IP IEEE ht&natimal Conference on Advanced Thermal Processing of Semiconductors- RTP 2002

HISTORY OF SEMICONDUCTORS DIFFUSION

ENGINEERING
B. Lojek
ATMEL Corporation
1150 E. Cheyenne Mtn. Blvd.
Colorado Springs, CO 80906
[email protected]

When creative men started working on

.. INTRODUCTION
CHRONOLOGY OF SEMICONDUCTORS
semiconductors by the late thirties and integrated
circuits at the end of the fifties, they did not
know that they were going to change the lives of

. DIFFUSION E N G I “ G
SEMICONDUCTOR DIFFUSION
ENGINEERING AT Bell Telephone
future generations. Very few people at that time
recognized the significance of perhaps the most
important invention of the century. Nobody

. Laboratories (BTL)
PROBLEMS WlTH APPLICATION OF
noticed that the key people behind the inventions
were frequently frustrated and disappointed.

. DIFFUSION TO TRANSISTOR AT BTL

DIFFUSION TRANSISTOR AT BTL &
Who remembers today names such as Russell
OM, Karl Lark-Horovitz, William Shwkley, Carl

. WESTERN ELECIRI(I. (ZN560)

DIFFUSION TRANSISTOR AT
Frosch, Lincoln Derick, Calvin Fuller, Kurt
Lehovec, Jean Hoerni, Shelton Roberts, Jay Last,

. S H W Y SEMICONDUCTOR
DIFFUSION “SISTOR AT
Tom Longo, and Dave Talbert?
In the beginning of the sixties the editors of

. FALRCHILD (2N696 and 2N697)

FIRST INTEGRATED CIRCUIT AT
Time-Life Books in Alexandria, VA published
“A golden Age of Entrepreneurship” with a
photograph (Fig. 1) accompanied by legend

. FAIRCHILD
FIRST INTEGRATED CIRCUIT AT
“1958-1959 Robert Noyce, Jean Hoerni, lack

.
Kilby, and Kurt Lehovec all took part in
TEXAS INSTRUMENTS developing the integrated circuit”.
FIRST INTEGRATED CIRCUIT AT

. SYLVANIA
COMMERCIALLYAVAILABLE

. DIFFUSION FURNACES (1960-1965)

CONCLUSIONS

PROLOGUE

On July I , 1948 The New York Times on page Fig. 1. Inventors of integrated circuit as they
48 in “News of Radio” section printed an were recognized in the early sixties.
announcement of NBC that it would broadcast
Waltz Time on Friday nights. The same section In Jack Kilby’s speech to The 2000 Nobel Prize
contains a brief report about a new invention, “a Committee the names reduced to just Robert
device called a transistor, which has several Noyce despite the fact that Hoerni and Lehovec
applications in radio where a vacuum tube ideas were so much more practical than Kilby’s
ordinaril.1; is employed. ’’ that even Texas Instrument adopted them.

0-7803-7465-71021$17.0002002 lEEE 209

Neither Hoerni nor Lehovec had the backing of a Research of human behavior consistently reveals
large company. PR f m s plant about 40% of all a significant difference between creative and
stories, which appear in newspapers, as a so- uncreative people, and sheds light on why highly
called Press Release. Because the radio and TV creative individuals frequently cause trouble.
news re-edit newspaper stories, a substantial Creative individuals exhibit more unusualness of
portion of the public’s “news” originates in PR thought processes and mental content, they are
information. Naturally the connection to the PR less constrained by conventional expectations,
source is edited out. Naturally some people who and they are less concerned with making the
created history are edited out. right impression on others. Highly creative
individuals do not respect common practices.
Creative individuals, not business leadership or Their methods, style, authoritarian control, and
government initiative, was the most important temperament are frequently at odds with
factor in creating the new device. The current conventional norms.
“politically correct” society and PR departments There are also differences in how these two-
of large companies alter historical events, and groups process information. Creative individuals
present a development of what is today called have a wider range of attention - they can think
microelectronics as a systematic effort or of more things at the same time than less creative
exceptional leadership. Many modern college people. They are also more open to new
business textbooks are listing companies such as information and willing to take a higher level of
Fairchild Semiconductor Corporation and risk. Creative people usually have a very deep
Geophysical Service Inc. (later re-named to knowledge of their subject. This high level of
Texas Instruments) as examples of exceptional expertise can very frequently lead to problems
business wisdom. The author of this article when the mind becomes “set”. Once their mind
would like to offer a different view of the history is set, creative people become persistent and try
of microelectronics. My view is based on to overcome any obstacle. The outcome of such
personal experience as a Diffusion Engineer for activities can result in phenomenal discoveries or
40 years. I experienced, unfortunately, too many colossal failures.
times, that the company establishment was Creative people are introverted, independent,
frequently one of the biggest, if not the biggest arrogant and hostile. Creative people are over-
obstacle, which needs to be overcome in the reactive. Highly creative individuals are usually
introduction of new ideas. high-maintenance employees. Creative people
When I was a younger engineer I was stupid are driven with a strong need for achievement
enough, not to look back in history. From history and they have the self-belief and energy to
I could have learned that certain human behavior challenge the practice of the system and their
has a character of axiom, and will never change. managers. Such behavior may become
My recollection may be biased. I met and increasingly disruptive. The highly competitive
worked with people I did not like, and I treated and ultimately detrimental interaction between
them accordingly. On the other side, I worked strong personalities eventually results in
with a much bigger group of individuals who destroying the system or the organization or
made my life extraordinilly and many of them more frequently in the separation of the creative
became my role models. person from the organization and creating a new
I regret that I did not save more historical spin off organization. This new organization is
material, did not ask more questions or did not set up by rules defined by a creative person that
spend more time with people who were has in the beginning top-notch knowledge of a
characterized as troublemakers, whistlehlowers particular problem. As the business succeeds and
or difficult person, and for whom 1 am using the expands the originally creative person become
term “creative individuals”. Regretfully, many of preoccupied with the business problems and
them are not here anymore. becomes increasingly disconnected from the
creative scientific work. Unless the founder of a
Creative individuals are critical to the success of start up is able to adjust to new situations, then
any innovative process. The common gradually, the original dynamic environment will
characteristics of creative individuals are their move to the same stagnant environment from
willingness to surmount obstacles and persevere. which the creative person originally separated.
Obstacles will almost undoubtedly face any At this time a new generation ofcreative people
creative endeavor because such endeavors will initiate a new cycle of unusual and
threaten some established or entrenched interest. unconventional thinking.

210
The monumental result, the integrated circuit, many young scientists, on the contrary to low
was a result of the creativity of high-maintenance appreciation of diffusion engineering today.
employees. Diffusion and thermal processing play a less
important role in state of the art semiconductor
manufacturing than thej. did fifty years ago. At
I dedicate this work to the memory of William B. that time the technique of diffusing impurities
Shocklev. from a gaseous state into the semiconductor
When I am feeling down at work I always crystal opened up the possibilities of device
remember Shock1e.v. In ntj, entire life I have not performance never before possible.
met a more creative and resilient person than This presentation describes the history of
William Shocklev. diffusion in semiconductors from the beginning
He created not only Silicon Valley and a new of the fifties, when Carl Frosch and Lincoln
industry, but he changed the wa.v we live on this Derick at BTL forgot to close a hydrogen valve
plan&. For erample, ifthe atomic bomb had not and by mistake discovered the wet oxidation of
been invented, the life of the majariry ofpeople silicon; through the first days of November 1959,
would not be affected. However, I cannot when planar, four transistors setireset flip-flop
imagine my life without the transistor, even circuit began to work; until 1961 when
though I d 0 not use a mobile phone. Electrnglas manufactured the fust commercially
The fundamental theory of the PN junction as available diffusion furnace for the semiconductor
used today was formulated by Shockley in a very industry. Contrary to the commnn myth
short period of time. Shockley became the most integrated circuits did not take the world by
respected and hated man bv the age of fort).. storm, but had a slow growth through the
What was the source of his genius or what some beginning of the sixties. The development of the
call evilgenius? My answer is that Shocklev was first integrated circuit was certainly far from a
a man that Nature rarelyprodures and who only mainstream activity of the semiconductor
appears on Erath at intervals of centuries. companies.
Interestingly enough, the majorit). of those who
sharply criticize Shocklev put into their resumes
or biographies a note that they worked with and
were trained by W. Shockley.
It is too sad that Shockley never shared in the
rewards that so many Silicon Valley pioneers
reaped.

