IEEE T’RANSACTIONS ON ELECTRON DEVICES, VOL. ED-23, NO.

7, JULY 1976 62 1

Single Crystals of Germanium and Silicon-Basic to the

Transistor and Integrated Circuit
GORDON K. TEAL, FELLOW, IEEE

HEN the transistor was discovered at Bell Labora- physicist Dr. Herbert E.Ives, noted T V pioneer, photo-
tories there were few even within the Labs who electric and optics scientist, prolific inventor, and artist.
knew about it for some months due to decision a of man- During the 1930’s I did research on electron emissive
agement to maintain close security for a period. John surfaces of photocells for television pickup, fairly com-
Bardeen and Walter Brattain [l],[2] made their earth- plicated electron multipliers, and camera tubes. Prepa-
shaki:ngobservations on December 23,1947; the first public ration, control of the composition and structure of elec-
demo’nstrationof the invention and announcementof the tronic materials to achieve special electron or photoem-
discovery was not made until June 30,1948. In the interim issive, conductive or photoconductive propertieswere the
period only those who wereassigned to theproject by re- major concern. I made several attempts during this period
search managementwere told the facts. to start germanium studies.
I was a t Bell Labs at this time and had been there for When I was in the Electro-Optics Department I sug-
many years. Though I was not in the physics semicon- gested a project of collaboration to Dr. C. J. Davisson which
ductor group, I learned a bit about the invention in the appealed tohim. It absorbed muchof our mutual energies
early part of 1948, due tomy having deposited pyrolytically. for a period of about four years. Davy and I carried on a
silicon, germanium, and germanium-siliconalloy films on research programinvolving very-low-energy electronop-
ceramic tubes duringWorld War11.Some of these exper- tical systems and determinations of low-energy electron
imental samples turned out to be of interest toShockley reflections from platinum. This began a short timebefore
in trying some of his ideas of how to make a field-effect he received the Nobel Prize in physics in 1937 for his earlier
tramistor. Bardeen and Brattain’s discovery was very demonstration, in i927, of the wave-particle aspects of the
exciting news, perhaps even more so for me than for many electron. It was also during thisearly period, from 1932 to
others because I had long nourished an interest in germa- 1935, when, due t o the depression, Bell Labs adopted a
nium.that had started with my undertaking graduate study 3%-daywork week,that I had anothervaluable experience
a t BrownUniversityunderProf.Charles A. Kraus, as researcherin my spare timein Nobel Laureate Harold
American ChemicalSociety president, andWillard Gibbs, C. Urey’s laboratory atColumbia. While was I associated
Franklin and Priestley Medalist, who was one of the two with Harold we undertook a critical review, covering a
outstanding U. S. experts on germanium at thattime. My multitude of disciplinary fields, of all the research done the
master’s anddoctor’s theses were on germanium.It was a world over on heavy hydrogen from the time of its dis-
material studied only for its scientific interest; its complete covery by Urey, Brickwedde, and Murphy in 1932 until
uselessness fascinated’ and challenged me. My concen- 1935 and published it: “Hydrogen isotope of atomic weight
tration on this shiny metallic-appearing material during two,” by HaroldC. Urey and Gordon K. Teal (Reviews of
my graduate school days resulted in continuinga personal Modern Physics, vol. 7, pp. 34-94, January 1935). A t the
sentimental attachmentfor germanium, which, to me, a t same time I did spectroscopic research in Columbia’s
least, was and is an exotic element. This deep but little Pupin Physics Laboratory.
known personal attachment influencedme from time to In 1940, while still working full time on T V research in
time over an eighteen-year period after leaving Brown New York,I made aspecial trip toHolmdel to try get to the
University to seek some way of capitalizing on this people in the laboratory there and the metallurgists to start
knowledge and interest acquired years before. some work on germaniumfor radar diodes. I carried with
The opportunity to enter a career of creative research me some films of germanium that I had deposited pyro-
and innovation, concerned particularly with electronic lytically from germanium hydride.
materials, I owe to Dr. Robert R. Williams and Dr. Robert With theofficial decision of Bell t o decrease its TV in-
M. Burns, who headed Bell Labs chemicalresearch. When volvement during World War 11, I turned quickly, in
they invited me to come to work a t Bell Laboratories, I February 1942, into semiconductor work, initiating the
accepted and went therein 1930. first germanium research inBell Labs, making the first
M.y initial studies a t Bell began as a member of the germanium devices ever made in Bell, specifically, excel-
Chemical Research Department. In1933 I transferred to lent rectifiersof germanium pyrolytically deposited from
the Electro-Optical Department reporting to its head, digermane onto tantalum filaments, the digermane gas
having been obtained from one of my Brown University
Manuscript received March 15, 1976.
associates of the period 1927-1930. My results proved of
The authoris at 5222 Park Lane,Dallas, TX 75520. sufficient interest to Harald Friis, who was in charge of
622 IEEE TRANSACTIONS ON ELECTRON DEVICES, JULY 1976

microwave research, that he encouraged me to continue Labs, this was a real disappointment in view of my earlier
it. I was authorized to conduct a research program on interest in the subject. However, it taught me a hard-
germanium diodes, which I did for a period, with Bill learned lesson; I decided that I had given up a good idea
Sharpless a t Holmdel making the microwave measu:*e- too easily and thatif ever again I believed I was working
ments on the diodes that I made. The germaniumdioc,es on a world beater of an idea, I resolved never to sell my
made by pyrolytic means were quite good, as judged by intuition short nor to neglect to explore the idea thoroughly
their microwave conversion characteristics as well as by even if there was no assistance available. This was a lesson
their staticlog I versus log V plots. When theywere filst which proved invaluable both for me personally and toBell
made we were all of the opinion that no other laboratcryLabs in the period 1948-1952 in insuring, through my
had been working on germanium devices. After a month all-out persistencein growing germanium single crystals,
or two, however, we learned that Sperry had also been that thejunction transistor was realized in Bell Labs ahead
working o n germ.anium,microwave ,diodes. Also Prof. of its achievementelsewhere.
Lark-Horovitz of Purdue University visited Bell sometime I often foundmyself returning consciously to thinking
later1 in the Springof 1942, on his way.toWashington,DCb about my resolve and mulling over what I could do with
to negotiate an NDRC contract to supportbasic a physics germanium that would have a major impact.
research program on the semiconducting properties of As Chemical Research Department coordinator in the
germanium and germanium devices. Joe Becker askedrne attenuator field, I greatly increased my experience and
to demonstrate the I-V characteristics of some of my knowledge of a broadfield of electronics of major impor-
germanium diodes to Dr. Lark-Horovitz, whichI did. Dr. tance to Bell. I participated in the development of a variety
Lark-Horovitz was impressed. of attenuator materials, attenuators, and terminations for
Soon, however,Bell Labs concentrated its effort on s:1- microwave and radar systemsin the X,S, and k regions
icon diodes, on which mostof its results had been obtaincd [3]. These materials anddevices came into quantity pro-
during thepreceding 2 or 3 years. Because of this concea- duction and extensive use during the war. Some of the
tration on silicon diodes, I started, with the collaboration devices were employed in the New York-Boston TD-2
of Keith Storks, experiments directed towards pyrolyticmicrowave communication system immediately after the
deposition of silicon on flat tantalum filaments by de- war. Because of the system use and production of these
composition of Sic14 vapor in hydrogen. We made silicon devices, we were alwaysattentive tomaintaining resonable
microwave diodes with the deposited silicon that we.re control of their electrical tolerances and reliability. These
excellent. Harald Friis arranged a small conference in 34. contributions have been noted in George C. Southworth’s
J. Kelly’s office, on June 12, 1942,for presentation of our book Forty Years of Radio Research (NewYork: Gordon
results on pyrolytically prepared germanium and silicon and Breach,1962, p. 230). The program of the American
microwave diodes. I was authorized to work on both p:y- Physical Society meeting a t Columbia University in New
rolytic types of diodes, but as thewar progressed within York City in February 1947, listed on page 17 an invited
a few months it was harder and harder for me to get mea- paper: “Attenuator materials, attenuators and termina-
surement help from the Holmdel Laboratories of Be:L tions for microwaves” by G. K. Teal,M. D. Rigterink, and
Since I was also unable to get an assistantin the Chemic.d C. J. Frosch-which I gave onFebruary 1,1947, along with
Research Department, of which I had again become a several well-known APS members, oneof them being J. R.
member in 1942, to help with the measurementsI became Oppenheimer. Newspaper accounts and photographs of
particularly discouragedby the lack of interest in germl- my presentation of an experimental microwave demon-
nium that I could arouse in others. Thisled me to doubt stration on the same subjectat thenational meetingof the
my intuition and I turned my attention to the development AIEE in Pittsburgh on January 28,1948, forecast for the
of some new types of resistors, including pyrolytically public the approaching important event of the transmis-
deposited alloy films of germanium and silicon to which sion of television to Pittsburgh.
I have already alluded. Discontinuance of my researchan As I have indicatedin several published papers[4], [5],
microwave germanium diodes turned out a. bitter disap- during 1948, startinginFebruaryandextending
pointment to me when the diodes later during the war throughout that year, I wrote several memoranda to re-
proved to have the outstanding importance that I had searchmanagementproposinginvestigation of other
initially envisioned for them. methods of processing germanium than formationof the
Among the activities undertaken in addition to the conventional metallurgical ingot. I emphasized the ne-
preparation and studyof pyrolytically deposited resistclr cessity of better control of materials preparation, of the
films, I requested and was given responsibility for ~ 0 0 1 ‘ -resulting electrical properties, and correlation of these with
dinating chemical assistance to Bell Labs radar attenuator composition, processing, and structure. I suggested spe-
and termination research. In the course of this latter a(:- cifically the need for higher purity germanium and the
tivity, as the resultof a laboratory accident,I developed need for germanium in single-crystal form. My proposals,
a serious case of pneumonia. This seemed abit of bad luck however, gained nosupport.
and when, during the same period the MIT Radiation In spiteof the large research effort underway on tran-
Laboratory requested Bell Labs to undertakegermanium sistors, no one in Bell or anywhere else was growingsingle
diode work as an urgent job and itwent to associates in the crystals of germanium norgiving any attention to purifying
TEAL: SINGLE CRYSTALS OF GERMANIUM AND SILICON 623

