editor's blog
Subscribe Now

Yield Correlations Get Continued Focus

Yield enhancement has never been easy, but it just keeps getting harder as process technologies get more complex. Figuring out where you’re losing dice actually takes a lot of number crunching and correlation between widely disparate types of data.

First you’ve got your basic yield information as embodied in a wafer map. But one wafer doesn’t a trend make; it takes lots to develop the statistics to suggest where systematic problems might lie.

Isolating a particular type of failure, you can then do things like figure out what possible causes might be – which requires both information on the failure mode and access to the design – and a further narrowing down based on the physical layout of the design, which requires a picture of the layout.

All under the guidance of a skilled engineer, of course.

Mentor announced what they called “DFM-aware diagnosis-driven yield analysis.” As you might guess, the focus here is on DFM issues. Unlike strict DRC rules that must pass, DFM rules are more “suggestions.” You may well end up with some that didn’t pass. But if you get systematic yield loss, the obvious question becomes, was that because of some of the DFM rules we blew off?

The Tessent DFM-aware analysis looks for correlations between failures and failed DFM rules. If you find some, you can decide whether to make changes so that they pass. On the other hand, there may be none that correlate: you may actually decide that a new DFM rule is required to fix the observed failures. So you can also test with the new DFM rule to see if there’s a correlation between that and the failure before adding the rule.

Synopsys, meanwhile, announced enhanced yield diagnostics and, in particular, tools to improve memory yields through similar kinds of correlation techniques. They show a loop between the design, from which vectors are generated and sent to the tester, and from which results are gathered. Those results are combined with the original design information in their STAR Silicon Debugger, from which maps of failing bits as well as the physical coordinates of those failures can be derived. From there, and engineer can look for actual failure mechanisms.

More info in the Mentor and Synopsys press releases…

Leave a Reply

featured blogs
Nov 19, 2018
At the Linley Fall Processor Conference recently, the keynote on the second day was by Cliff Young of Google titled Codesign in Google TPUs: Inference, Training, Performance, and Scalability . The TPU is Google's Tensor Processing Unit. It is actually a family, since, as...
Nov 16, 2018
If you aren'€™t from the packaging world, then chances are likely you might think '€œpacking'€ refers to objects being placed in to a box instead of packaging integrated circuits (IC'€™s) on a substrate. Packaging has advanced since it began, and as with many areas i...
Nov 14, 2018
  People of a certain age, who mindfully lived through the early microcomputer revolution during the first half of the 1970s, know about Bill Godbout. He was that guy who sent out crudely photocopied parts catalogs for all kinds of electronic components, sold from a Quon...
Nov 13, 2018
Podcast Interview with the authors of The Hitchhikers Guide to PCB Design eBook by Mike Beutow with PCB Chat....