Innovative New Uses for Ceramic Column Grid Arrays from TopLine
TopLine President and CEO Martin Hart sat down to discuss his paper on ceramic column grid arrays (CCGA) at the recent SMTA International show in Chicago. Hart begins by explaining the relationship between CCGA and ball grid arrays (BGA), and TopLine's drive to find new uses for CCGA.
Andy Shaughnessy: Martin, why don't you start by giving us a little background on CCGAs.
Martin Hart: The best way to talk about CCGAs is to reference their cousin, the BGA. Everybody is familiar with BGA—ball grid arrays. Few people realize that the CCGA, the ceramic column grid array, was actually invented about ten years before the first BGA. When BGA first came out, they were invented by Bell Laboratories; a full ten years after columns had already been in the field. The main purpose of having a column instead of a solder ball is to extend reliability. You get more temperature cycles out of a solder column because it will absorb the CTE mismatch between the materials of ceramic and substrate, such as an FR-4 PC board.
Shaughnessy: We just don't hear much about CCGA. Everybody seems to have settled on the BGA and flip chips and various other sorts of packaging. Why did the CCGA fall out of favor?
Hart: That's a good question. BGAs are lower cost and much easier to assemble. During the manufacturing process, attaching a solder ball is significantly easier than attaching a column. Solder balls roll and self-center onto the BGA pads. Columns are difficult to attach to the pads. They don't self-center because they are held rigidly in place, usually by a graphite tool, during the column attach process. Columns can easily get bend if handled incorrectly.
Shaughnessy: You have to literally stand this thing up.
Hart: You have to stand the column straight up during the attachment of the column to the substrate.
Shaughnessy: It sounds very risky, in a way.
Hart: There is an element of risk, but it’s more of a situation of having a correct manufacturing process. Your process has to be dialed to effectively use column grid arrays. Once you have a process in place, then CCGA seems to work well.
Shaughnessy: What is the new use that you seem to have found for them?
Hart: First I’ll tell you the traditional places where column grid arrays are used. Traditionally, they've been used in defense, military, aerospace and also in high-value computing. One new area is down-hole drilling, where you have electronics going down several miles. It's very hot as you go below the earth’s surface. There is also interest in specialized automotive where you have electronics in the firewall. Also, columns can be are found in some medical. Not medical implants in a human body, but in medical equipment. There is a benefit to use CCGA wherever you have a large ceramic package, and mounted on an FR-4 board or a polyimide board. CCGA is more reliable when the temperature swings dramatically.
If you have a very small ceramic, it's safe to use a BGA. I've researched and wrote some tables and calculations showing when it is safe to use a ceramic BGA and when you must use a CCGA. A small 15 mm square ceramic BGA will not cause any problem with solder ball delamination. But if you start to go larger, say a 35 mm square, 40 mm square, 45 mm or even larger, 50 mm square, then solder balls will delaminate. That will effectively create a catastrophic failure in the equipment.
I also want to point out that your equipment does not even have to be electrically switched on. Say, if your equipment is turned off, sitting in storage, then if you have a large ceramic part mounted with solder balls on an FR-4 board, then that equipment might fail. If your equipment is out in the elements experiencing alternating hot and cold temperatures, then your equipment might fail if it contains large ceramic BGA components. When the sun goes down, temperatures might drop -25 to -40°. And when the sun comes up temperatures might rise 100-125°. Just sitting in the box, you'll see the phenomenon of delamination with large ceramic BGA components.
Shaughnessy: What led you to focus more on CCGA?
Hart: The original business of TopLine was to manufacture process development kits of all kinds of parts, from BGA and QFP in a kit form with circuit boards. Customers, particularly in mil-defense, were asking for column grid arrays so they could perform their own process development. This has gone on for a number of years and we decided, "Okay, we need to jump in and address this market of producing CCGA and solder columns."
Shaughnessy: I guess CCGA would have its own completely different process.
Hart: Yes, it does. The process of attaching a column onto a ceramic substrate is totally different than attaching solder balls to a BGA. CCGA requires a graphite fixture to hold the column upright at a 90° angle. It involves having a controlled reflow process. We found that vapor phase reflow ovens are best solution because your maximum temperature is fixed. You can also use a twelve-zone oven, but you have to have that oven profile really dialed to attach solder columns.
