Jim Hisert’s Tech Support Blog
Recent Blog Postings
A passage from “Metals Handbook, Volume 6, 1983”
According to the Encarta Dictionary that popped up on the left side of my screen a moment ago, “inextricably” is defined as:
1) impossible to get free from
2) impossible to disentangle or undo
3) hopelessly involved or complex
The interaction of flux and alloys/surface finishes is involved and complex. I think we’ve gotten to the point where we can take “hopeless” out of the description of flux though.
It is still widely thought that a flux is ‘more powerful’ or ‘more active’ than another. That’s like saying “John is better than Joe” – John is better at what?!? Part of the confusion with fluxes came from the very chemical nature of the fluxes. Since certain chemicals react with certain oxides better over different temperature ranges, it would only be accurate to say that Flux A is better than Flux B for soldering alloy “X” to surface finish “Y” with “Z” profile. And even then you’d still be discounting the effects of Oxygen level during reflow, reflow equipment type, and other lesser variables.
When I began testing different fluxes to examine the effect surface finish had on solderability, I expected each flux to solder to copper differently. For example - lets call the wetting a function of that flux (f), and the difficulty of the surface a nominal value copper=1. It made sense to me that for a given alloy “A”, the result would be 1A(f), soldering to nickel would be something like 0.5A(f). That’s just not the way fluxes work though. Yes, it is more complicated than a simple formula, but flux activity is known through experience. Extensive testing has shown what works, and to what degree. There is no longer an excuse to guess at a preferred flux for ball-attach applications.
Electroplated Bump Formation
A more experienced engineer than myself offered a bit of advice for wafer bump plating with indium:
“Electroplating using the indium sulfamate plating bath is one method of depositing indium bumps onto wafers. However, using direct current often results in deposits with a high degree of surface irregularity that can result in electrical shorts between individual bumps. One solution to the problem is to use microprocessor controlled pulse plating where the current polarity periodically changes, resulting in the plating and de-plating of indium. Obviously, the positive portion of the cycle where indium is deposited, has to be greater than the negative portion of the cycle where indium is de-plated, to have a net deposit gain.
Pulse plating works because the negative portion of the cycle removes excess indium in high current areas that plate faster than low current areas, resulting in a leveling or smoothing of the deposit. There are three variables in pulse plating:
Wave shape - sine, sawtooth or square
Ratio of the amperage of the positive to negative cycle
Time on in both the positive and negative cycle
Determining these variables must be done empirically, and most companies that have successfully done so, consider the information proprietary.” We can suggest (offline) specific values as an initial starting point to establishing optimum settings for pulse plating of indium for a wafer bumping process.
I’ve always said that you learn twice as much when you are instructing – well today I had the opportunity to learn. A colleague here at Indium wanted to learn more about the assembly of solar cells, and asked me to answer some questions. As it happens, verbally explaining the assembly process brought up some questions for me to inquire about further. Usually the questions I answer on a daily basis are focused on one application for one technology, speaking about the differences connected some ideas that will help me explain things better the next time someone asks.
To become really knowledgeable about any topic, you need to share it. Whether it’s semiconductors or cooking, model train building or underwater basket weaving – if you’re passionate about your hobby you owe it to yourself to discuss it with others. We are lucky enough to live in a time when we can communicate quickly with people who share the same interests around the world. Take advantage of it.
This is something you may have run into if you’ve ever manually dipped and placed PoP components. Yesterday while trying to hold a conversation about the package-on-package process, I lost track of the very same process I was discussing. I neglected to dip one of the components into solder paste before placing it on the board. This is a picture of what happens. The stack itself soldered well, as would be expected, but fell off when the board was lifted from the conveyor at the end of the line.
If you’re wondering what a small error like this costs:
1 hour lost time + price of the board and components.
Luckily the other 14 PoP stacks can still be used for cross-sectioning and learning more about the PoP paste that was being evaluated. If this was a production board I would be able to simply dip the stack and re-place it on the pads, send it through a second reflow, and test the final assembly for functionality – but you just can’t get away with that during evaluation.
Fans of the Semiconductor Packaging Blog will realize that marketing charts are a strange sight on this blog. However, I thought I would share something that I found in a recent report from the Semiconductor Industry Association - because everyone is looking for some good news in light of the current economy.
I don’t like making decisions based on one or two data points, but seeing the start of a positive trend for semiconductor revenue brightened my day a little bit. I’m glad that the chart includes the “Crash of 2000” to put things in perspective. Young guys like me were still in college when that panic was upon us.
(Follows this post)
Microsphere quality is especially important for bumping wafers. The dimensional tolerance of microspheres impacts the bump co-planarity across a wafer surface. In short, more precise spheres directly influence the quality of bumps on the die. This will of course increase process yield since the spheres will all be closer to the pads they are being soldered to.
90um flux deposits on silicon
(Follows this post)
In most cases, all that is needed is about a 25um layer of flux on the bonding surfaces and solder – usually 2-4% mass of the solder. Flux should almost never be of equal volume as your solder (except for some solder pastes). The perfect amount of flux will be enough to form a good solder joint, but will clean well (for water or solvent soluble fluxes) or appear clearer with less residue (no-clean fluxes). In extreme cases, lowering flux volume can improve reflow cycle time, because complete activation can occur sooner – and thermal inertia is decreased. More volume is also more expensive. Keep this in mind and dial in your process.
In my opinion, microsphere bumping is the best combination of cost, simplicity, and precision for bumping.
Wafer level packaging shifts some of the common assembly house steps back to the wafer processing stage. This transfer of this responsibility allows all of the die on a wafer to be processed at once, saving time and reducing cost. This week we'll discuss the materials that are needed, and the benifits of using the proper materials.
A lot has changed in the world of package-on-package in the last few years. The most obvious change that I have seen is the development of specialized pastes for component dipping. If you haven’t tried one of these pastes, here are 8 reasons why you should:
8) More consistent transfer over time
7) Head-in-pillow elimination
6) Better wetting to a range of alloys
5) Optimized metal loading
4) Specially designed powder distribution
3) Halogen-free flux formulations
2) Maximized transfer volumes
1) Higher possible yields
It’s no secret that I love WS3622. The water soluble flux is a magic bullet for all types of semiconductor packaging applications. As a tech guy, you need to be able to trust a flux – I trust this one. It can be used for flip-chip attachment, sphere mounting, wafer-level bumping, and a variety of odd jobs.
Since I began describing the flux to coworkers and customers, I have always mentioned that it is red and it has the appearance of ketchup. Without going into my [over]use of ketchup, I guess that description stuck because it’s another material that is near and dear to my heart.
Today I noticed that WS3622 really bares more of a resemblance to red acrylic paint. I think it’s the smooth texture of the flux that really makes its consistency more like acrylic paint than ketchup.
Want to read more? Browse the archive of past entries.
