Download - How the Transistor Effect Invented

8/9/2019 How the Transistor Effect Invented

1/34


2/34


3/34

inority earners and the irst Two Transistors

flowing through it. By applying suitable voltages to two

circuits passing through the semiconductor, Shockleypredicted that an input signal applied to one circuit couldyield an amplified signal in the other.

Unfortunately all the attempts to fabricateShockley s field-effect device failed i as did Shockley stheoretical attempt to explain why. On the basis of theavailable semiconductor rectification theory by Nevill Mott

and Walter Schottky, his conceptual field-effect device didnot work s predicted.

In October 1945 John Bardeen joined the newsemiconductor group that was being formed under Shockleywithin Bell s larger solid state department. Shockley askedBardeen to check calculations that he had made to examinewhy attempts to demonstrate his design for a field-effectamplifier had failed. By March 1946 Bardeen had ananswer. He explained the lack of significant modulation ofthe conductivity using a creative heuristic model, the idea of surface states. 12 In this model, electrons drawn to thesemiconductor surface by the applied field become trapped inthe postulated states and are thus unable to act as chargecarriers. As Shockley later put it, the surface states blockedthe external field at the surface and shielded the interiorof the semiconductor from the influence of the positivelycharged control plate. 13

But were these postulated states real? And if so,how did they generally behave? These questions became ofprimary interest to the semiconductor group, which in thefollowing months responded to Bardeen s surface state

theory with n intensive research program to explore thesestates. Bardeen worked closely on the problem with thegroup s experimental physicists, Walter Brattain and GeraldPearson. 4

3


4/34

Riordan oddeson

Brattain made a discovery on 7 November 1947.Drawing on a suggestion by Robert Gibney, a physicalchemist in the semiconductor group, he found he couldneutralize the field-blocking effect of the surface states byimmersing a semiconductor in an electrolyte. I This newfinding was electrifying, observed Shockley. At long last,Brattain and Gibney had overcome the blocking effect of the

surface states. 16 It set in motion the events that wouldculminate one month later in the first transistor.

Four days after this discovery, Bardeen and Brattaintried to use the results to build a field-effect amplifier. Theirapproach was based on Bardeen s suggestion to use a pointcontact electrode pressed into a specially prepared siliconsurface. Rather than use the thin films employed in the 1945

experiments by Shockley and his collaborators, Bardeenproposed the use of an n type inversion layer a fewmicrons thick that had been chemically produced on theoriginally uniform surface of p type silicon. (The existenceof such inversion layers of n type semiconductor on a p-type substrate, or vice versa had been demonstrated byearlier experimental studies at Bell Labs by Russell Ohl andJack Scaff). Because charge c r r i e r ~ i nthis case,electrons would have a much higher mobility in such aninversion layer than they had in the vapor-deposited filmsemployed in the 1945 experiments, Bardeen believed thisapproach would work better in a field-effect amplifier. I? Inparticular, this layer would act as a narrow channel in whichthe population of charge carriers could be easily modulatedby an applied external field. I8 The device test ed on

November used a drop of electrolyte on the surface asone contact and a metal point as the other; Bardeen andBrattain obtained significant power amplification, but thefrequency response was poor. I9


5/34

Minority Carriers and the First Two Transistors

The next crucial step occurred on December 8. At

Bardeen's suggestion, Brattain replaced the silicon with anavailable slab of n type high-back-voltage germanium, amaterial developed during the wartime radar program by aresearch group at Purdue directed by Karl Lark-Horovitz.

They obtained a power gain of 33 but with a negativepotential applied to the droplet instead of positive, as theyhad expected. Although the slab had not been speciallyprepared, Bardeen proposed that an inversion layer wasbeing induced electrically, by the strong fields under thedroplet. Bardeen suggests that the surface field is so strongthat one is actually getting P type conduction near thesurface, wrote Brattain that day, and the negative potentialon the grid is increasing the P type or hole conduction. 21

Later that week Brattain evaporated a gold controlpla te onto a specially prepared germanium slab (whichtherefore had an inversion layer, a priori); he was trying toimprove the frequency response by eliminating the droplet.He thought the gold would be insulated from the germaniumby a thin oxide layer, but through a failure in his procedure itwas instead directly in contact with the semiconductorsurface. This f'Trersight proved to be a critical step towardthe point-contact transistor.

