Systems and methods for adaptive troubleshooting of semiconductor manufacturing equipment
Inventors
Gadre, Milind Jayram • Kumar, Prashanth
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12429846-B2
Publication Date
2025-09-30
Expiration Date
Abstract
A system includes a processing device, operatively coupled to the memory device, to perform operations comprising obtaining a plurality of sensor values associated with a deposition process performed, according to a recipe, in a process chamber to deposit film on a surface of a substrate. A machine-learning model is applied to the plurality of sensor values. The machine-learning model is trained based on historical sensor data of a sub-system of the process chamber and task data associated with the recipe for depositing the film. An output of the machine-learning model is generated that is indicative of a suspected failure of the sub-system and a corrective action is generated based on the suspected failure of the sub-system.
Core Innovation
The invention relates to adaptive troubleshooting for semiconductor manufacturing equipment by obtaining a plurality of sensor values associated with a deposition process performed, according to a recipe, in a process chamber to deposit film on a surface of a substrate. It applies a machine-learning model to the plurality of sensor values, where the machine-learning model is trained based on historical sensor data of a sub-system of the process chamber and task data associated with the recipe for depositing the film. The machine-learning model generates an output indicative of a suspected failure of the sub-system.
The invention further generates a ticket for the suspected failure and stores the ticket in a data structure accessible via a user interface presenting a manufacturing data graph. The approach uses output that can be indicative of a difference between expected behavior of the sub-system and actual behavior of the sub-system, and can compare the output to a predetermined threshold to determine whether the sub-system is experiencing a failure.
In addition, the invention describes training and fault data handling using historical sensor data mapped to task data associated with the recipe, including extracting failure data using natural language processing. It can calibrate the machine-learning model based on failure data to support expected behavior versus actual behavior outputs and fault-pattern comparison.
Claims Coverage
Independent claims cover three implementations of the same core workflow: a method, a system, and a non-transitory computer-readable storage medium. Each implementation includes five inventive features that obtain sensor values during a deposition process, apply a machine-learning model trained on historical sub-system sensor data and recipe task data, produce an output indicative of a suspected sub-system failure, create a ticket, and store that ticket for access via a user interface presenting a manufacturing data graph.
Deposition sensor value acquisition for recipe-based film deposition
Obtaining, by a processor, a plurality of sensor values associated with a deposition process performed, according to a recipe, in a process chamber to deposit film on a surface of a substrate.
Machine-learning model trained on historical sub-system sensor data and recipe task data
Applying a machine-learning model to the plurality of sensor values, the machine-learning model trained based on historical sensor data of a sub-system of the process chamber and task data associated with the recipe for depositing the film.
Suspected failure output from the machine-learning model
Generating an output of the machine-learning model, wherein the output is indicative of a suspected failure of the sub-system.
Failure ticket generation and storage for manufacturing data graph access
Generating a ticket for the suspected failure; and storing, in a data structure, the ticket to be accessible via a user interface presenting a manufacturing data graph.
System implementation with memory and processing device
A system comprising a memory and a processing device, operatively coupled to the memory, to perform the operations of obtaining sensor values, applying a machine-learning model, generating an output, generating a ticket, and storing the ticket for access via a user interface presenting a manufacturing data graph.
Non-transitory computer-readable storage medium instructions for ticketed suspected failure detection
A non-transitory computer-readable storage medium comprising instructions that, when executed by a processing device, performs the operations of obtaining sensor values, applying a machine-learning model, generating an output, generating a ticket, and storing the ticket for access via a user interface presenting a manufacturing data graph.
Across the independent claim set, the core claim coverage centers on machine-learning-based suspected failure detection for a deposition process chamber sub-system using sensor values mapped to recipe task data from historical records, followed by ticket generation and storage for user-interface presentation via a manufacturing data graph.
Stated Advantages
Reduced fault diagnosis/downtime.
Improved energy consumption.
Documented Applications
Adaptive troubleshooting for semiconductor process chamber equipment using sensor values and a machine-learning model to detect suspected failures and present manufacturing data graphs with tickets.
User-facing manufacturing data graph access via generated tickets for suspected failures.
Fault-pattern comparison to identify a type of suspected failure related to a process chamber sub-system.
Corrective action generation based on suspected failure of a process chamber sub-system [procedural detail omitted for safety].
Interested in licensing this patent?