Solder Voiding, Autonomous Autos, and Statistics—So Much to Learn from Dr. Ron

Estimated reading time: 20 minutes

While at the 2016 SMTA International conference and show, I had the lucky opportunity to talk with Indium Corporation’s Dr. Ron Lasky. He gave a great synopsis of the extensive testing done by Indium Corporation’s Chris Nash on solder voids, outlined in a paper co-authored by the two, which was being presented by Ron at the conference. Our discussion veered toward autonomous vehicles and the electronics and software that will be needed to make them a reality. Finally, the conversation morphed into the statistics of election polling and how it really works. Great stuff!

Patty Goldman: Hi, Dr. Ron. Let’s begin with a little intro about you and what you do.

Dr. Ron Lasky: I am a professor at Dartmouth College, and I work with the wonderful folks at Indium Corporation in Utica, New York. I've been involved in electronic assembly for about 30 years.

Goldman: I have been reading your columns, “Patty and the Professor.” And of course, I say to myself, "Is he talking about me?" (Laughs) But let's talk about the paper that you will be presenting today.

Lasky: I'm going to be giving a paper on minimizing voiding in bottom terminated components. It's a pretty important topic today because a lot of the amazing electronic devices that we use, like our mobile phones and computers, have bottom terminated components. If there is not a good solder connection between the ground pad on the printed circuit board and the thermal pad on the bottom terminated component, there can be overheating of the component. A major cause of having a poor thermal conduction path between the two pads is voids in the solder. We've done surveys of our customers and we think voids in bottom terminated components are the most critical issue in electronic assembly right now.

Goldman: You speak about smartphones and things, but there are also a lot of very critical applications as well where it's not just disappointment when it fails, but a lot bigger problem.

Lasky: Yes, it could be a disaster in mission critical applications.

Goldman: Maybe give us a little synopsis on the paper?

Lasky: It was some work done by one of my colleagues at Indium Corporation, Chris Nash. He collected all the data and, since statistical analysis is one of the things I teach at Dartmouth, I performed the data analysis, so we co-authored the paper. I think it's a pretty thorough paper because Chris did quite a few experiments where he covered the reflow profile, the thickness of the stencil, the powder diameter of the solder paste, and he tested several different solder paste formulations. We determined how each of these variables would minimize voiding. I think some of the results were sort of expected, but there was one at the end that was a big, pleasant surprise.

Goldman: Can you share that with us?

Lasky: Sure. Of course, to see all the details and the numbers you're going to want to come to my talk or see a copy of it in the proceedings. One of the things we found was that in the reflow profile, if you use a hot profile, you can dramatically reduce voiding. That wasn't too big a surprise, but again, what was helpful is that Chris collected a lot of data, and the statistical analysis is overwhelming. There's a significant drop in percentage of voiding by using a hotter profile with a little bit of a dwell before you go into the liquidus part of the profile. That probably wasn't too big of a surprise, but Chris also tested 4- and 5-mil stencils, and he found that a 5-mil stencil thickness reduced voiding. It's statistically significant, but it's not large—a few percent—and that was interesting.

We think the thicker stencil gives, physically, a little bit more relief in having some height to let the volatiles leave. The third thing Chris studied was the solder powder diameter. Solder paste is a mixture of solder powder and flux, and there are several sizes of solder powders, which are called types. For instance, a Type 3 solder paste has a powder that has an average particle size diameter that's a little larger, and then a Type 4 is a little smaller, and then a 4.5 is smaller still, and a Type 5 is the smallest of the four we looked at. What we found was there wasn’t much difference between Types 4, 4.5, and 5, but all of them were statistically significantly better than the Type 3 powder in reducing voiding. That one was a little bit of a pleasant surprise.

Goldman: Could you come up with a reason why the powder size made a difference?

Lasky: That's one that I'm not sure we've nailed down yet, but we probably will. At the very end of the paper, one of the things that we discuss is our experiments with a wide variety of solder paste formulations. Solder paste is solder powder mixed with flux, but the fluxes can vary. We studied approximately 20 different solder paste formulations, and we found that some of them will produce about 40% voiding. If you look between the bottom terminated component and the pad, about 40% of the area would be a void, which is not good at all. That's the bad news; there were some formulations that were that bad. The good news is we had some formulations that reduced the voiding almost to zero, down in the 5–6% range.

If you look at an X-ray of a bottom terminated component, you'll see the light areas where the voids are. When it's 40% that means 40% of the thermal pad has a void, essentially meaning that 40% of the area will not be conducting heat out. When you get to the better solder paste formulations that had about 4 or 5% voiding, that's close to 0. We think realistically it's hard to get much better than that. That was a big, pleasant surprise, because what this means to users is that if you work with your solder paste suppliers it's possible to get a solder paste that you can use and vary a lot of the other parameters and still get low voiding. You don't necessarily have to have the hot thermal profile and you don't have to have a perfect process and you can still get voiding into the less than 10% range.

