG. 1 , a method 100 for extracting critical process parameters is shown. Method 100 begins at step 102, where a device parameter is selected. Device parameters include a parameter related to the integrated circuit, including an electrical or physical parameter, which may indicate integrated circuit device performance. In one embodiment, the value of a parameter of the device is determined (eg, measured) upon wafer level testing of the integrated circuit device or portion thereof. Examples of device parameters include Iddq (change in saturation current), leakage parameters, speed parameters, and/or many other device parameters known in the art.

The method then proceeds to step 104 where wafer test parameters related to the selected device parameters are determined. In other words, a correlation analysis is performed to determine one or more wafer test parameters that affect the device parameters selected in step 102 . The effect of one parameter on another can be measured by statistical tools, such as the ^R2 value (coefficient of determination). Wafer test parameters related to device parameters may be considered herein as "critical" wafer test parameters. In one embodiment, the critical wafer test parameters are those having ^R2 values greater than the selected value; the selected value uses criteria such as process performance to be analyzed, resources used for correlation analysis , and/or other considerations known to those skilled in the art. Correlation analysis included gene map analysis, which is further described in Figure 3 below. In other embodiments, various statistical tools and methods may be used to determine one or more key WAT parameters. Wafer test parameters may also be considered wafer acceptance test (WAT) parameters. Wafer test (eg, WAT) processes include wafer-level electrical testing of integrated circuits formed on semiconductor substrates (eg, wafers), or wafer-level electrical testing of test structures. The WAT process typically includes the measurement of wafer test parameters at multiple test locations (eg, probe locations) on the wafer. Probe locations are placed on the dicing lines of the wafer or interspersed between integrated circuit devices. Wafer test parameters may include measurements of resistance, current, and/or other parameters known in the art. Values related to wafer test parameters are described herein as WAT data. For correlation analysis, WAT from multiple wafers and corresponding device performance (eg, values of device parameters) can be collected. In one embodiment, WAT data can be collected for over 200 wafers. In one embodiment, step 104 is omitted.

Method 100 then proceeds to step 106, where key process parameters are determined. The critical process parameters relate to critical wafer test parameters and/or selected device parameters, which are described with reference to steps 104 and 102, respectively. In other words, a correlation analysis may be performed to determine critical process parameters that affect the device parameters selected in step 102 . In one embodiment, correlation analysis may be performed to extract process parameters related to the critical wafer test parameters in step 104 .

Process parameters include equipment or measurement parameters related to the manufacturing process. The values of the process parameters may be determined during manufacture (eg on the production line). Values relating to process parameters are considered process data here. Process parameters can be active parameters or passive parameters. Active parameters include any manufacturing parameter that can be easily specified during the manufacturing process (such as by defining parameters of a device). Examples of active parameters include RF power, gas flow rate, concentration, and treatment time. Passive parameters include any manufacturing parameters that are not determined by the manufacturing method, but are, for example, the process itself in terms of other passive and/or active parameters, the type of equipment, the condition of the equipment, the conditions in which the wafers are processed, and/or other possible factors. Examples of passive parameters include reflected power, ambient conditions, contamination levels, and the temperature and/or pressure profile of the tool itself.

For correlation analysis, process data including data for plant parameters and metering parameters are collected, as well as corresponding plant performance (eg values of plant parameters) and/or WAT data. In one embodiment, process data is collected for more than 200 wafers. In one embodiment, the process data includes fault diagnosis and classification (FDC) data. In one embodiment, time-series process data related to one or more process parameters (eg, values related to equipment parameters) are collected. Time-series process data includes a stream of data collected at uniform time intervals, the number of wafers processed, and/or other intervals that may be determined. Time-series process data includes data before and after a preventive maintenance process, an equipment repair process, an equipment cleaning process, and/or other typical steps of a manufacturing process. In one embodiment, the timing process data may be collected using the known Generic Equipment Model/Semiconductor Equipment Communication Standard Communication (GEM/SECS). The time-series process data may be summarized with statistical analysis to provide summarized process data related to one or more process parameters.