The following Monday, December 15, Bardeen andBrattain were surprised to find that they could still modulatethe output voltage and current at a point contact positionedclose to the gold plate, however only when the plate wasbiased pos i t iv ly th opposite of what they ha dexpected 23 An increase in positive bias increased ratherthan decreased the reverse current to the point contact,wrote Bardeen almost ten years later. This finding, as heexplained in his 1956 Nobel lecture, suggested that holes

5


6/34

iordan oddeson

were flowing into the germanium surface from the gold spotand that the holes introduced in this way flowed into thepoint contact to enhance the reverse current. This was thefirst indication of the transistor effect. 24 Although Brattainand Bardeen failed to observe any power amplification withthis configuration, Bardeen suggested that it would occur iftwo narrow contacts could be spaced only a few milsapart. Brattain achieved the exacting specifications bywrapping a piece of gold foil around one edge of a triangularpolystyrene wedge and slitting the foil carefully along thatedge. He then pressed the wedge and the two closelyspaced gold contacts down into the germanium surfaceusing a makeshift spring (See Figures 1 and 2). In their firsttests, on December 16, the device worked just as expected.It achieved both voltage and power gains at frequencies up to1000 Hz. The transistor had finally been born. A weekafter that, on 23 December 1947, the device was officiallydemonstrated to Bell Labs executives in a circuit that allowedthem to he ar am pl if ied speech through a pair ofheadphones. (See Figure 3 for Brattain's record of thisevent in his notebook).

The Flo of harge Carriers

The crucial issue for us here is, how did Bardeen andBrattain conceptualize the flow of charge carriers while theywere developing the first transistor? Memory is imperfect,and later accounts are often subject to retrospectiverea lism, a te rm coined by historian and sociologist ofscience Andrew Pickering to describe the process wherebyconjectures are la ter imbued with an aura of certainty, orembellished with details that became known only at a later


7/34

Minority arriers and the irst Two Transistors

time. Fortunately, we have available several telling entries

Bardeen, Brattain and Shockley made in their laboratorynotebooks during those pivotal weeks before and afterChristmas 1947. 28

control@ lIoge _1Z. Y.Et

~ ~ : s i : °le : _n-type Qermorllum

bose

Figure 1: Photograph of the

point-contact transistor invented

by Bardeen and Brattain in

December 1947. A strip of gold

foil s li t a long one edge is

pressed down into the surface of a

germanium slab b= - polystyrene

wedge. (Reprinted with the

permission of AT T Archives)

Figure 2: Schematic diagram of

the first transistor (Figure 1).

The signal current flows

through the input circuit,

generat ing holes in a p typ

inversion layer that modulate the

flow of current in an output

circuit. (Reprinted from L.

Hoddeson, ''The Discovery of the

Point-Contact Transistor )

On December 19, three days after the first successfultest of their device, Brattain wrote: It would appear thenthat the modulation obtained when the grid point is bias is

7


8/34

iordan oddeson

D A n ~ 4 yt ~CASE No. 7

Figure : Entry in Brattain's lab notebook describing the circuit used in the

December 1947 demonstration the point-contact. transistor to Bell

Labs e x e c u t i v ~ s(Reprinted with the permission AT T Archives)

8


9/34

Minority arriers and the First Two Transistors

due to the grid furnishing holes to the plate point. 29 By

grid point and plate point, he was referring to what we nowcall the emitter and ol le tor he was obviously using afamiliar vacuum tube analogy. Although we cannotdetermine from this passage how he conceived the details of

their flow, we can be certain he understood that positivelycharged quantum-mechanical entities were the charge carriersresponsible for modulation.