INTRODUCTION

The original point contact transistor does not

require thermal processing ofthe semiconductor.
The manufacturing of the point contact and alloy
transistors required extensive manual labor that
was performed in a similar way as the assembly Fig. 2. An example of typical assembly line from
of vacuum tubes. Many vacuum tube companies the end of the fifties where point contact
such as RCA, General Electric, or Sylvania in transistors were manufactured.
the East started production of transistors. A
typical transistor assembly line is shown in Fig.
CHRONOLOGY OF SEMICONDUCTOR
2.
DIFFUSION ENGINEERING
Assembly of the early transistors was difficult, if
not impossible, to automate. 1826 OhmLaw
In the early fifties diffusion in semiconductors
caused a breakthrough in semiconductor device 1906 Vacuum Tube
fabrication. Diffusion processing enabled high
volume semiconductor manufacturing. At that 1941 March -Russell S. Oh1 of BTL discovered
time diffusion engineering and wet etch was all rectifying properties of PN junction.
that was needed to produce a transistor. At that
time being a diffusion engineer was the dream of

211
1945 William Shockley become coordinator ofthe containing impurities. Silicon transistors were
-
Solid State Phvsics Research Proeram at BTL.
~~
less exoensive to oroduce and onerated at
higher temperature than germanium transistors.
1947 Point Contact Transistor [Bardeen, Brittain]
December - L. Derick and C. Frosch
1948 Junction Transistor [Shockley] discovered masking properties of Si02against
the diffusion of impurities.
June -BTL Press Conference revealing the
transistor invention. 1955 C. Frosch and L Derick released BTL Report
“The Oxidation of Silicon to Prevent Surface
1950 Diffusion from Solid Source used to form a PN Erosion During High Temperature
junction [ Dunlap, General Electric]. Operations”, BTL Memorandum #55-ll>Z,
June 14, 1955.
W. Shockley published Magnum Opus
“Electron and Holes in Semiconductors” (D. Theory of Interaction between Defects in Ge
Van Nostrand Company, Inc., Princeton,N.l. and Si [H. Reiss, S.S. Fuller, F.J. Morin, BTL.
1950).
June - W. Shockley resigned from BTL
195 1 September bell Telephone Laboratories and
Western Electric Company organized the In late 1955 Amold Beckman paid $25,W.W
”First Symposium on Transistor Technology”. to license patent rights from Western Electric.
Centralab Division of Globe Union purchased
license and lack Kilby attended meeting. 1956 February - A. Beckman announced the
formation of Shockley Semiconductor
Texas Instruments renamed from Geophysical Laboratories .
Service Inc. founded in 1920.
June - April - Bell Telephone Laboratories and
1952 Semiconductor Diffusion Equation Solution Western Electric Company organized the
[Van Roosbroeck, BTL]. “Third Symposium on TransistorTechnology”
with Sony’s engineer Iwahe and first Shockley
April - Bell Telephone Laboratories and Semiconductor Lab employees Happ, Valdes
Western Electric Company organized the and Noyce in audience.
“Second Symposium on Transistor
Technology”. The U.S. Amy, Navy and Air Charles Kittel published “Introduction to Solid
Force chose about two hundred people, the State Physics”, (John Wiley & Sons, Inc., New
BTL selected the rest mainly from licensees of York, 1956).
transistor patents.
“Zone refining” invented by William Pfann in November I - Bardeen, Brattain and Shxkley
1950-51 disclosed. Teal and Buehler described won the Nobel price for physics.
in detail crystal pulling apparatus.
Attendees include among others December 16 -Work of C.J. Frosch and L.
representatives from IBM, GE, TI, Derick “Surface protection and Selective
GlobeUnion and fourteen foreign (all NATO Masking during Diffusion in Silicon” is cited
countries) companies. in letter from A.T. David of Western Electric
to W. Shockley. Shockleyattachedtotheletter
Single crystal silicon is fabricated at BTL. routing list that is initialed by Noyce, Moore,
Hoerni, and Last.
1954 Diffusion from Gaseous Compounds [Fuller,
Ditzenberger, Frosch, G . L. Pearson BTL]. 1957 Vic Jones, one of the first Shockley
Semicondcutors Laboratories employee
April 26 - BTL announced its Solar Battery. resigns.

March -G.C. Dacey, C.A. Lee, W. Shockley J. Andrus and W.L. Bond of BTL disclosed
developed Mesa Diffusion Transistor. photoresist patterning of transistor
(“Photoengraving in Transistor
First commercial silicon transistor (Texas Manufacturing”, News abstracts of
Instruments). On May I O , 1954, Texas Electrochemical Soc. Semiconductor
Instruments announced the commercial Symposium, Washington, D.C. 1957).
availability of the grown-junction silicon
transistors. These first silicon transistors were lune 10 - Meeting between Beckman,
constructed by cutting a rectangular bar from a Shockley and seven scientists results in an
silicon crystal that was grown from a melt agreement that Shockley will remain director
of the company and a new position would be Western Electric announced 2N560
created to whom everybody else would report. transistor.
(Shockley would move to the Spinko Data Sheet for Fairchild NPN 2N696
building). released.

August - Smoot Horsley and his technician in September 12 -Jack Kilby demonstrated at
Shockley Semicondcutor produced first 72 TI phase shift oscillator assembled from
working four-layer diodes. grown-junction transistor and capacitors cut
from diffused power transistors metallized on
September 18 - seventonel’ alias “Eight both sides
Traitors” resigned from Shockley Laboratories.
November I 2 -ground breaking for new
September 19 -The “Eight” signed agreement Fairchild facility at 545 Whishman Rd. in
with Fairchild Camera & Instrument Mountain View. Total number of employees
Corporation. at Fairchild Semiconductor reach 150.

October - Fairchild Semiconductor 1959 January -Jordan, Brook, and Donahue of

Corporation as Fairchild Camera & Inshument RCA developed Germanium planar transistor
Corporation division of started operation in process.
Palo Alto at 844 East Charleston Road.
January 23 -Robert Noyce wrote a patent
October 4 - Soviet Union successfully launch notebook entry entitled “Method of isolating
the first satellite. multiple devices”. (The patent filed July 30,
1959).
October 29 -Blueprint of Fairchild Laboratory
completed. Seven bench diffusion furnaces February 6 - Jack Kilby filled patent
were located in the corner of the laboratory. application “Miniaturized Electronic
No exhaust scrubber was required by city Circuits”
ordinance.
March 6 (or 24 ?tMark Shepherd of TI
November 6 - Projects assignments announced at a press conference at Waldorf-
distributed: NPN transitor -GEM (Gordon.E. Astoria in New York at IRE meeting the
Moore), PNP transistor -JEH(JeanE.Hoemi), concept of the integrated circuit (called “solid
PNiP, NPiN - JL (Jay Last). circuit”).

December - Fairchild’s DI water treatment March -Jean A. Hoerni invented planar

station completed. transistor after inspiration conceived during a
morning shower.
1958 January 6 Fairchild first silicon
~ Data sheet for Fairchild PNP 2N697 released.
crystal evaluation.
April 10- Kurt Lehovec filed patent
January 3 I - Gas pluming of Fairchild Lab application “Multiple Semiconductor
completed (02.H2,N,Ar, and compressed air) Assembly.”
Fairchild employed 15 professionals, 5 skilled,
and 14 semi-skilled employees. May I - Planar Transistor Patent by Jean A.
Hoerni filled. (method of manufacturing
February 6 - TI filed patent application semiconductor devices)
“Miniaturized Electronic Circuits.”
May 6 -Jack Kilby filled patent application
February- IBM placed order for 100 Fairchild Ser. No 81 1,486 “ Semiconductor Device”
mesa transistors for $ 150.00/piece.
May - Robert Noyce named General
April - Fairchild meeting ‘*NPNvs. PNP. Due Manager of Fairchild. Gordon Moore
to the problem with contact f o m i n g t o N w promoted to Director of R&D.
base, NPN transistor was determined as the
first Fairchild device. June 13 -BTL announced epitaxial transistor
at IRE Solid State Device Research
May -Jack Kilby hired by TI Conference.
August 19-22 -Jay Last announced first July 30 - Robert Noyce filed patent
Fairchild Mesa Transitor 2N696 at Wescon “Semiconductor device and lead structure”
1958 in Los Angeles. At the same meeting

213
 based on his patent notebook entry from formulated point defect (vacancy) mechanism
January of diffusion in Silicon.

August 18-21 -at WESCON 1959 show in January 31 -1. Hoerni, S. Roberts, 1. Last,
San Francisco, Robert Noyce asked Jay Last and Isy Haas resigned from Fairchild and
to initiate development of integrated circuit at formed Amelco as a part of Teledyne.
Fairchild. The main motivation behind this
decision was strong publicity campaign David F. Allison and Lionel Kattner form
generated by Texas Instruments. Signetics.

Summer - Shockley Semiconductor renamed Molectro -with Robert Widlar and Dave
to Shockley Transistor Corporation. Talbert.

1960 Spring - Fairchild’s operators collectively E. Kleiner left Fairchild and joined Amelco
burnt their hairnets and protested the wearing as industry consultant.
of smocks.
1962 Electroglas Inc started marketing the first
Furnace Loading Push and Pull Mechanism generation of diffusion fumaces developed by
[Clevite Corporation]. Leach with strong support from Hoerni.

April - Shockley Transistor Incorporated Solution of Diffusion Equation with Moving

