|
The excitement started with a masterpiece of understatement in the prestigious Physical Review of July 1948. J. Bardeen and W. H. Brattain wrote " A three element electronic device which utilizes a newly discovered principle involving a semi-conductor as a basic element is described. It may be employed as an amplifier, oscillator, and for other purposes for which vacuum tubes are ordinarily used". Here was the promise of an amplifying device that could replace the vacuum tube with its problems of size, power requirements, heat removal and limited life.
In retrospect, the rewards were greater than could be imagined in 1948. So were the problems and challenges that had to be overcome en route!
Si Ramo, Dean Wooldridge and Burt Miller of Hughes Aircraft Company appreciated the potential of semiconductors for advanced electronic systems. In 1948 they hired Dr. Harper Q. North, previously of the MIT Radiation Laboratory and of General Electric, to fulfill the possibilities of semiconductor devices. Starting with a lab bench in the corner of the Advanced Electronic Development Laboratory, Harper North built up a semiconductor operation that was essential to the Falcon missile program and to many following civil and defense systems.
In April, 1949, Dr. North hired a young man named Sanford H. (Sandy) Barnes, whose assignment was to duplicate BTL’s point-contact transistors. This was no easy task, because Bell told the world of the design, features and performance of transistors, but almost nothing about how to make them.
Bell’s coaxial point-contact transistors appeared ideal for Hughes systems applications, with possible advantages in geometry, ruggedness and circuit isolation. Sandy Barnes’ only roadmap was a two paragraph "Processing Procedure" section in an article "The Coaxial Transistor" by W. E. Kock and W. R. Wallace Jr. in the Bell Laboratories Record (Reproduced in this issue on page 19).
Barnes worked for six months in developing a useful double surface transistor. No single crystal germanium ingots were available, so he searched polycrystalline ingot slices for single crystal areas. This chip preparation process consisted of lapping germanium slices to a thickness of a few mils, then searching for single crystal areas. Small chips encompassing these areas were then sawn out of the slices, and mounted in fork-like fixtures. Dimples were then milled in opposite surfaces of the chip, and appropriately etched to remove surface damage. This was a very difficult process, with resulting base (germanium thickness) regions ranging from mils to complete punch-through. Some units had base regions of one mil or less in thickness, satisfactory if the germanium was indeed single-crystal and if the surfaces were damage-free.
Another problem was that the emitter and collector catwhiskers had to be directly opposite each other for proper transistor action. On top of all that, whisker contacts had to be moved slightly to find the ideal "sweet spot", since the crystal surfaces were not uniform. Electrical characteristics were observed as contacts were moved this way and that, not much different from tweaking in the old crystal set!
The process approached feasibility with jig-fabricated devices, which were potted in bead-type or "bathtub" packages. Coaxial packages would introduce a new dimension of difficulty, since one would lose the flexibility of diddling the opposed contact whiskers.
The Hughes opposed-contact program was suspended six months after it was begun. Hughes had learned all that it cared to know about opposed-contact transistor and their problems. Sandy Barnes moved on to become co-inventor of the subminiature glass diode that became the industry standard, and still is.
After the pivotal BTL-Western Electric transistor symposium of September 1951, Hughes was very anxious to embrace the junction transistor, with its great future potential. A patent agreement was consummated with Bell. A Hughes Semiconductor Division team was set up to absorb Bell’s junction transistor technology.
Hughes already had very good semiconductor chemical and metallurgical research capabilities. A good rapport existed with Bell Labs specialists. Morgan McMahon, device research engineer, was tasked to breathe life into the junction bar devices, and to determine their electrical properties.
A true fact of life quickly became apparent to the Hughes transistor team; generalized, or even specific, written processes were usually not adequate to duplicate semiconductor devices. There was an aura, a plasma, that was indefinable yet essential to the making of successful transistors.
Hughes’ laboratory work started with tiny n-p-n germanium transistor "bars" furnished by Bell. Morgan McMahon etched each bar, then attached leads to the n-type emitter and collector bar ends. After affixing the bar to a test jig, he contacted the base region with a p-doped gold wire which he electrically pulsed to establish a base lead connection. After a clean-up etch and rinse, McMahon passivated the junction regions with a substance called "glit" (a mysterious substance rumored to be made of glue and something unmentionable in print). If the transistor was worth keeping, it was potted and re-tested.
The process seems straightforward, but many late nights and weekends were spent in getting rid of process "bugs". Hybrid parameter testers had to be built since none were commercially available. Subtleties of etch-rinse processes had to be nailed down in order to make healthy devices. Attachment of the base lead was very critical, with a positioning and pulse process that defied optimization. One problem was that the base lead was larger than the width of the base region.. Perforce, the base contact region would have to overlap the collector region, yet be such that it didn’t ruin the electrical characteristics and carrier injection efficiency of the emitter. The waveform of the base lead pulse was itself a study in metallurgy, pragmatism and frustration. All this presumes that the n-p-n germanium bar was good to start with, which might or might not be true.
