# US 2021/0342991 A1 · Ml-enabled assured microelectronics manufacturing: a technique to mitigate hardware trojan detection

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

- Number: US 2021/0342991 A1
- Document ID: US20210342991A1
- Status: Published application
- Kind: A1
- Application number: 17/244,183
- Filed: 2021-04-29
- Published: 2021-11-04
- Earliest claimed: 2020-04-29
- Assignees: University of Southern California
- Applicants: University of Southern California
- Inventors: Ajey Poovannummoottil JACOB; John DAMOULAKIS; Akhilesh JAISWAL; Devanand Krishna SHENOY; Andrew RITTENBACH
- Web page: https://patentgrep.com/patent/US20210342991A1

## CPC

- G06T 3/4053: based on super-resolution, i.e. the output image resolution being higher than the sensor resolution
- G06T 7/001: {using an image reference approach}
- G06T 3/4038: Image mosaicing, e.g. composing plane images from plane sub-images
- G06T 3/4046: using neural networks
- G06T 7/30: Determination of transform parameters for the alignment of images, i.e. image registration
- G06T 12/00: Tomographic reconstruction from projections
- G06T 12/30: Image post-processing, e.g. metal artefact correction
- H01J 37/244: Detectors; Associated components or circuits therefor
- H01J 37/28: with scanning beams {(H01J37/268, H01J37/292, H01J37/2955 take precedence)}
- G06T 2207/10061: from scanning electron microscope
- G06T 2207/10072: Tomographic images
- G06T 2207/20084: Artificial neural networks [ANN]
- G06T 2207/30148: Semiconductor; IC; Wafer
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- H01J: ELECTRIC DISCHARGE TUBES OR DISCHARGE LAMPS (spark-gaps H01T; electric arc lamps H05B31/00; particle accelerators H05H)

## Abstract

A method for assuring that integrated circuits are free of malicious circuit insertions and/or IC design modifications through mask swapping/addition is provided. The method includes a step of comparing 3D tomographic images constructed from design GDS to the 3D tomographic images constructed from in-line fab metrology data.

## Claims

### Claim 1

A method for assuring that an integrated circuit is free of malicious circuit insertions and/or IC design modifications, the method comprising:
constructing a first set of one or more tomographic images constructed from design GDS for an integrated circuit layer or layers;
constructing a second set of one or more tomographic images from in-line fab metrology data for the integrated circuit layer or layers; and
comparing corresponding images from first set of one or more tomographic images and the second set of one or more tomographic images to detect malicious circuit insertions and/or IC design modifications to the integrated circuit, corresponding imaging being images of the same integrated circuit region, wherein the presence of detect malicious circuit insertions and/or IC design are indicated by one or more differences between the corresponding images.

### Claim 2 (depends on 1)

The method of claim 1 wherein the in-line fab metrology data is acquired during the time of manufacturing the integrated circuit.

### Claim 3 (depends on 1)

The method of claim 1 wherein mask swapping and/or mask addition is created by malicious circuit insertions and/or IC design modifications.

### Claim 4 (depends on 1)

The method of claim 1 wherein the first set of one or more tomographic images and the second set one or more tomographic images are 2D images.

### Claim 5 (depends on 1)

The method of claim 1 wherein the first set of one or more tomographic images and the second set one or more tomographic images are 3D images.

### Claim 6 (depends on 1)

The method of claim 1 wherein in-line fab metrology data include data generated by microscopy imaging tools selected from the group consisting of CDSEM tools, CD Uniformity tools, and e-beam imaging tools.

### Claim 7 (depends on 1)

The method of claim 1 wherein in-line fab metrology data is critical dimension scanning electron microscopy (CDSEM) SEM data.

### Claim 8 (depends on 1)

The method of claim 1, wherein during each step of IC fabrication flow, several in-line metrology measurements are taken layer-by-layer and used to create 2D tomographic images of each IC layer.

### Claim 9 (depends on 8)

The method of claim 8, wherein a 3D composite tomographic image of an entire IC is formed from the 2D tomographic images.

### Claim 10 (depends on 9)

The method of claim 9, wherein an OPC-modified GDS file of the entire IC are constructed from the metrology measurements.

### Claim 11 (depends on 9)

The method of claim 9, wherein an OPC-modified GDS file of a specific layer in an IC is constructed from the metrology measurements.

### Claim 12 (depends on 9)

The method of claim 9, wherein a wafer is accepted if the 3D composite tomographic image matches the 3D tomographic image constructed from in-line fab metrology data.

### Claim 13 (depends on 9)

The method of claim 9, wherein the composite 3D tomographic image enhances visualization for all IC layers and their routing.

### Claim 14 (depends on 9)

The method of claim 9, wherein the composite 3D tomographic image facilitates detailed examination of a suspicious area-of-interest.

### Claim 15 (depends on 1)

The method of claim 1 wherein machine learning algorithms are applied to register tomographic 3D images.

### Claim 16 (depends on 1)

The method of claim 1 wherein the resolution for images in the first set of one or more tomographic images and/or images in the second set of one or more tomographic images are upscaled.

### Claim 17 (depends on 1)

The method of claim 1 wherein the resolution for images in the first set of one or more tomographic images and/or images in the second set of one or more tomographic images are upscaled by a trained neural network.

### Claim 18 (depends on 1)

The method of claim 1 wherein missing regions in images in the first set of one or more tomographic images and/or images in the second set of one or more tomographic images are filled in by a trained neural network.

### Claim 19

A method for assuring that an integrated circuit is free of malicious circuit insertions and/or IC design modifications, the method comprising:
constructing a first set of one or more tomographic images constructed from design GDS for an integrated circuit layer or layers;
constructing a second set of one or more tomographic images from SEM data for the integrated circuit layer or layers; and
comparing corresponding images from first set of one or more tomographic images and the second set of one or more tomographic images to detect malicious circuit insertions and/or IC design modifications to the integrated circuit, corresponding imaging being images of the same integrated circuit region, wherein the presence of detect malicious circuit insertions and/or IC design are indicated by one or more differences between the corresponding images.

### Claim 20 (depends on 19)

The method of claim 19 wherein the SEM data is CDSEM data.

### Claim 21 (depends on 19)

The method of claim 19 wherein the first set of one or more tomographic images and the second set one or more tomographic images are 3D images.

### Claim 22 (depends on 19)

The method of claim 19, wherein during each step of IC fabrication flow, several in-line metrology measurements are taken layer-by-layer and used to create 2D tomographic images of each IC layer.

### Claim 23 (depends on 22)

The method of claim 22, wherein a 3D composite tomographic image of an entire IC is formed from the 2D tomographic images.

### Claim 24 (depends on 22)

The method of claim 22 wherein the resolution for images in the first set of one or more tomographic images and/or images in the second set of one or more tomographic images are upscaled.

### Claim 25 (depends on 24)

The method of claim 24 wherein the resolution for images in the first set second set of one or more tomographic images and/or images in the second set of one or more tomographic images are upscaled by a trained neural network.

### Claim 26 (depends on 22)

The method of claim 22 wherein missing regions in images in the first set second set of one or more tomographic images and/or images in the second set of one or more tomographic images are filled in by a trained neural network.