dissolved. Boundaries [M.O. Thurston, J. Tsai, Ohio
State University Research Foundation, Sixth
May -Jay Last and Lionel Kattner (formerly Quarterly Report- 1233-69.1 April I962 to
with TI), designed integrated circuit with 30 June 1962, Contract DA-36-039-sc81426,
physically etched device isolation. Isy Haas Columbus, Ohio].
produced fully operating device.
1965 TI and Motorola surpassed Fairchild in
August 15 - Jay Last filled a patent production of semiconductors.
application for integrated circuits with
physically etched isolation “Method of 1967 National Semiconductor established
making solid-state circuitry”.
1968 June - Robert Noyce resignedhmFairchid
October - Isy Hass, Lionel Kattner, and Jay July -Gordon Moore and Robert Noyce
Last demonstrated first ICs with P-type launched NM Electronics (“more noise
isolation difised through complete thickness electronics” renamed later to Intel).
of the wafer.
1979 Fairchild sold to Schlumberger.
October -Tom Bay, Fairchild marketing
manager, vigorously opposed manufacturing SEMICONDUITOR DIFFUSION
of integrated circuits at Fairchild
ENGINEERINGAT BTL
Management meeting with R.NoyceandG.
Moore in audience. , Tom Bay argued that
“Fairchild would loose all customers who are In 1947 Bell Telephone Laboratories divided the
using transistors now.” responsibility for the research and development
of materials between Chemical Laboratories and
November -first “Micrologic” devices with Apparatus Development Department. Calvin S.
diffused isolation produced by Last’s Group. Fuller was in charge of all work on plastics
including polymers at Chemical Laboratories.
December -Letter from Eugene Kleiner t o After the merge of both groups C. S . Fuller
Gordon Moore explaining that the staff in
characterized the new situation; “J.R. Townsend
Palo Alto were accusing people from
Mountain View of taking Palo Alto’s supplies was put in the same position with me and also
of paper and pencils. was assigned as head of the new organization,
my fate was not at all happy with the
Electroglas Incorporated established in Menlo arrangement”
Park. Cecil A. Lasch, Jr., the engineer who Fuller was gradually pushed away from his
developed diffusion furnaces atFakhild,joh successful work on polymers and eventually he
Electroglas. was offered two choices: I ) to go along with
direction of the group or 2) leave. By 1950,
1961 January - Baruch, Constantin, Pfister. and
Calvin Fuller decided to switch to solid state
Saintesprit of Ecole Normale Superieure Paris
chemistry.
He immediately began experimental work inside the tube with faces parallel to the gas flow
relating to the surface properties of germanium. during processing. The zone temperature was
At that time there was a problem with controlled to = +I- 2 O C
germanium that nobody seemed to understand
and which was called “thermal conversion.” 4BC13 + 3Si = 4B + 3SiCI,
Calvin Fuller described the problem as the
following: The germanium crystal was so
sensitive that if you went into the laboratory and
grabbed the doorknob, and then happened to
lightly touched the crystal, it would convert; that
is it would change type from N to P on
f
Y
x
subsequent heating above about 500 “ C . If the L
3
L

sample was etched and rinsed in “highly purified

water” conversion would not occur.
With the help of J.D. Struthers and Kathy
Wolfstrin, Fuller proved that copper was the
.. . . ..
Ba, N~
rapidly diffusing impurity creating “acceptors” in
the crystal and converting it from N type to P
type. P20, + Si = P + phospho-silicate glass
This work on copper led to investigating of other
fastdiffusing species and to an examination of
diffUsion in semiconductors in general. Fuller
with the help of John Ditzeuberger investigated
diffusion of Group Ill and Group V elements
and published a paper [ C.S. Fuller, J.A.
f
Y
Ditzenberger, “Diffusion of Boron and
Phosphorus into Silicon:, J. of Appl. Phys. Vol.
:
25 (1954), p. 14391
When developing the diffusion method, Calvin
Fuller felt that the BTL management was not
fully appreciative of the possibilities of Fig. 3. Diffusion furnacc used 111 cxly I95O.r in
diffusion. He wrote a letter to the Lab BTL
management [letter dated march 7, 1951, case
37939, Bell Laboratories], but received neither
response nor bigger support for his work.
When in 1954 Fuller’s work resulted in the
invention of the solar cell which Bell Labs
P=t//-\
p FURNACE

5‘ XK,
management highly publicized, he noted “It was
very interesting because people like to speak of
c ,’ FURNACE

teamwork when they talk about research. The CiSTyIcE UDNG TUBE
solar cell just sort of happened and had none of
the aspects of team work.” A photograph af the actual system schematically
shown in Fig. 3 s shown in Fig. 5.
The Apparatus used for diffusion from gaseous
compounds experiments in Bell Telephone In the mid fifties the first generation of top bench
Laboratories at the beginning 1950’s is shown in furnaces was improved and heating coils were
Fig, 3. added to the system. A very simple on-off
Most of the works reported by C.S. Fuller, J.A. controller and bubbler were added to the system.
Ditzenberger and C.J. Frosh were done in this Temperature was sensed by a thermocouple
system. The essential components are elongated inserted into a cavity in the middle of the wafer
fused quartz tubes with an inner diameter boat. All gas distribution was accomplished
approximately 25 mm extending through high through the quartz tubing and quartz valves. No
temperature controlled “glo-bar” furnace. The specialized gases for semiconductor industry
‘Glo-bar” furnace provides a constant existed at that time.
temperature zone approximately 100 mm long.
The semiconductor samples are placed vertically

215
 Fig. 4. Diffusion furnace used at Bell Telephone Laboratories during early fifties. In this type of furnace
Carl Frosch and Lincoln Derick discovered by accident the oxidation of silicon.

MFNSION

Fig. 5. Oxidation system used by Carl J. Frosch and Lincoln Derick at BTL in early 1950's

Regular technical quality gases were commonly PRORI.EMS WITH APPLICAIION OF

used. The K , N,9 and CO2 or their mixture DIFFL'SION TO TRAKSLSTOR AT BTL
was used as carrier gases. After the reaction in
the tube, carrier gases were vented through the Fuller's investig3tion provided the necessary
hood into the New Jersey atmosphere. infonn3rion about the high lemper3ture diffusion
The recipe sequence started with the stabilization parameters required tor controlling the impurity
of required temperature (no temperature ramping concentrations and depths of the diffused layers.
was available). The processing gases were turned There were, however. major problems with the
on after the temperature stabilized and then the application of diffusion to the Si iraitsistor:
boat with vertically positioned wafers were
inserted into the tube. The drastic decrease in lifetime, which
genr'rally occurred when silicon
s;lmples were heated to the high
tcmperalures.

216
 Pitting of the semiconductor surface
during the diffusion process when non-
oxidizing gases were used. The surface
pitting resulted in non-uniform junction
depth.

Fig. 6 shows an example of surface erosion of

silicon < I 1 I > surface after annealing at 1000 OC
in hydrogen ambient.
Fuller identified the decrease of lifetime as a
consequence of metallic contamination. Frosch
and Derick discovered that the use of an
oxidizing atmosphere provided surface
protection of silicon surface.

Fig. 9. An Assembly of GE alloyed junction

transistor.

Fig. 7 <111> silicon surface after annealing in

hydrogen ambient [C.J. Frosch, 19551

Fig. 10. Cross-section of GE alloyed junction

transistor.

It is interesting to note that CE discovered the

importance of diffusion earlier than BT when
they developed the alloyed junction transistor. A
photo in Fig. 9 illustrates the packaging used by
GE in the mid 1950s for the commercially
successful germanium devices, which were
manufactured using the “alloydiffusion”
Fig. 8. Alloyed junction transistors (1951) process developed by R.N. Hall and Crawford

217
Dunlap at the GE Research Labs. However, GE
did not follow up on the diffusion work except
for John Saby who extended this basic
technology to create a “control device” in 1951
which later become known as the alloy junction
transistor. (Fig. 8)

DIFFUSION TRANSISTOR AT BTL &

WESTERN ELECTRIC (2N560)

In early 1954, based on the work of Calvin S.

Fuller, and J. A. Ditzenherger, W. Shockley
suggested to Charles A. Lee, G.C. Dacey and
P.W. Foy, that diffusion of impurities into Ge or
Si from elemental impuritibs could be used to
form semiconductors junction with well
controlled junction depth.

In March 1954 they filed the patent application

describing the mesa diffusion transistor.
However, the theoretical concept in this patent
was still far away from the practical realization
because there was no available process and
equipment.
In a series of experiments P.W. Foy
demonstrated that the relatively high vapor Fig. 12. Ring-Dot and Stripe configuration of
pressure of elemental dopants could be reduced Mesa transistor suggested by Dacey, Lee, and
to a desirable range (about IO4 mm Hg) by Shockley.
diluting them in germanium.
shown in Fig. 11 with Arsenic doped
Germaniumand heated by a tungsten coil.
zii The main goal of Shockley and Lee was to
produce an impurity gradient in the base region
of the transistor. The doping gradient produced a
“built-in” electric field enhancing the transport
of minority carriers (Drift Transistor). This
transistor was described by John M. Early in
1954.
The typical sheet resistivity of the Arsenic
-
diffused region was in the order 200 Q/ and
-
junction depth 1.5 pn +/- 0.3 pn. The emitter
was formed by alloying the Aluminum film
approximately 1000 A thick was evaporated to
the base surface through a mask with emitter size
25x 50 p.The sample bar was then placed on a
Fig. 11 Apparatus for diffusion of elemental strip heater in a hydrogen atmosphere and heated
dopants into semiconductors. (BTL 1955) to the temperature sufficient to alloy aluminum.
The contact to the emitter was formed from 3000
C. A. Lee demonstrated in 1955 PNP Ge to 4000 A thick film of gold-antimony
transistor with diffused base. The 250 pn thick P evaporated and alloyed onto the surface of the
type Ge sample with dimension 5x1.5 mm was emitter.
placed into the evacuated molybdenum capsule

218
Almost at.the same time when Lee demonstrated through mask with size 50 x 150 pm to the base
his transistor, M. Tanenhaum and D.E Thomas region. An Aluminum layer was alloyed through
demonstrated silicon NPN transistor with the the emitter layer.
base and emitter regions produced by Alloying a film of gold with a small amount of
simultaneously diffusing impurities of antimony formed the contact to the emitter.
Aluminum and Antimony. Alloying of gold-antimony layer to the silicon
bulk formed the collector contact. The tungsten
AI LJYBNUS
ALUWNUM point pressure contacts were used to create an
EWnER GSOOLD-ANTIMONY electrical connection to the bulk.
The laboratory works of Lee and Tanenbaum
resulted in a transistor that was not easy to
manufacture. However, very good transistor
performance with diffused layers put into
question the main deterrent to the application of
diffusion to transistor technology - t h e drastic
decrease of lifetime (especially for silicon) when
the sample was exposed to the high temperatures
required for diffusion.

t COLLECTOR

Fig. 13 Mesa diffusion base transistor as

suggested by Dacey, Lee, and Shockley in March
23, 1955.
..
/-n+NPE DIFFUSED LAYER

I - T Y P E 1 H ) I G W N WrFERl

.. '-P-TIPE DIFFUSED LAVER

Fig. 15 An experimental BTL NPN diffusion

transistor

'P

Fig. 14 Direct alloying method of forming the

base contact to the diffused base.