germanium. As a matterof fact, in 1948 and early 1949 the ones, i.e., 1925-1933) of Irving Langmuir (G.E. Company)
people in Bell Labs who might have been expected to be and HaroldD. Arnold (Bell Laboratories)and of articles
the most likely ones to advocate the preparationof ger- concerning the long court fight waged between G.E. and
manium single crystals were taking the exact opposite Bell over G.E. patenting of an ultrahighvacuum in a vac-
point of view. For instance, certain materials experts said uum tube[6], [7], I was impressed by the tremendous im-
that such single crystals weren’t needed when I urged that pact of ultrahigh vacuum not only on beneficial the oper-
I be allowed to undertake such a goal. Even after John ating characteristics and practicality of vacuum tubes but
Little andI had grown large single crystals of germanium also by its ability to simplify the interpretation of the
and a,fter quite a bit had been learned about their excep- functioning of a design, thereby enhancing the designab-
tional properties,I was told by a well-knownexpert that ility of vacuum tubes toperform specific functions. Per-
mass-produced transistors would never use single-crystal sonal experience with gassey phototubes, electron’multi-
material-the costs of single crystals, according to him, pliers, and pick-up tubes made me especially conscious of
wouhl be much too great incomparison with ingots and, the difficulties of relating theoretically expected and actual
besides, the metallurgically prepared ingots were very electron ballistics in a vacuumtube device in which there
good material. Bill Shockley was opposed to work onger- might be present an unspecified number of unknown and
manium single crystals when I suggested it, because, as he unexpected electrical carriers in addition to the anticipated
has publicly stated on several important occasions, he electrons. Such a situation could easily thwart too the
thought that transistor science could beelicited from small development of the science of a solid-state electronic device
specimens cutfrom polycrystalline massesof material. and realizationof its maximum potentialities.
On the basis of knowledge gained from mypreviously The transistor inventiona t Bell Labs in late 1947 con-
mentioned researchon a varietyof T V vacuum tubes and vinced methat thiswas definitely the time to back get on
photo- and secondary electron emitting materials, I con- germanium work. In spite of my important challenging
cluded that the transistor would proliferate into many assignment of handling the chemical development of a
designs in order to perform a great varietyof functions, silicon carbide varistor for a new telephone handset, my
each design placing new requirements on the electrical most enthusiastic thoughtswere on the potentialof ger-
properties of the semiconductor materialand therebyalso manium and what I might do about it. I reasoned that re-
on the control of composition throughoutthe material. In moving the crystal boundaries and other undesirable de-
addition to the research mentionedon an earlier page, I fects from germanium would probably be as important to
grew single crystals of certain photoconductor materials the transistoras removing the last traces of gases from the
and keptwell acquainted with the current theoretical andvacuum tube [8],[9]. While I continued my assigned work
experimental research on conduction phenomena in solids. on siliconcarbide varistors for the telephone handset, I was
At the timeof the invention of the transistorby Bardeen determined also to get back into germanium research.
and Brattain my assignment was to work hand in hand Nothing that occurred to me seemed to have the poten-
with several electrical engineers (N.Y. Priessman, J. R. tialities as important as those of high-purity high-perfec-
Flegal, and H. F. Dienel), with mein charge of the mate- tion single crystals of germanium.
rials aspects of the development of two silicon carbide A number of times over the years after I had had asone
varistors for a new telephone handset.It actually was a job of nfy early jobs making single crystals from mercuricio-
that involved me in device design and device technology dide solutions I had thought would it be interesting to pull
as well as factors of control of composition, heat treatment, a single crystal of germanium, but I had never had an ad-
and firing atmospheres butalso studies of various kinds equate reason to do so. But now I had. The method in-
of silicon carbide, of various clay bonds and pressing trigued me notonly because I thought itwould be inter-
techniques. In this latter activity I had the continuous esting to pull a single crystal of germanium butbecause it
cooperation and advice of the ceramists(M. D. Rigterink, was not a method inuse at the timenor one tagged as a
J. R.Fisher, and others), as well as my fellow chemist, C. “metallurgical method” that I would have to yield to
J. Frosch. The devices, after we developed them, turned someone else as “their method.”
out to be the Bell System’s first close tolerance ( f 1 0 per- Late one afternoon around quitting timeI encountered
cent) solid-statedevices to be manufactured inlarge scale John Little and we got to talking about our work. He
(over 100 000 000) and at low cost (about 5q: each). Al- started by telling me how he needed a germanium rod
though we had not achieved these final results at mid-1948 small enoughin diameter tobe cut by a verysmall wheel
when the invention of the transistorwas announced, I had in order to minimizewaste. I could see that here was an
already by thenbecome deeply immersedin planning our opportunity tomake a rodfor someone who had areal job
experimental work, steeringmy part of it with these ends to do. So as we were getting on the bus for Summit, New
well in mind, including aswell the need for achieving at Jersey, I said, “SureI can make you a rod by pulling one
least 20-year reliability, and had had thoughts on experi- out of a germanium melt, and incidentally, it will be a
ment designs to assure this. single crystal too.” As soon as we got on the buswe started
With the previously discussed materials and device sketching. All we needed was something that would pull
baclkground anddetailed knowledgefromextensive the rod out smoothly and would wit,hstand the heat. A
reading of early research papers (as well as more recent graphite crucible seemed a suitable vessel in whichto melt
624 IEEE TRANSACTIONS ON ELECTRON DEVICES, JULY 1976

some germanium and aclock mechanism would serve 1;o materials, interest in single-crystal material picked up.
smoothly lift the rod from the surface of the melt. John he.d While some of the scientists were interested in these
a bell jar about 30 in high that was part of a large high- crystals in connection with transistor studies, most of,my
frequency heaterhe usedfor testing experimental vacuum associates continued to believe that single crystals were of
tube parts. The high-frequency heater coil was well up only limited scientific importance andwould never be of
inside the bell jar which was filled with hydrogen intro- any major use in devices that had to be produced on a large
duced at the top and excess the flowingout thebottom and scale [111.
into the ventilating systemof the room, the flow of the One reasonwhy metallurgically prepared polycrystalline
excess being adjusted sufficiently high to keep oxygen or germanium materials were noteasily displaced was that
other contaminating gas away from the heated crucible and they were serving large diode production requirements in
rod above the crucible. This would suffice for growing Western Electric. Some of my associates thought of it as
crystals within a hydrogen atmosphere.By the endof t h e the material aboutwhich we knew the most and thus the
3-mile ride into Summit we had sketchedthe equipment, material with which Bell should continue, striving, of
and two days later, on October 1,1948,we had completed course, to make small improvements and reductions in
our crude machine in John’s lab in New York City and manufacturing costs. Though many different possible
pulled our first single crystals of germanium. We did this conduction phenomena involving injection, flow, trapping,
without getting anyone’s permission orapproval and acted and recombinationof holes and electronswere under dis-
only on our own personal ideas. Most of the simple han - cussion and a degree of device complexity was being
dling and measurements on these crystals were made by envisioned far beyond that of the familiar diode, it was not
two “Kelly College” men who were moving through the at that timebroadiy recognized by the experts that this
various parts of Bell Labs to learn about the Labs as a implied an absolute necessity for a high-perfection, high-
whole before settling down to a specific job in a specific purity semiconductor medium.
department. These two men were electrical engineers. I I reasoned [21], [22] that polycrystalline germanium,
couldn’t afford to spend muchof my regular time on the with itsvariations in resistivity and its randomlyoccurring
single crystals since I was expected bymy department to grain boundaries,twins, and lattice defects that acted as
devote the major part of my time tomy officiallyassigned uncontrolled resistances, electron or hole emitters, and
job on the chemical aspects of silicon carbide varistors for traps would affect transistor operation in uncontrolled
the new telephone handset. ways. Additionally, it seemed to me that use of this mate-
About two months after John Little and I made our first rial to produce manycomplex units meant tobe identical,
germanium single crystals, in December 1948, I went t 1 3 with close performance tolerances,would be inconsistent
Jack Morton and suggested that a single-crystal program with high yields and, therefore,also with low costs. Even
on germanium andsilicon be set up.I spelled out the var-in developing complex transistor devices, it seemed to me
ious details of the program and asked him to supply som? essential to have a high-perfection, high-purity controlled
funds for John Little andme to get the program startedb:? composition semiconductor in order to achieve a separa-
first building a small duplicate bell-jar type of puller tion of various available electron and hole conduction
equipment that could be set up inmy laboratory on the processes in order to analyze and understand theoperation
third floor of Building I in Murray Hill. My main justifi- of the devices and thus to finally achieve an optimum
cation for the program that I was suggesting was to supply functional use of them.
a more uniform semiconducting material for large-scaXc My general aims for thesingle crystal researchwere as
manufacture. Jack,in agreeing to pay for the constructiorl follows [21], [22]:
of the equipment, said, “Gordon, you willget the scientific 1) to produce a conducting medium in which a high
credit for this.” I interpreted this ashis acknowledgmen; degree of lattice perfection, of uniformity of structure, and
at that time that I was providing, by my recent growtho F of chemical purity is attained;
germanium single crystals and by my personal suggestions, 2) to build into this highly perfect medium in a con-
a likely solution to one of his major problems-lack of ;L trolled way therequiredresistivitiesandelectrical
uniform semiconducting material for large-scale manu. boundaries to give a variety of device possibilities by
facture of devices. However, my interest in single crystals; control of the chemical composition (i. e., donor and ac-
of germanium and silicon was much broaderthan meeting; ceptor concentration) along the direction of single crystal
needs of manufactured devices, important as that is. I was; growth.
enthusiastic about single crystals of germanium andsilicor. The interplay of the variety of experiences that I had
both for research and development well as and thepossi. had with vacuum tubes and semiconducting or photo-
bility that theymight influence fundamentally thedesigr conducting solids, etc., was especially helpful to me in
of new and experimental devices and enlarge thescience suggesting possible important scientific and technological
surrounding new devices. insights relating to Bell’s transistor work. They mighthave
When it was found by measurement of injected minorit3 escaped me exceptfor the strong motivation Ithat had to
carriers by Haynes[lo] that our materiahwith large sin- work on germanium.
gle-crystal volumes in them hadminority carrier lifetimes Following our growth of the first germanium single
20-100 times greater thanfor polycrystalline germanium crystals, John andI sought to improve the equipment, and
TEAL: SINGLE CRYSTALS OF GERMANIUM AND SILICON 625

