Classifications

• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F11/00—Error detection; Error correction; Monitoring
  • G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
  • G06F11/0703—Error or fault processing not based on redundancy, i.e. by taking additional measures to deal with the error or fault not making use of redundancy in operation, in hardware, or in data representation
  • G06F11/0706—Error or fault processing not based on redundancy, i.e. by taking additional measures to deal with the error or fault not making use of redundancy in operation, in hardware, or in data representation the processing taking place on a specific hardware platform or in a specific software environment
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F11/00—Error detection; Error correction; Monitoring
  • G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
  • G06F11/0703—Error or fault processing not based on redundancy, i.e. by taking additional measures to deal with the error or fault not making use of redundancy in operation, in hardware, or in data representation
  • G06F11/079—Root cause analysis, i.e. error or fault diagnosis
• • G—PHYSICS
  • G05—CONTROLLING; REGULATING
  • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
  • G05B23/00—Testing or monitoring of control systems or parts thereof
  • G05B23/02—Electric testing or monitoring
  • G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
  • G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
  • G05B23/0224—Process history based detection method, e.g. whereby history implies the availability of large amounts of data
  • G05B23/024—Quantitative history assessment, e.g. mathematical relationships between available data; Functions therefor; Principal component analysis [PCA]; Partial least square [PLS]; Statistical classifiers, e.g. Bayesian networks, linear regression or correlation analysis; Neural networks
• • G—PHYSICS
  • G05—CONTROLLING; REGULATING
  • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
  • G05B23/00—Testing or monitoring of control systems or parts thereof
  • G05B23/02—Electric testing or monitoring
  • G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
  • G05B23/0259—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterized by the response to fault detection
  • G05B23/0275—Fault isolation and identification, e.g. classify fault; estimate cause or root of failure
  • G05B23/0278—Qualitative, e.g. if-then rules; Fuzzy logic; Lookup tables; Symptomatic search; FMEA
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F11/00—Error detection; Error correction; Monitoring
  • G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
  • G06F11/0703—Error or fault processing not based on redundancy, i.e. by taking additional measures to deal with the error or fault not making use of redundancy in operation, in hardware, or in data representation
  • G06F11/0751—Error or fault detection not based on redundancy
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F11/00—Error detection; Error correction; Monitoring
  • G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
  • G06F11/0703—Error or fault processing not based on redundancy, i.e. by taking additional measures to deal with the error or fault not making use of redundancy in operation, in hardware, or in data representation
  • G06F11/0766—Error or fault reporting or storing
  • G06F11/0778—Dumping, i.e. gathering error/state information after a fault for later diagnosis
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F11/00—Error detection; Error correction; Monitoring
  • G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
  • G06F11/0703—Error or fault processing not based on redundancy, i.e. by taking additional measures to deal with the error or fault not making use of redundancy in operation, in hardware, or in data representation
  • G06F11/0793—Remedial or corrective actions
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F11/00—Error detection; Error correction; Monitoring
  • G06F11/30—Monitoring
  • G06F11/34—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
  • G06F11/3452—Performance evaluation by statistical analysis
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F11/00—Error detection; Error correction; Monitoring
  • G06F11/30—Monitoring
  • G06F11/34—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
  • G06F11/3466—Performance evaluation by tracing or monitoring
  • G06F11/348—Circuit details, i.e. tracer hardware
• • G—PHYSICS
  • G01—MEASURING; TESTING
  • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
  • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
  • G01R31/28—Testing of electronic circuits, e.g. by signal tracer
  • G01R31/282—Testing of electronic circuits specially adapted for particular applications not provided for elsewhere
  • G01R31/2831—Testing of materials or semi-finished products, e.g. semiconductor wafers or substrates