Goldman: I'm thinking a 40% void—that's also 40% less strength in that connection, isn’t it?

Lasky: Yes, it's mechanical strength, thermal, and electrical conductivity.  All three of these important parameters are negatively affected by voiding. We estimate that the industry is living with probably something like 25% on average, and that most people in the past have considered that acceptable. Now with the work being done by people like Chris, we're heading toward a time where  it'll be more like five or 10% that will be considered  acceptable.

Goldman: I'm still on the percentage thing in my head, because I'm thinking, "Okay, is that 40% of every solder joint that you analyze?" It's not that 40% of the solder joints have some voiding, but rather that all of them have some percentage of voids.

Lasky: Say you analyzed 10 solder joints, some would have 40%, some would have 50%, some would have 30%, but the average was 40%.

Goldman: The likelihood that there were any without voids was probably slim.

Lasky: Essentially none.

Goldman: That's very interesting. I'm guessing the solder pastes were all lead-free?

Lasky: Yes, our experiments were all lead-free.  About 75% of all solder used today is lead-free.

Goldman: That's right, that's where all the testing is being done because that's the newest stuff. This is a nice synopsis of the paper.  You had some other topics you wanted to discuss, as I recall?

Lasky: Yes, I had suggested if there was interest, because I've been studying this quite a bit and blogging about it, I could talk about the reality of driverless cars, or autonomous vehicles. It's something that, first, you say, "Why should we in electronic assembly be interested in this?" Obviously, the autonomous automobiles are going to be requiring a tremendous amount of electronics. We will be making the electronics for it, and the reliability requirements will be unequaled in the history of electronics.

Goldman: Even more so, I think, than going to Mars.

Lasky: Well, that one may be tough to match!  But let’s think about when autonomous vehicles become common. I live in Woodstock, Vermont, and I'm going to use this as a little example, as it is a nice little rural town. I'm going to have an autonomous vehicle take me from my home in Woodstock to Logan Airport in Boston, and it's about a two-and-a-half-hour drive. I'm just going to sit back and relax, I'm going to read, I'm going to be on my cellphone, the car's going to be driving, and it's going to be going down Route 89 and Route 93 at 65 miles an hour. It had better work with perfect reliability!

Goldman: Not only must that one work, but every other vehicle on the road as well, and you've got to be able to avoid things like deer jumping out at you.

Lasky: Precisely. The reliability requirement is going to be beyond the space program. Not only the reliability requirement of electronics, but I think, at the very least, the software reliability needs to be seriously thought about. I don't know about you, but it's still not that uncommon for me to reboot my computer because it's just not working; there's a little conflict between Gmail and whatever, so I'm rebooting. When you're going down the highway at 65 miles an hour, you can’t have that.

Goldman: My cellphone got into this little loop last night, and couldn't get out of it so I had to shut it down completely, so it's the same kind of thing. You multiply that by your car and millions of other ones on the road and it's a big deal.

Lasky: One of the things I wanted to share is, you see literally, and I mean literally every day, an article about the fact that autonomous vehicles are right around the corner. It's overstated. There are different levels of autonomous vehicles. There are autonomous vehicles right now, but they're very, very primitive, and what they can do has been romanticized. I don't know another way to say it. What I mean by that is Google has had self-driving cars for years, but they must stay on the same route, they go about 20 miles an hour, and the route has markers on it that are specially designed for the cars. If it sees something like a paper bag in front of it, it panics because it thinks it might be a big stone. It doesn't know the difference, like you and I quickly would say, "It's a paper bag, and I’ll just run over it." The autonomous vehicle doesn't know.

There are five levels of autonomous vehicles, and right now there are some cars that are approaching level two. Volvo has been one of the champions of this, and I've driven one. You can set it on autonomous driving if the road has markers on the side and lane markers. It's fun. You let go of the steering wheel and the car starts to drift and then automatically it pulls it back. When there's a red light, you come up to it and you don't do anything—it slows down. That's level two, but true autonomous driving is level five, where the car wouldn’t have a steering wheel, a brake pedal, or a gas pedal.

People that study this more than me agree. There's an excellent article in the June issue of Scientific American for people who are interested in this topic. The author said he thinks it'll be at least 2045 before we have a fully autonomous vehicle. He then complained in the article that because he said that, people are now claiming that he's saying that the autonomous vehicle will be here in 2045. He said it'll be at least 2045 before we have a fully autonomous vehicle.

Page 1 of 2

• Next Page >