D2DB-PM Intro | Anchor Semiconductor

Die-to-Database Pattern Monitor

High resolution images provide increasingly valuable insight into the nature of yield killing events, so why are fabs wasting them?

SEM Tool - Vendor A

~1000 images per hour
Assume 50% utilization
18 hours a day
10 tools
9000 images per tool each day
90,000 images total each day

SEM Tool - Vendor B

~2000 images per hour
Assume 50% utilization
18 hours a day
3 tools
18,000 images per tool each day
54,000 images total each day

144,000 High Resolution Images Every Day!

But their potential is wasted…

My Image

A picture paints a thousand words, but the rich content of images is reduced to a single class code.

My Image

SEM Non-Visual images discarded. It is not unusual for 50% or more of the SEM images to be SNV.

Every 10 days…

1,440,000 (1.4 million) images are captured.
Valuable SEM and fab cycle time is taken.
Over 700,000 images are thrown away due to SNV.
The other 700,000 images are reduced to single class codes.
My Image

Is there a better way?


Is there a way to extract and retain more information from SEM images?
Is there a way to utilize SNV images as well, for a zero-waste policy?
Is there a way to use these images to continuously monitor patterning quality and process drift in production?

Introducing…
Die-to-Database Pattern Monitor (D2DB-PM)


Every image is analyzed in detail and compared to its reference design.
All critical and consequential features are identified, measured, and tracked.
My Image
Any significant deviation from the intended pattern is identified and highlighted automatically.
SEM Non-Visual (SNV) images are fully utilized. Nothing is thrown away.

Patterning quality and process drift are continuously monitored.

How it works…


My Image

Every image is thoroughly analyzed and decomposed into multiple sub-patterns.

My Image

Sub-patterns and measurements taken on all instances of the patterns are stored and tracked in a relational database.

My Image

Automatic excursion reports are produced, and comprehensive GUI-based analysis use cases are provided.

Some ways of visualizing results…


Summary Table, Pattern Strength and Weakness Box Plot, Contour Gallery, Trend of Selected Patterns.

D2DB Result Visualization