System, method and computer program product for calibration of metrology tools

A computer program product embodied on a non-transitory computer readable medium, the computer program product including code adapted to be executed by a computer to perform a method comprising:

• receiving one or more DoE (design-of-experiments) wafers for calibrating a metrology tool, the one or more DoE wafers including a plurality of metrology targets and the plurality of metrology targets including at least one of a periodic grating, a FinFet structure, an SRAM device structure, a Flash memory structure, and a DRAM memory structure;
• collecting a set of signals by measuring the one or more DoE wafers utilizing the metrology tool;
• analyzing the set of signals, utilizing analysis logic or an autoencoder, to determine a portion of principal components of the set of signals;
• determining a first transformation that converts the set of signals to the determined portion of principal components;
• converting the set of signals to components, using the first transformation;
• identifying a set of reference signals collected by either:
• a reference metrology tool measuring the one or more DoE wafers, or a simulation performed based on measurement parameters associated with the metrology tool and associated with the one or more DoE wafers;
• analyzing the set of reference signals, utilizing the analysis logic or the autoencoder, to determine a portion of principal reference components of the set of reference signals;
• determining a second transformation that converts the set of reference signals to the determined portion of principal reference components;
• converting the set of reference signals to reference components, using the second transformation;
• training a model, using a machine learning algorithm or a neural network, that takes as input the components and the reference components to map the components to converted components;
• and
• storing the model, first transformation, and second transformation in a memory associated with the metrology tool for use in calibrating the metrology tool.

A system, comprising:

• one or more DoE (design-of-experiments) wafers;
• a metrology tool configured to collect a set of signals by measuring the one or more DoE wafers, the one or more DoE wafers including a plurality of metrology targets and the plurality of metrology targets including at least one of a periodic grating, a FinFet structure, an SRAM device structure, a Flash memory structure, and a DRAM memory structure;
• a memory associated with the metrology tool;
• and
• a processor configured to:
• analyze the set of signals, utilizing analysis logic or an autoencoder, to determine a portion of principal components of the set of signals;
• determine a first transformation that converts the set of signals to the determined portion of principal components,
• convert the set of signals to components, using the first transformation, identify a set of reference signals collected by either:
• a reference metrology tool measuring the one or more DoE wafers, or
• a simulation performed based on measurement parameters associated with the metrology tool and associated with the one or more DoE wafers,
• analyze the set of reference signals, utilizing the analysis logic or the autoencoder, to determine a portion of principal reference components of the set of reference signals,
• determine a second transformation that converts the set of reference signals to the determined portion of principal reference components,
• convert the set of reference signals to reference components, using the second transformation,
• train a model, using a machine learning algorithm or a neural network, that takes as input the components and the reference components to map the components to converted components, and
• store the model, first transformation, and second transformation in the memory for use in calibrating the metrology tool.