The Story of the Infamous Crystal Grower

Taken from the R. VICTOR JONES, interview by DAVID C. BROCK via telephone, on 18 April 2006 Philadelphia: Chemical Heritage Foundation, Oral History Transcript # 0336 (Pages 13-16).

BROCK: When did the inklings of the crystal grower project first become visible?

JONES: In talking earlier about the organizing principles of the laboratory, I left out the third element: namely, the dictum that we would only work with silicon.  At that point silicon was a monster; the lifetimes in silicon were just terrible.  Because of the higher melting temperature it was a very much more difficult material to work with than germanium and, while silicon dioxide now contributes to making silicon the greatest material in the universe, initially it made working with silicon a horror story since the dioxide formed quickly on fresh surfaces and had to be removed with hydrofluoric acid—ugh.  For many of the folks working in semiconductor technology, silicon was an absolute monster so these folks were more comfortable making germanium devices.  In fact, there were many good germanium devices on the market at that point.  But Shockley wanted silicon. We were going to make it devices with silicon.  We were going to solve the problems of silicon.  One of the key problems was the notion that defects were being incorporated from the quartz crucibles in which the crystals were being grown.  The view was that one had to get away from graphite-heated quartz crucibles because the quartz-silicon interface supposedly generated oxygen defects in the crystals.

The defects had not as yet been identified, but Shockley always had in his mind the history of deathnium in development of germanium technology. [laughter]  He always had in mind the deathnium story while he was trying to solve the lifetime issue in silicon.  In fact, that was another thing I got deeply involved in; calculations on how to how to do floating zone crystal growth of silicon, which is, because of the higher melting temperature, much more difficult than floating zone growth of germanium. We did work up some pretty good designs for silicon floating zone techniques. These were vertical column setups very much like what later evolved in other silicon floating zone systems.  The problem was that the rf (radio frequency) heating had to be controlled to a very high degree.  The control problem is very nonlinear and it was difficult to build adequate control circuits in those days.  I spent a lot of time on that issue.

Shockley and I were on one of our trips and we started talking about the whole issue of how to grow silicon without a crucible.  Somehow in our conversation—I may have first posed it—we got into the idea, ―Why not use silicon as the crucible?   How would you do that?   What are the different ways you could do that?   I think I first put the idea forth because it had resonance with something else I had once done.  But, I really can't put that history together and I don't want to make any unwarranted claims.  In our discussion I said, ―If there was a heater that brought the whole silicon boule to a temperature close to the melting temperature, we could then use a small resistance heater above the surface to just heat a puddle of molten silicon from which a crystal could be pulled.  We spent the rest of the night talking about the idea. It was a transcontinental trip, and in those days that meant a long night.  By the time we got to Washington, [DC] which was where we were headed, we had pretty well scoped out what we were thinking about.  It seemed like a promising idea.
When we got back, I took the next step. A s mentioned before, I really enjoy down and dirty experimental work, so I took a simple tabletop chemical oven that went to a hundred and fifty degrees or something like that, and I built a little surface heater and got a pool of Wood’s metal within a boule of Wood’s metal. Wood’s metal is a good experimentalist’s trick since it melts at something like eighty degrees centigrade; I was obviously trying to impress Shockley.  I think in about a day of fooling around I was able to use a little drill motor to pull little cylindrical ingots of Wood’s metal up through the surface heater.  Voila!  There it was— of course, it wasn't a single crystal because that may not even be possible—but it was a nicely shaped polycrystalline cylindrical ingot grown from a boule of Wood’s metal. I took these ingots in and showed them to Shockley and of course he was ecstatic.  As I look back, this was one of the greatest mistakes I ever made in my life.  Shockley thought, ―It's so easy to do this.  We did this in one day.  We just go from here now to silicon.   I said, ―Well, no.  We really should work out some of the other issues.  He said, ―No. You haven't got the big picture here.  In order to move in this modern technology, one has to really be bold.  With that he pulled in Knapic [Dean Knapic], Kleiner (Eugene Kleiner), and Julie Blank [Julius Blank] who were at that point underemployed, I would say.  Shockley viewed them as people who were going to be doing production lines and things like that.  He asked Dean Knapic, ―Do you think you could do this?  Of course, Dean Knapic said, ―Yes, no matter what the job was. I cautioned them—this was a big shift from Woods metal's melting point to silicon’s, and there were some major problems.  But, Shockley was absolutely convinced that we were going to do this, so that began the adventure of building ― the crystal grower.   Of course, we rapidly started to work on some really difficult technologies.  We ended up with things like rolled molybdenum heat shields and titanium heater wire, some incredibly expensive stuff.  I spent a lot of time going to various kinds of vendors to see who could do this and who could do that.  I did a lot of traveling.  Of course, it all had to be secret because, of course, Shockley was a very secretive guy and he thought this was going to revolutionize the world.