Definitions

• This application relates to the use of process trace analysis for detection and classification of semiconductor equipment faults, and more particularly, to machine-based methods for predicting an equipment fail mode.
• FDC fault detection and classification
• the detection of equipment faults by monitoring time- series traces of equipment sensors is a long -recognized but very difficult problem in semiconductor manufacturing.
• an FDC method starts with breaking a complex trace into logical “windows” and then computing statistics (frequently called indicators or key numbers) on the trace data in the windows.
• the indicators can be monitored using statistical process control (“SPC”) techniques to identify anomalies, based primarily on engineering knowledge, and the indicators can be utilized as inputs for predictive models and root cause analysis.
• SPC statistical process control
• the quality of the indicators determines the value of all subsequent analysis. High quality indicators require high quality windows.
• the analysis of the indicators for anomaly detection is still primarily univariate in nature, with anomalies considered on a feature by feature basis, and is generally insufficient to identify equipment fail modes related to the detected anomaly.
• FIG. 1 is a process display interface illustrating a graphical collection of equipment sensor traces, with a first set of windows defined for a first type of anomalous traces.
• FIG. 2 is the process display interface of FIG. 1, with a second set of windows defined for a second type of anomalous traces in a different region of the sensor traces.
• FIG. 3 is a table with summary results for the traces shown in FIGS. 1 and 2.
• FIG. 4 is a flow chart illustrating one embodiment of a process for predicting an equipment fail mode.
• FIG. 5 is a flow chart illustrating another embodiment of a process for predicting an equipment fail mode.
• the term “sensor trace” refers to time-series data measuring an important physical quantity periodically during operation of a piece of semiconductor processing equipment, e.g., the sampled values of a physical sensor at each time point. The sampling rate can vary and the time period between samples is not always the same.
• trace or “equipment trace” refers to a collection of sensor traces for all the important sensors identified for a particular processing instance.
• step refers to a distinct device processing period, e.g., one of the steps in a process recipe.
• the model detects and identifies a current anomaly in trace data, calculates key features associated with the current anomaly, and searches for anomalies having those key features in a database of past trace data. If the same or similar anomalies are found in the past trace data, a likelihood can be determined as to whether or not the current anomaly can be accurately classified in accordance those past anomalies; e.g., the current anomaly is most like a prior anomaly in the past trace data. If so, then the type of anomaly, its root cause, and action steps to correct can likely be retrieved from the database of past trace data. If not, however, then the model returns an error, meaning it has not seen that anomaly before. The anomaly and its features will nevertheless be stored for future reference; and the database updated if a root cause and corrective actions are thereafter determined.
• an exemplary graph 100 of trace data is illustrated representing approximately two hundred individual traces, i.e., time-series values obtained from individual sensors taken during distinct steps of a semiconductor fabrication process run for producing semiconductor wafers. Sensor values are plotted on the y-axis and time in seconds is measured on the x-axis. It should be recognized that while process steps normally start at a specific point in time, the length of a process step may be variable. [0013] For most of the period between approximately 40 - 90 seconds, normal process operation is expected to yield trace data that is gradually falling off and is therefore relatively stable and consistent.
• a first set of traces 112 in the top grouping of traces 110 and a second set of traces 122 in the bottom set of traces 120 both show sensor readings that suddenly spike up in value, then down, then back up, and then settle back into the gradual falling off pattern.
• This trace behavior is unexpected and indicates some kind of problem with the process.
• windows 115 and 125 are defined over these Type I anomaly regions in the top group 110 and the bottom group 120, respectively, of the graph 100 where the unexpected Type I anomalies occur for some number of wafers.
• FIG. 2 shows the same graph 100 of trace data, but with the focus on a different portion of the trace data where a Type II anomaly is occurring for a third set of traces 132 only in the top group 110. In this case, some of the traces appear to fall off early, then straighten out, then fall off again, then straighten out again, before falling all the way to nominal as expected. Thus, for analyzing this type of anomalous behavior, window 135 is defined over the first falling off region in the top group of traces 110 and window 136 is defined over the second falling off region in the top group.
• a machine learning model is configured to detect anomalies using known methods including use of the data from window analysis. For example, a combination of wafer attributes and trace location features may be provided as inputs to a simple multi-class machine learning model, such as a gradient-boosting model, that is trained on datasets to detect anomalous behavior in the trace data.
• a simple multi-class machine learning model such as a gradient-boosting model
• indicators are calculated from the traces in each of the windows.
• the indicators are then stored as features associated with the window and the instance of trace data on those wafers, along with selected wafer attributes and the anomaly location in the trace.
• Feature engineering and selection can be performed to narrow a set of features to those key features determined to be most important to detecting and identifying the particular anomaly with the detection model.
• the predicted classifications from the detection model are summarized in the table 200 of FIG. 3, where five wafers have the anomalies identified as the Type I anomaly; seven wafers have the anomalies identified as the Type II anomaly; and one wafer has both the Type I and Type II anomalies.
• the small number of detected anomalies it is important to identify and characterize anomalies, particularly for use in training a predictive model to monitor trace data so as to minimize instances of process instability that may lead to defective wafers.
• the model is also configured for (i) searching a database of prior trace data for the same or similar anomaly, and (ii) either identifying one or more prior anomalies as most like the current anomaly, or indicating there is nothing like the current anomaly in the database.
• the model makes a determination of the likelihood that the current trace anomaly is most like one or more similar or same anomalies observed in past traces. If the likelihood exceeds a threshold, then the anomaly is classified, and prior knowledge regarding the root cause and corrective action is retrieved from the database.
• step 302 trace data is received into a predictive model and processed.
• step 304 at least one anomaly is detected in the received trace data and its location in the trace identified.
• a window is then defined in step 306 to contain the portion of the traces that include the anomaly, and features of the anomalous trace are calculated, including statistics, in step 308 and stored in step 310. Searches are then conducted in a database having past trace data in step 312 for anomalies having the same features associated with the current anomaly.
• a likelihood can be determined in step 316 as to whether or not the current anomaly can be accurately classified in accordance those past anomalies. If so, then in step 318, the type of anomaly, its root cause, and action steps to correct can be retrieved from database for the same or similar the past occurrences, and appropriate corrective action taken in step 320. If not likely, however, then the model returns an error, meaning it has not seen that anomaly before. The anomaly and its features will nevertheless be stored for future reference; and the database updated if a root cause and corrective actions are thereafter determined.
• FIG.5 presents a more generalized approach in process 400.
• trace data is received into a predictive model and processed.
• an anomalous pattern is detected in the trace data.
• features of the detected anomalous pattern are computed and in step 408 compared to features of prior anomalous patterns stored in a database of past trace data.
• step 410 if a features match is determined, then in step 412, information regarding the anomalous pattern from past trace data is retrieved from the database, including one or more root causes for the anomaly as well as corrective actions for the root causes.
• appropriate corrective action is taken.
• processor-based models for trace analysis can be desktop- based, i.e., standalone, or part of a networked system; but given the heavy loads of information to be processed and displayed with some interactivity, processor capabilities (CPU, RAM, etc.) should be current state-of-the-art to maximize effectiveness.
• the Exensio® analytics platform is a useful choice for building interactive GUI templates.
• coding of the processing routines may be done using Spotfire® analytics software version 7.11 or above, which is compatible with Python object-oriented programming language, used primarily for coding machine language models.

