# US 2021/0343000 A1 · Automatic selection of algorithmic modules for examination of a specimen

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

- Number: US 2021/0343000 A1
- Document ID: US20210343000A1
- Status: Published application
- Kind: A1
- Application number: 16/866,463
- Filed: 2020-05-04
- Published: 2021-11-04
- Earliest claimed: 2020-05-04
- Assignees: Applied Materials Israel Ltd.
- Applicants: Applied Materials Israel Ltd.
- Inventors: Ran SCHLEYEN; Eyal ZAKKAY; Boaz COHEN
- Web page: https://patentgrep.com/patent/US20210343000A1

## CPC

- G06T 7/0004: {Industrial image inspection}
- G06T 2207/20076: Probabilistic image processing
- G06T 2207/20081: Training; Learning
- G06T 2207/20084: Artificial neural networks [ANN]
- G06T 2207/30148: Semiconductor; IC; Wafer
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL

## Abstract

There is provided a system comprising a processor configured to obtain a set of images of a semiconductor specimen, (1) for an image of the set of images, select at least one algorithmic module MS out of a plurality of algorithmic modules, (2) feed the image to MS to obtain data DMS representative of one or more defects in the image, (3) obtain a supervised feedback regarding rightness of data DMS, (4) repeat (1) to (3) for a next image until a completion criterion is met, wherein an algorithmic module selected at (1) is different for at least two different images of the set of images, generate, based on the supervised feedback, a score for each of a plurality of the algorithmic modules, and use scores to identify one or more algorithmic modules Mbest as the most adapted for providing data representative of one or more defects in the set of images.

## Claims

### Claim 1

A system of examination of a semiconductor specimen, comprising a processor and memory circuitry (PMC) configured to:
obtain a set of images of one or more semiconductor specimens,
obtain a plurality of algorithmic modules, each algorithmic module being configured to provide, based on an image of a semiconductor specimen, data representative of one or more defects in the image,
(1) for an image of the set of images, select at least one algorithmic module MS out of the plurality of algorithmic modules,
(2) feed the image to the at least one algorithmic module MS to obtain data DMS representative of one or more defects in the image,
(3) obtain a supervised feedback regarding rightness of data DMS provided by the algorithmic module MS,
(4) repeat (1) to (3) for a next image of the set of images until a completion criterion is met, wherein an algorithmic module selected at (1) is different for at least two different images of the set of images,
wherein the system is configured to:
generate, based at least on the supervised feedback, a score for each of a plurality of the algorithmic modules, and
use scores associated with the algorithmic modules to identify one or more algorithmic modules Mbest as the most adapted among the plurality of algorithmic modules for providing data representative of one or more defects in the set of images.

### Claim 2 (depends on 1)

The system of claim 1, wherein selecting the at least one algorithmic module MS out of the plurality of algorithmic modules is provided by a selection module trained to select the at least one algorithmic module MS.

### Claim 3 (depends on 2)

The system of claim 2, wherein the score generated for an algorithmic module is representative of a ratio between a number of times a positive feedback has been obtained for this algorithmic module and a number of times this algorithmic module has been selected at (1).

### Claim 4 (depends on 3)

The system of claim 3, wherein the supervised feedback obtained at (3) impacts subsequent selection of the at least one algorithmic module MS at (1) by the selection module.

### Claim 5 (depends on 1)

The system of claim 1, configured to generate at (3) a score for the algorithmic module MS, which is representative of a probability that the algorithmic module MS is adapted to provide data representative of one or more defects in the set of images, wherein the score depends at least on the supervised feedback.

### Claim 6 (depends on 1)

The system of claim 1, configured to select at (1) the algorithmic module MS based at least on a score associated with the algorithmic module MS and generated based at least on a supervised feedback previously obtained at (3).

### Claim 7 (depends on 5)

The system of claim 5, wherein:
if the supervised feedback validates DMS at (3), the score associated with MS is increased with respect to a score previously associated with MS;
if the supervised feedback rejects DMS at (3), the score associated with MS is decreased with respect to a score previously associated with MS; and
the at least one algorithmic module Mbest is associated with a score which is the highest among the plurality of algorithmic modules.