Shaughnessy: How tall is this column?
Hart: Oh, good question. Traditionally, columns are 2.2 mm high. There's nothing totally magical about 2.2 mm high. Columns have been made 1.27 mm, which is 50 mills. Columns have been made bigger in diameter and smaller in diameter. IBM set the standard years ago with studies and of papers showing that 2.2 mm, is a good column height. The industry settled ever since. The secret is actually in the attachment process. You start with a longer column and then trim it down to 2.2 mm during the planarization process. After all the columns are attached in an array, you literally have to cut the columns shorter to planarize them. So you wind up with 2.2 mm, but you start off with a column that is 2.3 mm, up to 2.5 mm, in order to wind up with 2.2mm.
Shaughnessy: Do they cut it off with a laser?
Hart: Traditionally, planarization is not a laser cutting process. Typically it is a mechanical guillotine, or lapping and polishing process. TopLine uses a method involving lapping and polishing. The lapping wheel spins at about 125 RPM with silicon-carbide paper. After 20 seconds, the columns can be reduced to final length. Then we finalize it with a diamond polishing film to make the columns nice in cosmetic appearance. This again explains why columns are much more difficult to attach than solder balls. That's just one additional example showing why people have moved away from columns into BGAs.
Shaughnessy: Do you see this really making a comeback? Are you at the forefront of a major change here?
Hart: Actually, columns never went away. Mil-defense, aerospace industry and high-value computing prefer CCGA due to its ability to absorb tremendous swings in rugged environments. CCGA are not being designed out. The good news is that more and more applications are being designed that can benefit from the use of solder columns. In summary, there is an ongoing need for CCGAs.
Shaughnessy: Tell me about micro-coil springs. I understand that is a new innovation?
Hart: Over the past 40 years the industry has tended to keep using traditional solder columns. It was time for someone to invent a new type of column. In 2012, the engineers at NASA did a lot of research and they invented the micro-coil spring to replace solder balls. The springs are tiny and stand upright on the pads. Typically micro-coil springs are 1.27 mm high and 0.5 mm in diameter. Since NASA is not in the business of commercializing their inventions, they offered a technology transfer to TopLine. Today, TopLine manufactures and sell micro-coil springs.
Shaughnessy: Impressive. But why micro-coil springs?
Hart: A spring is able to absorb shock and compressions because it's a spring. A chip package can compress and then bounce back to its original size. NASA determined through exhaustive testing, that a chip carrier can attain considerably more thermal cycles using micro-coil springs. How many more cycles? Recently, NASA did a test where micro-coil springs survived more than 20,000 thermal cycles. Typical device data sheets state the number of hours a device will survive at a steady state, such as an operating temperature or storage temperature. However, the actual life of a device is measured in thermal cycles.
If you start at room temperature, and then you bake your chip to +125°C, then you freeze it down to -40°C, then bring it back to room temperature, that's called one thermal cycle. A large size BGA may survive only a small number of thermal cycles. Maybe a large ceramic BGA might survive only to 300 cycles before delamination. A traditional solder column can survive 2,000 to 3,000 thermal cycles. That’s a significant improvement. NASA tested micro-coil spring and some survived over 20,000 thermal cycles. The micro-coil spring have higher survival rate than the traditional solder column.
Shaughnessy: That's cool. This is all stuff that you guys have developed on your own?
Hart: It has been an ongoing process. TopLine is making it easier for customers who want CCGA packages by providing columns and the tools so that customer can make columns with their own packages. TopLine provides columns to do re-columnization process by taking the balls off and putting the columns on. TopLine's approach is to teach customers how to attach columns and provide them with the means to do it.
Shaughnessy: So, this is great for mil-aero, it's great for down-hole drilling, and that sort of thing. Any application that’s very high-reliability, with lots of vibration, and lots of thermal cycles.
Hart: Yeah, that summarizes the applications where columns are superior to solder balls.
Shaughnessy: I’m sure it will be an interesting class and paper session.
Hart: Anyone who wants to learn more about columns can go to our website, www.CCGA.co (and it’s not .com). There are a lot of pictures showing the process, movies showing how columns are attached, etc.
Shaughnessy: Martin, thanks for your time today.
Hart: Thank you, Andy.