Bardeen gave a more detailed explanation in anotebook entry dated 24 December 1947, the day after theteam made its official demonstration. After describing thesetup, which used a slab of n-type germanium speciallyprepared to produce a very thin inversion layer of p typeconductivity near its surface, he portrayed the phenomenonas follows (see Figure 4):

When A is positive, holes are emitted into the semi-conductor.

These spread out into the thin P-type layer. Those which

come in the vicinity of B are attracted and enter the electrode.

Thus A acts as a cathode and B as a plate in the analogous

vacuum tube circuit. 30

Again it was clear that Bardeen also attributed the transistoraction to the holes, basing his model on the same va ffi-

tube analogy Brattain used. But he went a step farther andstated that the flow of these holes occurred within theinversion layer.

This emerging theory of the transistor based on theflow of holes at or near the surface of the germanium

developed further during the following six months, theperiod in which Bell Labs kept the discovery of the transistor laboratory secret, while patent applications were beingdrawn up. A drawing found in Bardeen and Brattain s

9


10/34

iordan oddeson

DATE a c ~ - 1 1 1 7 tCASE No. Jtf j

Figure 4: Entry in Bardeen s lab notebook dated 24 December 1947, giving

his conception how the point-contact transistor functions. (Reprinted

with the permission AT T Archives)


11/34


patent application of 17 June 1948 (revised from a version

submitted on February 26 . suggests that although the flowof charge carriers was thought to occur largely within the p-type inversion layer, they were by this time allowing thatsome holes might diffuse into the body of the n-typegermanium (see Figure 5). The text of their applicationstates:

potential probe measurements on the surface of the block,

made with the collector disconnected, indicate that the major

part of the emitter current travels on or close to the surface of

the block, substantially laterally in all directions away from

the emitter 5 before crossing the barrier 4. 31

Oct. 3, 1950 1 H W : E L E c T £ o ~ A ~ W ~ W ,. UTILIZINC 2,524,035 ...SOIICONDUC :lVE IlATERIALS

rUed U e 1 7. 1948 S Sb•• t sh•• \ 1

FIG I

F igur e 5: Figures 1 and 1A from Bardeen and B rat ta in s U.S. patent (No.

2 ,524,035) on the point -con tac t trans is tor. The dominant flow between

emitter collector occurs through a p-type inversion layer on a slab of

type germanium (U.S. Patent Office).

In their famous letter submitted to the hysic l

Review on June 25 1948, they wrote that as a result of theexistence of the thin p type inversion layer next to thegermanium surface, the current in the forward directionwith respect to the block is composed in large part of holes,


12/34

Riordan oddeson

i.e., of carriers of sign opposite to those normally in excess

in the body of the block. 32 In a subtle shift from theirear li er concept ion they envisioned that holes flowpredominant ly in the p type inversion layer, but with aportion that can also flow through the n type layer beneath it.

It is not clear from these entries just how and whythis shift occurred. But both the revised patent applicationand the hysical eview letter are dated well afterShockley s disclosure of the junction tr ansi stor in lateJanuary and a crucial mid-February experiment discussedbelow) by John Shive.

Research a t Purdue

Bell Labs was not the only institution at whichresearchers were interested in the behavior of germaniumsemiconductors. Dur ing the wartime efforts to developcrystal recti fiers for radar receivers, Lark-Horovitz sresearch group at Purdue had greatly advanced understanding of the element germanium. As part of this study,the group explored a phenomenon·known as the spreadingresistance of germanium-the resistance to the current flowthat diverges from a metal point-contact pressed into thesurface of the semiconductor. Lark-Horovitz asked graduatestudent Ralph Bray to study this spreading resistance, in partto determine whether the local resistivity near a point contactmatched the bulk resistivity of the entire sample.