Landscapes

• Engineering & Computer Science (AREA)
• Theoretical Computer Science (AREA)
• Physics & Mathematics (AREA)
• General Engineering & Computer Science (AREA)
• General Physics & Mathematics (AREA)
• Quality & Reliability (AREA)
• Biomedical Technology (AREA)
• Health & Medical Sciences (AREA)
• Computer Hardware Design (AREA)
• Mathematical Physics (AREA)
• Automation & Control Theory (AREA)
• Life Sciences & Earth Sciences (AREA)
• Probability & Statistics with Applications (AREA)
• Bioinformatics & Cheminformatics (AREA)
• Bioinformatics & Computational Biology (AREA)
• Evolutionary Biology (AREA)
• Fuzzy Systems (AREA)
• Artificial Intelligence (AREA)
• Evolutionary Computation (AREA)
• Testing And Monitoring For Control Systems (AREA)
• Debugging And Monitoring (AREA)
• Testing Or Measuring Of Semiconductors Or The Like (AREA)
• Computational Linguistics (AREA)
• Data Mining & Analysis (AREA)
• Databases & Information Systems (AREA)

Priority Applications (4)

Applications Claiming Priority (2)

Publications (1)

Family

ID=79688227

Family Applications (1)

Country Status (7)

Families Citing this family (9)

Citations (4)

Family Cites Families (17)

• 2021
  • 2021-07-22 EP EP21845272.0A patent/EP4162395A4/en active Pending
  • 2021-07-22 JP JP2023504511A patent/JP7703011B2/ja active Active
  • 2021-07-22 CN CN202180058614.3A patent/CN116113942A/zh active Pending
  • 2021-07-22 US US17/383,334 patent/US11640328B2/en active Active
  • 2021-07-22 KR KR1020237005035A patent/KR20230042041A/ko active Pending
  • 2021-07-22 WO PCT/US2021/042842 patent/WO2022020642A1/en not_active Ceased
  • 2021-07-23 TW TW110127131A patent/TWI887455B/zh active

Patent Citations (4)

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events