Method 100 then proceeds to step 108, where a batch operation may be performed. In one embodiment, step 108 is omitted. The batch job processes one or more wafers using the various values of the critical process parameters determined in step 106 . Batch operations are quite useful in determining the range or optimum value of the key parameters determined in step 106 . For example, in one embodiment, one or more process conditions of the gate oxide formation chamber are identified as critical process parameters for a selected device parameter. In embodiments, batch operations may be used to determine specific process conditions (eg, values, settings).

Method 100 then proceeds to step 110, where process optimization is performed. The process can be optimized by adjusting the values of critical process parameters, determined in step 106, which are used in the manufacturing process. This value can be determined using the data collected in step 108 of the batch job. In one embodiment, the process can be optimized by varying, for example, the interval of preventive maintenance processes, cleaning processes, equipment repair processes, chamber idle time, process temperature, and/or equipment replacement intervals. By way of example, in one embodiment, device-specific sensor variation is extracted as a key process parameter that affects the equivalent oxide thickness of a device. Further analysis can provide process parameters (eg, related to sensor variations) that can be adjusted during the fabrication process to limit variations in a device's equivalent oxide thickness.

Accordingly, method 100 determines key process parameters related to selected device parameters. The method 100 can be considered as a two-tier hierarchical analysis (eg (1) extracting key WAT parameters and (2) extracting key process parameters). However, in other embodiments, other hierarchical analysis structures are possible.

Referring to FIG. 2, there is shown an embodiment of a computer system 200 for implementing embodiments of the present invention, including the systems and methods described herein. In one embodiment, the computer system 200 includes the function of extracting key process parameters such as described in the method 100 of FIG. 1 and the method 300 of FIG. 3 .

The computer system 200 includes a microprocessor 204 , an input device 210 , a storage device 206 , a system memory 208 , a display 214 and a communication device 212 , all of which are interconnected by one or more buses 202 . Storage device 206 may be a floppy disk drive, hard disk drive, CD-ROM, optical device, or any other storage device. In addition, storage device 206 may be capable of receiving a floppy disk drive, hard disk drive, CD-ROM, DVD-ROM, or any other form of computer-readable media containing computer-executable instructions. The communication device 212 is a modem, network card, or any other device that enables the computer system to communicate with other nodes. It is to be understood that any computer system 200 may represent a plurality of interconnected computer systems, such as personal computers, mainframes, personal digital assistants, and telephony devices. Communication device 212 may allow communication between computer system 200 and one or more tools or computer systems used for integrated circuit fabrication and/or testing.

Computer system 200 includes hardware capable of executing machine-readable instructions, and software for performing actions (typically machine-readable instructions) to produce a desired result. Software includes any machine code stored on any memory medium, such as RAM or ROM, as well as machine code stored on other storage devices, such as floppy disk, flash memory, or CD ROM. Software may include, for example, source or object code. Furthermore, software includes any set of instructions capable of being executed in a client machine or server. Any combination of hardware and software includes a computer system. The computer-implemented code includes code for extracting key parameters, including performing a correlation analysis and generating a correlation matrix and/or gene map.

Computer-readable media include passive data storage, such as random access memory, and semi-permanent data storage, such as compact disc read only memory (CD-ROM). An embodiment of the present invention can be implemented in a computer's RAM to transform a standard computer into a completely new concrete calculator. A data structure is a defined organization of data that can implement the invention. For example, a data structure provides an organization of data or an organization of executable code. Data signals can be carried across transmission media, and store and transmit various data structures, and thus can be used to transmit embodiments of the present invention. The microprocessor 204 can perform the correlation analysis described herein.

The display 214 is operable to display, respectively, in a human-readable form, eg, the hierarchical binary tree, correlation matrix and/or gene map of FIGS. 4, 5, and 6. Database 216 may be any standard or proprietary database software known in the art. Unlimited physical location of the database 216, can exist remotely from the server, can be accessed via the Internet or an intranet. The disclosed database 216 includes multiple database embodiments. Database 216 includes manufacturing data including, for example, process data such as equipment data, metrology data; WAT data; and device performance data.