The Purdue group continued this work after the war.In early 1947 Bray noticed that a spreading resistance muchlower than predicted by theory occurs when high fieldpulses of positive voltage were applied through a metal wireto certain samples. At the time, he did not realize that the

12


13/34


effect was being caused by holes injected into the n-type

germanium at the point where the wire touched thegermanium. The spreading resis tance was sort of amystery and nobody understood it, he admitted much later.But after reading Bardeen and Brattain s transistor paper, hecould finally explain everything they had seen by theproduction of holes. I guess we were on the verge of it inthinking about such [a possibility], Bray said, but it hadn t

really formulated itself in our minds. 35Even so, Brattain and Bardeen were concerned the

Purdue group might beat them to the discovery of thetransistor. They were especially nervous during the periodof laboratory secrecy.36 Jane Bardeen recalls that Johngave her hell for mentioning that John was working onsemiconductors in a letter to a friend who was acquaintedwith people in the Purdue group. He felt her friend couldpossibly let the cat out of the bag. 37

Similarly, Brat tain found himself in an awkwardposition when he heard Seymour Benzer, another member ofthe Purdue group, mention the spreading resistance in lateJanuary 1948 at the New York American Physical Societymeeting. He understood by then why the resistivity wasdecreasing near the point that it was the result of holeemission at the point. Brattain recalled listening quietly toBenzer in the corridor, until Benzer remarked, I think ifsomebody put another point contact down on the surface,close to this point, and measured the distribution of potentialaround the point, then we might be able to understand whatthis [effect] is about. To that Brattain replied, Yes, I thinkmaybe that would be a very good experiment and walkedaway.38

13


14/34

iordan oddeson

The Conception the Junction Transistor

In the weeks that followed the invention of the pointcontact transistor, Shockley was torn by conflictingemotions. Although Bardeen and Brattain s invention hadbeen a magnificent Christmas present to Bell Labs, he waschagrined that he had not h ad a direct role in this crucialbreakthrough. My ela tion with the group s success wastempered by not being one of the inventors, he recalled. Iexperienced frustration that my personal efforts, started morethan e ig ht years before, had not r s u ~ t in a sign ificantinventive contribution of my own. 39

Since the failure of his fie ld-effect idea more thantwo years earlier, Shockley had paid only passing attentionto sem ic ond uc tor research. D uring the several monthsbefore the invention, he had mainly been working cn thetheory of dislocat ions in solids, a problem he had becomeinterested in afte r attending a conference on the topic theprevious August. Shockley had, however, devo ted somethought to the physics of p junctions and their use in suchprac tic al devic es as lightning arresters and h igh -spe edthermistors. 4o

Brattain and Gibney s discovery in November 1947galvanized him into action. The breakthrough observationof 7 November that surface-states could be overcome, helater wrote , stimulated the will to th ink nd a c t - i nminds conditioned to search for se mi co nd uc to ramplifiers. 41 A few days later he suggested fabricating anamplifier using a drop of electrolyte deposited across a pjunction in s ilicon or germanium; this approach workedwhen Brattain and Pearson tried it. 4 On 8 December 1947,more than a week before the poin t-contact transistor was

4


15/34

Minority amers and the irst Two Transistors

invented, Shockley descr ibed an idea in his laboratory

notebook for an p sandwich that had current flowinglaterally in the p-Iayer and with the n-Iayers acting as controlelectrodes.

