12/12/09

Diodes,Transistors,LED's - Reminiscences

I once had a website in which I recounted some early engineering and manufacturing experiences with diodes , transistors and LED's. The website was inspired by an article in the Plain Dealer on 11/9/2001 which pointed out that "when two Bell Labs scientists invented the transistor in 1947, it was as tall as the face of a wristwatch. Now , another Bell team has made a transistor from a single molecule - small enough to fit about 10 million on the head of a pin."

Reading this article brought to mind the early days of semiconductor engineering - germanium diodes, rate grown transistors, junction transistors, zone refining, Czochralski single crystal growth, float zone single crystal growth, germanium vs. silicon as a material. The website is long gone but my reminiscences still haunt me, thus this lenghty post.
.

(For some vintage photos visit http://picasaweb.google.com/asrmjr/Semiconductors?feat=directlink )
===========================================


I remember a young engineer in 1948 whose interest in materials drew him into the pioneering days of semiconductor technology. I also remember a more mature individual who ventured from the world of diodes and transistors into the world of Solid State Lamps (GE Lamp Division's name for Light Emitting Diodes).

My introduction to semiconductors was the germanium whisker diode. Interestingly, GE chose to pilot the manufacture of germanium diodes in their Quartz Crystal operation at Electronics Park in Syracuse , N.Y. What the two products had in common was the equipment used for slicing quartz crystals and germanium billets. I use the term billets in reference to germanium rather than crystals for in 1948 germanium single crystals had not found their way into semiconductor manufacturing. The germanium pellets used in diodes were polycrystalline. Material preparation basically consisted of the reduction of germanium dioxide to germanium melts followed by a process which was known as progressive crystallization. The latter was designed to purify the melts by directional cooling. Progressive crystallization was later replaced by a more effective process called zone refining.

The point about polycrystalline germanium was made to recount some primitive engineering that was attempted. Shortly after the announcement of the transistor by Bell Laboratories we attempted to make a few transistors in our operation. Since single crystal germanium was required we modified the progressive crystallization process so as to produce billets with a larger than usual crystalline structure. The task then was to cut out "single crystals" from the germanium slices or wafers. Our quartz cutting specialists were equal to the task. A few working transistors were actually made using some of the quartz crystals metal cans.

GE's millionth diode was produced in Syracuse , N.Y. in May of 1950 . Shortly thereafter a manufacturing plant was established in Clyde, a small village in rural upstate New York. The village fathers proudly proclaimed Clyde " The Diode Capital of the World". Clyde would also be the manufacturing site for rectifiers - notably the controlled rectifier. (The Clyde Plant will always be special to me . I met my lovely wife in Clyde. Our first son was born there.) The pilot line operation in Syracuse evolved into the Semiconductor Products Department. The Semiconductor Products Department in Electronics Park , Syracuse would become the focus of transistor engineering and manufacturing. In later years it would establish a manufacturing facility in Buffalo , N. Y. and eventually in Dundalk, Ireland .The Clyde Plant became essentially a rectifier production facility with transistor manufacturing being concentrated in Syracuse, Buffalo and Ireland. Eventually the semiconductor business was split into the Semiconductor Products Department with headquarters in Syracuse and the Rectifier Products Department with headquarters in Auburn, N. Y. In the late 60's both departments were once again merged into a single Department, namely; the Semiconductor Products Department.

After more than fifty years, the details of products and processes escape me. Even dates in some cases become hazy. The internet is full of websites recounting the early days of semiconductor manufacturing and the history of the transistor. Some suggestions are http://www.pbs.org/transistor/, The Transistor Museum and Addison Sheckler's recollection of GE's role in semiconductors.
Getting back to diodes, I experienced my first lesson in the importance of product quality when a major problem developed in the field. It came as a shock to learn that diodes were "opening up". It turned out that under humid conditions, moisture was leaking thru the bakelite case and caused the whisker to separate from the pellet. The fix? Impregnate the diode with wax in a pressure cooker. Messy? Absolutely! A better fix - hermetically seal the diode.

On the more humorous side there was the problem of how to dispose of the reject (there were a few) diodes . The Clyde Plant Manager came up with a solution. He was putting in a new driveway and needed some fill. The reject diodes found their final resting place on Galen Street in Clyde. Why were there rejects? Device design engineers complained about the quality of the germanium or silicon and of course the materials engineers pointed out that poor design was perhaps a larger factor. Thus was born the constant struggle between device and materials engineers. Looking back it was stressful but rewarding in that from these struggles evolved better products.

I referred to the quality of germanium and silicon as an important factor in device characteristics. This brings to mind the germanium versus silicon debate which took place. Unfortunately for GE, those responsible for its engineering decisions stuck with germanium far too long. Little known Texas Instruments leapfrogged into silicon transistors and usurped the leadership position GE had long held.



Looking back, whether it was transistors, integrated circuits or light emitting diodes GE never had the entrepreneurial leadership which semiconductor technology demanded. In spite of outstanding research and inventions from the likes of Bob Hall, Crawford Dunlop, Hub Horn, John Saby , Nick Holonyak and others the early lead GE had was squandered and eventually GE was no longer a significant producer of semiconductor devices. Not withstanding the eventual demise of the GE semiconductor business the early years were exciting.

