Die Tracking

Track every die from wafer, to package, to the field, and back

Increase yield, reduce failures, understand your processes and designs

You want all your ICs to be identically perfect, like peas in a pod. But you would also like to track them through your manufacturing and distribution processes. ICID allows you to track otherwise identical parts.

Identical, untrackable die		Trackable die with ICID

Die tracking permits you to correlate field returns with test data, observe parametric shifts during manufacturing, and identify and discard the wafer neighbors of bad die after packaging. If you speed grade your parts, you can store packaged parts unmarked, and mark them by speed grade to match your orders.

Correlating field returns with test data

The ordinary industry "standard" defect rate for shipped parts is 100 DPM ( defects per million ). Extraordinary customers demand better. If you can correlate field returns with original test data, you can learn how to strengthen your tests so that parts are less likely to fail in the field. You can even identify tests that are too strict, that never correlate with field returns, and loosen them, increasing yield. Matching your tests to real defects, and ignoring imaginary defects, will increase yield, improve quality, and increase profitability.

Observing parametric shifts

ICs are usually tested at least twice, once on the wafer and once in the package. High reliability ICs are often tested after burnin. A die may pass all three tests, but if there are large parametric shifts between the tests, then the parametric shifts are likely to continue in the field, eventually leading to failure.

With ICID, you can watch the parameters shift for individual parts, and discard parts with excessive shift. You can also relax parametric requirements for stable parts, shipping more usable die. Again, with ICID you can eliminate field failures from your shipped parts, while increasing overall yield.

Identifying die from bad neighborhoods on the wafer

If a die fails in test, its neighbors on the wafer are far more likely to fail early, even if they test good and don't show parametric shifts. However, many tests can only be performed in the package. With ICID, you can locate "bad neighborhoods" on a wafer, even after the die have been separated and packaged, and discard or downgrade the parts from the bad neighborhood of the wafer.

Late marking of graded parts

You want to make the fastest ICs you can, and sell those for a premium. But speed test results can vary, and the parts that you bin probably won't match your orders. You sometimes fill orders for ordinary grade parts with premium parts, because that is what is on hand. But you don't want your customers to cherry pick the best parts, resell them at a better price, then reorder more ordinary parts at a lower price. If you mark packages with speed grades, or write speed information into EPROM on the die, you don't have much choice - the customers can easily figure out which are the fastest parts.

With ICID, you don't have to store speed information in the part, or mark it on the part before warehousing. Instead, you can read the ICID from the part with a simple fixture, look it up in the database, then mark it just before you ship it. You can also determine which parts went to which customers, which can help you provide better service to legitimate customers, or track theft and black market usage.

Knowledge is Power

If you know more about your die, and their behavior over time, you can design better processes and circuits. You can learn how parts evolve towards failure, and design around those failure mechanisms. ICID can be a powerful tool for understanding your design and manufacturing processes, and help drive continuous improvement.

A note about end-user privacy