Granted Patent B2

US 12,644,696 · App. 17/589,704 · Granted Jun 2, 2026

Inventors: Franz Zach (Los Gatos, CA); Mark D. Smith (San Jose, CA); Roel Gronheid (Leuven, BE)

Assignee: KLA Corporation

G01B11/2441G01B9/02G01N21/94G01N21/9501G01N21/956

Abstract

Claims (36)

1 . A wafer shape metrology system comprising:

a wafer shape metrology sub-system configured to perform one or more stress-free shape measurements on a bonded pair of wafers, wherein the bonded pair of wafers are bonded with a bonding tool; and

a controller communicatively coupled to the wafer shape metrology sub-system, the controller including one or more processors configured to execute a set of program instructions stored in a memory, the set of program instructions configured to cause the one or more processors to:

receive the one or more stress-free shape measurements of the bonded pair of wafers;

convert the one or more stress-free shape measurements of the bonded pair of wafers into one or more overlay distortion patterns for the bonded pair of wafers, wherein the one or more stress-free shape measurements are input into a machine learning algorithm to convert the one or more stress-free shape measurements into the one or more overlay distortion patterns;

detect one or more localized deviations in the bonded pair of wafers to identify one or more contaminant particles on the bonding tool; and

report the one or more localized deviations in the bonded pair of wafers.

2 . The system of claim 1 , wherein the one or more stress-free shape measurements are performed on the bonded pair of wafers using interferometric techniques.

3 . The system of claim 1 , wherein the one or more overlay distortion patterns includes variations exceeding 5 mm in length.

4 . The system of claim 1 , wherein the one or more stress-free shape measurements comprise at least one of local shape curvature (LSC) or in-plane distortion (IPD).

5 . The system of claim 1 , wherein the overlay distortion pattern is represented by at least one of a heat map or a generated vector distribution.

6 . The system of claim 5 , wherein a residual vector distribution is calculated by taking the delta between a model vector distribution and the generated vector distribution.

7 . The system of claim 6 , wherein the detecting one or more localized deviations in the bonded pair of wafers comprises a threshold algorithm, wherein the threshold algorithm operates on a length of the residual vector.

8 . The system of claim 7 , wherein the one or more localized deviations are reported when the length of the residual vector exceeds a threshold value.

9 . The system of claim 1 , wherein the one or more processors are further configured to determine a position of the one or more localized deviations, wherein the position of the one or more localized deviations are used to determine a position of the contaminant particle on the bonding tool via a position correlating algorithm.

10 . A system comprising:

a controller configured to receive one or more stress-free shape measurements from a wafer shape metrology sub-system, wherein the controller includes one or more processors configured to execute a set of program instructions stored in a non-transitory memory, wherein the set of program instructions are configured to cause the one or more processors to:

direct a wafer shape metrology sub-system to perform the one or more stress-free shape measurements;

receive the one or more stress-free shape measurements of a bonded pair of wafers;

convert the one or more stress-free shape measurements of the bonded pair of wafers into one or more overlay distortion patterns for the bonded pair of wafers, wherein the one or more stress-free shape measurements are input into a machine learning algorithm to convert the one or more stress-free shape measurements into the one or more overlay distortion patterns;

detect one or more localized deviations in the one or more overlay distortion patterns for the bonded pair of wafers to identify one or more contaminant particles on a bonding tool; and

report the one or more localized deviations in the bonded pair of wafers and instruct the bonding tool to shutdown.

11 . The system of claim 10 , wherein the one or more stress-free shape measurements are performed on the bonded pair of wafers using interferometric techniques.

12 . The system of claim 10 , wherein a machine learning algorithm is used to remove medium-to-long range overlay variations generating overlay residuals.

13 . The system of claim 10 , wherein the one or more stress-free shape measurements comprise at least one of local shape curvature (LSC) or in-plane distortion (IPD).

14 . The system of claim 10 , wherein the overlay distortion pattern is represented by at least one of a heat map or a generated vector distribution.

15 . The system of claim 14 , wherein a residual vector distribution is calculated by taking the delta between a model vector distribution and the generated vector distribution.

16 . The system of claim 15 , wherein the detecting one or more localized deviations in the bonded pair of wafers comprises a threshold algorithm, wherein the threshold algorithm operates on a length of the residual vector.

17 . The system of claim 16 , wherein the one or more localized deviations are reported when the length of the residual vector exceeds a threshold value.

18 . The system of claim 17 , wherein the one or more processors are further configured to determine a position of the one or more localized deviations, wherein the position of the one or more localized deviations are used to determine a position of the contaminant particle on the bonding tool via a position correlating algorithm.

19 . A method comprising:

directing a wafer shape metrology sub-system to perform one or more stress-free shape measurements;

receiving the one or more stress-free shape measurements for a bonded pair of wafers;

converting the one or more stress-free shape measurements of the bonded pair of wafers into one or more overlay distortion patterns for the bonded pair of wafers, wherein the one or more stress-free shape measurements are input into a machine learning algorithm to convert the one or more stress-free shape measurements into the one or more overlay distortion patterns;

detecting one or more localized deviations in the one or more overlay distortion patterns for the bonded pair of wafers to identify one or more contaminant particles on a bonding tool; and

reporting the one or more localized deviations in the bonded pair of wafers and instructing the bonding tool to shutdown.

Assignments (1)

Continuity (2)

References Cited (284)