Referring to FIG. 3 , there is shown a method 300 of correlation analysis to extract key process parameters related to selected device parameters. The shown method of correlation analysis can be used to extract one or more critical manufacturing parameters from a plurality of manufacturing parameters; the critical parameters are related to selected parameters. Fabrication parameters include, for example, process parameters such as equipment parameters, metrology parameters, wafer test parameters, final test parameters, or any parameter for which data (eg, fabrication parameter values) may be collected during or after fabrication of integrated circuits.

The correlation analysis of step 104 and/or step 106 of method 100 may be performed using method 300 . Method 300 further illustrates a correlation analysis that utilizes gene map analysis metrics. In the illustrated embodiment, method 300 extracts key process parameters related to device parameters. However, this illustration is for illustrative purposes only, and method 300 may be used to determine correlations between any parameters, including, for example, one or more layers of hierarchical analysis such as described in method 100 of FIG. 1 . For example, in one embodiment, method 300 may be used to extract WAT parameters related to device parameters. In another embodiment, method 300 may be used to extract process parameters related to WAT parameters.

Method 300 begins at step 302, where device parameters are selected. Step 302 is substantially similar to step 102 of the method 100 described above. Method 300 then proceeds to step 304 where process data related to a plurality of process parameters is collected. In one embodiment, sequential process data and corresponding wafer results (eg, device parameter values) are collected for a plurality of wafers. Process data includes values for one or more equipment parameters and/or metrology parameters determined during the manufacturing process of a known wafer; corresponding wafer results include values determined for device parameters and/or WAT parameters for known wafers. In one embodiment, sequential process data and corresponding wafer results are collected for at least 200 wafers.

The method 300 then proceeds to step 306 where the process data or portions of the process data are summarized. In one embodiment, summarizing data includes determining one or more statistical values representative of time-series process data (eg, equipment data). Statistical values include a maximum value, a minimum value, a standard error value and mean value, and/or other possible statistical values. The timing data can be summarized to determine one or more of maximum, minimum, standard error, and mean values for each process step. By way of example, in one embodiment, time-series data of process parameters related to RF power is collected. Timing data may be summarized as one or more of maximum values, minimum values, standard error values, and average values for each process step that the wafer undergoes in RF power-related fabrication.

Method 300 then proceeds to step 308 where the process parameters are grouped. Hierarchical grouping can be used to group parameters. In one embodiment, hierarchical grouping includes determining highly correlated first and second parameters, and grouping the first and second parameters into a single group for further analysis. A group can include any number of process parameters that are highly correlated. In one embodiment, highly correlated parameters have an ^R2 value of at least 0.8. Thus, in one embodiment, a group includes parameters that have an ^R2 value of at least 0.8 relative to each other. In one embodiment, one or more group parameters are grouped together to form a larger group.

In one embodiment, a hierarchical binary tree may be used to group parameters. FIG. 4 provides a hierarchical binary tree 400 . Hierarchical binary tree 400 includes a vertical axis 402 containing relative distances, and a horizontal axis 404 containing groups of specified parameters. A small correlation distance represents a high ^R2 value among the parameters of this group. Cutoff point 406 shows a correlation distance chosen to provide proper correlation between parameters. In one embodiment, clusters with correlation distances below the cutoff point 406 may be further analyzed into clusters, eg, clusters associated with a selected device parameter may be determined.

Method 300 then proceeds to step 310, where the correlation of the clusters to the selected device parameters is determined. Minor groups, key groups (eg, those having a greater correlation with selected device parameters) are also identified. In one embodiment, for one or more groups, group-to-device parameter ^R2 values are calculated by key component transformations plus stepwise regression. In one embodiment, the group parameters are related to device parameters, which have different mechanisms at different manufacturing process steps. In this example, regression and ^R2 calculations are performed separately at each step.

As noted above, in one embodiment, a population of correlations (eg, ^R2 values) with selected device parameters is determined by key component transformation plus stepwise regression. For example, critical component transformations can transform highly correlated process parameters into smaller data sets (eg clusters). For example, the ^R2 value of X1 (parameter) to the selected device parameter is 0.1, the ^R2 value of X2 (parameter) to the selected device parameter is 0.1, and the ^R2 value of X1 and X2 group can be (1) >0.2, (2)=0.2 or (3)<0.2. The outcome of case (1), (2) or (3) depends on the data analyzed. Stepwise regression uses ^R2 satisfying the condition to select process parameters that fall under the specified condition (1). In other words, the method is able to find parameters that, when grouped together, have a greater correlation with device parameters than the sum of their correlations when grouped individually.