### Claim 8 (depends on 1)

The system of claim 1, configured to generate a score for each algorithmic module based on:
a first probability provided by a selection module trained to select the at least one algorithmic module MS, and
a second probability generated based at least on the supervised feedback.

### Claim 9

A method of examination of a semiconductor specimen by a processor and memory circuitry (PMC), the method comprising:
obtaining a set of images of one or more semiconductor specimens,
obtaining a plurality of algorithmic modules, each algorithmic module being configured to provide, based on an image of a semiconductor specimen, data representative of one or more defects in the image,
(1) for an image of the set of images, selecting at least one algorithmic module MS out of the plurality of algorithmic modules,
(2) feeding the image to the at least one algorithmic module MS to obtain data DMS representative of one or more defects in the image,
(3) obtaining a supervised feedback regarding rightness of data DMS provided by the algorithmic module MS,
(4) repeating (1) to (3) for a next image of the set of images until a completion criterion is met, wherein an algorithmic module selected at (1) is different for at least two different images of the set of images,
generating, based at least on the supervised feedback, a score for each of a plurality of the algorithmic modules,
using scores associated with the algorithmic modules to identify one or more algorithmic modules Mbest as the most adapted among the plurality of algorithmic modules for providing data representative of one or more defects in the set of images.

### Claim 10 (depends on 9)

The method of claim 9, wherein selecting the at least one algorithmic module MS out of the plurality of algorithmic modules is provided by a selection module trained to select the at least one algorithmic module MS.

### Claim 11 (depends on 9)

The method of claim 9, wherein the score generated for an algorithmic module is representative of a ratio between a number of times a positive feedback has been obtained for this algorithmic module and a number of times this algorithmic module has been selected at (1).

### Claim 12 (depends on 11)

The method of claim 11, wherein the supervised feedback obtained at (3) impacts subsequent selection of the at least one algorithmic module MS at (1) by the selection module.

### Claim 13 (depends on 9)

The method of claim 9, comprising generating at (3) a score for the algorithmic module MS, which is representative of a probability that the algorithmic module MS is adapted to provide data representative of one or more defects in the set of images, wherein the score depends at least on the supervised feedback.

### Claim 14 (depends on 9)

The method of claim 9, comprising selecting at (1) the algorithmic module MS based at least on a score associated with the algorithmic module MS and generated based at least on a supervised feedback previously obtained at (3).

### Claim 15 (depends on 13)

The method of claim 13, wherein:
if the supervised feedback validates DMS at (3), the score associated with MS is increased with respect to a score previously associated with MS;
if the supervised feedback rejects DMS at (3), the score associated with MS is decreased with respect to a score previously associated with MS;
the at least one algorithmic module Mbest is associated with a score which is the highest among the plurality of algorithmic modules.

### Claim 16 (depends on 13)

The method of claim 13, including generating a score for each algorithmic module based on:
a first probability provided by a selection module trained to select the at least one algorithmic module MS, and
a second probability generated based at least on the supervised feedback.

### Claim 17

A non-transitory computer readable medium comprising instructions that, when executed by a processor and memory circuitry (PMC), cause the PMC to perform operations comprising:
obtaining a set of images of one or more semiconductor specimens, upon obtaining a plurality of algorithmic modules, wherein each algorithmic module is configured to provide, based on an image of a semiconductor specimen, data representative of one or more defects in the image:
(1) for an image of the set of images, selecting at least one algorithmic module MS out of the plurality of algorithmic modules,
(2) feeding the image to the at least one algorithmic module MS to obtain data DMS representative of one or more defects in the image,
(3) obtaining a supervised feedback regarding rightness of data DMS provided by the algorithmic module MS,
(4) repeating (1) to (3) for a next image of the set of images until a completion criterion is met, wherein an algorithmic module selected at (1) is different for at least two different images of the set of images,
generating, based at least on the supervised feedback, a score for each of a plurality of the algorithmic modules,
using scores associated with the algorithmic modules to identify one or more algorithmic modules Mbest as the most adapted among the plurality of algorithmic modules for providing data representative of one or more defects in the set of images.
