# US 2021/0202204 A1 · System and method for low-noise edge detection and its use for process monitoring and control

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

- Number: US 2021/0202204 A1
- Document ID: US20210202204A1
- Status: Published application
- Kind: A1
- Application number: 17/200,305
- Filed: 2021-03-12
- Published: 2021-07-01
- Earliest claimed: 2017-04-13
- Assignees: Fractilia, LLC
- Applicants: Fractilia, LLC
- Inventors: Chris Mack
- Web page: https://patentgrep.com/patent/US20210202204A1

## CPC

- H01J 37/222: {Image processing arrangements associated with the tube}
- G01Q 30/02: Non-SPM analysing devices, e.g. SEM [Scanning Electron Microscope], spectrometer or optical microscope
- G01Q 30/06: for error compensation
- G06T 5/70: Denoising; Smoothing
- G06T 7/0004: {Industrial image inspection}
- G06T 7/13: Edge detection
- G06T 7/40: Analysis of texture (depth or shape recovery from texture G06T7/529)
- G06T 7/42: using transform domain methods
- G06T 7/49: based on structural texture description, e.g. using primitives or placement rules
- H01J 37/28: with scanning beams {(H01J37/268, H01J37/292, H01J37/2955 take precedence)}
- G06T 2207/10061: from scanning electron microscope
- G06T 2207/30148: Semiconductor; IC; Wafer
- H01J 2237/223: Fourier techniques
- H01J 2237/2814: Measurement of surface topography
- H01J 2237/2817: Pattern inspection
- H01J: ELECTRIC DISCHARGE TUBES OR DISCHARGE LAMPS (spark-gaps H01T; electric arc lamps H05B31/00; particle accelerators H05H)
- G01Q: SCANNING-PROBE TECHNIQUES OR APPARATUS; APPLICATIONS OF SCANNING-PROBE TECHNIQUES, e.g. SCANNING PROBE MICROSCOPY [SPM]
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL

## Abstract

In one embodiment, a method includes generating a model trained to predict a low-probability stochastic defect, calibrating, using unbiased measurement data, the model to a specific lithography process, patterning process, or both to generate a calibrated model, using the calibrated model to predict the low-probability stochastic defect; and modifying, based on the low-probability stochastic defect, a variable, parameter, setting, or some combination of a manufacturing process of a device.

## Claims

### Claim 1

A method, comprising:
generating a model trained to predict a low-probability stochastic defect;
calibrating, using unbiased measurement data, the model to a specific lithography process, patterning process, or both to generate a calibrated model;
using the calibrated model to predict the low-probability stochastic defect; and
modifying, based on the low-probability stochastic defect, a variable, parameter, setting, or some combination of a manufacturing process of a device.

### Claim 2 (depends on 1)

The method of claim 1, wherein:
the stochastic defect comprises a bridge or a break in a pattern of lines and spaces manufactured in a specific way,
the calibrating is performed by measuring a stochastic behavior of a sampling of lines and spaces manufactured in the specific way, and
the stochastic behavior is characterized by a standard deviation of an edge position or linewidth, by a power spectral density of measured deviations, by a local critical dimension uniformity of short line segments as a function of a length of the line segments, by another metric, or some combination thereof.

### Claim 3 (depends on 1)

The method of claim 1, wherein:
the stochastic defect comprises a merged contact hole or pillar feature, a missing contact hole or pillar feature, or both, and
the calibrating is performed by measuring a distribution of contact hole or pillar feature sizes on a sample of having a certain size.

### Claim 4 (depends on 1)

The method of claim 1, further comprising:
generating a yield model trained to predict an amount of yield based on one or more stochastic defects; and
determining, using the yield model, the amount of yield by inputting the stochastic defect into the yield model.

### Claim 5 (depends on 1)

The method of claim 1, wherein the model comprises a probability density function that determines a probability that any given contact hole will have a critical dimension between a critical dimension value (CD) and that value plus an infinitesimal change in the critical dimension (dCD).

### Claim 6 (depends on 5)

The method of claim 5, wherein the probability density function comprises a Gaussian probability distribution model or a skewed probability distribution model or an empirical model approximated over a range of critical dimensions with an equation.

