You may have heard that it rains in Ireland. Even when the sun shines, drizzle is common, creating photogenic rainbows, which often highlight part of a hill in the distance that you’ve never noticed before. Take a picture and share it instantly on Facebook or Instagram with your friends and followers. One of them in California, points out that what’s highlighted by the rainbow, is a brook and bridge above a valley that you frequented together as children. That’s part of the beauty of social media. Not only connecting people, but informing, reminding and enhancing real-life experiences. I’ll draw a parallel below for applying this type of engagement to Yield Management in an IC manufacturing enterprise.
Email Threads are Old School
Most of the time it’s hard in the daily work of a semiconductor product or test engineer to see the nugget of information in the deluge of data that arrives to their desktop. Tools today can allow you to analyze data, but where do you start? And, if you discern something is up, how do you share with the right expert to get to the core issue? Will you be lucky enough to have them in your email thread? Someone in your company you may not even know will be able to explain about an issue which is evident from the data because they know, for example, that the probe card was changed that week. But the probe card ID was not stored with the datalog so it’s not obvious to you; and the probe technician who would spot this with just one glance at the data, will inevitably not be in your email thread.
Are email threads actually the best idea any more for resolving high volume manufacturing issues? Are they used very much any more by our children for their social networking? No. Perhaps we can learn something from them.
A familiar sounding fail bin comes up one day and is affecting your yield. You have no way of knowing right now if and when it came up before and what the root cause for that bin was. But you can view it in a sophisticated looking chart, which breaks down the bin by site. This chart shows you that the bin is consistent across sites (parallel testing was in place). You know you can get to the heart of the issue if given enough time. You might email a few people then start looking at some old data stored on your desktop.
The pot of gold at the end of the rainbow is hours if not (more likely) days away. And the clock is ticking faster all the time.
You May Not Know Who Can Help
Data and information in Yield Management are only good when they end up with people who can help convert them into positive action before further unnecessary scrap is created. So the key of course is to get the information in the shortest time to those who can and will contribute to resolving the situation.
In large companies there are hundreds, even thousands, of engineers working on multiple projects for potentially tens of thousands of products. Even in smaller companies there are dozens of engineers all with different experience and slants and angles who can provide unique insights into issues. A QA engineer will be able to tell you that an issue you have brought up with a test program is actually more serious than you even thought because the program is not even qualified yet. A fab engineer can tell you that a particular pattern, which you see in a wafer parametric analysis, is because of a reticle issue, which was fixed (or not) subsequently. You may never have these people in a thread because you don’t know them. Imagine, though, if you don’t even need to have them on the thread but they will still get notification anyway? To paraphrase an old adage: “Everyone you meet no matter who they are knows something you don’t know”. Let’s take that seriously.
Fig 1. Who on your email thread will know that the probe card was changed to explain the above test’s step change in performance (x-axis is wafer id)?
“Following” Transforms Manufacturing
To try and solve this data deluge and the “who do I ask?” conundrum for high volume manufacturing in a big company, let’s go back to Social networking, but with a twist. Instead of people being “followed” for their latest updates, equipment and material
could be followed by those responsible in a “socially enhanced” Yield Management System (“se-YMS”).
Fig 2: Will the engineers you know be able to help you explain these issues? Maybe not. But no doubt others will.
Take the example of a test floor supervisor in Shanghai who “follows” the testers under his watch. A colleague unknown to him in California observes that one of the testers on that floor is affecting the yield of their new product, so starts to mention this tester in the se-YMS. This has the effect of alerting the supervisor who straight away takes part in the conversation on-line, even though these two far flung colleagues never met or heard of each other before. So in a matter of minutes the supervisor on the test floor in Shanghai is aware of a potential tester issue and is starting to get to know an internal customer. Two very positive outcomes it has to be said.
What's at the End of the Rainbow?
In the proposed se-YMS, experts follow topics they are expert in, product engineers follow products, and technicians follow equipment. Test engineers follow test programs, reliability engineers follow unusually performing wafers, probe technicians follow probe cards and probers. And the list goes on.
In each case, problems that come up in their area of interest will be routed to them straight away and they will get to work with people they probably never knew.
We all know that the ends of rainbows are elusive, but the journeys will be shorter, more interesting and more colorful when we all work in closer collaboration using a far more effective system, no matter how far apart we are geographically.
About the Author John O’Donnell is CEO of MFG Vision Ltd, the County Limerick, Ireland-based Company sharply focused on the gap in the market for productivity-enhancing analytics of big data generated in semiconductor testing.