The starting material was NType Silicon with 3

n c m and after diffusion the NPN shucture with
base thickness - 2 pn was formed. Both
diffusion layers completely covered all surfaces
of the Si sample.
Fig. 16 Junctions of the simultaneousiy diffused
The electrical contacts are formed on the sample
base and emitter
surface by alloying a aluminum layer evaporated

219
The L.E. Martin of BTL in Laureldale, PA An Aluminum contact to the ring base is
further advanced Tanenbaum’s silicon transistor evaporated through the mask and alloyed. The
with several refinements in fabrication silver is alloyed to the N-type emitter layer. The
techniques. These refinements included an individual transistor dies were bonded to the
independent emitter and base diffusion and high gold plated heater. Thermocompression bonding
vacuum evaporation technique for producing the of gold wires was used to wire electrical
ohmic contacts with thermocompression bonding contacts.
of attached lead wires. Martin’s design of the transistor enabled for the
L. Martin designed the silicon NPN transistor fust time production of a large number of
with mesa “ringdot” structure shown in Fig. 17 transistors simultaneously. However, the 2N560
which was manufactured by Western Electric as did not use photoresit masks. The metal
2N560 for military logic application. “shadowing masks” were used for:
The base diffusion was performed in an open a) emitter diffusion,
tube used in Frosch and Derick experiments b) base contact
with the galium sesquioxide (Ga203) as the c) emitter contact
diffusion source in wet hydrogen atmosphere. The aligned (at that time so called “indexed”)
The surface concentration of the diffusion layer mask was used to etch the mesa structure.
was approximately 10’’ cni3 with thickness Fig. 19 shows a slice of silicon with transistors
approximately 0.5 p m ready for assembly.

Fig. 17 Beveled section of 2N560 ringdot mesa Fig. 19. Wafer with transistors 2N560 ready for
silicon transistor. assembly.

-
Fig. 18 Cross-section of 2N560 ringdot mesa
silicon transistor.

The N-layer with surface concentration IOzo cm.’

and thickness approximately 0.25 pm was
performed in the same type of open tube
diffusion furnace using P205with dry oxygen as Fig. 20.2N560 Western Electric Transistor -
the carrier gas. collector and emitter junctions.

220
 Several of Shockley's former employees
publicized that Shockley had an exceptional
ability to simplify and solve problems of physics,
and, also, exceptional ability to find good and
talented people. History clearly proved, that this
myth is not hue as he hired several employees of
questionable integrity.

One of Shockley's problems was that at that

time he already understood the transistor quite
well and he did not consider the transistor
challenge to be big enough. His scientific interest
shifted to much more complicated four-layer
structure which required a type of processing not
known and available at that time.

I clibll*r
Bqcpen.
........ /
hvlnrrtrhr
. .. ., y;,.~:.
' .~... . .., ... ....~... . .... . ~, . . ....,..
z;,
~ ~ ~

I
6 !
-
.I
-

Fig. 2 1. More details of 2NS60 cross-section

BTL announced the 2NS60 at the WESCON

Convention in August 1958 in Los Angeles in
the same session where Victor Grinich and
Robert Noyce announced the Fairchild mesa
uansistror 2N696.

D1FR'SK)N IRAHSISTOR AT
SHOChT.E\ SE3IICONDL'CTOR

Equipped and encouriged with the latest data

from ciprriment with mesa transistors M'.
Shockley rcsigned from BTL and decided IO
reNm IO Palo Alto. With backing of Arnold
B e c h a n the) pdy in late a i 1955 S ?S.OOO.OU IO
license transistor parent rights from Western
ElccIris and in February 1056 A. Beckman
announced the formation of Shockley
Semiconductor I.aboratories. Shockley hired
group of young scientisrs all of them in their late
twentics \vtth the goal to improw the BTL mesa
tranmtor and establish a production capabilit! of Fig. 22. Letter from Backman Company attorney
Shockle? . Semiconductor L.aborarortes to tow. Shockley.
mnnufacture a cheap silicon trmsistor.

221
While at BTL Shockley did not need to be Shockley, however, was not interested in any
worried about any facility issues, tooling or legal dispute and no legal action was initiated.
payroll. In Palo Alto the mesa transistor was still Eventually, other very capable employees of
in the early stages of development with no Shockley’s joined the “Eight” at Fairchild.
equipment suitable for production available. Between them, for example, D. Allison. “Tom”
Shockley underestimated the problems of start C.T. Sah .
up organization and he concentrated mostly on
the scientific aspects of company. Shockley never understood the reasoning of the
The four-layer diode was the key to W. group of “eight traitors,” although he never gave
Shockley’s plan to revolutionize phone systems. up, he also never fully recovered from this set
It was great device in theory, but not in practice back. He moved his interests back to science and
- at least not at the time when mesa transistor Shockley Semiconductor Laboratories
was still not completed. The four-layer diode is disappeared.
basically a combination of P and N type layers
of semiconductors. The four-layer diode could Shockley accepted a professorship at Stanford
replace both rectifier and transistors necessary to and at the time when the nation was going
connect long distance phone calls. through desegregation he picked the work of his
big idol - Alexander G. Bell, who was, beside
The disillusioned group of Shockley’s young the telephone, also interested in eugenics and
scientists, with the strong ego and desire to prove served as a honorary chair of World Eugenics
themselves were running into more frequent Conference in 1921. Nobody really proved that
conflicts with Shockley. It was almost Shockley was wrong with his eugenics opinions,
predictable that once they had more confidence however, also nobody joined him in support. He
and more knowledge about transistors they gave to his detractors easy ammunition against
would quit. They had access to all the him and it was not too difficult for incoming’
information kom BTL “Diffusion Conferences”, “politically correct” society to minimize his
and they were aware of the latest work on oxide contribution to mankind.
masking and resist “photoengraving.” There is a
memo dated December 16, 1956 describing the
latest BTL development work. Shockley attached DIFFUSION TRANSISTOR AT
a routing list to the memo and Robert Noyce, FAIRCHILD (2N696 and 2N697)
Gordon Moore, Jean Hoemi, and Jay Last
initialed the list. On November 6, 1958 The Fairchild “eight”
formed several transistors design teams. One,
After a night of wrestling between Robert under Gordon Moore, to develop NPN’s, and
Noyce, who originally did not want to leave second under Jean Hoemi, who was working for
Shockley, and G. Moore, R. Shelton, .I.Blank, J. Moore, to develop PNP’s.
Last, E. Kleiner, V. Grinich, and J. Hoemi all Jean Hoemi and David Allison developed the
eight scientists quit Shockley Semiconductor diffusion processes in diffusion furnaces they
Laboratories. designed with Cecil ‘‘Art’’ Lasch, Jr. Former
Paramount Pictures consultant and Shockley
Everybody who worked at Shockley Semiconductor Laboratory chemist, Gordon E.
Semiconductor Laboratories have described their Moore, managed all the early diffusion work at
personal experiences and portrayed all the Fairchild.
negatives of Shockley’s personality. The side The group extended BTL’s research work on
that was not heard was from Shockley himself. diffusion at Shockley Labs and then refined the
process at Fairchild to enable the production of
Historical documents only suggest what type of commercially manufactured transistors.
transistor Shockley originally intended to design. Transistor technology used by transistor
When the “Eight” left Shockley, the Beckman companies on the east coast was relatively well
Company’s legal department wanted to file a evolved in 1957. However, the mesa structure
suit and stop them from using information and silicon diffusion technology was still a
acquired during their association with Shockley. novelty. By the end of 1957 only Bell Telephone
(Fig. 22) Labs and Shockley Labs had some experience
with this device. From all other transistor
manufacturers, only Motorola hired a Western

222
Electric engineer to start the Mesa Transistor was used to control the temperature of the single
PrOgEUn. zone. The tube was equipped with a quartz
network of pipes, quartz valves, and flow meters.
The Fairchild “eight” made two important All systems resembled research laboratory
decisions regarding the mesa transistor: Kodak equipment rather than industrial manufacturing
photoresist was used for patterning structure and equipment.
both base and emitter were formed by difision. The wafers were pushed into and pulled out of
Although the photoresist patterning of the tubes manually. Cooling wafers usually took
semiconductors structures was developed at place in a cool zone within the diffusion tube and
BTL, Western Electric engineers considered this allowed the wafers to cool to a temperature
process as non-manufacturable and they suitable for handling. Programmed heating and
preferred the metal shadow masks. There was no cooling was not used because of the difficulty in
photoresist for semiconductor masking maintaining good temperature profiles.
operations generally. All the .work at BTL and The used gases were just emitted into
Fairchild was done with photoresist developed atmosphere by vents placed just above diffusion
by Eastman Kodak for panerning printed tube exhausts.
circuits. There was no “clean room etiquette” and the
chain smoker, Robert Noyce, frequently smoked
Initially, the “eight” had a big motivation. They in the Laboratory.
disagreed with Nobel Price winner and they
wanted to prove that they were right. They had at Jay Last who had a background in optics paired
the beginning a remarkable relationship. They with Robert Noyce and designed contact
knew all the tasks necessary to make the masking equipment that alligned masks to the
transistor: they divided them up and worked wafer flat. Fairchild I” diameter wafers had quite
together. At the start up each of “eight” did his a long “flat edge” (Fix. 23) so alligning the mask
part without a whole lot of formal supervision, to the end point of the flat and to the wafer
and they worked mostly with their hands. They crown was quite satisfactory.
did not even have a copy machine. They
depended completely on their own
resourcefulness and they were not worried about
Wall Street ratings.

