As if in another life, almost another world, and certainly when I was a much different man, I wore many different hats. They have been ball caps, hard hats and hats that weren’t hats, like suit jackets and ties. Each has been a composition of time, place, circumstance and opportunity and each has formed who I am today. And, today, one of those lives has decided to make a reappearance. It was a “white-collar” time, a time when I moved, not exactly comfortably, in and among the Silicon Valley brain trust, the engineers, primarily, who would drive innovation and determine future production.
I wasn’t part of it “at the beginning,” not of Silicon Valley proper, but I do remember when the Santa Clara Valley became Silicon Valley. It was in the 70s, I was still a kid. My dad, a Ph.D. chemist by trade, worked for SRI International (formerly, the Stanford Research Institute). He had contacts and friends who were very much involved, in some way, in the semiconductor industry. While chemists are certainly employed by these companies, they are mostly after the engineers. One of those contacts at some time in the 70s asked my dad if he could solve a problem – a chemistry problem – that was within his realm of expertise.
It was not for a semiconductor, exactly, but rather for a means of handling their semiconductors. They were called “beam-lead diodes” and were made from the far more expensive semiconducting compound of GaAs (gallium-arsenide) because silicon would not operate effectively at the microwave frequencies required for the military applications these diodes were needed for. (I know I am getting into the technical weeds here, but bear with me). The name of the diode, “beam-lead,” is an appropriate description of their tiny leads; they look like beams protruding off of the diode. These tiny diodes are fragile enough, their even tinier “beams,” so easily knocked off, are even more-so. One broken lead renders the entire device worthless. At about $1,000 per diode, so small they are hard to see without magnification, every damaged beam was a lot of money.
The way they came up with handling them was to use a rubber-like “gel,” a silicone (not silicon, that is something entirely different) applied to a two-inch square glass slide that the diodes would “stick” to, but still release from with a dab of alcohol. It seemed to work, when it worked. The problem was that they could not get the gel to cure consistently. It is a two-part resin and when applied that thin, it is difficult to get the curing agent to react properly. Someone from either Hewlett-Packard or Raytheon (I don’t remember who was first) asked for help and my dad solved the problem. Then, since they had no desire to be in the chip handling business in the first place, they asked if he would just make these damned things for them. After crunching the numbers, my dad and mom saw it would be a profitable part-time endeavor. “Beam-Pak” was born – in our garage. And our kitchen. And our family room.
This is where I am going to fast-forward – a lot. I don’t remember exactly when the first Beam-Paks went out the door, although I do, very distinctly, remember the first Beam-Paks going out the door. Eventually, though experimentation, the “gel” in the Beam-Pak found its way into other applications and, as the market grew to more than just beam-lead diode manufacturers, the name was changed to “Gel-Pak.” While our market was still primarily high-value GaAs semiconductors – FETs (field-effect transistors) and MMICs (monolithic microwave integrated circuits) – the business outgrew our garage and our home. Over the years, the business moved to larger and larger facilities in Mountain View and Sunnyvale, employed more than 20 full-time employees and employed several family members, myself included - two or three different times.
At one point, my dad was faced with not so much a chemistry problem, but an engineering one, a physics one. It had to do with automation. While Gel-Paks did an outstanding job holding these fragile, high-value chips in place during handling and transport, they still had to be manually transferred out of the package to wherever they needed to be. Automation used vacuum wands to handle chips, and a vacuum is not enough to break the surface tension formed between the bottom of the chip and the gel. Between my dad and his partner at the time, they came up with, and patented, a unique way to temporarily break that surface tension when needed, on demand. The new system was called “Vacuum-Release” because a vacuum drawn under the gel would pull it away into recesses that would allow the chip to only be in contact with bumps of gel, not the entire surface. And it was reversible, when the vacuum is released, the gel becomes flat again.
Armed with this new technology, Gel-Pak (eventually incorporated as Vichem Corp, still family owned) was able to sell the advantage of complete safety during shipping and handling along with ease of removal for automated assembly. It was the best of both worlds. However, because our product was considerably more expensive than the competition, we were still limited to only very expensive chips and, at the time, that meant not silicon, but GaAs. However, in the late 80s, Intel was getting ready to introduce its new processor, the i486. I ran into an Intel engineer at a trade show (probably Semicon West) and found that, in the early production, these chips had very low wafer yields and were being packed and transported prior to assembly. And that was a problem. This was damaging what was left of their already low yields.
Our containers would have solved the problem, but their chips were too big and our packs were too small. We got busy and came up with a larger format VR package that would accommodate their processors and, sometime thereafter, Intel became our largest customer. We might have been selling packages for silicon chips before (and we had a variety of other applications, too), but this time it was big, literally and figuratively. We continued to grow, continued to explore new markets, continued to adapt and develop our technology for new applications and, no matter what the economy was doing, we remained very profitable.
Eventually, a combination of personal issues in my life, corporate changes in the company and a bunch of other bullshit – some of it my doing, some not – led me to depart the company for the last time. My title when I left was “marketing manager,” but I did a lot of everything. I, still, have seriously mixed feelings about it all. A couple of years later, my dad sold the company, consulted for it for another year, and retired. It’s still around, still making the same products we developed. I don’t know if it still seeks the opportunities the way we did, I don’t even know if that world still exists. That time in Silicon Valley was heady; it was fast, new things were happening all the time. It was a full-time job just keeping up with what was going on, staying ahead of the curve… finding out what the Intels needed before they knew they needed it.
So, what brought all this back? Well, in the news recently, Google unveiled its new quantum chip, “Willow.” I don’t pretend to understand what all it means other than to understand that quantum computers are the next big thing, and this is a step towards that. In stories about it, there is a photo of this new chip, on a gloved hand. Between that chip and that hand is a Gel-Pak VR tray. It is exactly the same thing I used to show to potential customers back in the late 80s and early 90s. It is still doing what it was designed to do – protecting a high-value, fragile chip while still allowing it to be released from its captive state on demand with a simple vacuum. The technology is no longer protected by its patent, but it is the same patented technology developed all those years ago.
***Correction: Beam-lead diodes were actually fabricated from silicon, not GaAs. They were still fragile and expensive, but not GaAs. Fets and MMICs were GaAs. Some other small details are also subject to minor errors due to my aging memory.