The 16 December invention of the point-contacttransistor and Bardeen s interpretation of its action in termsof the flow of holes also stimulated Shockley s thinking.Bardeen s above-quoted analogy with the operation of a

vacuum t u b e i n which the current carriers were holesinstead of electrons was in fact due to Shockley,44 whoapplied it in his first attempt at a junction transistor, writtenin a room in Chicago s Hotel Bismarck on New Year s Eveof 1947. A letter he wrote to his mother a few days earlieraboard the New York Central s 20th Century Limitedcaptures some of the flavor of his frenetic activity:

I shall attend the Physical Society Meeting Tues and Wed andthen dig in at a Hotel and try to write some articles until about

Monday or Tuesday when I shall come out to give a lecture atthe Institute for the Study of Metals a t the Univers ity of

Chicago. 45

Holed up in his hotel room, Shockley recorded his

ideas on a padof

paper, pagesof

which he mailed to BellLabs, where they were witnessed by Bardeen and group coleader Stanley Morgan before being pasted into his labnotebook. In this first stab at a junction transistor, one cansee a clear analogy with a vacuum tube; its controlelectrode acts as a grid to control the flow of holes from a source to a plate (see Figure 6 46 About this disclosureof

a p p device, Shockley admitted that he failed torecognize the possibility of minority carrier injection into abase layer. . What is conspicuously lacking [in thesepages] is any suggestion of the possibility that holes might

15


16/34

Riordan & Hoddeson

t

P u : . w ~ : : f f : .~ l r ; / , ~ ¥ r. I t ,:t fL.I. . ~ o ~ ~ 1 1 ~ ~

~ ~ ,.oov ·T -

..vot ~ . G - . - l.J ( . ~ .J- J . ~ . : F l I i ~ - --rJ II I I, , Al P ~ ~ - .-ff ::

F7 7, III J e - J df? - / 1~ V

b.J r-;-; J ,. i l - ~ JlI P . I ~ -+ U j ~ / , I l. L 1 1 f ~ ..1 ;1\1\. t I ~ J.. ~ I : .1 d ~ , ~ I

] , ~ ~ r l . ~~ j :

~ / ) J, 1~ : > r A ~ ? V I h? AI 1. - . . ~ A.U

1/ J , 1,1 1-// / j I t. D: ~ J I/ - / / I ~r 7 ~- / / 1 2 / w-LLl. b f k - - Y I \ ~

d5P.:Jj; ~ I . I ~ .L.. ~ ,. ri-

a f IVIA--t-I l a Al ~ . 'B .4Lt- ~ ~ I .I k , ~-(1 jf t 0

t:tl A l . J, _ r: ~ J I t . Ai -;:;L. ~ It...ik r LJ Iu.:., ;r, 'f, rI-;;- ~ ~ ~ I D ~ft,1 . ~ _

1 I dII\ J ,6-, It. --p ~ 1 - ~ 11 ....

\... I I tV / ~ I~ ~ itt. ;;YL. /AfI.i ~ rJI ~ J ; ; ; ;........ ~ - O~~ - ;

a~ -, 1..160 JJ .. ·1 ~ ~ fAI. ,..-1:Ir } /.2 .... 1 . ~

1 t~ ~ rt1.. . . -:g: k _1. ~ i 9 L ,..: ~ ~ - L~3 ~ J V\\:l . \ J n J I I ~ ~ - - ~ , n q/ --._--...._ _.....

Figure 6: Page in Shockley's lab notebook showing one of his first

attempts at a junction transistor. This entry was written by Shockley inChicago on a piece of paper that he later (after Bardeen and Morgan had

witnessed it) pasted into his notebook. (Reprinted with the permission of

AT&T Archives)

16


17/34


be injected into the n-type material of the strip itself, thereby

becoming minority carriers in the presence of electrons. 47A little more than three weeks later, this time

working at his home in Madison, New Jersey, Shockleyconceived another design in which n-type and p type layerswere reversed and electrons rather than holes were thecurrent carriers (see Figure 7). Applying a positive potentialto the p-Iayer should then cause holes to flow into it and

thereby lower its potential for electrons; this he realizedwould increase the flow of electrons over the barrierexponentially. 48 As Shockley noted in an account writtenalmost thirty years later, this p device finally containedthe key concept of exponentially increasing minority carrierinjection across the emitter junction. 49 Minority carriers, inthis case the electrons, had to flow in the presence of the