Fig. 1. Type I Teal-Little crystal growing equipment.

on completion of the movable puller equipment we con-

tinued working with it at nights in Building I. It turned out
impossible to get a separate small laboratory assigned to
the equipment. The best bargain thatI could make was
permission to set the puller in the middle of a large met-
allurgical shop, on a lowerfloor, for use a t night aslong as
I would promise to roll the puller equipment intocloset a
when we or I finished each nightso it wouldn’t be in the
way of the metallurgists during the daytime.
This meant that frequently around 2 or 3 o’clock in the Fig. 2. An early germanium single crystal grown in the type I crystal
morning I had t o disconnect fromthe wall approximately puller equipmentby G. K. Teal andJohn Little.
30-ft hydrogen, nitrogen, and water-cooling lines leading
to the puller as well as high-power electric lines to the
high-frequency heater.Then thepuller, whichwas about cut short had thebaffling proved insufficientor had the
2 square f t and 6 or 7 f t high, with the above line attach-ventilation of the laboratoryfailed. Fig. 2 shows oneof the
ments, was rolled into the storage closet and the high- earliest germaniumcrystals pulled on the aforementioned
frequency heatermoved some25-30 ft into an unused shop equipment.
corner. About 4:30 P.M. when the technicians started get- After longer minoritycarrier lifetimes and higher
ting ready to go home, I could reverse the process and begin mobilities were knownto have beenachieved in our pulled
crystal-growing experiments again. single-crystal materials, as compared withpolycrystalline
This became pretty much a way of life for me during germanium, Karl Olsen and Henry Theuerer, in the Met-
almost all of 1949. Naturally, this led to unusually long ,allurgical Sub-Department, started trying to grow ger-
work days andweeks in view of my carrying my regularly manium single crystals by other methods than the pulling
assigned project on silicon carbide during the days. My techniques that John Little and I were using. Someof their
family, i.e., my wife and three small sons, of course, bore results were encouraging but improvements in minority
the burden of my long hours a t work away from home. carrier lifetimeswere sporadic and controlof resistivity,
Though I am sure they were lonelya t times becauseof the particularly as a functionof the geometrical coordinates
little timeI had to participate in their activities, they were of the solid, was poor. By mid-1949 or early fall of 1949
interested andexcited by their thoughtsthat theresearch everyone investigating the semiconducting propertiesof
I spent so much time on seemed that important. germanium preferred touse pulled crystals, e,xcept Mor-
Fig. 1shows a diagramof the movable crystal puller. The ton’s group, I believe, was still studying point contact
germanium was melted in the graphite crucible by means transistors madefrom polycrystalline material as well as
of the large high-frequency heater that we have already from pulled single crystals of germanium in order to
discussed. The crystal was pulled in an atmosphere of compare reliability and reproducibility for the two mate-
hydrogen, theexcess hydrogen flowing freely out the bot- rials.
tom of the open bell jar. Baffling was provided aroundthe During the early part of 1949 and to about mid-1949 we
puller to lead the excess hydrogen towardsthe laboratory purified our germanium by selection of bottom slices of
air exhaust. Qur crystal growing careers might have been polycrystalline germanium ingots, and by fractional re-
 IEEE TRANSACTIONSONELECTRONDEVICES, JULY 1976

DIAMOND STRUCTURE INTRINSIC GERMANIUM

Fig. 3. The diamond structure of germanium and silicon.

crystallization in thepulling equipment of selected por-

tions of ingots. The results were encouraging in giving
higher specific resistivities-as high as 45 O-cm in some
cases and longer minority carrier lifetimes and higher
carrier mobilities. We continued purification experiments
ELECTRON ACTIVATION
ENERGIES
in view of their proven value.
I continued to exert as much effort as possible in grow; ng
germanium single crystals, justifyingit as essentialfor 1) CONDUCTION
BAND
the elicitationof the science surrounding semiconducm
devices, 2) the design and interpretation of experiments
with new types of transistors, and 3) eventually, the
t ++
+ + +lAEa
technology of mass-produced devices. IAEi= hva
As the work proceeded into late 1949 and into 19’50,
practically all scientific and development studies a t 13ell
concerned with transistors began to make use of single-
crystal material. In an invited paper, “Germanium and
silicon single crystals,’’ presented by G. K. Teal on Mal. 3, aE, -
AE2 -
DIAMOND 6 eV 0.35 0.35
1952, at theSpring Washington meeting of the America:n SILICON 1.12 ev 0.06 0.075
GERMANIUM 0.72 ev 0.0125 0.0 I
Physical Society [23], I summarized fairly completely the GREYTIN 0.10 ev
TIN OVERLAPPING
BANDS
scientific advances madepossible by the preparationof a LEADOVERLAPPINGBANDS

great variety of highly perfect single crystals of germanium Fig. 5. Electron acJivation energies of group IV elements.
and silicon, briefly referring to specific interesting nl3w
findings resulting from experimental studies made bya Fig. 4 is a different two-dimensional representationof
number of Bell Labs associates who were usingthe crystids germanium showing the electron pair bonds of intrinsic
that we had made. I pointed out in this paperthat notonly germanium, withan electron-hole pair created bylight or
were the single crystals of interest because of improved thermal vibration of the latticegiving two carriers to drift
uniformity of composition, but also because they h u i in the solid.
electrical properties that were strikingly different frcm n-type germanium can be caused by addition of, per-
polycrystalline germanium and of improved usefulnessdue haps, only one atomof arsenic to lo8 atoms of germanium.
to thehigh purity and perfectionof the single crystals. I The electroncan drift away, leaving the arsenic bound as
drew what seemed to me an important conclusion: “. . . . an ion in the lattice, and the electron giving an easily de-
these crystals have provided uswith semiconducting mediatectable n-type conductivity to the germanium.
having a degreeof perfection sufficientthat a more pow- p-type germanium can be created by addition of, per-
erfultype of experimentationhas become possib’le. haps, only oneatom of gallium to 108 atoms of germanium.
Avoidance of complicating features in the material has The gallium remains fixed in the lattice as ion an and the
permitted direct observationsof theoretical parameters hole can drift away contributing a detectable p-type con-
to be made and certain parts of a picture of electroric ductivity to the germanium.
conduction to be established that might otherwise remain A further part of the physical picture with which we
a matterof speculation.” This, as a matter of fact, was the work isshown in Fig. 5 where the electronic bands of ger-
main themeof the paper other than to discuss some of the manium, silicon, etc.,are as shown,the statesin the valence
important practical devices that had been achieved in bond bandbeing nearly fully occupiedand the states in the
single crystals. conductionbandbeingpracticallyunoccupied. The
It is well known, of course, that both germanium artd amount of energy to extract an electron from the covalent
silicon crystallize in the diamond structure, and this is due bond to give a freeelectron-hole pair to the lattice corre-
to the similarity of their four electron pair or covalent sponds to theenergy AI31 required to raise an electron from
bonds. This structureis illustrated in Fig. 3. the topof the valence band to thebottom of the conduction
TEAL:SINGLECRYSTALS OF GERMANIUMANDSILICON 627

band, thus leaving a free hole in the valence bond band.