Noteworthy was participation in the Signal Corps Industrial Preparedness contracts in the mid fifties. Contracts were awarded to prepare for the manufacture of specified semiconductor devices meeting the requirements of military specifications at specified rates of production, and to reduce the time required for the delivery of devices in large quantities in the event of an emergency. Concurrently there was the Air Force Production Refinement Program which played a large role in the engineering of silicon production refinement processes and equipment. Government support was vital in accelerating the growth of the semiconductor industry.

A byproduct of the Production Refinement and Industrial Preparedness contracts was the realization that a successful semiconductor business could not be built solely around "top hat" or metal can transistors. Consumer electronics required a much lower cost product. A brief period of entrepreneurship took place and GE came up with the plastic transistor design together with high volume production equipment. To further reduce cost, manufacturing was moved offshore to Ireland - a novel idea in those days.

Perhaps one of the most rewarding experiences during the heyday of GE's semiconductor business was participation in the various licensing agreement GE had developed with foreign firms. Interaction with representatives of firms such as Telefunken, Siemens, French Thompson - Houston and Sony was fulfilling both professionally and socially. The highlight of this precursor to globalization was my experience with the Sony Corporation. Sony had made a bold entry into the consumer electronics market and was eager to soak up any semiconductor technology which GE had to offer. ( A visit to the Sony Corporation in 1958 is covered in a previous post - How Times Have Changed )


Another sidelight of the Sony experience was being introduced to Leo Esaki who had recently presented a paper on tunnel diodes. I later had the pleasure of entertaining Mr. Esaki at my home. Little did I know that in 1973 Leo Esaki would be awarded the Nobel prize in Physics for his discovery of tunneling in semiconductors.

A more humorous but again indicative aspect of our visit occurred when we were invited to Akio Morita's home for dinner. We were expecting a traditional Japanese dinner and were surprised to be served Salisbury steak cooked by Mr. Morita's wife.

Our visit to Sony culminated in a meeting with Sony's Board of Directors where we were asked to comment on Sony's semiconductor progress. A flattering event for two young American engineers.

By 1971 economic clouds hovered over GE's semiconductor business. Domestic manufacturing was almost non-existent. The Clyde Plant had been shut down and the building was donated to the Village of Clyde ( a grateful village named a street after Berkley Davis , the GE VP who reigned over the semiconductor business ). The Buffalo Plant would cease operation in 1972 and the Auburn Plant (rectifier products) was on it's way out. In addition SPD's foray into integrated circuits proved to be non-spectacular. In short GE's commitment to semiconductor production had waned considerably.

About this same time GE's Lamp Division in Cleveland, Ohio was piloting Solid State Lamps a.k.a. light emitting diodes or LED's. Nick Holonyak had invented the first visible spectrum light emitting diode while at GE's Electronic Laboratory in Syracuse, NY. The Semiconductor Products Department within whose charter fell light emitting diodes never was able to commit the funds to develop a manufacturing capability for LED's. It gallantly acquiesced to the manufacturing and marketing of LED's or Solid State Lamps by the Lamp Division. Amongst the rank and file it was rumored that the Lamp Division had deep pockets and would undoubtedly be successful with its Solid State Lamp Project. So it was off to Cleveland to be on the ground floor of another semiconductor venture. Indeed by 1972 a manufacturing facility was established in Chesterland, Ohio.

Almost as suddenly, the plant was shuttered in 1975. The deep pockets were gone! The Lamp Division was not as independent as it once was and had to more closely adhere to the dictates and management philosophy of "headquarters". The Solid State Lamp project was not cutting it and was dragging down the bottom line of it's parent department. Why? Trying to do too much too fast in order to catch up with more established producers of LED's, the failure of mass production equipment designed and built by people unfamiliar with smiconductor technology, insisting on using gallium phosphide rather than gallium arsenide


At the onset I pointed out that these reminiscences were inspired by an 11/9/2001 Plain Dealer news story announcing that a Bell Labs team had made a transistor from a single molecule. These reminiscences will end on an ironic note - on 9/26/2002 The Plain Dealer ran another story under the banner of "Bell Labs Research Rings False". Apparently the breakthrough claimed the previous year was based on fraudulent data . Quite a comedown for the birthplace of the transistor. My reminiscences? Fraudulent only to the extent that some of the details may have become a bit clouded
EPILOGUE
Never in my wildest dreams did I envision L.E.D’s illuminating the Empire State Building. Such is the case per a New York Times story on 04/20/2007 reporting a contest that “bathed the building in “intelligent illumination,” employing a new generation of computer-controlled high-brightness light-emitting diodes — or L.E.D.s — that are capable of producing millions of different colors and an infinity of patterns”.

(GE making big move into LEDs for homes , GE's Hate /Love Affair With LED's, bring the GE position on LED's up to date.)

On 11/30/2006 it was announced that French telecom gear maker Alcatel SA completed its purchase of U.S. rival Lucent Technologies Inc. for $13.4 billion in stock. Bell Laboratories is part of Lucent. The headquarters of the new company is in Paris. Thus France has become the foster parent of the laboratory that gave us the transistor. How sad!

No comments: