Semiconductor run-to-run control system with state and model parameter estimation

A method for controlling a manufacturing process, includes:

• processing using a process input and producing a process output;
• storing the process input in a database;
• storing at least one measurement of the process output in the database associated with each process input;
• iterating over data from the database to estimate one or more coefficients for a model;
• if one or more measurements is missing from the database during model coefficient estimation, replacing the missing measurements for the database based on a prediction from said model;
• updating said model with said coefficient estimates;
• iterating over data from the database to estimate a process state;
• if one or more measurements is missing from the database during process state estimation, replacing the missing measurements for the database based on a prediction from said model;
• and
• updating said model with said process state estimate;
• and
• wherein if actual measurements become available a time period after model coefficient estimation or process state estimation, for said one or more measurements which was missing, replacing the predicted measurements for the previously missing measurements with said actual measurements, storing said actual measurements in said database, and utilizing said actual measurements in subsequent model coefficient estimation and process state estimation.

A method for controlling a manufacturing process, includes:

• processing using a process input and producing a process output;
• storing the process input in a database;
• storing at least one measurement of the process output in the database associated with each process input;
• iterating over data from the database to estimate one or more coefficients for a model, wherein a constrained quadratic optimization is solved at each iteration of said data;
• if one or more measurements is missing from the database during model coefficient estimation, replacing the missing measurements for the database based on a prediction from said model;
• updating said model with said coefficient estimates;
• iterating over data from the database to estimate a process state, wherein a constrained quadratic optimization is solved at each iteration of said data;
• if one or more measurements is missing from the database during process state estimation, replacing the missing measurements for the database based on a prediction from said model;
• and
• updating said model with said process state estimate.

A method for controlling a manufacturing process, includes:

• processing using a process input and producing a process output;
• storing the process input in a database;
• storing at least one measurement of the process output in the database associated with each process input;
• iterating over data from the database to estimate one or more coefficients for a model, wherein a constrained quadratic optimization is solved at each iteration of said data;
• if one or more measurements is missing from the database during model coefficient estimation, replacing the missing measurements for the database based on a prediction from said model;
• updating said model with said coefficient estimates;
• iterating over data from the database to estimate a process state, wherein a constrained quadratic optimization is solved at each iteration of said data;
• if one or more measurements is missing from the database during process state estimation, replacing the missing measurements for the database based on a prediction from said model;
• and
• updating said model with said process state estimate;
• and
• wherein if actual measurements become available a time period after model coefficient estimation or process state estimation, for said one or more measurements which was missing, replacing the predicted measurements for the previously missing measurements with said actual measurements, storing said actual measurements in said database, and utilizing said actual measurements in subsequent model coefficient estimation and process state estimation.

A method for controlling a manufacturing process, includes:

• providing a database with process input data and associated process output data;
• iterating over data from the database to estimate one or more coefficients for a model;
• if one or more measurements is missing from the database during model coefficient estimation, replacing the missing measurements for the database based on a prediction from said model;
• updating said model with said coefficient estimates;
• iterating over data from the database to estimate a process state;
• if one or more measurements is missing from the database during process state estimation, replacing the missing measurements for the database based on a prediction from said model;
• and
• updating said model with said process state estimate.

A method for controlling a manufacturing process, includes:

• providing a database with process input data and associated process output data;
• iterating over data from the database to estimate one or more coefficients for a model;
• if one or more measurements is missing from the database during model coefficient estimation, replacing the missing measurements for the database based on a prediction from said model;
• updating said model with said coefficient estimates;
• iterating over data from the database to estimate a process state;
• if one or more measurements is missing from the database during process state estimation, replacing the missing measurements for the database based on a prediction from said model;
• and
• updating said model with said process state estimate;
• and
• wherein if actual measurements become available a time period after model coefficient estimation or process state estimation, for said one or more measurements which was missing, replacing the predicted measurements for the previously missing measurements with said actual measurements, storing said actual measurements in said database, and utilizing said actual measurements in subsequent model coefficient estimation and process state estimation.