The activation energyfor this process differs in diamond,
silicon, and germanium as indicated in the first column.
We are primarily interested in the values 1.12 and 0.72 eV
for silicon and germanium,respectively, for the present.
The energies A E 2 to extract electronsfrom donor atoms
and place them in the conduction band A E 3 to create a free
hole by raising an electron to theacceptor level are indi-
cated in thesecond and thirdcolumns, respectively. The
somewhat lower A E 2 and A E 3 values of germanium (about
0.01 eV) as compared to silicon accounts for the almost
complete ionizationof donor or acceptors for germanium
a t room temperature in contrast to lowera degree of ion-
ization in silicon. Fig. 6. View of cross section of polycrystalline germanium, cut per-
pendicular to axis of ingot.
Actually, it should be emphasized that these physical
pictures with which we work and uponwhich theories must POSSIBLEENERGYLEVELSINEXPERIMENTAL
SAMPLES OF GERMANIUM
be built are highly idealized and assume a perfectionin
structure and control of purity and composition of ger-
manium and silicon that were not present in the actual
materials under study until the work discussed here was
done. This perhaps is made more evident by the photo-
graph of Fig. 6. This is a photograph of a specially treated
cross section of polycrystalline germanium, the cross sec-
tion being taken perpendicular to the growth axis of a
standard germanium ingot. The obvious imperfections
that can be seenare accompanied by more subtle ones that
are notvisible, such as lattice vacancies, interstitial atoms,
unidentified foreign atoms, allof which may act as traps
for the carriersof electric charge. Moreover, uncontrolled GRAIN
BOUNDARY
variation in donor and acceptor concentrations occurred
Fig. 7 . Possible energy levels in an experimental sample of germanium.
in suffficient degree to be disturbing. Thesewere, however,
the best materials available to us until the improvements
ffrom the present studybecame available. germanium. Experiments in 1951 byW. I?.Slichter of my
The electron bands in the previously available experi- group showed that highly pure multiply recrystallized
mental samples of germanium might more properly be germanium could be produced with lifetimes of carriers
represented by the diagram in Fig, 7 . Here is one typeof of several thousand, 12 000 &being the maximum ob-
grain 'boundary that acts like a narrow p region in the solid; served.
even within one crystalin the solid composition changes The higher purity of our germanium, obtained by re-
might cause wiggly energy levels and be accompanied by peated recrystallization prior to growing the crystal of
n-p regions or have uncontrolled variations in a given interest for specific study, and the higher perfection of the
carrier density as indicated here. structure and greater uniformity of composition led to
Single crystals simplified the study of injection phe- higher mobilities; the values of 1100 [27] measured for
'nomema and the determination of mobilities by the method electrons in germaniumin 1945 gave way to 3800 [28] in
of Haynes and Shockley [24] in which the time takenfor 1952.
an injlected minority carrier to move down a filament is The longer lifetimes of minority carriers were of vital
measured by a pulse technique [25]. Such experimentsby importance in the design of transistor devices. In addition,
Shockley and Haynesgave reality to the concept of injec- they greatly facilitated basic studies of surface recombi-
tion of excess minority carriers which then drift withan nation rates andof the influence of various treatments on
electric field to give a very different type of conduction the rates.
from that normally thoughtof as metallic conduction. The Later, thesingle crystals were crucial in realizing the first
high purity and perfectionof structure of single crystals junctiontransistor. About thetime of publication of
lengthened the path before detection of the carrier, re- Shockley's paper on the structure and theory of the junc-
duced spurious generators and provided bars of uniform tion transistor (June 1949), I suggested to Morgan Sparks,
resistivity. Someof the earlier measurements referred towho was working with Shockley attempting to make a
above showed germanium single crystals to have lifetimes junction transistor, thathe and I collaborate.We started
of carriers of 120-140 p s or more [26], which was 20-100 not so long thereafter. John Little and I had by then been
times greater than was being found in polycrystalline developing germanium single crystals about a year, with
628 IEEE TRANSACTIONS ON ELECTRON DEVICES, JULY 1976

Uec. 20, 1955 G. K. TEAL 2,727,840

METHODS OF PRODUCING SEMICONDUCTIVE BODIES
Filed June 15. 1950 2 SheetsSheet 1 DONOR A
ACCEPTOR
q/ ORIENTED SINGLE
CRYSTAL SEED

GERMANIUM
SINGLE CRYSTAL

CRUCIBLE

Ge 1
Fig. 9. Schematic of grown junction single-crystal technique of making
p-n junctionsand junction transistor structures.

APPROXIMATE

n
CONCENTRATIONS
OF DONORS
ANDACCEPTORS

NG.3
I9
p-n

T3
JUNCTION 2 PARTS Sb

+ 4 PARTS Ga/lo8 Ge
Fig. 10. Germanium single-crystal p-n junction made by the grown
junction technique.

method is shown in Fig.9. During the growth of an n-type

germanium single crystal addition of a tiny pellet of an
IN VENTOR
G. K TEAL
acceptor element in suitable amount causes the crystal
-+e
A TTORNC-I’
then grown to be p-type, thus resulting in asingle crystal
with a p-n junction perpendicular to the direction of
Fig. 8. The “double doping” or grown junction single-crystal method
of making p-n junctions and junction transistor structures. Cover shoet growth of the crystal (Fig. 10). The amount of impurity
of U.S. patent 2 727 840, filed on June 15,1950, and granted to G. K. required to be added is so minute that the thermal equi-
Teal onDecember 20,1955.
librium of the growing crystal is not disturbed and the
crystalline perfection of the crystal containing the p-n
very promising results. During the last month or two of junction is preserved throughout. Regarding this crystal,
1949 the crystal growing activities were assigned a SElidl Shockley has said:
laboratory in BuildingI1 and ErnieBuehler was assigned
to helpme with these crystal growing activities. Thisnew The recent advancesof transistor science and technolo-
gy are founded on these crystals. The addition of acceptors
location was very much more convenient in being mu:h to an n-type melt during pulling changed the solidifying
closer to themajority of people working on the transistor material from n- to p-type and gave the first good p-n
project, as well as not having the drawbacks of trying to junctions.
work in space belonging to someone else. As a resultof our
collaborative effort, within a few months we prepared h e Fig. 11shows a single crystal p-n junction rod under zero
first high-perfection single-crystal p-n junctions [29], [eo] bias. Shockley will discuss other bias conditions.
and the first junction transistor [31]. The key technolo,gy The flow of carriers in a small single-crystal rod cut to
usedin these achievements was the “doubledoping” include a p-n junction from the larger junction areais il-
technique otherwise known as the grown junction method lustrated in Fig. 12.The currentfor a p-n junctionis given
of making junction structures. It was accomplished in the by this equation originally devised by Wagner [32] afor one
movable single-crystal puller that John Little andI hiid carrier rectifier theory. k is Boltzmann’s constant, T is
been using, by slight modifications to facilitate addition absolute temperature,q is the electronic charge,and Vis
of impurity pellets to themelt duringthe growth of a sing,le the voltage applied across the junction being for the +
crystal of germanium. The method is described in U. S. forward current and- for the reverse current. Ro is a di-
Patent 2 727 840 issued to G. K. Teal, thecover sheet of mensional factor determined as the experimental resis-
the patent being shown inFig. 8. Schematically, the tance a t low applied voltages. The solid lines are the the-
TEALSINGLE CRYSTALS OB GERMANIUMAND SILICON 629

SINGLE CRYSTAL GERMANIUM RECTIFICATIONCHARACTERISTICS OF

P
A GERMANIUM pn JUNCTION

10

---I---
Lo

1.5

o<o
IO

1 1
IO’ 10-3 162 Id‘
AMPERES/CM~

Fig. 12. Rectification characteristics of a germanium single-crystal p-n

junction.

NO NET F L O W
Fig. 1.1. Single crystal germanium p-n rod. Zero bias condition.

experimentally observedvalues. The two arein excellent n .001”

agreement upto 20- or 30-V reverse. This is the firstrec-

tifier inwhich such agreement has been possible without
resorting to an arbitrary shape factor in the exponen-
tial.
An algreement with theorv which is still more illumi- Fig. 13. Germanium single crystal containing constant-resistivity re-
nating in revealingthe detailsof conduction in p-n junc- gions and an n-p-n
transistor structure.
tions has been obtained by determination of Ro from the
measured propertiesof the p and nregions by Shockley’s As shown in Fig. 13 the crystal producedis one in which
p-n junction theory[33].,The following theoretical equa- a thin p-layer about 1mil.thick and perpendicular to the
axis of growth is “sandwiched” in between the n-type re-
tion gives the total current of a p-n junction:
gions. Just as inthe p-n junctioncrystal, the perfection of
the crystal is maintained throughout thevolume in which
the changes of conductivity type take place, along the
length of the crystal. The constant resistivity regions
-, shown in this crystal were achieved later in our type I1
1
crystal puller.
The preparation of such highly perfect single crystal
We see that theintrinsic conductivityof germanium, which n-p-n structuresin which donor-acceptor concentrations
we can measure, entersto thesquare power. The mobility are controlled madeit possible to investigate the electrical
of holes in n material enters asinverse square root: This properties of such structures and to find that have theythe
factor includesb , which is the ratioof mobilities of holes electrical properties expected on the basis of Shockley’s
and electrons. We can determine theseby measurement. theory [33].Here (Fig. 14) I show such structure with zero
The sigma’s and tau’s are conductivities and lifetimes in bias. Bill Shockley will show a structurebiased as an n-p-n
the n and pregions and we can measure them.The restof amplifier and will discuss in his paper the effect of bias on
the expression is the same as in the Wagnerian equation the electron energy bands, the flow of carriers, and other
[32]. M7e can now put the experimentally determined theoretical aspects deserving of attention. One of his
values for the p and the nregions into this equation and conclusions is that the electrical characteristics of such
calculate the properties of the p-n junction between the structures biased in the way he is discussing them show
two regions. The agreement obtainedbetween theory and agreement with diffusion theory.
experiment in this case is’within a factorof 2, the shape fit This is an importantscientific contributionachieved in
being the same asbefore. This agreement is remarkably the high-purity high-perfection single-crystal materials
good and is superior to anything that has been obtained that we made.
for any other type of rectifying junction. Though we succeeded in making the first n-p-n tran-
630 IEEETRANSACTIONSONELECTRONDEVICES,JULY 1976

&
NO BIAS

DISTANCE e

Fig. 14. A germanium single-crystaln-p-n structure andthe energy level

scheme. Zero biascondition.