3 and 5, method 300 proceeds to step 312, where a correlation matrix for one or more clusters is provided. FIG. 5 shows a dependency matrix 500 . Correlation matrix 500 provides (eg, shows) correlations (eg, ^R2 values) of a group of parameters. Correlation matrix 500 may be provided via display 214 of FIG. 2 described above. Correlation matrix 500 may be provided for groups that are determined to be critical (eg, have sufficient correlation with device parameters). Correlation matrix 500 includes a vertical axis 502 that includes a number of parameters within the clusters that provide the correlation matrix. The horizontal axis 504 includes the same number of parameters (for ease of use, the parameter names are represented by numbers). Therefore, the diagonal line shows the same value, ie ^R2 is equal to 1. The entries of the correlation matrix 500, such as those shown by reference numeral 506, provide the ^R2 values between the parameters. For example, reference numeral 506a shows that the ^R2 value between parameter B and parameter C is 0.9. Correlation matrix 500 can be color coded to easily show relative correlation values. In one embodiment, R values are classified as 1, <0.9, <0.75, <0.5, <0.25, <0.1, <0, <-0.1, <-0.25, <-0.5, <-0.75, <-0.9 , and <-1 categories, one or more categories are displayed in different colors. Correlation matrix 500 includes headings 508 that provide group designations and correlation values (eg, ^R2 values) for selected device parameters.

Referring to FIGS. 3 and 6, the method 300 then proceeds to step 314, where a gene map 600 shown in FIG. 6 is generated. The genetic map 600 shows the relative ranking of the correlations between process parameters or group parameters and selected plant parameters in a plurality of process steps. Gene map 600 may be shown on a display, such as display 214 described above with respect to FIG. 2 . A benefit of the genetic map 600 is that the process parameters most relevant to the selected device parameters are highlighted to the user in a relative ranking. In one embodiment, each device parameter selected in step 302 is displayed on a genetic map.

Gene map 600 includes a plurality of manufacturing process steps 602 labeled 1 , 2 , 3 , 4 , and 5 . Vertical axis 604 represents a number of process parameters. After the correlation of each group parameter with the selected device parameter has been determined, as described above in step 310, each group may be ranked according to the strength of its correlation and assigned a relative number. For example, the group of parameters with the highest correlation to a device parameter (e.g., the highest ^R value) can be assigned a "1," and the group of parameters with the next highest correlation to the device parameter (e.g., the second ^highest R value) can be assigned a value of "1." are designated as "2" etc. Gene map 600 displays parameters for a group at a specified level, as shown by item 606 . Gene map 600 shows rankings of the top seven high clusters, however any number of rankings may exist.

In one embodiment, the user can select one of the items 606, and the parameter name will be displayed. In one embodiment, the user can select the item 606 and all parameters of the related group will be displayed, as shown in block 608 . In one embodiment, the user can select a group to review in more detail. In this way, a correlation matrix is generated, such as the correlation matrix 500 shown in FIG. 5 .

Advantages of the genetic map 600 include an efficient and effective system for users such as process engineers to review process parameters one by one according to the degree to which each process parameter affects selected device parameters in a systematic manner. In other embodiments, a genetic map can be used to show the relative correlation of any number of parameters (eg, manufacturing parameters) to a selected parameter (eg, indicative of device performance).

Thus, method 300 provides a method to perform correlation analysis including gene map analysis. As previously described, method 300 is used to select a device parameter and select one or more process parameters that are highly correlated with the device parameter. However, other embodiments of adjusting the method 300 to determine the relative correlation of any number of parameters to another parameter are possible, examples provided below.

In one embodiment, a correlation analysis similar to method 300 is performed to determine one or more WAT parameters that correlate with selected device parameters. The correlation analysis used to extract key WAT parameters can be used in combination with the correlation analysis used to extract key process parameters. For example, in one embodiment, device parameters are selected. Correlation analysis, such as an analysis involving one or more steps of the genetic map analysis of method 300, can be used to extract key WAT parameters related to selected device parameters. Correlation analysis, such as an analysis including one or more steps of the genetic map analysis of method 300, can be used to extract key process parameters related to the extracted key WAT parameters.

