Methods, systems, and computer programs are presented for predicting the performance of semiconductor manufacturing equipment operations. One method includes an operation for obtaining machine-learning (ML) models, each model related to predicting a performance metric for an operation of a semiconductor manufacturing tool. Further, each ML model utilizes features defining inputs for the ML model. The method further includes an operation for receiving a process definition for manufacturing a product with the semiconductor manufacturing tool. One or more ML models are utilized to estimate a performance of the process definition used in the semiconductor manufacturing tool. Additionally, the method includes presenting, on a display, results showing the estimate of the performance of the manufacturing of the product. In some aspects, the use of hybrid models improves the predictive accuracy of the system by augmenting the capabilities of data-driven models with the reinforcement provided by the physics-based models.

An orientation chamber is provided. The orientation chamber includes a substrate holder, an orientation detector, and a purging system. The substrate holder is configured to hold a substrate. The orientation detector is configured to detect an orientation of the substrate. The purging system is configured to inject a cleaning gas into the orientation chamber and remove contaminants from the substrate. The purging system includes a gas regulator adjusting a volume of the cleaning gas supplied into the orientation chamber according to a detection signal output from a gas detector which indicates a content of a specific gas contaminant outgassed from the substrate.

A method of detecting and classifying anomalies during semiconductor processing includes executing a wafer recipe a semiconductor processing system to process a semiconductor wafer; monitoring sensor outputs from a sensors that monitor conditions associated with the semiconductor processing system; providing the sensor outputs to models trained to identify when the conditions associated with the semiconductor processing system indicate a fault in the semiconductor wafer; receiving an indication of a fault from at least one of the models; and generating a fault output in response to receiving the indication of the fault.

A method to simulate operations of a manufacture plant comprising a plurality of machines, the method including receiving a capacity function and an elapsed time function associated with a first machine of the plurality of machines, wherein the capacity function and elapsed time function is defined by one or more parameters characterizing the first machine, receiving a record of historical production data associated with the first machine, calculating, based on the capacity function and the record of historical production data, an augmented capacity function and an augmented elapsed time function that is defined by the one or more parameters and a quantity relating to parts waiting for processing (WIP), and simulating the operations of the plant based on the augmented capacity function and augmented elapsed time function.

To facilitate evaluation of a predicted process shape in process recipe development using machine learning, a process recipe search apparatus that searches for an etching recipe that is a parameter of a plasma processing apparatus set so as to etch a sample into a desired shape displays, on a display device, the predicted process shape of the sample by a candidate etching recipe predicted by using a machine leaning model, by highlighting a difference between the predicted process shape and a target shape.

An object capturing device includes light emission, receiving, and scanning units, and distance calculation, and object determination units. The scanning unit measures light from the emission unit to head toward a measurement target space to perform scanning, and to guide reflected light from the object with respect to the measurement light to the receiving unit. The distance calculation unit calculates a distance to the object in association with a scanning angle of the scanning unit. The object determination unit determines whether the object is a capture target based on whether a scanning angle range within which a difference between distances is equal to or less than a predetermined threshold value corresponding to a reference scanning angle range of the capture target, and a determination of whether intensity distribution of the reflected light within the scanning angle range corresponds to reference intensity distribution of the reflected light from the capture target.

Disclosed are methods of advanced process control (APC) for particular processes. A particular process (e.g., a photolithography or etch process) is performed on a wafer to create a pattern of features. A parameter is measured on a target feature and the value of the parameter is used for APC. However, instead of performing APC based directly on the actual parameter value, APC is performed based on an adjusted parameter value. Specifically, an offset amount (which is previously determined based on an average of a distribution of parameter values across all of the features) is applied to the actual parameter value to acquire an adjusted parameter value, which better represents the majority of features in the pattern. Performing this APC method minimizes dimension variations from pattern to pattern each time the same pattern is generated on another region of the same wafer or on a different wafer using the particular process.

A substrate processing system comprises one or more transfer chambers; a plurality of process chambers connected to the one or more transfer chambers; and a computing device connected to each of the plurality of process chambers. The computing device is to receive first measurements generated by sensors of a first process chamber during or after a process is performed within the first process chamber; determine that the first process chamber is due for maintenance based on processing the first measurements using a first trained machine learning model; after maintenance has been performed on the first process chamber, receive second measurements generated by the sensors during or after a seasoning process is performed within the first process chamber; and determine that the first process chamber is ready to be brought back into service based on processing the second measurements using a second trained machine learning model.

A system for monitoring one or more conditions of an automation system of a semiconductor factory includes one or more instrumented substrates, one or more sealable containers and one or more system servers. The one or more instrumented substrates include one or more sensors. The one or more sensors measure one or more conditions of the one or more instrumented substrates as the one or more sealable containers transport the one or more instrumented substrates through the semiconductor factory. The one or more sealable containers also receive sensor data from the one or more sensors included on the one or more instrumented substrates. The one or more system servers are configured to receive the sensor data from the one or more sealable containers. The one or more servers are configured to identify one or more deviations in the measured one or more conditions.

Implementations described herein generally relate to a method for detecting anomalies in time-series traces received from sensors of manufacturing tools. A server feeds a set of training time-series traces to a neural network configured to derive a model of the training time-series traces that minimizes reconstruction error of the training time-series traces. The server extracts a set of input time-series traces from one or more sensors associated with one or more manufacturing tools configured to produce a silicon substrate. The server feeds the set of input time-series traces to the trained neural network to produce a set of output time series traces reconstructed based on the model. The server calculates a mean square error between a first input time series trace of the set of input time series traces and a corresponding first output time series trace of the set of output time-series traces. The server declares the sensor corresponding to the first input time-series trace as having an anomaly when the mean square error exceeds a pre-determined value.