Dr. Bortrum

---

Dr. Bortrum

07/02/2011

Impurities and Other Matters

When we moved to New Jersey and Bell Labs, we settled in a garden apartment complex, Leland Gardens, in Plainfield. Another Bell Labs physical chemist, Morris Tanenbaum, and his family lived in a nearby garden apartment development. Morry and I occasionally golfed together and our wives knew each other through pushing baby carriages down the street etc. For a brief period, Morry would be my boss in the Metallurgy Department at Bell Labs. I lost track of Morry as he took off in the high management ranks of Western Electric, New Jersey Bell Telephone Company and finally, AT&T, where he became Vice Chairman and Chief Financial Officer. We got back together some years ago and I had lunch with him recently. At lunch, I learned that I should correct something I wrote in a recent column in which I mentioned that Gordon Teal and his team at Texas Instruments made the first silicon transistor.  

Morry, working with Ernie Buehler at Bell Labs, had also made a silicon transistor. Ernie was a crystal grower par excellence and managed to grow a silicon crystal in which the impurity distributions in the crystal suggested the possibility of making a transistor. Morry made a device using a sample cut from the crystal and it did indeed perform as a transistor. This was in January of 1954. However, it was not until 1997, with the publication of the book "Crystal Fire" by Michael Riordan and Lillian Hoddeson, that Tanenbaum found in the book that Teal's transistor was fabricated in April of that year. Hence, Morry actually had the first silicon transistor.  (In 1997, the 50th anniversary of the invention of the transistor, Bell Labs had a big anniversary celebration at which Riordan appeared and we attendees all received signed copies of the book. Morry called Riordan's attention to the matter and, in an article by Riordan published in IEEE Spectrum in May 2004, Riordan credits Morry with being the first.) 

According to Morry, neither his nor Teal's silicon transistors were good transistors and it would await Morry, working with Cal Fuller, a diffusion expert, to fabricate the first "diffused- base" silicon transistor in 1955. Diffusing impurities into slices of silicon to form p-n junction structures was vastly superior to trying to control impurity distributions in a crystal pulling machine. It was news of this diffused-base transistor that caused the head of Bell Lab's development area, Jack Morton, to cancel the rest of a trip in Europe in order to return and order the immediate dropping of all germanium work and work on anything other than diffusion to make silicon transistors. Silicon was on its way!  

All this does not diminish the fact that Texas Instruments plunged ahead with their grown-crystal silicon transistors and for several years led the way in volume production of these devices. The military branches were the leading customers for these transistors and years later, Jack Kilby at TI would invent the integrated circuit, leading to the ubiquitous silicon chip and a Nobel Prize. (Robert Noyce, then at Fairchild Semiconductor, is credited with independently inventing the integrated circuit but died before the Nobel Prize was awarded to Kilby.) 

The above discussion gives a flavor of the revolutionary semiconductor technology innovation going on during my early years at Bell Labs. Let me get back to my own, less earth-shaking memoirs. Having survived a near-disastrous first year or so (see last month's column), I was off probation and settled down to what would become a more than 36-year career at Bell Labs. Coming from Cleveland to the New York metropolitan area was somewhat of a cultural shock. For example, many native Jerseyans and New Yorkers considered Pittsburgh and Cleveland to be distant lands "Out West". You've probably seen that famous map of the USA showing New York writ large with everything west of the Hudson River compressed into a small area. One of the things I dreaded on moving to New Jersey was the idea of driving in the horrible traffic in New York City but I quickly found myself driving to Columbia Presbyterian Hospital and then into the heart of Manhattan to attend the theater.  

My introduction to Broadway was attending a performance of the play "My Three Angels" starring Walter Slezak. I really loved Slezak's comedic performance as one of the prisoners in French Guiana. In fact, when my former lab partner at NACA, George Fryburg and his wife Mary Lou came to visit I got front-row tickets to Angels and, in the greatest parking achievement of my life, found a free parking spot right on Times Square. Our guests weren't nearly as impressed with the play as I was and Mary Lou slept through most of the performance. Not only that, but we took them to Mama Leone's, which at the time we considered an upscale dining establishment, for a post-play repast and Mary Lou ordered a hamburger! 

