Fabrication fingerprint for proactive yield management
Systems and methods for improving wafer fabrication. Wafers may be inspected at various points in the fabrication process to generate inspection data. The inspection data and wafer-in-progress data may be used to identify defect patterns and tools and/or processes that cause wafer defects. The inspection data may be stacked to form virtual wafer maps that amplify signals to detect patterns more easily. Defect patterns and tools and/or processes may also be identified through machine learning models receiving artificial defect visualizations as input.
1 . A method for identifying a tool or process causing wafer defects, the method comprising:
generating first inspection data by inspecting, by a first inspection device, a plurality of wafers after each of the plurality of wafers has been processed by a first fabrication tool performing a first fabrication process on each of the plurality of wafers and not by a second fabrication tool;
generating second inspection data by inspecting, by a second inspection device, the plurality of wafers after each of the plurality of wafers has been processed by the second fabrication tool performing a second fabrication process on each of the plurality of wafers;
generating a plurality of first wafer maps corresponding to the plurality of wafers based on the first inspection data;
generating a plurality of second wafer maps corresponding to the plurality of wafers based on the second inspection data;
stacking the plurality of first wafer maps together to provide a first stacked virtual wafer map;
stacking the plurality of second wafer maps together to provide a second stacked virtual wafer map;
performing spatial pattern recognition (SPR) operations on the first stacked virtual wafer map and the second stacked virtual wafer map; and
based on an output of the SPR operations, identifying at least one of the tool or the process that caused a defect of at least one of the plurality of wafers.
2 . The method of claim 1 , wherein one of the plurality of first wafer maps or the plurality of second wafer maps represents locations of defects of one of the plurality of wafers identified by the first inspection device or the second inspection device.
3 . The method of claim 2 , wherein each of the plurality of first wafer maps and the plurality of second wafer maps is represented as an image.
4 . The method of claim 2 , wherein each of the plurality of first wafer maps and the plurality of second wafer maps is represented as an artificially created visualization.
5 . The method of claim 4 , wherein the artificially created visualization includes stacked wafer probe data or stacked metrology data.
6 . The method of claim 1 , further comprising:
providing an image or an artificially created visualization of one of the plurality of first wafer maps or the plurality of second wafer maps as input into a trained machine learning model;
processing the image or the artificially created visualization by the trained machine learning model to generate another output; and
based on the another output from the trained machine learning model, generating an indication of a type of the defect.
7 . The method of claim 1 , further comprising:
generating recognized fingerprint patterns based on an analysis of the plurality of first wafer maps and first wafer-in-progress (WIP) data for first wafers processed in a first fabrication facility, wherein the first WIP data includes process data and tool data for the first wafers corresponding to the plurality of first wafer maps; and
storing the recognized fingerprint patterns into a fingerprint library.
8 . The method of claim 7 , wherein performing the SPR operations includes comparing the first stacked virtual wafer map to the recognized fingerprint patterns in the fingerprint library.
9 . The method of claim 7 , further comprising:
generating other recognized fingerprint patterns based on an analysis of other wafer maps and second WIP data for second wafers processed in a second fabrication facility; and
storing the other recognized fingerprint patterns in the fingerprint library.
10 . The method of claim 7 , wherein the fingerprint library is stored in a cloud-based server accessible by a first computing device in the first fabrication facility and a second computing device in a second fabrication facility.
11 . The method of claim 1 , wherein the SPR operations include using a fingerprint library generated from wafer-in-progress (WIP) data.
12 . The method of claim 1 ,
wherein a first of the plurality of wafers and a second of the plurality of wafers are from different wafer lots, respectively.
13 . The method of claim 1 , wherein the first inspection device is the same as the second inspection device.
14 . A method for identifying a tool or process causing wafer defects, the method comprising:
generating first inspection data by inspecting a wafer after the wafer has been processed according to a first fabrication process and not according to a second fabrication process;
generating second inspection data by inspecting the wafer after the wafer has been processed according to the second fabrication process;
generating a first wafer map based on the first inspection data;
generating a second wafer map based on the second inspection data;
performing operations on the first wafer map and the second wafer map; and
based on an output of the operations, identifying at least one of the tool or the process that caused a defect of the wafer.
15 . The method of claim 14 , wherein the first inspection data and the second inspection data are generated, respectively, by different inspection devices.
16 . The method of claim 14 , wherein the operations include spatial pattern recognition operations.
17 . The method of claim 14 , wherein the first fabrication process and the second fabrication process are performed by different fabrication tools.
18 . The method of claim 14 , wherein the first fabrication process and the second fabrication process are different fabrication processes.
19 . The method of claim 14 , wherein each of the first wafer map and the second wafer map is represented as an image.
20 . The method of claim 14 , wherein each of the first wafer map and the second wafer map is represented as an artificially created visualization.
21 . The method of claim 20 , wherein each artificially created visualization includes stacked wafer probe data or stacked metrology data.
22 . The method of claim 14 , further comprising:
providing an image or an artificially created visualization of the first wafer map or the second wafer map as input into a trained machine learning model;
processing the image or the artificially created visualization by the trained machine learning model to generate another output; and
based on the another output from the trained machine learning model, generating an indication of a type of the defect.
23 . The method of claim 14 , further comprising:
generating recognized fingerprint patterns based on an analysis of wafer maps and wafer-in-progress (WIP) data for wafers, wherein the WIP data includes process data and tool data for the wafers corresponding to the wafer maps; and
storing the recognized fingerprint patterns into a fingerprint library.
24 . The method of claim 23 , wherein performing the operations includes comparing the first wafer map and the second wafer map to the recognized fingerprint patterns in the fingerprint library.
25 . A system for identifying a tool or process causing wafer defects, the system comprising:
at least one processor; and
non-transitory computer-readable storage storing instructions which, when executed by the at least one processor, cause the system to:
generate first inspection data by inspecting a wafer after the wafer has been processed according to a first fabrication process and not according to a second fabrication process;
generate second inspection data by inspecting the wafer after the wafer has been processed according to the second fabrication process;
generate a first wafer map based on the first inspection data;
generate a second wafer map based on the second inspection data;
perform operations on the first wafer map and the second wafer map; and based on an output of the operations, identify at least one of the tool or the process that caused a defect of the wafer.
26 . The system of claim 25 , wherein the operations include spatial pattern recognition operations.
27 . The system of claim 25 , wherein each of the first wafer map and the second wafer map is represented as an image.
28 . The system of claim 25 , wherein each of the first wafer map and the second wafer map is represented as an artificially created visualization.
29 . The system of claim 28 , wherein each artificially created visualization includes stacked wafer probe data or stacked metrology data.
30 . The system of claim 25 , wherein the non-transitory computer-readable storage stores further instructions which, when executed by the at least one processor, cause the system to:
generate recognized fingerprint patterns based on an analysis of wafer maps and wafer-in-progress (WIP) data for wafers, wherein the WIP data includes process data and tool data for the wafers corresponding to the wafer maps; and
store the recognized fingerprint patterns into a fingerprint library,
wherein performing the operations includes comparing the first wafer map and the second wafer map to the recognized fingerprint patterns in the fingerprint library.