One hundred and ten days after completing

blueprints for the Fairchild Lab, the gas pluming
(O&,N,Ar, and compress air) of Fairchild Lab
in Palo Alto was completed on January 31, 1958
by 34 Fairchild employees (15 professionals, 5
skilled and 14 semi-skilled employees).
The first seven bench diffusion furnaces used in
the corner of Fairchild Palo Alto facility were
very similar to the furnaces used by Shockley
Semiconductor. The main improvement of
furnaces designed by Moore’s group was using
the FeCrAl Kanthal heating alloy developed by Fig. 23 Fairchild 1” diameter wafer
AB Kanthal in Sweden.
The idea to use resistive heating with Kanthal
wire very likely originated from Shelton Roberts. The geometry feature of the first Fairchild
Roberts was responsible for crystal growth and transistor was 750 pm. Alignment error was
switched from induction heating to resistive probably not better than 200 pm
heating in his crystal growth equipment because The structure and dimension of the Fairchild
the resistive heating made the temperature mesa transistor was identical to the BTL
control much easier. transistor, except that Fairchild group decided to
Fairchild’s diffusion furnace quartz tube had an use Kodak photoresist instead of metal
inner diameter of 2” and a quartz liner between shadowing masks.
the processing tube and heating element. A very Three masks were necessary to complete the
simple “variac” adjustable variable transformer transistor:

223
I) Emitter region Process Flow: MESA PNP 2N697
2) Base and emitter contacts
3) Metal 1) 200 pm thick P-type Si substrate (0.7-1.3
The wafers were placed by the front side (resist) Ricm)
on the emulsion side of the masking plate and 2) Substrate etch and polish to thickness120
exposed for 60 sec by a mercury lamp. The P
exposed wafers were developed and baked. 3) Base Deposition
Source Sb203 at 650 O C , wafers zone
temperature 1120 OC, 250 cm'imin Nitrogen,
Process Flow: MESA NPN 2N696 deposition time 25 min
-
4) Base Drive-in (XI 6 pnQ
1) 200 pm thick N-type Si substrate (1-1.4 1200"C/ 15%hours, 250 c&min Oxygen
Rim) Thickness of oxide grown during drive-in -
2) Substrate etch and polish to thickness120 I
P
P 4) MASK-Emitter
3) Surface oxidation (wafer flat on quartz boat) 5) Wet etch oxide (oxide grown during base
1200°C/16hoursdry0~,- lOO008, diffusion)
4) Base Deposition 6) Emitter diffusion
Source Ga203 is inserted into furnace during Wafers are placed vertically into the quartz
surface oxidation, nitrogen flush 1200 "Ci boat
5minutes, 500 cm3/minNitrogen Wafer zone temperature 11230 O C , 400
-
5 ) Base Drive-in (XI 3.5 m) cm3/min Nitrogen, 3 cm3/min Oxygen, flush
1200 "C/ 30 minutes, YO crn'imin Nitrogen, time 5 minutes
10 cm3/min Hydrogen Wafers are inserted into processing tube
Flash - 1200 "Ci 3 minutes, 500 c d i m i n Additional 15 cm'imin Hydrogen (thin
Nitrogen oxide)
Base Deposition and Drive-in is performed E l 3 is added to the gasses (3 minutes)
in the same tube by switching gases 400 cm3/minOxygen (3 minutes)
6) MASK - Emitter Wafers are removed from the processing
Wet etch oxide, emitter region and back side tube
(17 OC HF, W F solution) 7) Front side of wafer is covered by wax and
Strip resist (acetone) glass plate
7) Emitter Deposition 8) Back side of wafer cleaned in HF
Wafers flat on quartz boat 9) Wafer backside sand paper scratch
P20s Source temperature 200 OC, Wafer IO) Evaporation of Nickel on the back side
zone temperature 1000 O C , 200 m3/min 11) Nickel diffusion
Hydrogen, deposition time 60 minutes Wafers flat on quartz boat
8) Evaporation ofNickel on back side Wafer zone temperature 11230 OC, 400
-
9) Emitter Diffusion (XI 2.6 pm)
e Wafers flat on quartz boat
Wafer zone temperature 1100 OC, 200
. cm3/min Oxygen, diffision time 11 minutes
Slow cool to 200 o c
12) MASK - Base and Emitter Contacts
cm3/minOxygen, diffusion time 45 minutes 13) Vacuum Evaporation of Aluminum and
10) MASK- Base and Emitter Contacts Phosphorus
1 1) Vacuum Evaporation of Aluminum 14) MASK-Metal
12) MASK-Metal 15) Metal etch
13) Metal etch 16) Metal alloy (-600 ' C i Smin, Argon, rapid
14) Metal alloy (-600 OC/ Smin, Argon, rapid cool)
cool) 17) Back grinding to thickness 60 pm
15) Backgrinding to thickness 60 pm 18) Nickel backside plating
16) Nickel backside plating 19) Apiezon wax deposition on front side
17) Wax deposition on front side through a glass through a glass screen covering mesa region
screen covering mesa region 20) Mesa etch ( - 15 pm)
18) M e s a e t c h ( - l S P ) 21) Wafer dicing
19) Wafer dicing 22) Thermocompression wire bonding
20) Thermocompression wire bonding
 Both transistor projects were successful, but the Good luck strikes the Fairchild “eight” for the
NPN transistor was easier to manufacture, second time: Fairchild Camera received an
mainly because of contact to the base . To form important contract from Autonetics, a Division
an Aluminum ohmic contact to a moderately of North American Aviation, who was
doped Antimony base of a PNP transistor is developing the guidance control system of the
much more difficult than in the case of a Boron Minuteman missile.
doped base of NPN transistor. In addition, G. The 30 kg and 0.05 m’ computer was consuming
Moore as head of group had the power to choose 250 W of electric power, mainly for hundreds of
NPN as the first Fairchild transistor. Fairchild transistors used in computer designed
with discrete Transistor - Diode Logic.
It was not very difficult for Sherman Fairchild,
who was IBM’s largest shareholder and member Shortly after the contract was signed, Autonetics
of IBM’s board committee to get an order for a discovered that the Fairchild double diffused
1
hundred transistors at a price one hundred fifty mesa transistors had a reliability problem: an
dollars per piece. unpassivated base-collector junction was very
“.ll”..l...” .,.. __x
susceptible to electric breakdown.

Volatile, often sarcastic, Swiss born, and not

necessarily always nice, scientist Jean A. Hoerni,
was clearly unhappy that his PNP transistor did
not prevail. Significant differences between
Moore’s and Hoerni’s personalities only
increased tension between Hoerni and his boss.
At Shockley Laboratories Jean Hoemi was
hired by William Shockley as a theoretical
physicist and worked for the first months in
Shockley Labs on the diffusion theory and then
started laboratory work on diffusion. He became
a very skillful experimentalist and did most of
the Fairchild work on Boron diffusion, which
Fig. 24. Sherman Fairchild was at that time extremely difficult.

Despite the fact that the yield of mesa transistors

was very low, Fairchild sales reached in July and
August 1958 $65,000.00 because of the high
transistors price.

Fig 26. Jean A. Hoemi

One of the Hoemi’s boron tubes exploded during

the experimental work he conducted. But the
persistent Hoerni was determined to design the
best transistor ever.
Fig. 25. The guidance computer of the Hoemi was aware of the work done by C. Frosch
MINlTTEMAN intercontinentalballistic missile. and L. Derick passed on by Shockley: the work
of C.J. Frosch and L. Derick “Surface Protection
and Masking during diffision in Silicon” from
the time when he was associated with Shockley.

225
Hoerni also attended the meeting of the
Electrochemical Society in 1958, where
Mohamed “John” Atalla of BTL presented a
paper about passivation of PN Junction by oxide.

Hoerni started work on the new device and in an

extremely short period of time, when he worked N-Type Si Substrate
alone as was his custom, he finished the planar
silicon transistor and the device was
Base Mask
demonstrated in March 1959.
--c
Hoerni’s planar process was the major change in
transistor structure that enabled the high volume 4

production of commercially manufactured

transistors.
The main processing steps of Hoerni’s Planar
Process are shown in Fig. 27 and they may be
summarized as:
Emitter Mask ,a
N-type starting material
Base masking oxide
MASK- Base
Base diffusion and oxidation
MASK - Emitter
Emitter diffUsion Emitter .Diffusion
MASK - Base and Emitter cnntacts
Vacuum evaporation of Aluminum
MASK Metal

The significant advantage of the planar process is

oxide passivation of the semiconductor substrate
which improved the electrical parameters such as
reverse leakage current, breakdown voltage,
noise figure and low current hw.
The planar process removed the main
disadvantage of the mesa transistor - an exposed
collector-base junction, which is particularly
vulnerable to contamination during contacting Metal plating and contacts
and assembling. The reliability of the planar
devices was unprecedented.

