Tokyo Electron Limited

Sanjeev Kaushal

Kenji Sugishima

Sukesh Janubhai Patel

Pradeep Pandey

Wolfgang Polak

Aaron Archer Waterman

Orion Wolfe

Anthony Dip

Yuichiro Morozumi

Raymond Joe

Tetsushi Ozaki

Sairam Sankaranarayanan

Pradip Kumar Roy

Takanori Saito

Seungho Oh

Toshiyuki Makiya

Allen John Leith

Swaroop Ganguly

David Smith

Sundar Gandhi

Robert Filman

Karuki Eickmann

Jessie Burger

David L O'meara

Donthineni Ramesh Kumar Rao

A system and method for learning and/or optimizing processes related to semiconductor manufacturing is provided. A learning component generates a set of candidate process models based on process data associated with one or more fabrication tools. The learning component also selects a particular process model from the set of candidate process models that is associated with lowest error. An optimization component generates a set of candidate solutions associated with the particular process model. The optimization component also selects a particular solution from the set of candidate solutions based on a target output value and an output value associated with the particular solution.

System and method for learning and/or optimizing manufacturing processes

A system and method autonomously determines the impact of respective tool parameters on tool performance in a semiconductor manufacturing system. A parameter impact identification system receives tool parameter and tool performance data for one or more process runs of the semiconductor fabrication system and generates a separate function for each tool parameter characterizing the behavior of a tool performance indicator in terms of a single one of the tool parameters. Each function is then scored according to how well the function predicts the actual behavior of the tool performance indicator, or based on a determined sensitivity of the tool performance indicator to changes in the single tool parameter. The tool parameters are then ranked based on these scores, and a reduced set of critical tool parameters is derived based on the ranking. The tool performance indicator can then be modeled based on this reduced set of tool parameters.

Method and apparatus for autonomous tool parameter impact identification system for semiconductor manufacturing

Systems and techniques to facilitate tool failure analysis associated with fabrication processes are presented. A monitoring component determines a candidate tool failure associated with one or more fabrication tools based on sensor data generated by a set of sensors associated with the one or more fabrication tools. A signature component generates a signature dataset for the candidate tool failure based on data associated with the one or more fabrication tools. A comparison component compares the candidate tool failure to at least one previously determined tool failure based on the signature dataset and at least one other signature dataset associated with the at least one previously determined tool failure.

Tool failure analysis using space-distorted similarity

A system and method for autonomously tracing a cause of particle contamination during semiconductor manufacture is provided. A contamination analysis system analyzes tool process logs together with particle contamination data for multiple process runs to determine a relationship between systematic particle contamination levels and one or more tool parameters. This relationship is used to predict expected contamination levels associated with regular usage of the tool, and to identify which tool parameters have the largest impact on expected levels of particle contamination. The contamination analysis system also identifies process logs showing unexpected deviant particle contamination levels that exceed expected contamination levels, and traces the cause of the deviant particle contamination to particular process log parameter events.

Method and apparatus for autonomous identification of particle contamination due to isolated process events and systematic trends

A system and method autonomously determines the impact of respective tool parameters on tool performance in a semiconductor manufacturing system. A parameter impact identification system receives tool parameter and performance data for one or more process runs of the semiconductor fabrication system and generates a separate function for each tool parameter characterizing the behavior of a tool performance indicator in terms of a single one of the tool parameters. Each function is then scored according to how well the function predicts the behavior of the tool performance indicator, or based on a determined sensitivity of the tool performance indicator to changes in the single tool parameter. The tool parameters are then ranked based on these scores, and a reduced set of critical tool parameters is derived based on the ranking. The tool performance indicator can then be modeled based on this reduced set of tool parameters.

