Back to results
US 10,853,534 B2Granted
Fault related FDC feature extraction
AssigneeTaiwan Semiconductor Manufacturing Company, Ltd.
InventorHsin-Chao Mi
Life of this patent
Estimating the term…
Abstract
A system includes at least one tool, a storage device and a processor. The at least one tool performs semiconductor fabrication processes on at least one wafer, in which the at least one tool includes sensors. The storage device stores computer program codes. The processor executes the computer program codes in the storage device for: modeling profiles from the sensors to generate a modeling result; extracting features from the modeling result corresponding to the modeled profiles; based on the extracted features, extracting scores each representing a degree of the at least one wafer being processed by the at least one tool; and based on the extracted scores, displaying a ranking for fault detection of the at least one wafer.
Claims20
A system, comprising:
- at least one tool configured to perform semiconductor fabrication processes on a plurality of wafers, wherein the at least one tool comprises a plurality of sensors;
- a storage device configured to store computer program codes;
- and
- a processor configured to execute the computer program codes in the storage device for:
- identifying whether there is any target tool in the at least one tool that is related to a fault event;
- when there is at least one target tool identified to be related to the fault event:
- modeling profiles from the sensors, to generate a modeling result;
- extracting features from the modeling result corresponding to the modeled profiles;
- based on the extracted features, extracting scores each representing a degree of distinguishing the plurality of wafers being processed by the at least one tool;
- based on the extracted scores, displaying a ranking of the plurality of sensors for fault detection of the plurality of wafers;
- and
- based on the ranking of the plurality of sensors for fault detection of the plurality of wafers, identifying that at least one normal wafer of the plurality of wafers is not defective,
- wherein extracting scores comprises calculating correlations between values of the extracted features corresponding to at least one defective wafer of the plurality of wafers and the at least one normal wafer of the plurality of wafers,
- wherein in response to the at least one normal wafer of the plurality of wafers not being defective, the at least one tool performs semiconductor fabrication processes on the at least one normal wafer of the plurality of wafers.
11 more elements
A non-transitory computer readable medium comprising computer executable instructions which, when executed by a processor, cause a system to perform operations, the operations comprising:
- identifying at least one tool to determine whether there is any target tool that is related to a fault event;
- when there is at least one target tool identified to be related to the fault event:
- modeling profiles from sensors of the at least one tool to generate a modeling result, wherein the at least one tool is configured to perform semiconductor fabrication processes on a plurality of wafers;
- extracting features from the modeling result corresponding to the modeled profiles;
- based on the extracted features, extracting scores each representing a degree of distinguish at least one defective wafer of the plurality of wafers from at least one normal wafer of the plurality of wafers;
- based on the extracted scores, displaying a ranking of the sensors for fault detection of the plurality of wafers;
- based on the ranking of the sensors for fault detection of the plurality of wafers, identifying the at least one normal wafer of the plurality of wafers is not defective;
- and
- based on the at least one normal wafer of the plurality of wafers not being defective, controlling the at least one tool to perform semiconductor fabrication processes on the at least one normal wafer of the plurality of wafers,
- wherein extracting scores comprises calculating correlations between values of the extracted features corresponding to the at least one defective wafer and the at least one normal wafer of the plurality of wafers.
7 more elements
A method comprising:
- sensing parameters by sensors of at least one tool configured to perform semiconductor fabrication processes on a plurality of wafers;
- based on the sensed parameters, generating profiles;
- identifying the at least one tool to determine whether there is any target tool that is related to a fault event;
- when there is at least one target tool identified to be related to the fault event:
- modeling the profiles to generate a modeling result;
- extracting features from the modeling result corresponding to the modeled profiles;
- based on the extracted features, extracting scores each representing a degree of distinguishing at least one defective wafer of the plurality of wafers from at least one normal wafer of the plurality of wafers;
- based on the extracted scores, displaying a ranking of the sensors for fault detection of the plurality of wafers;
- and
- based on the ranking of the sensors for fault detection of the plurality of wafers, identifying that the at least one normal wafer of the plurality of wafers is not defective,
- wherein extracting scores comprises calculating correlations between values of the extracted features corresponding to the at least one defective wafer and the at least one normal wafer of the plurality of wafers, and
- wherein in response to the at least one normal wafer of the plurality of wafers not being defective, the at least one tool performs semiconductor fabrication processes on the at least one normal wafer of the plurality of wafers.
9 more elements
Dependencies are inferred from the claim text.