The timing for Hnerni could not have been

better. Noyce and Moore, who emerged as
Fairchild managers, were under pressure from
Autonetics. They recognized immediately the
benefits of the Planar process and dedicated very
substantial resources to further improve the
planar process. Fig. 27. The main processing steps of Hoemi’s
After the invention of the transistor, the Planar planar process
Process was the second most important invention
of microelectronics. Who then, was responsible If Hoerni had followed the “politically correct”
for such great achievements? Occurrence of such way, developments of the transistor might have
a constellation of facts and imperatives, are taken a different route. The hardheaded Hoerni
interpreted many years later in textbooks, as the motivated by rejection of his ideas by his boss,
results of exc eptinnal business management. had only one alternative to succeed - create an
engineering masterpiece. According to Hoerni,

226
the masterpiece idea struck one morning when be The company grew significantly and as always
was taking a shower. growth brings problems. Transfer from the
In the history of transistors and integrated development facility in Palo Alto to
circuits, here are only a few cases when well manufacturing facilities was slow and difficult.
planned and managed projects resulted in For example, in December 1960, Eugene Klemer
success. Transistor, Diffusion Technology, and asked Gordon Moore to solve a problem when
the Planar Process were not the result of the staff in Palo Alto were accusing people from
coordinated and supervised effort. They ware the Mountain View of taking Palo Alto's supplies of
result of personal individuality and creativity. paper and pencils.

The significant reduction of leakage cnnent of The photographs of planar transistors, showing
planar devices led Fairchild to new diode the contacts to the base and emitter is in Fig. 29,
manufacturing facility in San Rafael. The and decorated emitter and base diffusion layer is
Fairchild parts were superior to all other showed in Fig. 30, are both from the time when
manufacturers and the market demand the author did reverse engineering work on
skyrocketed. Fairchild management was Fairchild transistors.
preoccupied with manufacturing problems and
paid very little attention to development of future
products.

Fig. 28. Jean A. Hoemi's Planar 'Transistor

(Fairchild, March 1959). Outside diameter is -
754 um

Fig. 30. Decorated Emitter and Base Diffusion of

Fairchild planar transistor.

FIRSTINTEGRATEDCIRCUITAT
FAIRCHILD

Kilby's patent demonstrated that transistors,

resistors and capacitors could be formed in the
same piece of silicon substrate.
However, there are only a very few electronic
circuits which can be assembled from transbtors
with all collectors common. The majority of
electronics circuits required an electrical
Fig. 29. Emitter and Base Contact
isolation of the individual electronics

227
 component, therefore the concept of appropriate
device isolation was one of the principal
problems of integrated circuits. The second
problem, equally important for integrated
circuits, was the interconnection of individual
isolated components. Although none of these
issues were addressed in Texas Instruments
invention, the company launched a large
publicity campaign promoting the Kilby patent.
In response to this campaign, Robert Noyce
asked at WESCON 1959 (August 18-21) Jay
Last to start a development program with a goal
to create a Fairchild integrated circuit. Except for
this encounter, Noyce paid no attention to Jay's
work. The management of Fairchild
Semiconductor, a division of Fairchild Camera
and Instrument Corporation, was preoccupied
with exploiting the new exciting product - t h e
planar transistor, including a new plant in San
Rafael making diodes based on the Hoerni's
planar principle. The development of the first
planar integrated circuit was certainly not the
mainstream activity at the company.

O"lBU1, Wwrp Fig. 32. Layout of RS Flip-Flop prepared by

Lionel Kattner

coated by Apiezon wax and the whole wafer was

IRDUt, input2 mounted on the quartz plate. The isolation mask
was applied on the back side of the wafer. Using
selective wet etch the complete wafer thickness
was removed in the isolation regions. The etched
trenches were filled with epoxy resin. After resin
curing the wafer was separated from quartz plate
and wax removed from the front side of the
Fig. 31. RS Flip-Flop used as test vehicle for wafer.
first Fairchild integrated circuit.
The very first integrated circuit was produced by
Jay Last's team included Lionel Kattner, Isy Isy Haas in May 1960. The maximum operating
Haas, and Robert Norman. The group selected
clock speed was 1 megahertz and the delay 50
the simple bistable RS (ReseUSet) Flip-Flop
nanoseconds.
constructed using four NPN bipolar transistors
and two resistors diffused into a single
Although it is difficult to contemplate production
monolithic chip of silicon as test vehicle. (Fig.
of a circuit with this technology, the circuit with
31) etched isolation demonstrated the feasibility of
The layout prepared by Lionel Kattner is shown
the planar integrated circuit in the form as used
in Fig, 32.
in modern integrated circuits.
The main processing steps are shown in Fig. 33.
After metal patterning identical to the Hoemi
planar devices, the front side of the wafer was

228
 Fig. 34 The world's first planar integrated
circuit designed and produced by Jay Last,
Lionel Kattner, and Isy Haas of Fairchild
Semiconductor. Functional units were
demonstrated in May 1960.

In October 1960 Isy Haas and Lionel Kattner

tried long boron diffusion necessary to produce
a P-type isolation region. This diffusion step
required eighteen hours diffusion through a
patterned oxide of the front side ofthe wafer. All
of Fairchild's experts in boron diffusion thought
that oxide would not be able withstand the long
diffusion time. However, the experiment was
successful and first devices were made by this
technique in November, 1960. Further
improvement of this technology was used for
Fairchild Micrologic circuits produced at 1961.

Jay Last approached Fairchild management team

and proposed a new product based on the latest
successful research work. Between the attendees
of the meeting was Fairchild VF' of Marketing,
Tom Bay. Tom Bay was vigorously arguing that
Fairchild could not get into integrated circuit
business because Fairchild would lose all main
valuable transistor customers. Although Noyce
and Moore were also at the meeting, no one
recognized the potentials of integrated circuits
and no one supported Jay Last's proposal.

Two months after this meeting, on January 3 1,

1961, lay Last, Jean Hoerni. Shelton Roberts
and Isy Haas resigned from Fairchild
Semiconductor.
Pig. 33. Process used for first Fairchild
integrated circuit with etched isolation.

229
 ,..l.._".".. . . . . . . ........................

Micrologic process flow: ----- B%.G 1

I. N-Si substrate polishing (80 m +/ - 5 m)

2. Oxidation (wet oxide 8000 A ) -uIp(I.Dllftu
...... -...... .......
3. MASK 1 (Isolation) . . .- .
4. Wet etch oxide . .-.
. . , * "
I *

5. Boron Deposition and Drive-in ........... ...............

Li**E i
6. MASK 2 (Base and P-Resistor) ..................................... " ......................................................... " ................................. ~

7.
8. -
Boron diffusion ( 6000 A oxide, -150

9. MASK 3 (Emitter and Collector Contacts)

10 Phosphorus Deposition and Drive-in (-2 ...... ............................... .j
/sqandXj - 1.4-1.6p) p-5- 9
i
1 1 Resist (front side) *----- ___^

12 Wet etch oxide (hack side only)

13 Vacuum Evaporation of Gold on the hack ..
- -
...... ....
side ( 400 A) 1. *mn
~:.-
-
14 Gold Diffusion (- 1050 'C/ 15 min with . .
<..-PP
...................
. .Z.'
" .
..
.
.
.
.
.......
I

fast cool) i =Li

p *
15 MASK4 (Contacts) " ...... ..... ..................... .....................
16 Evaporate Aluminum(front side, 0.01 /sq) K E -en
~ WDZ a

17 MASK 5 ( Metal) di. cot%€.lrm

~ l l c l
18 Wet etch metal (25 % solution of sodium
hydroxide)
19 Metal alloying (- 600 OC/ Argon)

L." ........................................... .......................... " ...................... 2

230
Fig. 35.RIS Flip Fop circuit produced with technology later used for Fairchild Micrlogic DTpL 900 series

The integrated circuit activity, directed by Jay

Last, was separated from the rest of Fairchild
R&D. Robert Noyce was involved with the
general company management and, contrary to
common myth, played no part in actual design or
fabrication of the first integrated circuits. There
is a mystery about the entry in Noyce’s notebook
dated January 30, 1962 where he discussed
bonding Pyrex glass to the surface of the wafer
as a support for subsequent silicon etch from the
backside of the wafer. This is essentially the
technique used in the very first Fairchild
integrated circuit designed by Last’s group
almost three years ago.
Noyce’s reasoning for this type of isolation was
based on the thought that this type of isolation
would eliminate the extra capacitance introduced
Fig. 36. Bonded circuit. into a circuit by the isolating PN junction, which

231
 ’I?
would deteriorate the circuit perfomiance. The
fact that Noyce considered this type of device
isolation of some importance is supported by a
note Gordon Moore on December 12, 1962
added to the notebook where he wrote “I do
remember discussing this idea with RNN early
this year.”

Why did not Noyce patent application tilled on

Fig. 37. Robert N. Noyce (1927-1990) July 30, 1959 which become the U.S. patent #
2,981 877 contain the ideas about junction
There is some mystery about notes in Noyce’s isolation as described in Noyce’s lab notebook?
The mystery is even more underlines by the fact
lab notebook The pages are not dated. But Noyce
himself identitiy the date of notes January 23, that recollection of people involved in actions at
1959. Noyce said that as he began thinking about that time are contradicting each other.
the Kilby’s circuit “all the bits and pieces come
together one day.’’
Noyce’s notes describes “multiple devices on a
single piece of silicon in order to enable to make
interconnections between devices as part of the
manufacturing process, and thus reduce size,
weight, etc., as well as cost per active elements.”
The next entry mentioned “connections are made
by evaporating metal through the holes in the
oxide to interconnect the diodes as desired for a
particular circuit.”