Performance of a manufacturing tool is optimized. Optimization relies on recipe drifting and generation of knowledge that capture relationships among product output metrics and input material measurement(s) and recipe parameters. Optimized recipe parameters are extracted from a basis of learned functions that predict output metrics for a current state of the manufacturing tool and measurements of input material(s). Drifting and learning are related and lead to dynamic optimization of tool performance, which enables optimized output from the manufacturing tool as the operation conditions of the tool changes. Features of recipe drifting and associated learning can be autonomously or externally configured through suitable user interfaces, which also can be drifted to optimize end-user interaction.

Method and apparatus for self-learning and self-improving a semiconductor manufacturing tool

An autonomous biologically based learning tool system and a method that the tool system employs for learning and analysis are provided. The autonomous biologically based learning tool system includes (a) one or more tool systems that perform a set of specific tasks or processes and generate assets and data related to the assets that characterize the various processes and associated tool performance; (b) an interaction manager that receives and formats the data, and (c) an autonomous learning system based on biological principles of learning. The autonomous learning system comprises a memory platform and a processing platform that communicate through a network. Both the memory platform and the processing platform include functional components and memories that can be defined recursively. Knowledge generated and accumulated in the autonomous learning system(s) can be cast into semantic networks that can be employed for learning and driving tool goals based on context.

Autonomous biologically based learning tool

System(s) and method(s) are provided for adjustment and analysis of performance of a tool through integration of tool operational data and spectroscopic data related to the tool. Such integration results in consolidated data that enable, in part, learning at least one relationship amongst selected portions of the consolidated data. Learning is performed autonomously without human intervention. Adjustment of performance of the tool relies at least in part on a learned relationship and includes generation of process recipe parameter(s) that can adjust a manufacturing process in order to produce a satisfactory tool performance in response to implementation of the manufacturing process. A process recipe parameter can be generated by solving an inverse problem based on the learned relationship. Analysis of performance of the tool can include assessment of synthetic performance scenarios, identification of spectroscopic condition(s) that affect performance, and extraction of endpoints based at least on time dependence spectroscopic data.

Tool performance by linking spectroscopic information with tool operational parameters and material measurement information

The subject disclosure relates to automatically learning relationships among a plurality of manufacturing tool parameters as applied to arbitrary semiconductor manufacturing tools and a graphical user interface that is supported, at least in part, by an autonomous learning system. The graphical user interface can create one or more matrixes based on received data and can further generate additional matrices by transforming the one or more matrixes. A series of windows can be output, wherein the series of windows, provide performance analysis that comprises a matching between a focus chamber and a reference chamber. In an aspect, the focus chamber and the reference chamber can be different chambers. In another aspect, the focus chamber and the reference chamber can be the same chamber, which provides analysis of the deterioration in performance of the same chamber over time.

Biologically based chamber matching

Autonomous biologically based learning tool system(s) and method(s) that the tool system(s) employs for learning and analysis of performance degradation and mismatch are provided. The autonomous biologically based learning tool system includes (a) one or more tool systems that perform a set of specific tasks or processes and generate assets and data related to the assets that characterize the various processes and associated tool performance; (b) an interaction manager that receives and formats the data, and (c) an autonomous learning system based on biological principles of learning. Objectively generated knowledge gleaned from synthetic or production data can be utilized to determine a mathematical relationship among a specific output variable and a set of associated influencing variables. The generated relationship facilitates assessment of performance degradation of a set of tools, and performance mismatch among tools therein.

Method and system for detection of tool performance degradation and mismatch

System(s) and method(s) for optimizing performance of a manufacturing tool are provided. Optimization relies on recipe drifting and generation of knowledge that capture relationships among product output metrics and input material measurement(s) and recipe parameters. Optimized recipe parameters are extracted from a basis of learned functions that predict output metrics for a current state of the manufacturing tool and measurements of input material(s). Drifting and learning are related and lead to dynamic optimization of tool performance, which enables optimized output from the manufacturing tool as the operation conditions of the tool changes. Features of recipe drifting and associated learning can be autonomously or externally configured through suitable user interfaces, which also can be drifted to optimize end-user interaction.

Growing community of inventors

San Jose, CA, United States of America

Sanjeev Kaushal