The first junction transistors fabricated in the Hughes lab were undeniably bad. Real bad! Then, as processes were improved, it appeared that there might be transistor action going on, but masked by other problems. Then suddenly it happened; an emitter-collector current gain of 0.85 was observed (1.0 was the theoretical maximum). The project was on its way toward gains of 0.98. Then came the next big challenge; making transistors from germanium bars made in the Hughes labs, rather than donated by Bell. This was accomplished over a period of some months.
Although the project was successful, it never went into production. Here was a lesson that must be heeded in the semiconductor field; if you take the time to duplicate a device and processes developed by someone else, there’s a good chance that some other technology will render your efforts obsolete by the time you get there.
Neither the Hughes point-contact transistor nor the bar-type transistor went into production. Yet, these projects more than earned their way by developing experience, skills and teams that helped propel the semiconductor industry to its fabulous success.
About Morgan McMahon
Morgan E. McMahon received his BSEE and MSEE degrees at UC Berkely in 1950 and 1951. He helped build Hughes Aircraft’s semiconductor Division in the 1951-54 period, advancing from research engineer to manager. During the same period, he co-organized and taught the first transistor electronics course in the West, at UCLA.
Mr. McMahon then joined the team setting up Pacific Semiconductors, Inc., which grew into the TRW Electronic Components Group with 20,000 employees. Mr. McMahon was Group Technical Director at the time of his retirement in 1986. During the same time he wrote and published the Vintage Radio book series. These books have become the prime references for radio historians.
Mr. McMahon is now a Disaster Emergency Services worker with the Red Cross as well as an active radio-electronics historian.
EXPLORATORY DEVELOPMENT CHARACTERISTICS FOR THE BTL M1752 TRANSISTOR
The BTL M1752 transistor is an NPN junction transistor
which can give high power gain at high efficiency, even at very low operating voltages and currents. In addition, the noise
figure of the device is very much lower than that of point contact transistors of conventional design.
A limited number of development models of the BTL M1752 have been made for preliminary study in several possible
circuit applications. Figure 18-1 shows outline dimensions and lead arrangement of these models, and a development
specification is attached.
SPECIFICATIONS BTL M1752 TRANSISTOR
Reference - JAN 1A
Description - Transistor
Ratings: Vc Ic Ve Ie Load Dissipation Ambient Temp.
Pe Pc degrees C
Units Volts mA Volts mA ohms milliwatts
Maximum *+50 5 5 50 50
Test Conditions
Small Signal
+4.5 1.0 Approx. 25
Dimensions: per Outline Dwg. Fig. 18-1
Connection: per Outline Dwg. Fig. 18-1
Test Conditions Min. Max.
Handling Note 2
Holding Period 48 hrs., Note 5
Vibration and Shock Note 3
Collector Voltage 0 +50 volts
Input Resistance - 1000 ohms
Although the project was successful, it never went into production.
Here was a lesson that must be heeded in the semiconductor field; if you
take the time to duplicate a device and processes developed by someone
else, there's a good chance that some other technology will render your
efforts obsolete by the time you get there.
Neither the Hughes point-contact transistor nor the bar-type transistor
went into production. Yet, these projects more than earned their way by
developing experience, skills and teams that helped propel the semiconductor
industry to its fabulous success.
Test Conditions Min. Max
Output Resistance 100,0000
Current Amplification
Factor 2 .95 1.0
Note 1 Subscripts c and e refer to collector and emitter respectively. Voltages are measured with respect to the
base. D-C ratings are on the basis of any duration longer than about 5 microseconds. Transients of shorter
duration somewhat in excess of the ratings may not injure the transistor.
Note 2 The transistor should not be subjected to excessive transients such as may occur on plugging in or out with
power on. This may be done if in any case the ratings are not exceeded; base contact should be made first.
When solder connections are made, heat sink protection on the transistor side of the joint should be provided,
as with flat-nose pliers.
Note 3 The same degree of ruggedness experienced with germanium diodes may be expected.
Note 4 For ambient temperatures from 50 C up to 80 C the maximum collector dissipation should not exceed 25
milli-watts.
Note 5 All tests should be made after holding period.
Reference F3, JAN 1A.
Note 6 Operating point for measurement of circuit parameters is Ie = 1 mA, Vc = +4.5
K. D. SMITH
SMEC UPDATE
This spec sheet was done by K.D. Smith for the book that resulted
out of the 1951 Transistor Symposium. This is also referred to as the $25,000
book by Morgan McMahon in his article that appears in this issue on the
symposium.
We would like to thank Bob Ryder for sending us this book!
This book, that started the world out on the new era is here in the
museum for your viewing pleasure!
|