If Noyce dates are correct he knew a better

solution to the problem which Last’s group tried
to solve. Why this idea was not passed to Jay
Last? Why did Noyce, at the time when
Micrologic with PN junction isolation was
already in production, consider an idea which
was filed to the Patent Office on August 15,
1960 by Jay Last? Fig. 39. Robert Noyce in 1966

232
THE FIRST INTEGRATED CIRCUIT created N-type base layer. From this wafer he cut
AT TEXAS INSTRUMENTS a bar 1.6 mm wide and alloy contacts to the back
of this semiconductor bar to provide contacts to
Texas Instrument was very slow in introducing the bulk resistor. Aluminum evaporated through
diffusion to transistor manufacturing. According an appropriate mask was used to form an emitter
to Kilby in 1958 there was still no working alloyed region. The gold-alloyed contact was
silicon mesa transistor. formed to the Ntype base. Wax was applied by
hand to mask the mesa of the transistor and
Obviously, at the time when TI hired Jack Kilby, region forming a dishibuted RC network. Loose
the company had no program or plan to explore Gold wire was then thermally bonded to the
an integrated circuit. appropriate areas to complete the
There are plenty of stories describing how Jack interconnection.
Kilby developed his circuit during a mandatory The transistor used by Kilby was identical to Lee
vacation. In a recent interview Kilby said: and Shockley’s transistor developed in 1955 in
“After several interviews, I was hired by Willis BTL.
Adcock of TI. My duties were not precisely
defined, but it was understood that 1 would work
in the general area of microminaturization. Soon June 23. 1964 J. 5. KILBY 3,138,743
after starting at TI in May 1958, I realized that ul”IATmI7m -w:c ClRNm
since the company made transistors, resistors, Fli.6 I.D. 6. 19s I Sk..t&“i.

and capacitors, a repackaging effort might

provide an effective alternative to the Micro-
Module. I therefore designed an IF amplifier
using components in tubular format and built a
prototype. We also performed a detailed cost
analysis, which was completed just a few days
before the plant shut down for a mass vacation.
As a new employee, I had no vacation time
coming and was left atone to ponder the result of
the IF amplifier exercise. The cost analysis gave
me my first insight into the cost structure of a
semiconductor house. The numbers were high -
Fig. 40. Phase shift oscillator using one
very high - and J felt it likely that 1 would be
transistor, one resistor, and distributed RC
assigned to work a proposal for the Micro
network connected together by loose wire.
Module program when vacation was over unless
I come up with a good idea very quickly. In my The main contribution of Kilby’s invention is a
discouraged mood, 1 began to feel that the only
demonstration that all electronic components
thing a semiconductor house could make in a
could be produced from the semiconductors.
cost-effective way was a semiconductor. Further
Kilhy’s network does not resemble the integrated
thought led me to the conclusion that
circuit as used today at all. In fact the note about
semiconductors were all that really required -
the inductor in Kilby’s patent: “Small
that resistors and capacitors, in particular, could
inductances, suitable for high frequency use, any
be made from the m e material as the active
also be made by shaping the semiconductor as
devices.
evident by Figure 5a” (Fig. 41) indicates that
Although the test showed that circuits could be
Kilby was still considering microminiturization
built with all semiconductor elements, it was not
as used in Micro-Module program under Signal
integrated.”
corps.
Because of the unavailability of silicon diffusion
Kilby’s patent does not solve or suggest any
transistors at TI Kilhy used the germanium
isolation between multiple devices.
transistor. Tews Instruments used a square 10 by
In the transistor anniversary issue of Transaction
10 mm germanium wafers. In five rows and five
of Electron Devices, Kilby described twenty
columns there were 25 mesa transistors with
years later, that he omitted to include into the
emitter and base contacts evaporated through
patents a more sophisticated version of the
metal masks.
circuit with diffused resistors and oxide
Kilby used such Ptype wafer with resistivity
capacitor. There is absolutely no evidence that
about 3 Rcm and used Antimony diffusion

233
Texas Instruments had at that time more tactics were the reason why Texas Instruments
sophisticated devices. lost in legal litigation with Sprague and Fairchild

The importance of Kilhy’s invention and proof

that all electronic devices can be designed by the
use of semiconductors was significant. The
Molecular Electronics program launched by
Texas Instruments put in question the established
status of miniaturization of electronic
assemblies.
Texas Instruments was not leading the
microelectronics revolution. At ihe end of fifties
and at the beginning of sixties they were at least
12-24 months behind Fairchild and Tom Longo’s
Sylvania know-how.
Fig. 41. “Spiral semiconductor inductance” as Jay Last of Fairchild still has in his possession
suggested by Kilby in 1958. one of the first Fairchild products with label
“Open Here”. Fairchild marketing people knew
Kilby’s reputation was severely damaged by that T1 bought some quantity of each new
Texas Instruments legal maneuvering with Fairchild p“t. As a joke Jay Last labeled some
Sprague’s Kurt Lehovec and Fairchild’s Robert of these parts “Open Here” instead of with
Noyce. product label.
The fact, that almost the same or similar
engineering ideas and solution originated at It took Texas Instruments another five to eight
different development organizations years before h e y consolidated their R&D and
independently, is not unusual. production activity to surpass Fairchild.
However, the unethical and unprofessional legal

Fig. 42. Jack S. Kilby’s phase shift oscillator with germanium mesa transistor, hulk resistor, and istributed
RC network (Texas Instruments, August 1958).

234
 THEFIRST~GRA'IEDCIR~AT The other major Lehovec invention had a very
SPRAGW ELEClWC COMPANY interesting background. Lehovec attended a
workshop at Princeton University in the end of
At the beginning of the fifties the manufacturers 1958. Torkel Wallmark of RCA presented in this
of transistors resided on the east coast. The
manufacturing of point contact and alloyed
germanium transistors was very similar to the
assembling of vacuum tubes. Vacuum company
tubes such as RCA, Westinghouse, General
Electric, Sylvania, Raytheon and others saw the
transistor as a similar product. The management
of vacuum tube companies had no knowledge of
semiconductors and.they were not able to extend
the design of the transistor beyond the
knowledge they bought with license rights from
Western Electric. Gradually all of them quit the
semiconductor business.
The Sprague Electric Company is an example of
a different business decay. Two man in charge of
the company, two brothers, struggled for control
of the company, each of them trying to put his
son into the company leadership. As could have
been predicted, neither of them achieved his
goal. However, this internal fighting drove the
company slowly out of business.
In this environment Research Director of
Sprague Kurt Lehovec launched the
semiconductor business in 1952.
Kurt Lehovec started his career in former
Czechoslovakia. At Charles University in Prague
as a member of a group of Prof. B. Gudden
(student of Prof. Pohl), Lehovec worked on
selenium rectifiers. Suddeutsche Apparate Fabrik
in Nuremberg sponsored research. During the
course of this research he noticed that thallium
diffused into selenium very rapidly and the
"blocking effect" ' o f the rectifier was
significantly improved.
After the war, the CIA transferred Kurt Lehovec
to the USA during operation "Paper Clip", and
he started work for the U.S. Signal Corps in Fort
Monmouth in New Jersey. There he worked on
the problems of light emission from solids.
In 1952 A.C. Sprague hired Lehovec with the
intention to bring the company into the
semiconductor business. One of the first major
improvements of transistor technology was the
introduction of new alloying -'*capillary
alloying" at the end of the fifties. This method
with the surface melt technique enabled the " .
.. ,. ._
. , ,~ .. m. e*
production of multiple junctions in a single slice .I_,.
I .',, L*2;s'~~.rL*
.. . '. , , ., . .. .:
of semiconductor. The method was superior to
the BTL and TI grown junction transistor
Fig. 43 Transcript from Kilby vs. Lehovec final
process, which could produce, only PNP or NPN
hearing march 16,1966
structures.

235
workshop a visionary paper about the next Sprague Electric Company in Concord, NH
generation of electronics. Wollmark presented a where UNICIRCUIT series was manufactured.
list of problems that needed to be solved before
integrated circuits could be designed. One of
these limitations was device isolation.
Lehovec, educated from his work on the surface
melt multiple junction devices found a solution -
PN junction isolation.
The engineering problem was solved; the bigger
problem remained to be overcome. At that time
nobody had heard anything about integrated
circuits, so only visionaries could have
appreciated the importance of Lehovec’s
invention. Sprague Company did not have
visionaries and they did not want to file a patent
application.
After months, a persistent Lehovec finally
prevailed. His patent was approved and issued by
the Patent Office without a major objection.
Immediately after the patent was issued Texas
Instruments appealed causing the interference
Kilby vs. Lehovec (Fig. 43 ).
After an attorneys’ meeting, which counted a
double digit of TI’S attorneys and one attorney
for Lehovec, in the interference proceeding, the Fig. 45. Packaged Sprague integrated circuit
Patent Office held that Lehovec was entitled to (1962).
all claims.
Lehovec moved to Los Angeles and retired in
1978. He becomes a poet and play writer. He
remains a loner and enjoys lone walks during the
night paying no emphasis to his previous career
of physicist.

THEFIRsTlhTEGRATEDClRClJlTAT
SYLVANIA

In the early days of integrated circuits process

development was not driven by device
requirements. The circuit designer had to use
devices with parameters given by available
processes at the time.
Robert H. Norman joined Fairchild in 1959 as a
section head for device evaluation. He developed
the industry’s first monolithic resistor-transistor
logic (Micrologic). The reasoning for this choice
was simple: Fairchild transistors at that time
were large with big variations of electrical
parameters. The resistor was considered a simple
Fig. 44. First Integrated Circuits produced by device.
Sprague Electric Company in 1962. [R. Norman, J. Last, I. Hass, “Solid State
Micrologic Elements”, IRE Solid State Conf.,
Sprague successfully completed the design of February 19601
their first integrated circuit in 1962. The Orville Baker, a former Fairchild employee and
technology, after further modification, was Signetics employee number 5 , developed the
transferred into the Semiconductor Division of DTL series SE100.