### Claim 7 (depends on 1)

The method of claim 1, wherein:
the stochastic defect comprises a line-end or space-end feature merged with another feature, and
the calibrating is performed by measuring a distribution of distances from the line-end tip to the other feature..

### Claim 8 (depends on 1)

The method of claim 1, wherein the model comprises a lithography model configured to predict an impact of exposure dose on a critical dimension, wherein the impact of exposure dose is combined with a photon shot noise or dose variation model to predict the occurrence of the stochastic defect.

### Claim 9 (depends on 1)

The method of claim 1, wherein the model is a non-Gaussian model.

### Claim 10 (depends on 1)

The method of claim 1, further comprising controlling a lithography tool based on the low-probability stochastic defect.

### Claim 11

A system comprising:
a memory device storing instructions; and
a processing device communicatively coupled to the memory device, wherein the processing device executes the instructions to:
generate a model trained to predict a low-probability stochastic defect;
calibrate, using unbiased measurement data, the model to a specific lithography process, patterning process, or both to generate a calibrated model;
use the calibrated model to predict the low-probability stochastic defect; and
modify, based on the low-probability stochastic defect, a variable, parameter, setting, or some combination of a manufacturing process of a device.

### Claim 12 (depends on 11)

The system of claim 11, wherein:
the stochastic defect comprises a bridge or a break in a pattern of lines and spaces manufactured in a specific way,
the calibrating is performed by measuring a stochastic behavior of a sampling of lines and spaces manufactured in the specific way, and
the stochastic behavior is characterized by a standard deviation of an edge position or linewidth, by a power spectral density of measured deviations, by a local critical dimension uniformity of short line segments as a function of a length of the line segments, by another metric, or some combination thereof.

### Claim 13 (depends on 11)

The system of claim 11, wherein:
the stochastic defect comprises a merged contact hole or pillar feature, a missing contact hole or pillar feature, or both, and
the calibrating is performed by measuring a distribution of contact hole or pillar feature sizes on a sample of having a certain size.

### Claim 14 (depends on 11)

The system of claim 11, wherein the processing device is further configured to:
generate a yield model trained to predict an amount of yield based on one or more stochastic defects; and
determine, using the yield model, the amount of yield by inputting the stochastic defect prediction into the yield model.

### Claim 15 (depends on 11)

The system of claim 11, wherein the model comprises a probability density function that determines a probability that any given contact hole will have a critical dimension between a critical dimension value (CD) and that value plus an infinitesimal change in the critical dimension (dCD).

### Claim 16 (depends on 15)

The system of claim 15, wherein the probability density function comprises a Gaussian probability distribution model or a skewed probability distribution model or an empirical model approximated over a range of critical dimensions with an equation.

### Claim 17 (depends on 11)

The system of claim 11, wherein:
the stochastic defect comprises a line-end or space-end feature merged with another feature, and the calibrating is performed by measuring a distribution of distances from the line-end tip to the other feature.

### Claim 18 (depends on 11)

The system of claim 11, wherein the model comprises a lithography model configured to predict an impact of exposure dose on a critical dimension, wherein the impact of exposure dose is combined with a photon shot noise or dose variation model to predict the occurrence of the stochastic defect.

### Claim 19

A tangible, non-transitory computer-readable medium storing instructions that, when executed, cause a processing device to:
generate a model trained to predict a low-probability stochastic defect;
calibrate, using unbiased measurement data, the model to a specific lithography process, patterning process, or both to generate a calibrated model;
use the calibrated model to predict the low-probability stochastic defect; and
modify, based on the low-probability stochastic defect, a variable, parameter, setting, or some combination of a manufacturing process of a device.

### Claim 20 (depends on 19)

The computer-readable medium of claim 19, wherein:
the stochastic defect comprises a bridge or a break in a pattern of lines and spaces manufactured in a specific way,
the calibrating is performed by measuring a stochastic behavior of a sampling of lines and spaces manufactured in the specific way, and
the stochastic behavior is characterized by a standard deviation of an edge position or linewidth, by a power spectral density of measured deviations, by a local critical dimension uniformity of short line segments as a function of a length of the line segments, by another metric, or some combination thereof.
