06/26/2008
Passing of a Pioneer
For me, the most exciting news of the past week is that the Phoenix Mars Lander has indeed found convincing evidence that it has achieved one of its objectives, confirming the presence of ice on Mars. On June 15, Phoenix exposed some dice-size chunks of whitish material in the trench they’ve dubbed “Dodo- Goldilocks”. The Phoenix team couldn’t be sure that the chunks were ice, a salt being another possibility. However, when they looked again on June 19, some of the chunks were gone, as anticipated if the water ice sublimed into vapor on exposure to the Martian atmosphere. A NASA/JPL news release on June 19 quotes Phoenix Principal Investigator Peter Smith of the University of Arizona, “It must be ice. These little clumps completely disappearing over the course of a few days, that is perfect evidence that it’s ice. … Salt can’t do that.”
Last month marked the passing of one of the early pioneers of the transistor era at Bell Labs. Morgan Sparks, who died at the age of 91, was my first boss when I came to Bell Labs in 1952. Without the transistor and its electronic offspring, notably the ubiquitous silicon chip, Phoenix and the many other NASA spacecraft missions would not have been possible. Our Editor, Brian Trumbore, called my attention to Sparks’ obituary in the May 8 New York Times. The picture of Morgan in the Times reminds me somewhat of famed actor Alan Alda. In the book “Crystal Fire” by Michael Riordan and Lillian Hoddeson, Sparks is described as a “handsome, low-key, soft-spoken man with close-cropped hair and deep smiling eyes” who charmed William Shockley’s secretary and married her in 1949. The marriage lasted 57 years until her death in 2006.
Until I read Sparks’ obituary, I didn’t realize we shared a number of things in our backgrounds. We were both born in Colorado and both of us skipped two grades, Morgan first and fifth while I skipped first and twelfth. We also both worked on batteries at Bell Labs – he worked on them before his semiconductor efforts while my battery work followed work on semiconductors. To appreciate Sparks’ major contribution to the transistor, let’s take a look at some of the intrigue surrounding the invention of the transistor. I’ve probably written before about some of what follows but hopefully there will be additional bits of interest. Crystal Fire is a wonderfully detailed account of those early days and of the personalities involved.
The first person I was introduced to upon arriving for work at Bell Labs was Bill Shockley. Completely out of my league in meetings involving Shockley and other superstars of the semiconductor field, it was clear to me that Shockley was not only brilliant, but also had a substantial ego. Five years before I arrived at Bell, in 1947, Shockley headed a group working on semiconductors. In December of that year, two members of his group, John Bardeen and Walter Brattain, put some point contacts down on a piece of germanium and observed amplification of an electrical signal. They had invented the transistor. Shockley was of the opinion that work he had done a few years earlier proposing a so-called “field effect” paved the way for the transistor and that he should share in the patent.
However, the Bell patent attorneys turned up an earlier 1930 patent by Polish-American immigrant Julius Lilienfeld; that patent also involved some sort of “field effect”. The attorneys concluded that if they filed the transistor patent application citing Shockley’s earlier work, the patent would be denied! Bardeen and Brattain’s transistor was clearly new and novel and Shockley was crushed to find Bell’s patent department arguing against his inclusion. He essentially withdrew from working with Brattain and Bardeen and furiously began working on the theory of another approach to making a transistor. He published a groundbreaking paper on his work and proposed a new sandwich structure type of transistor.
We’ve talked before about p-n junctions, where p-type material has a deficiency of electrons (hence is positive or p-type) while n-type material has a surplus of electrons (hence, negative or n- type). One sandwich structure suggested was an n-p-n sandwich, in which a very thin p-layer is like a slice of bologna between two slices of rye bread (n-type). Instead of the point contacts, Shockley proposed making a sandwich of germanium in which the current would pass through the crystal. There was some question then as to whether the current could actually pass through a crystal of germanium. Bardeen apparently had his doubts, feeling that all the current travels on the surface.
Here’s where Morgan Sparks entered the picture. In addition to marrying Shockley’s secretary, Sparks took up the challenge of actually making a sandwich p-n junction transistor. By April of 1949, he and his technical assistant, Bob Mikulyak, made a structure that was not a sandwich but did show transistor action of the p-n junction type that Shockley predicted. (Bob was an interesting individual I’ve mentioned before. I heard my first stereo sound in his home on stereo speakers he made himself. He also for years printed our Christmas cards using the drawings made by our cartoonist, Harry Trumbore.)
One of the strengths of Bell Labs back in those golden years was the ability to reach out and cooperate with experts in other areas to accomplish an objective. Morgan Sparks reached out to Gordon Teal. Teal and John Little had bootlegged a crystal pulling effort, receiving no encouragement from Shockley for their efforts. Teal and Little wanted to grow single crystal germanium by constructing a crystal puller based on a 1917 idea of a little known Polish scientist named Czochralski. On a bus between Bell Labs and Summit that I was to ride for many years, Teal and Little sketched up a model for a crystal puller and I am amazed that only two days later they had constructed a puller using a bell jar and an induction heating coil Little happened to have. They soon pulled their first crystal by dipping a seed crystal into molten germanium and slowly pulling the seed out of the melt.
Morgan Sparks got together with Teal and they collaborated on growing and measuring p-n junctions made by adding doping materials during the growth of the crystals. Shockley finally realized his mistake and Teal and Little got a lab and a wonderful, talented guy, Ernie Buehler, to handle the crystal pulling. By March and April of 1950, Sparks and Teal had made their sandwich n-p-n transistor and its electrical properties were in very good agreement with the predictions of Shockley’s theory. Their big challenge was to get that p-type “bologna” in the middle of the sandwich very thin. They finally achieved this as they were growing n-type germanium by dropping a little pills of germanium doped with gallium into the melt to convert the growing crystal to p-type and then, only ten seconds later, dropping in a pill of germanium doped with arsenic to turn the germanium n-type. This gave them a p-type layer of only 30 mils (thousandths of an inch) thickness.
The achievement marked the beginning of the end for the point contact transistor, which was being produced and supplied to various companies and researchers. The devices were electrically noisy and the electrical characteristics were not reproducible from one batch to another. The 30-mil thickness of the p-layer was still not thin enough but the achievement of the sandwich structure (and its electrical properties agreeing with Shockley’s theory) was a feasibility experiment that helped pave the way for the zillions of p-n junction devices that populate our world today.
As for Sparks, he was sent to New Mexico in 1972 to be the president of Sandia National Laboratories, then managed by AT&T. Following his retirement in 1981, he became dean of the Robert O. Anderson School of Management at the University of New Mexico. I remember him as a very decent man and a boss for whom I had the utmost respect.
Allen F. Bortrum
|