sistor inApril 1950, by use of high-perfection pulled sin-

gle-crystal germanium p-n junctions, we devoted n m e
than a year to its development before making it puhlic
information. A t no time during this period did the entl~u‘
siasm for or the activity on the double doping program :.ag.
On July 5,1951,by publication of “p-n Junction Triln-
sistors” in Physical Review, vol. 83, pp. 151-162 (Ju1.y 1,
1951) byits codevelopers W. Shockley, M. Sparks, and 6.
K. Teal, Bell Laboratories announced thedevelopmen,;of Fig. 15. The type I1 crystal pullin machine designed and made by G.
K. Teal a n f J . B. Little.
the junction transistor, revolutionary firstamplifier ca-
pable of operating on as little as1yW of power such t:mt
tiny signals of the level of a millionth of a watt, of;en
characteristic of modern electronic equipments,could be
amplified without wasteof power and space. Previoudy,
a million times thispower, i.e., about 1 W, had ordinady
been used byconventional vacuum tubes to amplify.tiny
signals in the microwattlevel.
During the firsthalf of 1950, John Little and I designed
and built more
a sophisticated crystal puller, the so-called
type 11,shown in Fig. 15. Notice that the inertgas atrno-
sphere was now more positively contained within an en-
closure of quartz, with the cooled high-frequency hee.ter
coil outside the quartz tube for added cleanliness. The
single-crystal seed was attached to the lower end of a rod
with well-designedmechanical meansfor rotation and easy
vertical movementof the crystal as desired.
The improved symmetryachieved in the typeI1 crylrtal
puller gave improved temperature control and mads it Fig. 16. Small diameter germanium single crystal with bulbar lower
end-grown in the typeI1 machine with rotating seed. Smallpin for
possible to control the diameters of crystals within ve:ry companson.
narrow limits. Notice that theneck of the crystalin Fig. 16
is not much larger than the straight pin which we have lifetimes of carriers than obtainedfor materials produced
included as asize reference. in the earlier puller [34],[35]. These variations correlated
Shown in Fig. 17 are threecrystals prepared in the t,ype with crystal rotation rate and periodic variations of im-
I1 puller, giving goodevidence of a greaterease of conx-ol purity concentration detected by use of radioactive dopant
of the crystal shapes than was possible in the typeI cryl3tal in crystals grown in collaboration with J. D. Struthers, J.
pullers, where the growing crystal could not be rotated. A. Burton, and ErnieBuehler. After a shortinvestigation,
While the crystalsof Fig. 17 were beautiful ones,peric,dic we found we could overcomethese troubles by using higher
variations in resistivity showed up (in measurements by crystal rotation rates andby vibrating thecrystals up and
J. R. Haynes and G. L. Pearson) accompanied by lower down an amplitudeof about5/looo ofan inch during growth.
T E A L SINGLECRYSTALS OF GERMANIUM AND SILICON
Fig. 17. Three germanium single crystals grown while slowly rotating
in the typeI1 crystal pulling machine.
Fig. 18. 11/4-in-diamgermanium single crystal grown in the (111) di-
rection and ata very slow rate.
Improved uniformity of resistivity was also obtained across
a slice cut perpendicular to the directionof crystal growth.
In Fig. 18 is shown a beautiful lY4-in-diam germanium
single crystal grown in the typeI1 crystal puller soon after
constructing thepuller. It was grownin the (111)direction
at a very slow rate.
In Fig. 19 is shown by comparison cross sections of a
germanium single crystal grown inthe typeI1 crystal puller
and of a polycrystalline germanium ingot, thecross sec-
tioning being made along the axis of the crystal and of the
ingot. An uncut germanium single crystal is shown for
comparisonwith thecross-sectionedmaterials.This
photograph shows the very nonuniform structureof the
germanium ingot and reveals a complete absence of grain
boundaries and twinning in the cross-sectional view of the
germanium single crystal.
In Fig. 20 comparative measurements of the specific
resistivity along rods of polycrystalline germanium and
single-crystal germanium are exhibited. The improvement
631
Fig. 19. Com arison of cross sections of germanium single crystal on
the right a n i o f polycrystalline ingot onthe left. The
polycrystalline
ingot cross section reveals many grain boundaries. The cross-sectional
viewof the germanium single crystal shows the absence of grain
boundaries. An uncut germanium singlecrystal is shown inthe middle
for comparison with the cross-sectioned materials.
SPECIFIC RESISTIVITY MEASUREMENTS
__ POLYCRYSTALLINE
DISTANCE
Fig. 20. Comparison of specific resistivity measurements of polycrys-
talline germaniumand germanium single-crystal material.
I
I 2 3
DISTANCE IN INCHES
Fig. 21. Resistivity profiles of normal and programmed single crystals
of germanium.
of single-crystal germanium as compared with polycrys-
talline germanium is striking.
In Fig. 13, which showed a n-p-n transistor structure,
homogeneous regions havingresistivities p1 and p2 indicate
that we found it possible to achieve a leveling out of the
resistivity along the crystal growth direction by automatic
programming of crucible temperatures andpulling rates.
Ernie Buehler and did I this withempirical programs using
the typeI1 crystal puller.W. P. Slichter was later able to
achieve excellent resultswiththeoreticallycalculated
programs based on theoretical pictures of crystal growth
developed with J. A. Burton [36], [37].
Fig. 21 shows results obtainedby Ernie Buehlerand me
[38] by controlling the impurity distributioncoefficients
632 IEEE TRANSACTIONS ON ELECTRON DEVICES. JULY 1976

number of p-n junctiondevices was the use of single crys-

,tals having a high degreeof lattice perfection and compo-
sition control. To achieve suitable materials for a wide
range of devices, extensivework on specifyingthe crystal
growing processes was undertaken byW. W.Bradley [43],
[44] of my crystal growing group.
In February1951, BellLabs management interest in the
transistor project rapidly reacheda peak because of our
success in achieving greatly improved point contactdevices
and outstanding microwatt junction transistors. As a re-
sult, outresearch effort was enlarged, starting immediately
with the addition of more people to our groups and ef-
fecting, over a period of about nine months,consolidation
a
of electronic materials research.I, together with several
of’myassociates, began reporting into a new laboratory,
“ChemicalPhysicsResearch,”headed by Leland A.
Fig. 22. Germanium single crystal grown in the type I1 crystal pulling Wooten and Addison H. White.
machine by Teal and Little andoften used in Bell system advertise- With the developmentof useful devices of a varietyof
ments. kinds, the demands for high-purity materials increased
by changingthe growth rate during the growth of a crq stal strikingly.Satisfaction of thesedemands was greatly
to achieve a “hand tailoring” of the resistivity along the simplified in 1951 when W. G. Pfann used the same prin-
crystal growth axis. ciples of recrystallization that we had used in producing
The germanium single crystal shown in Fig. 22 waa one
‘7 high-purity single crystals, but in a method particularly
of the most beautiful crystalsthat John Little and I grew appropriate for production known as zonerefiningin which
in the typeI1 crystal puller relatively early the in his3;ory a molten zone is repeatedly’ passed through a crystal by
of the machine. It proved a popular one for use in :Bell relative motion between the heater and the crystal [45],
System Advertisements. We were veryproud of this cqstal [461.
at the time we grew it, long before we realized it would During 1951 Ernie Buehler and I turned our attention
enjoy such wide publicity asan importantexample of :3ell more and more towards silicon because of its theoretically
research efforts. expected improved high-temperature capability. I was
Ernie and I later modified this design to a so-called more than a little familiar with the material,having made
improved type I1 crystal puller to facilitate its us(! in pyrolytically deposited silicon microwave diodes experi-
Western Electric production, and added the best h?m- mentally with Keith Storks in 1942 and later, in 1943,
perature controls that we could improvisefrom purche sed pyrolyticallydepositedfrom silicon tetrachlorideand
commercial items. The best control promised by menu- hydrogen precisely vapor doped and purefilms of silicon
facturers of temperature control equipment, even for and some of its alloys, with someof the silicon appearing
well-insulated tube-type furnaces, was f3.0’. To pre\.ent to be the highest purity ever prepared, applying it to in-
meltback of just-formed p-layers and thus to assure ac- frared photoconducting cells of interest to the war effort
ceptable production yieldsof n-p-n transistor structures and handling it underan NDRC project that I headed. My
by the crystal growing process, the Labs undertook de- associates and I later published someof this work in the
velopment of a temperature controller which for the first paper “A new bridge photo-cell employing a photo-con-
time achieved for the short periods needed for p-leyer ductive effect in silicon. Some properties of high purity
preservation a controlto f0.2’ atapproximately 1000” C. silicon,” G. K. Teal, J. R. Fisher, and A. W. Treptow, J.
Controllers made by the Leeds and Northrup Company Applied Physics, vol. 17, pp. 879-886, November 1946. On
using these Bell Labs circuits developed largelyby J. C. the firstpage of this paper we discuss idiochromatic and
Lozier [39], [40], who, assisted by D. C. Weller, worked allochromatic materialsand the fact that materials of the
closely with membersof my group, becamean important first class exhibit photoconductivity when “chemically
aid tocompanies going into thisnew industry. pure and in the form of perfect or nearly perfect single
A good deal of the interest in the single-crystal work crystals”-although the films we prepared and studied
resulted from recognition, by device development engi- were polycrystalline. We concluded that thephotoelectric,
neers, of their needfor improved materials.An extenrjive thermal conductance, and optical data all had common
a
program was thus carried outin the development are,2of origin-electron transistions from the upper filled band
the Labs to determine the suitability of single-crystal to theconducting band.
materials that we were making experimentally for new Fig. 23 isa photographof a cross sectionof a polycrys-
devices under development. Here again single crystals. talline silicon ingot typicalof the period of our studies.
proved themselves in the achievement of better reliability John Little and I first triedto grow siliconsingle crystals
and reproducibility in point contact transistordevices [ 1x1, in our movable open-bottom bell-jar equipment but we
[42]. A key factor in this and in the development of a failed becauseof excessive oxidation that occurred to the
TEALSINGLE CRYSTALS OF GERMANIUMANDSILICON 633

Fig. 23: Polycrystalline silicon ingot cross section revealing a large

number of grain boundaries.

Fig. 24. Early single crystals of silicon grown by G. K. Teal and E. Buehler.

Fig. 25. Single crystal of silicon # 136, Length: 9-10 in.

Fig. 27. Highly symmetrical silicon single crystal. Length 4%in.

