# US 2022/0092242 A1 · Virtual metrology for wafer result prediction

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2022/0092242 A1
- Document ID: US20220092242A1
- Status: Published application
- Kind: A1
- Application number: 17/025,651
- Filed: 2020-09-18
- Published: 2022-03-24
- Earliest claimed: 2020-09-18
- Assignees: Tokyo Electron Limited
- Applicants: Tokyo Electron Limited
- Inventors: Jun SHINAGAWA; Megan WOOLEY; Toshihiro KITAO; Carlos FONSECA
- Web page: https://patentgrep.com/patent/US20220092242A1

## CPC

- G03F 7/705: {Modelling or simulating from physical phenomena up to complete wafer processes or whole workflow in wafer productions}
- G06F 30/33: Design verification, e.g. functional simulation or model checking
- G03F 7/70525: {Controlling normal operating mode, e.g. matching different apparatus, remote control or prediction of failure}
- G03F 7/70616: {Monitoring the printed patterns}
- G06F 2111/10: Numerical modelling
- H10P 74/23: {characterised by multiple measurements, corrections, marking or sorting processes}
- G03F: PHOTOMECHANICAL PRODUCTION OF TEXTURED OR PATTERNED SURFACES, e.g. FOR PRINTING, FOR PROCESSING OF SEMICONDUCTOR DEVICES; MATERIALS THEREFOR; ORIGINALS THEREFOR; APPARATUS SPECIALLY ADAPTED THEREFOR; (phototypographic composing devices B41B; photosensitive materials or processes for photographic purposes G03C; electrophotography, sensitive layers or processes therefor G03G)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- H10P: GENERIC PROCESSES OR APPARATUS FOR THE MANUFACTURE OR TREATMENT OF DEVICES COVERED BY CLASS H10

## Abstract

Aspects of the disclosure provide a method for wafer result prediction. The method includes determining predictor parameters of a semiconductor process using domain knowledge that includes knowledge of the semiconductor process, a processing tool associated with the semiconductor process, a metrology tool, and/or the wafer. The method also includes removing collinearity among the predictor parameters to obtain key predictor parameters, and selecting a subset of the key predictor parameters based on metrology data of the wafer obtained from the metrology tool. The method further includes building a virtual metrology (VM) model on the subset of the key predictor parameters and may include predicting wafer results using the VM model.

## Claims

### Claim 1

A method for wafer result prediction, comprising:
determining predictor parameters of a semiconductor process using domain knowledge including knowledge of the semiconductor process, a processing tool associated with the semiconductor process, a metrology tool, and/or the wafer;
removing collinearity among the predictor parameters to obtain key predictor parameters;
selecting a subset of the key predictor parameters based on metrology data of the wafer obtained from the metrology tool; and
building a virtual metrology (VM) model on the subset of the key predictor parameters.

### Claim 2 (depends on 1)

The method of claim 1, wherein determining predictor parameters comprises:
determining a first subgroup of predictor parameters using domain knowledge, the first subgroup of predictor parameters expected to affect wafer results; and
determining a second subgroup of predictor parameters based on Design of Experiments (DOE) and domain knowledge, the second subgroup of predictor parameters being a subset of the first subgroup of predictor parameters and known to affect wafer results.

### Claim 3 (depends on 2)

The method of claim 2, further comprising:
obtaining a third subgroup of predictor parameters by processing manufacturing data collected from the processing tool, the third subgroup of predictor parameters associated with the semiconductor process; and
processing the third subgroup of predictor parameters to remove error and variance.

### Claim 4 (depends on 3)

The method of claim 3, further comprising:
determining interaction terms of the first, second, and/or third subgroups of predictor parameters using domain knowledge so that the interaction terms are non-linear and are used as a fourth subgroup of predictor parameters.

### Claim 5 (depends on 4)

The method of claim 4, wherein at least one interaction term is a product, a division, and/or another mathematical operation of two or more predictor parameters of the first, second, and third subgroups.

### Claim 6 (depends on 1)

The method of claim 1, wherein selecting a subset of the key predictor parameters comprises:
calculating correlations between the key predictor parameters and the metrology data, the subset of the key predictor parameters meeting a correlation threshold requirement.

### Claim 7 (depends on 1)

The method of claim 1, wherein removing collinearity among the predictor parameters to obtain key predictor parameters comprises:
calculating correlations between a predictor parameter and other predictor parameters;
grouping predictor parameters based on correlations, wherein one or more predictor parameters are grouped together with respective correlations meeting a threshold requirement; and
selecting a representative predictor parameter from each group to form the key predictor parameters.

### Claim 8 (depends on 1)

The method of claim 1, wherein building the VM model on the subset of the key predictor parameters comprises:
performing regression analysis using one or more regression models based on the subset of the key predictor parameters; and
selecting the VM model from the one or more regression models based on wafer result prediction variation and the number of parameters that are used for the corresponding regression model.

### Claim 9 (depends on 1)

The method of claim 1, further comprising:
predicting wafer results using the VM model.