I was enchanted by Broadway and I still treasure performances such as those by Vivian Blaine and Stubby Kaye in "Guys and Dolls" and Judy Holliday in "Bells Are Ringing". I remember them for the songs they sang but there was one performer in those days who stands out in my memory. That was Mary Martin in "Sound of Music". I don't know if it was the lighting or not, but when she first came on stage there was a glow in her eyes and a radiance that I'll never forget - she had me without a word being spoken or a note being sung. 

Back at Bell Labs, I was beginning to find a niche in the field of impurities in germanium and silicon. Making a transistor requires exquisite control of the amounts and distributions of impurities (dopants) in germanium or silicon. Extremely pure germanium or silicon was required and a fellow named William Pfann had developed a most elegant, yet simple way of purifying these semiconductor materials called zone refining. To understand zone refining, let's consider something called the distribution coefficient. The distribution coefficient is simply the ratio of the concentration of an impurity in the solid phase to that in the liquid phase. For most impurities, it's significantly less than 1.  

Let's say that we have a long ingot of germanium in which one out every hundred atoms (1%) is an impurity X. Let's assume that the distribution coefficient is one-tenth (0.1). Now suppose we pass that ingot through a setup with a heating coil that creates a molten zone as the ingot passes through the coil. As the ingot emerges from the coil and freezes, the solid germanium now only contains 0.1% of X impurity, while the molten zone carries away the rest. Pass it through another hot zone and the solid contains only 0.01% X. Make an apparatus with a bunch of hot zones and Voila! You've got a helluva pure material! Pfann was able to make germanium with only parts per billion or less of some various. Most impurities have distribution coefficients significantly less than 0.1.  

I began a series of experiments in which I would grow crystals of germanium or silicon doped with various impurities under various conditions at different temperatures and would then collaborate with Bell Labs' wonderfully talented analytical chemists who would analyze the resulting crystals for the impurity of interest. I would also interact with physicists who could make various optical and electrical measurements that correlated with the chemical analyses. Aside from crystal pulling, I often used a thermal gradient method to grow crystals. This process involved a simple principle - namely, that the solubility of germanium or silicon in a particular impurity, say molten tin, is higher at higher temperatures. I would seal tin and an excess of germanium in a fused quartz tube and place the tube in a furnace where the top of the tube was hotter than the bottom. The tin would melt and the germanium would float in the molten tin to the top of the tube, which was hotter than the bottom of the tube. The germanium dissolved in the tin and precipitated out as a crystal at the cooler bottom of the tube. This was a slow process and in one of my papers I showed a picture of one crystal that took three months to grow. 

This work culminated with a review paper with the stirring title "Solid Solubilities of Impurities in Germanium and Silicon "published in January 1960 in the Bell System Technical Journal (yes, Virginia, there once was a Bell System). In that paper I reviewed all available data on germanium and silicon and in the paper had two curves, one for silicon and the other for germanium, that summarized the so-called solidus curves for impurities in Ge and Si. These curves were of interest to those involved in fabricating devices and were apparently pretty popular all over the world, not for any intellectual achievement on my part!  

Well, that fairly well covers my main research efforts in the 1950s. On the home front I should note one quite significant event that is responsible for you reading this column. I'm speaking of the birth of our second son, Brian Trumbore, in 1958. Brian, of course, is the founder and editor of this stocksandnews.com Web site. We were still living in Plainfield at that time and were to move out of our garden apartment into a house we purchased for $20,000. My wife was working part time as a nurse in Muhlenberg Hospital, which recently was one of those hospitals closing its doors in the current economic climate. She also was becoming increasingly involved in painting and would take our other son, Harry with her painting "on location".  Even then, Harry, creator of the Lamb cartoons on this site, was showing signs of artistic ability. I found it interesting that when he would draw a figure of a person he typically would start with drawing a foot first. 

Finally, I must congratulate our editor Brian on his sparkling 2-over-par 29 on our local 9-hole par 3 course this week. Sadly, my 36 that day was actually one of my better rounds this year.  

---