We were soon able to grow large highly perfect single

crystals of silicon such as those shown in Figs. 25-27.
Some of the improvementsin carrier mobilit,ies both for
germanium andsilicon can beseen in Fig. 28 in which the
mobilities are plottedin square centimeters/volt-second.
Hall effect measurements by G. L. Pearson for holes in
germanium are shownas 2000. Some drift measurements
Fig. 26. Three silicon single crystals. Lengths: 4$, 3, and 2y4 in. by Haynes and Shockley appear somewhatlower, 1600 or
1700. As already mentioned previously, in the case of
silicon at the elevated temperatures. Following this and electrons, Hall mobility appears to have increased from
using crystal pullers with improved atmosphere control 1100 in 1945 to 3800 as measured by Peter Debye in single
and adding a quartz liner to the graphite crucible, Ernie crystals of exceptionally high purity and uniformity of
Buehl.er and I pulled our earliest silicon rods, shown in Fig. composition of donors and acceptors. This is a value
24, which containedlarge volumes of single-crystal mate- somewhat greater than obtained by Haynes by a drift
rial. It is obvious that these rodswere much more homo- method on a somewhatless pure and uniformcrystal. For
geneaas and contained larger crystals than the ingot of materials to which donors have beenadded to obtain lower
polycrystalline silicon shown in Fig. 23. resistivities themobility decreases. Using the very high-
634 IEEE TRANSACTIONS ON ELECTRON DEVICES, JULY 1976

M O B I L I T I E S OF CARRIERS IN GERMANIUM
AND SILICON the possibilities of epitaxial formation of thin single-crystal
transistors of controlled layer thicknesses by pyrolytic
decomposition of germanium halide vapors. Introducing
desired concentrations of impurities during theuse of this
method seemedto us, in principle, to have advantages over
a process in which layers are achieved by some method of
dedonorizing or deacceptorizing of an already prepared
layer or layers in order to achieve film compositions needed
for a specific functioning of the whole layeredsingle-crystal
structure. We used germanium iodide in our experimen-
tation,.expecting to use other halides in further investi-
gations. We formed both p- and n-type germanium sin-
YEAR
gle-crystal films on single-crystal material of the opposite
ELECTRONS HOLES conductivity type. U.S. Patent 2 692 839, issued to thetwo
HALL EFFECT
DRIFTMEASUREMENT *0 0
0 of us in 1954 [51], proved to be a basic and important
Fig. 28. Mobilities of carriers in germanium and silicon single crystals. patent, broadly covering epitaxial growth on semicon-
ductor bodies, thus including all high-perfectionsingle-
crystal controlled composition p-n junction and/orcon-
purity crystals of uniform acceptor-donor concentrations, ductivity structures of germanium or silicon in which
Debye and Miss Conwell obtained quantitative agreemmt successive layers of different conductivities or of opposite
of the theoretical and experimental values relating rno- conductivity types are formed pyrolytic by decomposition
bility and resistivity in germanium [47]. of germanium or silicon chlorides, bromides, iodides, hy-
Now in the case of silicon, too, one can see that the drides, or mixtures of these with hydrogen with added
measured values of mobilities of the carriers have increased donor or acceptor impurities of elements such as alumi-
with the availability of improved single-crystal materi,sls. num, gallium, phosphorus, or arsenic in an amount to de-
The high value shown by Haynes and Westphal [48] for termine the layer conductivity or conductivity type as
electrons is four times the former polycrystalline value rmd desired. The impurity may decompose by virtue of its
is shown here to be 1200, the value for holes being 250. being a halide,through influence of hydrogen present or
These values were later verified by Pearson by Hall m 3a- by virtue of the influence of both. Essentially, all high-
surements on the same samples used by Haynes. ‘I’he performance transistors and most transistors incorporated
higher values have been thought tobe due in part to1;he into integrated circuits are of this type.
higher purity of the material with respect to chance i m - I will not attempt todiscuss all of the different crystal
purities and also acceptors and in part to a structurElly pullers that my associatesand I designed and built. Earlier
more perfectsolid. These higher values greatly influence I referred to theimproved type I1 crystal puller. We made
all considerations of conduction in silicon and germanic.m. a numberof them. Fig. 29 shows our first one of these and
On May 1,i952, we presented a paper reportingto the the first temperature controller built from purchased parts.
American Physical Society, meeting in Washington, IC, In Fig. 30 is shown the Leeds and Northrupversion of Bell
on growth of the firstlarge high-perfection single crystals Labs precision temperature controllerwhich held a melt
of silicon and single-crystalsilicon p-n junctions[49]. A t constant tof 0.2O at approximately 1000°C. We furnished
that timewe showed slidesof several of the silicon sing1.e several improvedtype I1 pullers to Western Electric, but
crystals shown in Figs. 24-27. In some of our silicon single with aprecision temperature controller in each case. The
crystals lifetimes of holes wereas high as 200 p s , compa~ed crystal puller in the right of Fig. 30 is a very flexible one
with less than 1 p s for polycrystalline silicon. We made with servodiameter control and adigital programmer, both
public both of these values on presenting our paper ref- designed by Fred Buhrendorf, whowas in my crystal
erenced above. Diodes made from our grown silicon 1,-n growing group a t Bell Labs.
‘junctions by Kenneth McAfee and Gerald Pearson [50]E.ad Because of a very critical family need I left Bell Labs at
the low reverse saturation currents expected on the basis the end of 1952 and joined Texas Instruments Incorpo-
of the band separation in silicon. From the equation rated, at that time a smallaggressive engineering, elec-
(shown previously), one can understand the reverse cur- tronics, andgeophysical company, in Dallas, Texas, with
rents lower by a factor of 104-105 obtained in thecase of only 1770 employees. It had shortly prior to that time be-
single-crystal silicon p-n junctions as compared with sin- come a licensee of the Western Electric Company, withthe
gle-crystal germanium junctions. The low reverse currer.ts, aim of going into the transistorbusiness.
good saturation of the reverse current, and sharpZener I went to Texas Instruments to organize and direct its
voltage breaks are due to the silicon single crystals having research in a central research laboratory. My job was to
a high degreeof lattice perfection and chemical purit~,. assemble aresearch staff and toestablish the environment
In addition, Ernie Buehler and I made the first silicon for and to direct innovation in contrast to performing
single crystals having in them n-p-n and p-n-p structures.personal invention. I gradually recruited a group of re-
During 1950-1952 I obtainedthe collaboration of searchem and concentrated our efforts on programscal-
Howard Christensen in investigating, as a general method, culated to comprehend and extend the forefront technol-
TEAL:SINGLECRYSTALS OF GERMANIUMANDSILICON
Fig. 29. Improved type I1 crystal pulling machine.
Fig. 30. Crystal pulling machine with servo diameter control and digital grown junction barwe used to make the first
programmmg.
ogy and tolead as quickly as possible to important prod- more evident.
ucts thatwould exert a valuable impacton the company.
Shortly afterarriving at TI, I set up a program
crystal growing aimed at producing high-perfection silicon
single-crystal p-n junction structures t o facilitate our de- ry, for with this advance it gained
velopment of a silicon transistor withuseful amplification
properties. I persuaded Dr. Willis A. Adcock, an able young
scientist, toleave his catalysis studiesin one of the oil in-
dustry laboratories andjoin TI to undertake these crystal the big leagues.
growing investigations, whichI believed to be the key to
the achievement of a silicon transistor. I reasoned that search discussed above contrasts strikingly with the ex-
635
going the grown junction route would avoidthe differential
expansion difficulties between silicon and an alloying
electrode inherent in use of alloy junctions. Most com-
panies took the alloy route. Another possible approach was
to use the thenrecently discovered 111-V compounds. Some
scientists at the time were advocating “leapfrogging” sil-
icon and going to the111-Vcompounds to achieve an even
higher temperature capability than potentially available
in silicon. Certain companies followed this advice and
concentrated on 111-V intermetallics. I argued that thiswas
the wrong approach because of the added complexity of
the 111-V’s and because of the superior chemical and
physical stability of silicon.
Our program succeededin 1954 and the first commer-
cially feasible silicon transistor became a reality.In addi-
tion to the contributions of Willis Adcock, I should men-
tion thoseof Dr. Morton Jones relating to the solution of
electrode attachment andof chemical problems, and his
transfer of the technology to theSemiconductor Products
Division. The excellent performance of the crystalgrower
designed and constructedby Boyd Cornelisonof the S-C
Division was an important factoralso. The experimental
results of the development were published in the Pro-
ceedings of the Institute of Radio Engineers by w. A.
Adcock, M. E. Jones, J. W. Thornhill, and E. D. Jackson
[51],[52]. As director of this research, I announced the
achievement at an IRE NationalConference in Dayton,
Ohio, onMay 10,1954 [53]. The circumstances were rather
dramatic. During the morning sessions, the Speakers had
unwittingly set thestage for us. Oneafter another they had
remarked abouthow hopeless it was to expect thedevel-
opment of a silicon transistor in less than several years.
They advised the industry to be satisfied with germanium
transistors for the present. We of TI listened with great
respect-andmountingexultation-because I hada
handful of excellent silicon transistors in my pocket.
Somehow, the last sentence of my31 page prepared speech
announcing that TI was in commercial production of the
silicon transistor suddenly seemed incomplete, so I quickly
added a handwrittenclosing statement. The first and last
pages of my paper aregiven in Fig. 31. In my excitement
I even omitted the preposition “to.”
Fig. 32 shows a dra.wing of the type of n-p-n silicon
grown junction transistor thatwe announced on May 10,
1954. The photo of Fig. 33 shows the size of the silicon
silicon tran-
sistors. It was almost y4 in long. In spite of its small di-
mensions the importance of its benefits steadilybecame
Some of the economic results were highlighted in an
on silicon issue of Fortune (November 1961, p. 226):
T h e silicon transistor wasa turning point inTI’s histo-
a big headstart over the
competition in a critical electronic product; there was no
effective competition in silicon transistors until1958. TI’s
sales rose almost vertically; the company was suddenly in
The tiny size of the silicon transistors made in the re-
636 IEEE TRANSACTIONS ON ELECTRON DEVICES, JULY 1976

Fig. 31. On May 10,1954 theIXectronics Industry was startled by Dr.