### Claim 10

A method for predicting a wafer characteristic in a plasma process, comprising:
determining predictor parameters of the plasma process using domain knowledge including knowledge of the plasma process, a plasma tool associated with the plasma process, a metrology tool, and/or the wafer;
removing collinearity among the predictor parameters to obtain key predictor parameters;
selecting a subset of the key predictor parameters based on metrology data of the wafer obtained from the metrology tool; and
building a VM model on the subset of the key predictor parameters.

### Claim 11 (depends on 10)

The method of claim 10, wherein determining predictor parameters comprises:
determining a first subgroup of predictor parameters using domain knowledge, the first subgroup of predictor parameters expected to affect the wafer characteristic;
determining a second subgroup of predictor parameters based on DOE and domain knowledge, the second subgroup of predictor parameters being a subset of the first subgroup of predictor parameters and known to affect the wafer characteristic;
collecting manufacturing data from the plasma tool, the manufacturing data including time trace data associated with settings of the plasma tool and/or the plasma process;
obtaining a third subgroup of predictor parameters by processing the manufacturing data, the third subgroup of predictor parameters associated with the plasma process;
processing the third subgroup of predictor parameters to remove error and variance; and
determining interaction terms of the first, second, and third subgroups of predictor parameters using domain knowledge so that the interaction terms are non-linear and are used as a fourth subgroup of predictor parameters.

### Claim 12 (depends on 11)

The method of claim 11, wherein:
knowledge of the plasma process includes physical characteristics and/or chemical characteristics of the plasma process;
knowledge of the plasma tool includes a recipe for the plasma process and/or information of a plasma chamber in the plasma tool; and
knowledge of the wafer includes information of a film and/or a substrate of the wafer.

### Claim 13 (depends on 10)

The method of claim 10, wherein removing collinearity among the predictor parameters to obtain key predictor parameters comprises:
calculating correlations between a predictor parameter and other predictor parameters;
grouping predictor parameters based on correlations, wherein one or more predictor parameters are grouped together with respective correlations meeting a threshold requirement; and
selecting a representative predictor parameter from each group to obtain the key predictor parameters.

### Claim 14 (depends on 10)

The method of claim 10, wherein selecting a subset of the key predictor parameters comprises:
calculating correlations between the key predictor parameters and the metrology data, the subset of the key predictor parameters meeting a correlation threshold requirement.

### Claim 15 (depends on 10)

The method of claim 10, wherein building the VM model on the subset of the key predictor parameters comprises:
performing regression analysis using one or more regression models based on the subset of the key predictor parameters; and
selecting the VM model from the one or more regression models based on wafer result prediction variation and the number of parameters that are used for the corresponding regression model.

### Claim 16 (depends on 10)

The method of claim 10, further comprising predicting a wafer characteristic, wherein:
the wafer characteristic is an etch rate or a critical dimension.

### Claim 17

A method for root cause analysis of an excursion event, comprising:
collecting data in the absence of the excursion event and data in the presence of the excursion event;
analyzing the data using a VM module that is constructed using domain knowledge;
obtaining, from the VM module, a first model with an excursion event and a second model without an excursion event; and
determining a unique term by comparing the first model and the second model, the unique term being in only one of the first model and the second model.

### Claim 18 (depends on 17)

The method of claim 17, wherein the VM module comprises:
determining predictor parameters of a semiconductor process using domain knowledge including knowledge of the semiconductor process, a processing tool associated with the semiconductor process, a metrology tool, and/or the wafer;
removing collinearity among the predictor parameters to obtain key predictor parameters;
selecting a subset of the key predictor parameters based on metrology data of the wafer obtained from the metrology tool; and
building a VM model on the subset of the key predictor parameters.

### Claim 19 (depends on 18)

The method of claim 18, wherein determining predictor parameters comprises:
determining a first subgroup of predictor parameters using domain knowledge, the first subgroup of predictor parameters expected to affect wafer results;
determining a second subgroup of predictor parameters based on DOE and domain knowledge, the second subgroup of predictor parameters being a subset of the first subgroup of predictor parameters and known to affect wafer results;
obtaining a third subgroup of predictor parameters by processing manufacturing data collected from the processing tool, the third subgroup of predictor parameters associated with the semiconductor process;
processing the third subgroup of predictor parameters to remove error and variance; and
determining interaction terms of the first, second and third subgroups of predictor parameters using domain knowledge so that the interaction terms are non-linear and are used as a fourth subgroup of predictor parameters.

### Claim 20 (depends on 19)

The method of claim 19, wherein:
removing collinearity among the predictor parameters to obtain key predictor parameters includes calculating correlations between a predictor parameter and other predictor parameters, grouping predictor parameters based on correlations, wherein one or more predictor parameters are grouped together with respective correlations meeting a threshold requirement, and selecting a representative predictor parameter from each group to form the key predictor parameters;
selecting a subset of the key predictor parameters includes calculating correlations between the key predictor parameters and the metrology data, the subset of the key predictor parameters meeting a correlation threshold requirement; and
building the VM model on the subset of the key predictor parameters includes performing regression analysis using one or more regression models based on the subset of the key predictor parameters, and selecting the VM model from the one or more regression models based on wafer result prediction variation and the number of parameters that are used for the corresponding regression model.
