Methods and evaluation devices for analyzing three-dimensional data sets representing devices

Patent metadata

Field	Value
Publication number	US-12148139-B2
Application number	US-202117536662-A
Country	US
Kind code	B2
Filing date	Nov 29, 2021
Priority date	Nov 29, 2021
Publication date	Nov 19, 2024
Grant date	Nov 19, 2024

Abstract

Methods and evaluation devices for evaluating 3D data of a device under inspection are provided. A first machine learning logic detects target objects, and a second machine learning logic provides a voxel segmentation for the target objects. Based on the segmented voxels, a transformation to feature space is performed to obtain measurement results.

First claim

What is claimed is:

1. A method for evaluating 3D data of a device under inspection, comprising:
detecting target objects in the 3D data using a first machine learning logic,
applying a voxel classification to the detected target objects using a second machine learning logic to provide classified voxels that represent a segmentation of voxels depending on a material of the device the respective voxel represents,
applying a transformation to feature space to the classified voxels, and
obtaining measurement results based on the transformation to feature space.

2. The method of claim 1 , wherein the device is a semiconductor device.

3. The method of claim 2 , wherein the target objects are interconnects between chip dies.

4. The method of claim 1 , wherein the first machine learning logic comprises a hough forest model.

5. The method of claim 1 , wherein the second machine learning logic comprises a 3D random forest segmentation model.

6. The method of claim 1 , wherein the transformation to feature space includes a transformation to linear feature space.

7. The method of claim 1 , wherein the transformation to feature space comprises providing one or more functions describing a dependency of a first dimensional variable to a second dimensional variable, or derivatives thereof.

8. The method of claim 7 , wherein the first dimensional variable includes an area or a diameter.

9. The method of claim 7 , wherein the second dimensional variable includes a position variable.

10. The method of claim 7 , wherein obtaining measurements includes identifying deviations of the functions from nominal functions.

11. The method of claim 7 , wherein the one or more functions are user configurable.

12. The method of claim 11 , furthermore comprising predicting a desired user configuration.

13. An evaluation device for evaluating 3D data of a device under inspection, comprising one or more processors and a non-transitory machine-readable medium storing instructions that when executed by the one or more processors cause the one or more processors to:
detect target objects in the 3D data using a first machine learning logic,
apply a voxel classification to the detected target objects using a second machine learning logic to provide classified voxels that represent a segmentation of voxels depending on material of the device the respective voxel represents,
apply a transformation to feature space to the classified voxels, and
obtain measurement results based on the transformation to feature space.

14. The evaluation device of claim 13 , wherein the evaluation device is configured to perform a method for evaluating the 3D data of the device under inspection, the method comprising:
detecting the target objects in the 3D data using the first machine learning logic,
applying the voxel classification to the detected target objects using the second machine learning logic to provide the classified voxels that represent the segmentation of voxels depending on the material of the device the respective voxel represents,
applying the transformation to feature space to the classified voxels, and
obtaining the measurement results based on the transformation to feature space.

15. The evaluation device of claim 13 , wherein the first machine learning logic comprises a hough forest model.

16. The evaluation device of claim 13 , wherein the second machine learning logic comprises a 3D random forest segmentation model.

17. A method for training the evaluation device of claim 13 , comprising:
training the first machine learning logic based on training data with annotated target objects, and
training the second machine learning logic with training data including annotated voxels.

18. A system, comprising:
a measurement device configured to obtain 3D data of a device under test, and
the evaluation device of claim 13 .

19. A tangible non-transitory storage medium storing a computer program including a program code which, when executed on one or more processors, causes execution of the method of claim 1 .

20. The tangible non-transitory storage medium of claim 19 , wherein the device is a semiconductor device.

Assignees

• Zeiss Carl Smt Gmbh

Inventors

• Pichumani Ramani
• Korb Thomas
• Klochkov Dmitry
• Neumann Jens Timo

Classifications

• G06T2207/30148

  Semiconductor; IC; Wafer · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06N20/20

  Ensemble learning · CPC title

• G06T7/11

  Region-based segmentation · CPC title

• G06T2207/10061

  from scanning electron microscope · CPC title