Gordon K. Teal's announcem a t at a technical conference that Texas
Instruments was in commerc;all production of the silicon transistor.

SINGLE CRYSTAL
(ACCEPTOR TYPE SILICON BAR
MATERIAL,1.E.ALUMINUM)

COLLECTOR EMITTER
LEAD

9
n - TYPE P-TYaE n - TYPE
LR'ER

Fig. 32. The May 10, 1954n .p-n silicon grown junction transistor.

of this material tothe chemical and electronic industries.

Fig. 33. Early single-crystal grown junction silicon transistor bar
pectations for silicon materials and devices and, mcrl'e
generally, for silicon technology that it generatedin us, as
evidenced by the words of my paper of 1954, that we ha\,e
just scanned. We in the Central Research Laboratories,as
a result of these expectations, almost immediately U Y I -
dertook, in 1954, the development of a chemical reduction
method for high-purity silicon production in anticipation
of TI supplying itsown need for silicon for its own device 3,
and that started TI towards becomingthe leading supplier
I have previously discussed some of my experience during
World War I1 in making ultrahigh-purity silicon by hy-
drogen andsilicon tetrachloride decomposition. This ex-
perience influenced Willis Adcock and me towards the
chemical method we proposed. In undertaking thedevel-
opment of a production method in1954, a number of as-
sociates in CRLalso became involved. The work was car-
ried out for the most part in the CRL Materials and
Components Department headedby Willis Adcock, who
was especially active inthe effortin the early stages, with
the latterstages being spearheaded by Ray Sangster.John
Ross and JamesL. Fischer were also deeply involved. Over
a periodof years TI furnished industry over half its total
high-purity silicon requirements.
TEAL:SINGLECRYSTALS OF GERMANIUMANDSILICON

Fig, 35. Large silicon single crystal, graphite crucible, and quartz liner.
Fig. 34. LSI chip, 4096-bit Read Only Memory.

