US-12623316-B2 - Chemical mechanical polishing system for a workpiece, arithmetic system, and method of producing simulation model for chemical mechanical polishing

US12623316B2US 12623316 B2US12623316 B2US 12623316B2US-12623316-B2

Abstract

The present invention relates to a cyber-physical system for optimizing a simulation model for chemical mechanical polishing based on actual measurement data of chemical mechanical polishing. The chemical mechanical polishing system includes a polishing apparatus ( 1 ) for polishing the workpiece (W) and an arithmetic system ( 47 ). The arithmetic system ( 47 ) includes a simulation model including at least a physical model configured to output an estimated polishing physical quantity including an estimated polishing rate of the workpiece (W). The arithmetic system ( 47 ) is configured to: input polishing conditions for the workpiece (W) into the simulation model; output the estimated polishing physical quantity of the workpiece (W) from the simulation model; and determine model parameters of the simulation model that bring the estimated polishing physical quantity closer to a measured polishing physical quantity of the workpiece (W).

Inventors

Norikazu Suzuki
Hozumi Yasuda
Yoshihiro Mochizuki
Yohei Hashimoto

Assignees

EBARA CORPORATION
NATIONAL UNIVERSITY CORPORATION TOKAI NATIONAL HIGHER EDUCATION AND RESEARCH SYSTEM
NATIONAL UNIVERSITY CORPORATION KANAZAWA UNIVERSITY

Dates

Publication Date: 20260512
Application Date: 20210310
Priority Date: 20200324

Claims (4)

1 . A chemical mechanical polishing system comprising: a polishing table configured to support a polishing pad having a polishing surface; a polishing head configured to press a workpiece against the polishing surface; a slurry supply nozzle configured to supply slurry onto the polishing surface; and a computing system and a memory coupled thereto and storing a simulation model, the computing system communicatively coupled to the chemical mechanical polishing system, the computing system configured to: cause the simulation model to instantiate at least a physical model configured to output an estimated polishing physical quantity including an estimated polishing rate of the workpiece, wherein the physical model includes a polishing-rate model configured to calculate the estimated polishing rate of the workpiece, and a polishing-torque model configured to calculate an estimated value of torque generated due to a sliding resistance of the polishing pad, and wherein the polishing-torque model includes a head rotation torque model configured to calculate an estimated value of a polishing-head rotation torque for rotating the polishing head and the workpiece on the polishing pad about an axis of the polishing head, and a pad rotation torque model configured to calculate an estimated value of a polishing-pad rotation torque for rotating the polishing pad about an axis of the polishing pad; input polishing conditions for the workpiece into the simulation model; output the estimated polishing physical quantity of the workpiece from the simulation model; determine model parameters of the simulation model that bring the estimated polishing physical quantity closer to a measured polishing physical quantity of the workpiece; and update at least one polishing control parameter of the chemical mechanical polishing system during polishing based on polishing process parameters and model output from the simulation model.
2 . The chemical mechanical polishing system according to claim 1 , further comprising a dresser configured to dress the polishing surface, wherein the polishing-torque model includes a dresser rotation torque model configured to calculate an estimated value of a dresser rotation torque for rotating the dresser on the polishing pad about an axis of the dresser, and a dresser oscillation torque model configured to calculate an estimated value of the dresser oscillation torque around an oscillation axis required for the dresser to oscillate on the polishing pad.
3 . The chemical mechanical polishing system according to claim 1 , wherein the simulation model further includes a mathematical model representing a deterioration of the polishing pad with a passage of a polishing time.
4 . The chemical mechanical polishing system according to claim 1 , wherein the arithmetic system is configured to: input the polishing conditions to the simulation model having the determined model parameters to calculate an updated estimated polishing physical quantity; and evaluate a difference between the updated estimated polishing physical quantity and the measured polishing physical quantity.

Description

TECHNICAL FIELD The present invention relates to a chemical mechanical polishing system for polishing a surface of a workpieces, such as wafer, substrate, or panel, and particularly relates to a cyber-physical system that optimizes a simulation model for chemical mechanical polishing based on actual measurement data of chemical mechanical polishing. BACKGROUND ART In manufacturing of semiconductor devices, various types of films are formed on a wafer. After the film forming process, the wafer is polished in order to remove unnecessary parts of the film and surface irregularities. Chemical mechanical polishing (CMP) is a typical technique for the wafer polishing. CMP is performed by rubbing the wafer against a polishing surface while supplying slurry onto the polishing surface. A film that forms the surface of the wafer is polished by a combination of a chemical action of the slurry and a mechanical action of abrasive grains contained in the slurry. Simulation techniques for wafer polishing have been developed for the purpose of estimating a film thickness of a wafer and detecting an end point of wafer polishing. Machine learning, such as deep learning, is a typical technique for polishing simulation. For example, a model constituted of a neural network is created by machine learning, and polishing conditions for a wafer are input into the model, so that an estimated value of a polishing result is output from the model. Polishing estimation by such machine learning is expected as a technique capable of obtaining an estimation result closer to actual polishing. CITATION LIST Patent Literature Patent document 1: Japanese laid-open patent publication No. 2012-74574 SUMMARY OF INVENTION Technical Problem However, a large amount of training data (so-called big data) is required for the operations of producing a model by the machine learning. In particular, in order to create a model that can output more accurate polishing result, a larger amount of data is required. As a result, it takes a long time to create the model. Furthermore, since the model itself has a complicated structure, it takes a relatively long time for the model to output the polishing result. In addition, the model constituted of the neural network is a so-called black box, and it is unknown what structure it has (or what weight parameters it has). Therefore, if an actual polishing result and the polishing result output from the model are different, it is impossible to identify a part to be corrected in the model. In order to correct the model, additional training data is required and it takes a long time to correct the model. Therefore, the present invention provides a chemical mechanical polishing system and an arithmetic system capable of accurately and quickly estimate a result of polishing of a workpiece, such as a wafer, by using a physical model. The present invention also relates to a method of producing a simulation model for chemical mechanical polishing. Solution to Problem In an embodiment, there is provided a chemical mechanical polishing system comprising: a polishing table configured to support a polishing pad having a polishing surface; a polishing head configured to press a workpiece against the polishing surface; a slurry supply nozzle configured to supply slurry onto the polishing surface; and an arithmetic system having a memory storing therein a simulation model including at least a physical model configured to output an estimated polishing physical quantity including an estimated polishing rate of the workpiece, the arithmetic system being configured to input polishing conditions for the workpiece into the simulation model; output the estimated polishing physical quantity of the workpiece from the simulation model; and determine model parameters of the simulation model that bring the estimated polishing physical quantity closer to a measured polishing physical quantity of the workpiece. In an embodiment, the physical model includes a polishing-rate model configured to calculate the estimated polishing rate of the workpiece, and a polishing-torque model configured to calculate an estimated value of torque generated due to a sliding resistance of the polishing pad. In an embodiment, the polishing-torque model includes a head rotation torque model configured to calculate an estimated value of a polishing-head rotation torque for rotating the polishing head and the workpiece on the polishing pad about an axis of the polishing head, and a pad rotation torque model configured to calculate an estimated value of a polishing-pad rotation torque for rotating the polishing pad about an axis of the polishing pad. In an embodiment, the chemical mechanical polishing system further comprises a dresser configured to dress the polishing surface, wherein the polishing-torque model includes a dresser rotation torque model configured to calculate an estimated value of a dresser rotation torque for rotating the dresser on the polishing pad abou