Via Reliability and Robustness in Today’s Environment
I had the pleasure of talking with my old friend Bob Neves, who visited the I-Connect007 booth at IPC APEX EXPO 2020. Bob had some startling and exciting updates on the testing of microvias and through-hole technology.
Happy Holden: Bob, let’s start with the biggest news: a new coupon you’ve developed.
Bob Neves: For the past couple of years, a working group with IPC has been busy developing a new test method to try to quickly test vias for conformance requirements that would sit in some of the IPC documents like the 6012 or the flexible performance documents. We decided that it’s a good idea to test electrical performance during reflow simulation. So, we came up with a methodology that exposes IPC “D” coupons to multiple reflow simulations while making electrical measurements during the simulations and trying to electrically detect vias and microvias that fail at the high-temperature portion of the reflow cycle and might electrically heal themselves again when they cool off.
I updated the HATS technology that we originally started in the early 2000s to be able to do this new IPC-TM-650 2.6.27B test. Along with that, we have been doing reliability testing at Microtek Labs in China for the automotive business in microvias and in single-via testing. In 2009, a ZVEI working group decided that testing daisy chains of vias only showed them the end of the via failure and not the beginning of it. They wanted to understand the beginning of a via’s failure, so they came up with a test for the automotive industry, where you test one via at a time. If the via starts to fail, you see the electrical effect of that right away.
That came into being in 2011, and we put that capability into Microtek Labs, and we currently have 14 dual chambers in China doing single via testing for the automotive industry. When I made this update to HATS2 technology, I decided I wanted to also add that capability to try to speed up that process of testing single vias, whether they’re through/buried vias or microvias. Now, we’re able to rapidly test single vias in the HATS2 tester for via reliability and via robustness.
Holden: There are some exciting variants on the coupon.
Neves: We came up with a coupon that would test seven single vias on a test coupon because people still want to get some a sense of multiple vias, and the HATS tester will test 252 of these single vias in a single run. You get some quantity of data from your vias. That’s the one thing you lose when you take away daisy chains is the multiple vias and more opportunities for failure.
Holden: But you don’t have to take away the daisy chain.
Neves: Correct. The other feature we added is that the HATS2 single-via coupon allows you to replace any of these seven single-via nets with a daisy chain, so you could have one or two daisy chains as well as five or six single vias on the same coupon. Let’s say that for a complex via structure, we monitor a microvia or stack and then a buried via and the same microvia structure on the other side of the coupon.
You could make a daisy chain of the entire via structure, a single hole with that via structure where you measure all the way through, and then all of your other single vias could be manufactured with that structure, but you only measure one part of the structure—like the microvia structure on the top side of the coupon. Number two, you would measure the microvia portion on the bottom side of the coupon; number three via, you would measure the buried via; and number four, you would do the entire via structure.
You could look individually at the parts of your complex via structure from a complete via manufacturing process. You can get all aspects of multiple vias, a single via, and then each section of your via structure all from one coupon.
Holden: This could also be an IPC standard “D” coupon.
Neves: Correct. The machine also will test standard IPC “D” coupons, which are daisy chains now. You can also work those into single holes, and the way the IPC “D” coupon is set up, there is that opportunity between net one and net two to create a single via or even another daisy chain net. The HATS machine will allow people to add a third net to their IPC “D” coupon where, historically, there have only been two nets up to this point.
Holden: This is exciting because now we have a way of doing both kinds on the same coupon in the same test.
Neves: Yes. You could put your two daisy chains on your IPC “D” coupon and then put one single hole in the middle where you could measure a single via structure all the way through with one hole and get an understanding of where that via might start to fail.
Holden: And I bet you’ll have come up with ways that we can put five single vias or something like that in the “D” coupon.
Neves: There are a lot of possibilities. I think we can come up with a way to mix both the new HATS technology with the 20-pin connector and the traditional IPC eight-pin connector on the other side.
Holden: And then where are these coupons tested at?
Neves: Right now, I’m placing HATS2 Systems into certified A2LA, 17025, or ITAR laboratories around the world. We have NTS in North America, Microtek Labs China in Asia, and we’re working on a partner right now in Europe to be able to put HATS2 Systems there. We’ve had some interest in Europe from the European Space Agency and done some preliminary work with them in the microvia roundtable to try and help them understand their via structures.
Holden: I work with the IPC special counsel or committee on the weak microvia interface problem, and Bob Willis and his testing have been a significant provider of the performance information.
Neves: One of the problems with testing daisy chains for looking at weak microvias is by the time you see a resistance change in a daisy chain, one or more of your microvias is significantly damaged, and they’ll probably 80–90% damaged. Small changes in one via or two vias are going to fall into the noise of a daisy chain from all the resistance, which is mostly from all the circuits interconnecting the daisy chain.
By the time you get to failure analysis, your microvia is blown apart, you have this big wide chasm in most cases, whereas with a single via, you could stop it before catastrophic failure. If it reached 10% resistance change in your single via, you could stop the test, and you would be able to find the beginnings where the failure is starting, where 10% resistance change would equate to a 10% crack or separation in your microvia.
Holden: This will be great because as we focus on 5G and new materials, and the new materials are going to have to go through this kind of reliability component testing.
Neves: Exactly. One of the things we have, and it’s a little bit confusing to people, is that we clump everything into the word “via reliability,” and via reliability is one path to go through that relates to reliability in the environment that your product is in, but we don’t want to wait that long.
We moved toward what I like to call “via robustness,” where we take the coupons and test them above the Tg of the material, take them to solder temperatures multiple times, and run them as hard and as fast as we can until they break to try and see differences between these new materials and differences between processes. People tend to confuse the results from via robustness with via reliability.
Holden: Since we’re on the topic of reliability, why don’t you give us an idea of the European Space Agency where they have a full program including the leakage testing and CAF.
Neves: They have CAF and via reliability, and they want to keep the materials below the Tg point of the laminate system because at or near the Tg point, you start getting some extreme expansion in the material that causes degradation to your vias that doesn’t exist in real life. You never see that stress in your product’s life cycle. Even in a satellite, you don’t see that material going over the Tg of the material during operation. They also want to get a very high-temperature gradient in their testing. They want a 200°C delta gradient, so we go from -55 to 145°C to give them that 200°C gradient. We cycle through that, which allows them to then take that information and relate it to real-life reliability and see how long it’s going to last in their environment in the field. Meanwhile, a lot of the testing that’s out there right now is, “Let’s take it to 220°C, 230°C, 250°C, or 260°C and cycle it until it breaks.”
This via robustness approach may tell you which material is better than the other, but it’s not necessarily going to relate to what’s happening in your end life use situation. You could have two materials with one that is significantly worse in a via robustness test that may be as reliable in the intended end-use environment, and there’s going to be cost differences between materials of different via robustness as well. People are going to have to decide, “How much am I willing to pay for this additional via robustness that may or may not have any influence on the via reliability in the end product environment that I’m using it in?”
Holden: Reliability is getting more and more important. If you don’t understand a lot about that, I’d recommend the seventh edition of The Printed Circuits Handbook, where Chapter 53, "The Acceptability and Quality of Fabricated Printed Boards," is written by Bob Neves. It’s a great place to start, but there are six chapters on bare board and assembly reliability written by Dr. Reza Ghaffarian, which are outstanding and provide you with the fundamental knowledge about modeling, reliability, and testing. If you want to know about these topics, get the handbook, read it, and then they can contact your organization for further information.
Neves: Thanks, Happy. It’s always a pleasure, and congratulations on your “night of Happy-ness” at IPCA APEX EXPO 2020.
Holden: Thank you very much.