236
At that time James Buie at Pacific November of the same year, Tom Longo
Semiconductor (later TRW Electronic presented a paper at the IRE group meeting
Components Group) developed logic that (PGED) in Washington. It was from that paper
became later known as transistor-transistor that Texas Instruments copied this device and
logic. introduced two years later as Texas Instrument
TTL series 5400 with only a single difference -
Tom Longo started his career at General different pin out.
Telephone Labs in 1958. GTL merged with Actually, Tom Longo visited TI after he
Sylvania Electric Products in Wohurn, MA. presented the paper. TI’S Howard Moos who
Sylvania was one of the original companies was in charge of the 5400 series development
licensing transistor technology from BTL and said to Tom: “we are not too far behind you.”
producing alloy transistors. Tom become However, TI added to Tom’s circuit a very
Director of R&D and managed advanced devices important feature. TI developed the new plastic
development. package. This was the reason why the very low
By 1959 the transistor technology was well priced logic series became the most popular
evolved and Tom’s group developed the fust family of logic circuits.
Sylvania mesa double diffused transistor. The
next project in 1960 was the first gjgahertz The case of Sylvania TTL is a typical example of
silicon planar epitaxial transistor. During this advances in integrated circuits. Except for Jean
project Sylvania developed an accurate masking Hoerni’s idea of the planar transistor, which
procedure, and in 1961 started development of stood out, most of the achievements of
integrated circuits. The established goal was to monolithic integrated circuits are results of
develop a gate with sub 10 nsec speed (at that contributions of many participants: James Buie
time competitors offer 70 nsec which was the sparked the idea, Thomas Longo make it
typical gate delay). functional and manufacturable, Texas
Tom Lnngo was the first to recognize that the Instruments lowered price make a product
old myth that the resistor is the cheapest favorable over the existing devices on the
electronic components was not true any longer in market.
monolithic circuits. He used a large number of
transistors in his logic gate: multi-emitter input Tom Longo left Sylvania in 1966 and moved to
transistor, fust inverter, level shifter, Darlington start IC business at Transitron where he
and cascade - totem pole. The gate speed was developed the first 16 hit bipolar RAM.In 1970
about 6 nsec. In comparison with Fairchild Longo joined Fairchild.
circuits the major difference was that Sylvania
transistors were small, with very tight
dimensional control.

Fig. 46 Sylvania planar epitaxial transistor.

Fig. 47. Thomas A. Longo
The product was first introduced to Litton
Guidance and Control as Sylvania ultra high-
level logic circuit (SUHL) in January 1963. In

237
 COMMERCULLY AVAaABLE
FURNACES (1960-1965)

Cecil A. Lasch, Jr., a technician who worked at

Fairchild with Gordon Moore and Jean Hoerni
on development diffusion furnaces established
Electroglas in Menlo Park in 1960.

The first Electroglas diffusion furnaces were

identical with furnaces developed by Fairchild.
The main improvement of furnaces introduced
by Fairchild was using of the heating element
coil from the FeCrAl Kanthal heating alloy
developed by AB Kanthal in Sweden.

The furnace quartz tube had an inner diameter of

2” and quartz liner between the processing tube
and heating element. Very simple “variac”
adjustable variable voltage transformer was used
to control the temperature of the single zone. The
tube was equipped with a quartz network of
pipes, quartz valves, and flow meters. All
systems resemble laboratory equipment rather
than industrial manufacturing equipment.
Fig. 49. ‘‘QuartzJungle”- gas distribution
The main problem of such a furnace was the system of an early diffusion furnaces.
variation of line voltage, which resulted in direct
variation of the processing temperature. In late 1962 two former Lindberg Company
employees, Karl Lang and John Dannelly wound
The second generation of Electroglas furnaces heating wire and did preliminary tests in Karl’s
had three zones, each zoce with a separate garage in Stanton.
transformer controlled by an analog controller Encouraged by experimental results they
manufactured by Honeywell. This furnace had a established in 1963 in Garden Grove, CA a new
flat zone approximately 12” long (later 22”). The company: Thermco Products Company.
later version of this type furnace was equipped
with the proprietary Electroglas controller.

The biggest Electroglas customer, IBM, had at

the mid sixties about 500 diffusion furnaces.

Electroglas was sold to new owner in 1963. The

development of semiconductor diffusion
furnaces was stagnant, and only minor
improvements were introduced during this
period of time. The new competition from
Thermco and Heavy Duty Lindberg gradually
expanded the number of vendors. Most furnaces
were quite similar and the price was the main
selling factor.
Fig. 50. The beauty of the first generations of
After Electroglas developed the first automatic diffusion furnaces (Electroglas 1963).
prober (with help of another ex-Fairchild
employee) the probers becomes the main
Electroglas busness and diffusion furnace were They know that Lindberg diffusion furnaces and
discontinued in 1968. furnaces developed at Fairchild had problems
with temperature variation as a consequence of

238
line voltage changes. They decided that instead Nevertheless, a change in the power requirement
of powering the heating elements directly by of the center zone changes the power of two
line voltage they would use a step down other zones.
transformer and use low voltage heating The insertion of the boat into the center zone
elements. caused an increase of temperature of the end
zones. When the center zone reaches the set
point temperature, the power applied to the end
zones decreased.
For long processing times this was not a very
significant problem.
This type of furnace was marketed by Thermco
Products as Pacesetter 1, and an early units were
shipped to Westinghouse in Canoga Park, CA,
Motorola in Phoenix, and Hughes
Semiconductors in Newport Beach.
As other vendors, Thermco was using the quartz
pipes and quartz valves to distribute gases. The
quartz assemble called ‘Quartz Jungle” was
produced by hay area Hank Schroder’s company.
The processing time was “controlled” by
mechanical “Green Labs” timer. One turn of
timer was 60 minutes, when the timer rings, “the
girls’’ adjusted the timer again.
Fig. 51. Fairchild’s one inch wafers in
Electroglas difksion furnace (1962).

Lang and Dannelly designed a single tube

diffusion system with a 2” inner diameter. The
Temperature of approximately 18” long zone
was controlled by a single thermocouple placed
in ceramic slurry contacting the quartz tube at
the center of the zone.
Fig. 52. Single point temperature control of the
first generation of Thermco Products diffusion
furnace.

Fig. 53. Second generation of Thermco Products

diffision furnaces (- 1965).

In 1965 Thermco introduced a second generation

of diffusion furnaces with hand helically wound
three zone heating elements. Heaters were
wound on a custom designed mandrill with
variable pitch to provide extra power at the load
end and gas end ofprocessing tube.
Fig. 52. Single point temperature control of the An analog Dialette controller manufactured by
Thermco products diffusion furnace (1963). Honeywell controlled the temperature. This
MasteriSlave control system automatically
The voltage difference between Reference (set adjusted power to the end zones. The transformer
point) and TC was processed by the Wheelco powered a low voltage heating elements. The
controller which drowe the power to the three proprietaly firing circuit controllmg SCR devices
heating zones. The manual trim can (Westinghouse or International Rectifier)
independently adjust the power of each zone. eliminated the dependence of temperature on the

239
variation of the line voltage and allowed the second biggest producer of equipment for
Thermco to guarantee temperature uniformity semiconductor manufacturing, started production
over 2 0 zone with +/- 0.25 OC in the range of of Thermco diffusion furnaces under a license
processing temperatures of 8OO-IZSO ‘C with strictly limited to just the Japanese market only.
virtually no drift.
The furnace was marketed under the name
Pacesetter I1 as a two stack model, and become
industry workhorse for many years processing 2”
and 3” wafers

During the sixties, diffusion furnaces had no gas

scrubbing. There was just a vent hanging over
tube mouth, which usually covered a quartz lid
with opening pointing to the vent.
All operations and wafer handling in the
diffusion area were strictly manual. The
automatic boat loader, although patented by
Clevite Transistor Corp. in 1960 become used in
the beginning of the seventies.
Vendors of the diffusion furnaces had no process
application engineers. Thermco hired the first
application engineer at the beginning of
seventies when the company moved into LPCVD
nitride business. Fig. 54. The quartz boat loaded with one inch
diameter wafers.
Not a)well known fact of the Themco business
was a joint venture that started in 1968 between
TEL and Themco. Tokyo Electron Limited, now

Fig. 55. Clean room “etiquette” in diffusion area (1960).

240
 CONCLUSIONS

In May 1952 the British scientist G.W.A.

Dummer made the following prediction at the
IRE annual electronic components meeting in
Washington, D.C.; “With the advent of the
transistor and the work in semiconductors
generally, it seems now to be possible to
envisage electronic equipment in a solid block
with no connecting wires. The block may consist
of layers of insulating, conducting, rectifying and
amplifying materials, the electrical functions
being connected by cutting out areas of the
various layers”.
Except for a few individuals, nobody was really
concerned. The few concerned, hard headed and
hard working scientists and engineers
transformed the vision into reality - into reality
that exceeded their vision.
Once W. Shockley said “We will h i l d these
transistors for a nickel a piece.” He was overly
pessimistic.
If we are looking back today on the work of a
relatively small group of men who liked to work
and were not even thinking of the financial
aspects, the integrated circuit does not look as a
big deal. However, the transistor and integrated
circuit, it is a big deal.

In this ankle an author offers a different opinion

about several myths and common folklore
knowledge of events, which took place almost
fifty years ago.

During the course of this work I was relying

mostly on my recollections and life long interests
in the history of engineering. I accumulated a
significant amount of samples, wafers,
photographs and news clips. I noticed that
history is like solid-state diffusion process that is
evolving with time.
The states of the art Rapid Thermal Processing
reduce the diffusion evolution of dopants to
minimum. Unfortunately, there is no similar
process known for treatment of history. The
politics, the power, and special interest define
how the history is presented to us. My attempt
was to eliminate such effects from my work. I
understand that I may be wrong in my arguing. If
I am wrong, once the new evidence and facts are
brought to my attention I will gladly correct my
opinion.

Colorado Springs, March-August 2002

241