About this time Valdes left. Valdes and Shockley didn't get along because Valdes thought he knew more about semiconductors than Shockley.  That was not a healthy working relationship with Shockley.  After Valdes left all crystal growing for the lab fell into my lap which was amazing because I had no previous experience with that kind of thing.  But, I had read articles on conventional techniques and so we started pulling crystals—not great quality crystal, but silicon that was reasonable and useful for early development work.

While I was doing conventional Czochralski crystal growth as a service function for the rest of the lab, I modified one the conventional growers to test further the ―puddle growth idea.   Of course the problem was that the surface heater suspended above the silicon in the conventional configuration was heated both by a current source and uncontrollably by the rf generator. Somehow, I roughly shielded the surface heater from the rf.  I've forgotten exactly how I did this, but it was a kluge for sure. I could make a puddle in this modified grower, and I was able to reach down and pull up a crystal.  It showed the feasibility that something like this could be done. The real grower couldn't be built with an rf heater because there was no way to control what was being heated.  The crystals from the modified grower looked really funny because funny things were going on.  But, I was able to make them slower and then hotter, and so forth.  So, we did grow these little crystals.  Of course, that again made Shockley ecstatic.  I was so deeply mired in this project at this point that I needed this as some kind of reassurance that it wasn't total nonsense. So I grew those early, or only, puddle-grown crystals (In fact, I actually grew these puddle-grown crystals before I had the convention crystal growing program underway.). A number of years ago one of my kids found one these feasibility crystal in a drawer and he said, ―What's the heck is that?  A picture of that crazy crystal is now on one of my web sites.

Meanwhile, the full-scale experimental
puddle crystal grower just grew like topsy. I don't know if this was a punishment or something, but it was mounted next to my desk in the old garage. I
almost lost my hearing because a worker putting up a steel structure for the grower popped off a powder actuated gun near my ear—I lost my hearing for about seven months.

BROCK: Oh my gosh.

JONES: This machine just grew and grew.  Of course, it was very complicated to build things in Palo Alto at that point, because of the limited technological resources in the Bay Area.  The engineers (Knapic, Blank, and Kleiner) had come from the East Coast where they knew that there was always a guy to make a vacuum system over there and another guy to do complex machining over here.  It was pretty frustrating for them since there just wasn't that kind of technological base in California at that time.  A lot had to be built and imported from the East.  However, I think the grower’s huge vacuum tank was fabricated at a shipyard in South San Francisco.  The most difficult problem was the design and fabrication of the heat shielding for the resistance heated core.  Resistance heating was used in the system as the only way to solve difficult temperature control issues and the resistance heater had to be shielded to minimize heat loss.  But, of course, all these structures had to be stable at silicon’s melting temperature (1410 oC). Nothing was simple.  Everything was complex.  In conventional Czochralski crystal growth, rf heating makes things pretty simple since only the crucible needs to be at the silicon melting point.  The shielding issues are relatively simple and it is very easy to get power in and out.  Resistance heating was a much more complicated business.  The resistance wire available when we first started the heater core had to be used in a vacuum.  As we designed the system there was just one problem after another.  These difficulties certainly contributed to my wanting to leave since I was stuck with this monster.  I spent a lot of time with patent attorneys developing the patent that I saw for the first time yesterday after you sent it to me. [laughter]   The patent people were very good—and I've always enjoyed working with patent attorneys because they can make you expand your thinking to increase the coverage.  I really haven't read the patent yet but I would suspect that, from the guys we had, the coverage was pretty good on it. [laughter]  They always said, ―Could you use it for this?  Why can't you use it for that?

That is the whole history of the grower as far as I was concerned. It did go on after I left, but at that point I wasn’t really central to it.  It became an engineering project.  It was certainly a lesson that I learned for future work; namely that one can’t have a half-cooked idea and then take it immediately to the endpoint.  New ideas have to be worked up slowly. Again, it was Shockley’s mistake – I had little autonomy at that point.  He was out to beat the world in the shortest possible time.  He just wasn't patient.