Automated process control using parameters determined with approximation and fine diffraction models

A method of controlling a fabrication cluster using a machine learning system, the machine learning system trained developed using an optical metrology model, the optical metrology model comprising a profile model, an approximation diffraction model, and a fine diffraction model, the method comprising:

• (a) developing a metrology model of a structure, the metrology model including a profile model, the profile model having profile parameters;
• (b) optimizing the metrology model, the optimized metrology model including an optimized profile model,
• (c) calculating a simulated approximation diffraction signal based on the approximation diffraction model of the structure;
• (d) generating a set of simulated fine diffraction signals from a set of profile parameters, the simulated fine diffraction signals generated using the optimized profile model of the structure;
• (e) calculating a set of difference diffraction signal by subtracting the simulated approximation diffraction signal from each of the simulated fine diffraction signals of the set of simulated fine diffraction signals and pairing each difference diffraction signal with the corresponding profile parameters;
• (f) training a machine learning system using the pairs of difference diffraction signal and corresponding profile parameters, the machine learning system trained to process the difference diffraction signal as input and generate the profile parameters as output;
• (g) subtracting the simulated approximation diffraction signal from a measured diffraction signal resulting in an adjusted measured diffraction signal;
• (h) using the trained machine learning system, inputting the adjusted measured diffraction signal and generating profile parameters;
• (i) if one or more matching criteria are met, accessing at least one generated profile parameter;
• (j) transmitting at least one profile parameter of the generated corresponding profile parameters to a fabrication cluster wherein the fabrication cluster was used to create the structure, the fabrication cluster having process parameters and equipment settings;
• and
• (k) adjusting one or more process parameters or equipment settings of the fabrication cluster based on the at least one profile parameter.

A method of controlling a fabrication cluster using a machine learning system and an optical metrology model, the optical metrology model comprising a profile model, an approximation diffraction model, and a fine diffraction model, the method comprising:

• generating a set of simulated fine diffraction signals from a set of profile parameters using a fine diffraction model;
• calculating a simulated approximation diffraction signal based on the approximation diffraction model of the structure;
• calculating a set of difference diffraction signals by subtracting the simulated approximation diffraction signal from each fine diffraction signal of the set of simulated fine diffraction signals;
• training a first machine learning system with the set of difference diffraction signals and corresponding profile parameters, the first machine learning system trained to process profile parameters as input and generate a difference diffraction signal as output;
• generating a library of simulated fine diffraction signals and profile parameters using the trained first machine learning system and using ranges and corresponding resolutions of the profile parameters;
• training a second machine learning systems using the generated library, the second machine learning systems trained to process measured diffraction signals as input and generate profile parameters as output;
• determining at least one profile parameter from a measured diffraction signal using the generated library or the trained second machine learning system;
• transmitting the at least one profile parameter to a fabrication cluster wherein the fabrication cluster was used to create the structure, the fabrication cluster having process parameters and equipment settings;
• and
• adjusting one or more process parameters or equipment settings of the fabrication cluster based on the at least one profile parameter.

A system for controlling a fabrication cluster using a machine learning system developed using an optical metrology model, the optical metrology model comprising a profile model, an approximation diffraction model, and a fine diffraction model of a structure having a profile, the profile having profile parameters, the system comprising:

• an optical metrology tool configured to illuminate the structure with an illumination beam and detect the diffraction signal off the structure;
• a processor configured to generate a simulated approximation diffraction signal off the structure based on an approximation diffraction model of the structure and approximation algorithm for diffraction simulation, generate a simulated fine diffraction signal off the structure based on a fine diffraction model of the structure and the profile parameters, and calculate a difference diffraction signal by subtracting the approximation diffraction signal from the simulated fine diffraction signal;
• a machine learning system trained with pairs of difference diffraction signal and corresponding profile parameters and configured to process a difference diffraction signal as input and generate profile parameters as output;
• and
• a fabrication cluster configured to perform a process on the structure, the process controlled by process variables and equipment settings;
• wherein the structure is measured by the optical metrology tool generating a measured diffraction signal, the simulated approximation diffraction signal is subtracted from the measured diffraction signal generating an adjusted measured diffraction signal, the adjusted measured diffraction signal is input into the trained machine learning system, the trained machine learning system generates profile parameters as output, and at least one generated profile parameter is used to modify a process variable or equipment setting of the fabrication cluster.

A system for controlling a fabrication cluster using a machine learning system developed using an optical metrology model, the optical metrology model comprising a profile model, an approximation diffraction model, and a fine diffraction model of a structure having a profile, the profile having profile parameters, the system comprising:

• an optical metrology tool configured to illuminate the structure with an illumination beam and detect the diffraction signal off the structure;
• a processor configured to generate a simulated approximation diffraction signal off the structure based on an approximation diffraction model of the structure and approximation algorithm for diffraction simulation, generate a simulated fine diffraction signal off the structure based on a fine diffraction model of the structure and the profile parameters, and calculate a difference diffraction signal by subtracting the approximation diffraction signal from the simulated fine diffraction signal;
• a first machine learning system trained with pairs of difference diffraction signal and corresponding profile parameters and configured to process profile parameters as input and generate difference diffraction signal as output;
• a fabrication cluster configured to perform a process on the structure, the process controlled by process variables and equipment settings;
• and
• a second machine learning system trained with pairs of profile parameters and associated simulated diffraction signals and configured to process measured diffraction signals as input and generate profile parameters as output;
• wherein the processor creates a set of pairs of difference diffraction signals and corresponding profile parameters, trains the first machine learning system with the set of pairs of difference diffraction signals and corresponding profile parameters to process profile parameters as input and generate difference diffraction signal as output, generates a library of simulated fine diffraction signals and profile parameters, trains the second machine learning systems using the library to process measured diffraction signals as input and generate profile parameters as output;
• and
• wherein the structure is measured by the optical metrology tool generating a measured diffraction signal, the measured diffraction signal is input into the second machine learning system, and the second machine learning system generates at least one profile parameter as output, the at least one profile parameter is transmitted to the fabrication cluster, and the at least one profile parameter is used to modify at least one process variable or at least one equipment setting of the fabrication cluster.