In one embodiment, a group (eg, a key group) determined to be associated with selected device parameters may be validated through the device history. Equipment history records may provide information on when preventive maintenance, repairs, installation of new equipment or parts, cleaning and/or other equipment processes occurred. For example, a time graph of device performance data (eg, device parameters) may show changes in device performance trends attributable to processes encompassed by the device history. For example, after a preventive maintenance process, device performance improves immediately.

Although only some exemplary embodiments of the present invention have been described in detail above, those skilled in the art will readily understand various modifications to the exemplary embodiments without departing from the innovative teachings and advantages of the present invention.

Explanation of main component symbols

100 Method 200 Computer System

202 Bus 204 Microprocessor

206 storage device 208 system memory

210 Input device 212 Communication device

214 Display 216 Database

400 Hierarchical Binary Tree 402 Vertical Axis

404 Horizontal Axis 406 Cutoff Point

500 Correlation Matrix 502 Vertical Axis

504 horizontal axis 506 reference number

508 Title 600 Gene Map

602 Manufacturing Process Steps 604 Vertical Axis

606 items 608 blocks.

Claims (15)

1. A method for extracting process parameters, comprising: selecting a device parameter, wherein the device parameter is related to the performance of an integrated circuit; collecting process data, wherein the process data includes time-series process data, and wherein the process data includes values related to a plurality of process parameters; summarizing the time series process data; and performing a correlation analysis, wherein the correlation analysis identifies a critical process parameter included in the plurality of process parameters, wherein the critical process parameter is related to selected device parameters, and wherein the correlation analysis includes generating a genetic map for selected device parameters, wherein the plurality of process parameters are grouped into a first group and a second group, and wherein the genetic map includes the first group and the second A ranking of groups, wherein the ranking provides a relative correlation to selected device parameters. 2. The method of claim 1, further comprising: collecting wafer test data, wherein the wafer test data includes values related to a plurality of wafer test parameters; and Wherein the correlation analysis includes identifying a critical wafer test parameter of the plurality of wafer test parameters related to the selected device parameter. 3. The method of claim 1 , wherein summarizing the timing process data comprises determining a statistical value from a group of statistical values consisting of a maximum value, a minimum value, a standard deviation value, a mean value, and combinations thereof value. 4. The method of claim 1, wherein said correlation analysis includes grouping said plurality of process parameters to provide at least a third group and a fourth group. 5. The method of claim 4, wherein the coefficient of determination of the third group with respect to the selected plant parameter is greater than the coefficient of determination of any one of the process parameters included in the third group. 6. The method of claim 4, wherein grouping the plurality of process parameters includes determining correlations among the plurality of process parameters. 7. The method of claim 6, wherein said correlation among said plurality of process parameters of said third group comprises a coefficient of determination of at least 0.8. 8. The method of claim 4, further comprising: The correlation of the third population with the selected device parameters is determined using a principal component transformation and a stepwise regression. 9. The method of claim 4, further comprising: A correlation matrix is provided to the third group. 10. The method of claim 1, wherein the genetic map shows the relative correlation of the plurality of process parameters with selected device parameters at a plurality of process steps. 11. The method of claim 1, further comprising: A batch operation is performed to determine a value for the critical process parameter. 12. A semiconductor processing system comprising at least one subsystem to: collect manufacturing data during a manufacturing process, using the collected manufacturing data to determine a correlation between a first manufacturing parameter and a second manufacturing parameter; determining a correlation between a group comprising the first fabrication parameter and the second fabrication parameter and performance of an integrated circuit during the fabrication process; and A genetic map is generated showing a relative correlation of the population to the performance of the integrated circuit. 13. The system of claim 12, wherein the manufacturing data includes fault detection and classification data. 14. The system of claim 12, further comprising a subsystem to: A correlation matrix is generated for the group. 15. The system of claim 12, further comprising a subsystem to: receiving time series data; and Summarizing the received timing data.

Images