Personal Tales of Innovation: Impossible things….
One of my buddies used to say “America was built by salesmen and engineers.” I think there’s a good argument for that. Edison and the light bulb, Bell and the telephone, Ford and mass production, the Wright brothers and flight, just to name a few. You can find books written on these endeavors. But there are many patents that exist that are not known or studied, but that have commercial impact. Finding stories about those is a little tougher. As an Engineer, I took part in creating several of those commercially successful patents. Here’s one of them, where my team worked a persistent problem that no one else had cracked.
I have already mentioned one assignment where I was working on a revolutionary technology call Chemical Mechanical Polishing (CMP). One of my patents (now expired) from this time is US5985045 “Process for Polishing a Semiconductor Substrate.” This patent goes to mixing slurry on the polishing tool. CMP requires that you take a wafer, and rub it against a pad in the presence of a slurry or polishing fluid. The fluid and pad combination help remove material preferentially from the high spots, thus, as you polish, the wafer gets flatter. If you’ve ever seen a jeweler’s wheel, or similar contraption with a feed tube where a polishing fluid is delivered to a moving wheel, and then an object is pushed against the fluid covered pad, that’s what our machine did – only our polish object was an 8 inch (200 mm) diameter silicon wafer.
As basic as that seems, we were so early in the development process the state of the art at the time was to take a slurry concentrate, dilute it with water (1:1) to form our polishing fluid, slosh it about to mix it, pump the resultant fluid and use it for polishing. We were hand-mixing slurry in batches. This was a lot of work, mixing using 5L graduated cylinders, and big Nalgene carboys to make 20 liters of fluid at a time. (That makes a batch weigh about 45 lbs.) Just moving the full Carboys requires special carts and lifting tools. We would need each operator to spend about 2 hours mixing stuff to keep the tool running. If we went to production, I might need 3 techs working full time just mixing stuff. Additionally, once mixed, the polishing fluid had a shelf life of only a few hours, so you couldn’t just premix a week’s worth at a time. This decision had big implications for training, space, numbers of people, etc….
It seemed like automating this was a logical step as we headed into production. It not only would free up people, but properly done, would also help us manage the timing between mixing and usage. I looked at the tool and they had a bank of pumps that were traditionally run one at a time, so you could have more than one slurry source available. The pumps were programmable, so there was not any reason other than “we've never done that before” that they could not be used at the same time.
But I asked the chemical provider, the tool provider, and the company that we were transferring the process in from and ALL of them told me that it would not work. The common wisdom for the day was that the mixing would be poor quality and result in a lot of scratches on the wafer if it was used in the process. My background was different than most of my peers. I was a Chemical Engineer in a world of mostly Electrical Engineers and Solid State Physicists. In my world, all kinds of stuff gets mixed at point-of-use every day. While this was a demanding use case, I could not figure out why it would not work in an automated way, considering we were already mixing exactly the same chemicals by hand and having success. Additionally, this technology area was dominated by trade secret. Any technical advance was deemed to be a huge competitive advantage so no one was publishing technical details of their techniques. That meant everyone would tell me that it wouldn’t work, but no one could point me any data to support that position, or point me at an alternative that would work.
It probably says something about me that despite the advice of more experienced, world experts, I decided to try automating mixing anyway. I had my team modify the tool, and we tried it out. It was very anticlimactic. Everything was fine. We did not have scratching and the rate was high and stable. In fact, it was both stable and faster than normal at the same conditions because the chemical activity in the slurry was higher than from the premixed material. No catastrophe ensued. We recouped 2 hours a shift for my team that we could then use to do other development. No inventory or storage control issues (that I’m pretty sure that everyone else was dealing with). No need for a separate chemical handing area…. Pretty good outcome.
But the story does not stop there.
Here I learned one of my early lessons about patenting and patentability. Because of the importance of the CMP project to Motorola, a patent attorney was assigned to stop by my cube pretty regularly to ask what was going on. After relaying the story of the automated slurry mixing his eyes get really big and he asks me a bunch of questions. Is this easy to implement on existing equipment? Yes. Is this cheaper? Yes. (It polished at a higher, more stabile rate so used less slurry than the existing method AND it saved a bunch of labor.) Turns out our process was actually faster, better AND cheaper than the state of the art.
The attorney begged me to submit it. “This is stupid, if I submitted a paper on this, it would not pass peer review as there is nothing new or exceptional.” “This is a patent and not a paper. The rules are different.” Eventually, I did as he asked. Sheepishly, I applied for a patent on “mixing two things together automatically.” True to his word, (and to my astonishment) he got it to issue. To me this was an everyday problem with a clear, obvious solution. How could that be patentable?
I later learned, in the patent sense, part of the legal definition for obviousness, excludes simple solutions to long standing problems. The reasoning is that if it was actually obvious, then someone else would have already tried it because the problem is there to be solved. Commercially important problems (making things faster, better or cheaper) don’t just sit around; they get solved because folks want that efficiency. The existence of the problem was enough to justify the solution was patent eligible. The emails from the manufacturer refusing to modify the tool as I requested turned out to be the key to getting it issued.
So if you know a simple way to do an “impossible thing”, you just may be sitting on a patentable idea….