It is hard tooverestimate the impactof TI’Scommercial

lead in silicon transistors. Therehave been many benefits
for the company in addition to the dollars from salesof
devices. It contributed to a broadeningof the company’s
military electronics, where high-temperature performance
of the silicon transistor was essential to meeting military
equiplment specifications. The unique expertise in silicon
was also important in successful negotiations in setting up
fully owned TI subsidiaries in England and France. Of
greatest importance, perhaps,was that silicon technology
led into the developmentof integrated circuits.
The availability of hi.gh-perfection single-crystal silicon
and germanium and the successful commercial introduc-
tion of the silicon transistor opened theflood gates to new
Fig. 36. Modern crystal pullers.
technologies [4], [5],making use of single-crystal materials.
Diffusion, mesa, planar, and epitaxial techniques produced
useful new discrete devices. A specially significant advance cessing low,the trend has been to make the silicon crystals
was made in 1958, whenJack Kilby of Texas Instruments of 3 or 4-in diameter. Eighty-85 percent of the marketfor
fabricated the firstworking integrated circuit, employing single crystals is therefore served by silicon singlecrystals
a concept that has madepossible the implementation of pulled from quartz crucibles. Fig. 35shows alarge silicon
many functions on singlea chip of single-crystal silicon. single crystal that has been pulled from a quartz liner in
In Fig. 34 is a large-scale integrated circuiton a chipof ,graphite crucible. The photograph of Fig. 36 shows a
‘/s
single-crystal silicon approximately in2. It is capable of number
production
of crystal
line a t
pullers
Texas
in use in a
Instruments.
silicon single-crystal
storing 4096 bits of data in a Read Only Memory. Jack
Kilby will discuss this integrated circuit along with others. In his 1962 Presidential Address to the BritishAssoci-
Today’s silicon singlecrystals are agood deal larger than ation fortheAdvancement of Science [57] SirJohn
those we produced 25 years ago, but most of them arestill Cockcroft (Nobel Laureate, 1951), commenting on his visit
produced from silicon melted in a graphite crucible fur- t o Bell Laboratories in 1950, alluded to the single-crystal
nished with a quartzliner entirely similarto theones we work and forecast its impact:
developed. In the intervening years, however, an alternate Twelve years orso ago I visited the Bell Telephone Lab-
method of silicon refining, float zone refining, was devel- oratories in New Jersey and saw scientists drawing single
oped by P. H. Keck [54] and independently by R. Emeis crystals of verypuregermaniumfromamoltenmass.
[55] and H. C. Theuerer [56]. These experimenters con- These very pure crystals were seeded with small quantities
of vital impurities and then cut up form the transistors
trived to form astable molten zone in a vertical rod of sil- which are now most familiar to us in our portable wireless
icon such that zone refining can be extended tosilicon. In set. Their influence on the development of electronics has
silicon this technique has the advantage of removing any been very great. Their great reliability and small power re-
residual impurity of oxygen fromthe silicon. This has been quirements have led to a reduction in size and a great in-
of value in single-crystal material usedin power devices creaseinreliability of electronicequipment.Without
them, space science and space travel would hardly have
such as the thyristor. beenpossible.Theyhavealsomadepossiblethevery
Integrated circuits arevery demanding with respect to much more powerful next generation of computers, such
low cost of the material and the utmost in performance and as Atlas . . . .Computers are playing a creative role in the
reliability. In integrated circuits to keep the costs of pro- development of sciences such as molecular biology, high
638
energy nuclear physics, the studyof stellar evolution, plas.
m a physics, theoretical chemistryand many other b r a n c h
es .. ..
They are becoming an indispensable tool for design
of complex equipment such aircraft, nuclear power sta.
tions and are even penetrating ship design. They are be-
coming important in the control of industrial organiza.
..
tions. and studies of our economy. . ..
Shockley has said (W. Shockley, Chapter 7, “Transis-
tors,” p. 148 of the book The Age of Electronics, Lincoln
Laboratory Decennial Lectures,edited by Carl F. J. Ov-
erhage, McGraw-Hill, NY,1962):
There was probablyno more important scientific de.vel.
opment in the semiconductor field in the early days fol.
lowing the announcement of the transistorthan t h e devel.
opment of high-quality single crystals of germanium at
Bell Telephone Laboratories.’’
In this lecture Shockley also identified G. K. Teal as i ~ l i -
tiating and carrying out the research, publishing with J.
B. Little the first announcement of success in preparing
germanium single crystals, the method used being a pulling
process.
In their “History of semiconductor research,” G. L.
Pearson and Walter H.Brattain (Proceedings IRE, vol.
43, pp. 1803 and 1804, December 1955) wrote:
Duringthistimevariousimprovementswerebeing
made in semiconductor materials.A big step forward was
taken when G. K. Teal and J. B. Little succeeded in grow-
ing single crystals of germanium. Sometime later single
crystals of silicon were also obtained byG. K. Teal and E.
Buehler. In all of this work steps were continually being
taken to produce even more perfect crystals both as to lat-
tice perfection and degree of chemical purity . . . We are
getting so close to the forefrontthat the perspective is in-
sufficient to give a good general picture.It can certainly be
said, however, that the availability of such pure andper-
fect single crystals as we have in present-day silicon and
germanium amounts toa major revolution in the physids
of solids. New phenomena are turning up all around us
....
Shockley, in November 1952 in the Transistor Issue of
the Proceedings of the IRE ( “ T r a n s i s t o r electronics: Im-
perfections, unipolar and analog transistors,” Proceedings
IRE, vol. 40, p. 3) said:
...Teal and Little undertook a program of growing
large crystals of germanium. These single crystals, weigh-
ing up to several hundred grams, havethe same orienta-
tion throughout and no grain boundaries. For the last few
years, practically all advances at Bell Telephone Labora-
and transistor physics have
tories in transistor electronics
been based on the availability of single crystal material.
Writing in May 1968 PhysicsToday (p. 54), with
added time for perspective in evaluating “twenty years of
physics, solid state,” John M. Ziman said:
Freedom from extraneous interference is achieved with
very pure and perfect crystals. New techniques for crystal
I I
manium and silicon semiconducting devices, but they have
-
-growth. . . were needed orieinallv inthe fabrication of eer-
profoundly altered the solid-state physicist’s attitude to
the objects of his regard; for they allow the observation of
IEEE TRANSACTIONS ON ELECTRON DEVICES, JULY 1976
phenomena that would otherwise be completely obscured
by the scattering of electrons or phonons from static im-
purities and imperfections.
REFERENCES
J. Bardeen and W. H. Brattain, “Thetransistor, a semiconductor
triode,” Phys. Rev., vol. 74, pp. 230-231, July 15, 1948.
J. Bardeen and W. H. Brattain, “Physical principles involved in
transistor action,” Phys.Rev., v.01. 75, pp. 1208-1225, April 1949.
G. K. Teal, M. D. Rigterink, and C. J. Frosch, “Attenuator materials,
attenuators andterminations,” Trans. AIEE, vol. 67, pp. 754-757,
1948; Bell Labs Monograph B1571.
G. K.Teal, “Roots of creative research,” IDEA, vol. 11, pp. 1-6,
Spring 1967 (published by The Patent Trademark and Copyright
Institute of the George Washington University).
G. K. Teal, “Roots of creative research,” Nurturing New Ideas:
Legal Rights and Economic Roles,pp. 18-22. Edited by J. L. Harris
(published by the Bureau of National Affairs, Inc., Washington, DC,
1969).
“Tribute to Harold de Forest Arnold,’’ article published shortly after
his death, July10,1933, Summit, NJ. Published in BellLaboratories
Record, vol. 11, pp. 350-359, August 1933.
“A quarter century of transcontinental telephone service, 1915-
1940,” Bell Telephone Quarterly, vol. 19, .. pp. 2-58, January 1940.
See [4] and 151.
G. K. Teal, “Reflections on early germanium and silicon single
crvstal research.” Invited Paver. IEEE 1968 International Electron
Devices Meeting, Washington, DC, October 1968 (18 pages).
J. R. Haynes, private communication to G. K. Teal in early 1949.
The references from (111 to 1191 are for general information. Up to I
the end of World War I1 the reader is referred to H. C. Torrey and
C. A. Whitmer, Editors, Crystal Rectifiers, McGraw-Hill, NY,1948.
This book describes the contributions of Lark-Horovitz and his
associates at Purdue University, Scaff, Schumacher and associates
at Bell Laboratories, North at theGeneral Electric Co., and Seitz
at TheUniversity of Pennsylvania. Several other references [12]
-[19] are good general references. [20] summarizes some of Bell’s
World War I1 work.
W. Shockley, “Electrons and Holes in Semiconductors,” Van
Nostrand, NY, 1950.
K. Lark-Horovitz, “The new electronics,” The Present State of
Physics, AAAS, Washington, 1954.
E. Spenke, “ElektronischeHalbleiter,” Springer, Berlin, 1955.
R. L. Petritz, “Contributions of materials technology to semicon-
ductor devices,” 50th Anniversary issue of Proc. IRE, vol. 50, .. pp.
1025-1038, May 1962.
N. B.Hannay, Ed., Semiconductors, Reinhold,NY, 1959.
G. L. Pearson and W. H. Brattain. “Historv of semiconductor re-
search,” Proc. IRE, vol. 43, pp. 1794-1806, December 1955.
W. R. Runyon, Silicon Semiconductor Technology, McGraw-Hill,
NY, 1965.
Special Issue on Transistors, Proc. IRE, vol. 40, pp. 1-1283, No-
vember 1952.
J. H. Scaff,H. C. Theuerer, and E. Schumacher, “P-type and N-type
silicon and theformation of the photovoltaic barrier in silicon in-
gots,” Trans. Met. SOC.AIME, vol. 185, pp. 383-388, June 1949.
G. K. Teal, “Chapter 4, germanium single crystals: Introduction,”
pp. 69-71, Transistor Technology, vol. 1, published by Bell Tele-
phone Laboratories and Western Electric Co., September 1952.
G. K. Teal, “Chapter4, preparation of germanium single crystal by
the pullin method: Introduction,” Transistor Technology, vol. 1,
8.
edited by E. Bridgers, J. H. Scaff and J. N. Shive, published by
Van Nostrand, NY, 1958.
G. K. Teal, “Germanium and silicon single crystals,” Invited paper,
Phys. Rev. vol. 87, p. 221, 1952.
J. R. Haynes and W. Shockley, “The mobility and life of injected
holes and electrons in germanium,” Phys. Rev., vol. 81, pp. 835-843,
March 1951.
W. Shockley, G. L. Pearson, and J. R. Haynes, “Hole injection in
germanium-Quantitative studies and filamentary transistors,”
BSTJ, vol. 28, p. 344, July 1949.
G. K. Teal andJ. B. Little, “Growth of germanium single crystals,”
Phys. Rev., vol. 78, p. 647, 1950. Abstract read by title only at the
APS meeting in Oak Ridge,TN, March 18,1950.The abstract gave
the upper lifetime of 200 F S but did not publicly reveal a definite
com arison with 7 found in polycrystalline germanium. In the in-
vitex .
paper
. . - I gave
of 1231, - 1-5 FS to be normal for .polycrystalline
. .
germanium.
A mobility of 1156 cmz/Vs for electrons in ermanium as a mea-
surement of K. Lark-Horovitz and V. A. Joanson in 1946. At the
same time they obtained a mobility of 385 cm2/Vs.This information
comes from v. 63 of Torrev and Whitmer 1111.
[28] This value ;as obtained 6y Hall measurements by P. Debye after
TEAL SINGLE CRYSTALS OF GERMANIUM AND SILICON
special purification and by achieving great donor and acceptor
uniformity in germanium.
[29] G. K.Teal, M. Sparks, and E. Buehler, “Growth of germanium single
crystals containing p-n junctions,”Phys. Rev., vol. 81, p. 637, Feb-
ruary 1951.
[30] F. S. Goucher, G. L. Pearson, M. Sparks, G. K. Teal, and W.
Slhockley, “Theory and experiment for a germanium p-n junction,”
Phys. Rev.,vol. 81, p. 638, February 15,1951.
[31] W. Shockley, M. Sparks, andG. K. Teal, “p-n junction transistors,”
Phys. Rev., vol. 83, pp. 151-162, July 1, 1961.
[32] C. Wagner, “On the theory of rectification,” Physikalische Zeits-
chrift, vol. 32, pp. 641-645, 1931.
[33] W. Shockley, “The theory of p-n junctions insemiconductors and
p-n junction transistors,”B S T J , vol. 28, pp. 435-489, July 1949.
[34] W . P. Slichter and J. A. Burton, “Chapter 6, the distribution of solute
elements: Transient conditions,” Section6.4, Transistor Technol-
ogy; uol. 1, pp. 119-142, Bell Telephone Laboratoriesand Western
Electric, September 1952.
[35] W . P. Slichter and J. A. Burton, “Chapter 6, the distribution of solute
elements: Transient conditions,” Section6.4, Transistor Technol-
ogy, Vol. 1, pp. 107-129, edited by H.E. Bridgers, J. A. Scaff, and
J. N. Shive, Van Nostrand Co., NY, 1958.
[36] See chapter 6 of [34] and [35] and J. A. Burton andW. P. Slichter,
“Chapter 5, thedistribution of solute elements: Steadystate
growth,” pp. 79-118, “Transistor Technology; vol. 1,” Bell Labo-
ratories and Western Electric Co., September 1952.
[37] See chapter 6 of [34] and [35] and J. A. Burton andW. P. Slichter,
“Chapter 5, the distribution of solute elements: Steady state
growth,” pp. 71-107, Transistor Technology, vol. 1, Van Nostrand
Co., NY, 1958.
[38] G, K.Teal andE. Buehlier, “The electrical properties of germanium
single crystals of single and mixed conductivity types,”IRE-AIEE
Device Research Conference, Durham, NH, June1951 (6 pages).
[39] J. C. Lozier, “Chapter 8, controls for the n-p-ncrystal growing ma-
chines,” pp. 181-182, Transistor Technology, uol. 1, Bell Telephone
Laboratories and Western Electric Co., September 1952.
[40] J. C. Lozier, “Chapter 8, controls for the n-p-ncrystal growing ma-
chine,” pp. 155-164, Transistor Technology,uol. l , Van Nostrand
Co., NY, 1958.
- . J. A. Morton, “Present status of transistor development,” B S T J ,
(411
vol. 31, pp. 411-442, 1952.
[42] G. K. Teal,M. Sparks, andE. Buehler, “Single crystal germanium,”
PTOC. IRE, V O ~ .40, pp. 906-909, August 1952.
639
[43] W. W. Bradley, “Chapter 7, Preparation of germanium single
crystals,” pp. 143-170, Transistor Technology, vol. 1, Bell Labo-
ratories and Western Electric Co., 1952.
[44] W.W. Bradley, “Chapter 7, Preparation of ermanium single
crystals,” pp. 130-154, Transistor Technology, vof. 1, Van Nostrand,
New York, 1958.
[45] W. G. Pfann, “Segregation of two solutes, with particular reference
to semiconductors,” J. Metals, vol. 4, pp. 861-865, August 1952.
[46] W. G. Pfann, “Techniques sf zone melting and crystal growing,”
Solid State Physics,edited by F. Seitz and D. Turnbull, Academic
Press, Inc., vol. 1, pp. 423-521, NY, 1957.
[47] P. Debye and E. Conwell, “Mobility of electrons in Ge,” Phys. Rev.,
vol. 87, p. 1131,1952.
[48] J. R. Haynesand W. C. Westphal, “The driftmobility of electrons
in silicon,” Phys. Rev.,vol. 85, p. 680, February 15,1952.
[49] G. K. Teal and E. Buehler, “Growth of silicon single crystals and
of single crystal p-n junctions,” Phys. Rev., vol. 87, p. 190,1952,
presented at the APS meeting in Washington, DC, May 1;1952.
[50] K. B. McAfeeand G. L. Pearson, “The electrical properties of silicon
p - n junctions grown from the melt, ” Phys. Reu., vol. 87, 190,
1952, presented at the APS meeting in Washington, DC, Mayl,
,nco
lili).G.
[51] Patent a plied for March 7,1951; granted to us on October 26th,
1954-U.f 2 692 839.
I521 W. A. Adcock. M. E. Jones. J. W. Thornhill. and E. D. Jackson.
“Silicon transistor,” Proc. IRE, vol. 42, p. 1192, July 1954.
[53] G. K. Teal, “Some recent developments insilicon and germanium
materials and devices,” resented National IRE Conference, Day-
ton, OH, May 10,1954. Zonference sponsored by IRE Navigational
and Airborne Electronics Group, (31 pages).
[54] P. H. Keck and M. J. E. Golay, “Crystallization of silicon from a
floating.liquid zone,” Phys. Rev.,vol. 89, p. 1297, March 1953.
[55] R. Emels, “Growing silicon single crystals without a crucible,” 2.
Naturforschung, vol. 9a, p. 67, January 1954.
[56] H. C. Theuerer, “Removal of boron from silicon by hydrogen-water
vapor treatment,” J.Metals, vol. 8, pp. 1316-1319, October 1956.
[57] J:D. Cockcroft, Presidential address given on Augusk 22, 1962 to
the British Association for the Advancement of Science, “Invest-
ment in Science,”Advancement of Science, vol. 19, September 1962.
Nature, vol. 195, pp. 875-880, September 1,1962.
[58] I regretthat space does not permit acknowledgment to all individ-
uals who contributed.