dominant majority carriers, in this case the holes of the p-type layer. Yet at the time he did not seem to have beenterribly excited with the significance of this invention, giventhat it was written on Friday, 23 January 1948, andwitnessed four days later by his assistant John Haynes. I f I had r lly appreciated the impact that the junctiontransistor would have, he wrote over a quarter century later, I would have driven the few miles necessary to obtain awitnessing signature the same day. 5o

rucia l ExperilDent

It is a curious fact that almost another month went bybefore Shockley revealed his breakthrough idea to anyone inhis group other than Haynes. It is especially curious giventhat Bardeen and Morgan witnessed Shockley s earlierjunct ion transistor design. Why did Shockley keep the

17


18/34

Riordan Hoddeson

r A ~ - w N 4 I . 2 . J....

U : J ~ L -\..u Au, ~ ~ t L . - ~ ~I / ~f1 a /

II . IL L t fh ~ . I I .. ~ - J 8 aI- ~, v q~ ~ - - iL--.. ..:. ~ ,I . ~ _ oJ- ... ., - ~ ~ u () 1j i - - --v '- --d- . - ._-It:II .. ~ -- t:L ...f 11lI. . . . Le L - v - , l ' Jr ~ , , F} D r, i; : Cl -:1 4: ..lewd , , - c . . , P ~ L .:::I JJ .. . - ...

~ a v 0 I7 ; - jl 1.... - ~ ~ D R ~ ~ 7:-

1- , - '-

.

, ~

N-. v II 1~ _..... f.., t : -..u ;t:Z ~ I --- I • -.i ') . ~

_·v -~ ~ ~ II tT' _A

-I- _ ..~ : ~ ; ,b 'a.a.. L.d.. •~ I. . . . f \ ~ -fJ_~ _ .__. IJiI AF P - C If+:-:- ; L i - - : ~ -.:a;- M I. _~ .. •.. .1 z~ ~~ - - 7+- .


19/34


information to himself? Did he recognize that he had made a

major conceptual breakthrough but decide to keep it quiet togive himself the additional time to follow up its theoreticaland practical ramifications? Was he afraid Bardeen andBrattain were so close to making the discovery themselvesthat knowledge of his idea would push them to publishbefore him?51 Or was Shockley simply so unsure of thework that he avoided showing it to colleagues until he could

think about it further? Unfortunately, documentary evidenceprovides no firm answer to these questions.

It s true that in order to function, Shockley s n p ndevice required something more than the point-contacttransistor did. To work, Shockley s device needed newphysics namely it was crucial to understand that minoritycarriers are in fact able to diffuse through the base layer in

the presence of majority carriers. In late January suchbehavior was not obvious from the work of Bardeen andBrattain, who at that time were preoccupied with preparingpatent documents dealing with the point-contact device. The very next week, in fact, was the New York APSmeeting at which Brattain had his discussion with Benzer.)As noted above, Bardeen and Brattain still believed then that

essentially all the hole flow occurred in a micron-deep p typelayer at the semiconductor surface.

Evidence for the required diffusion of the minoritycarriers into the bulk material was not long in coming. In aclosed meeting at Bell Labs on February 18, John Shiverevealed that he had just tested a successful point-contacttransistor using a very thin wedge of n-type germanium, but

with the emitter and the collector placed on two oppositefaces of the wedge see Figure 8).52 At the place where thetwo contacts touched it, the wedge was only 0.01 cm thick,

19


20/34

iordan oddeson

while the distance between them along the germanium

surface was much larger. Shockley immediately recognizedwhat this revelation meant. In this geometry, the holes wereflowing by diffusion in the presence of the majority carriers,the electrons in the n-type germanium, through the bulk ofthe semiconductor; they were not confined to an inversionlayer on the surface, as Bardeen and Brattain had beensuggesting occurred in their device. Shockley recounted that

as soon as I had heard Shive s report, I presented the ideasof my junction transistor disclosure and used them tointerpret Shive s observation. 53

I O O

.. .. .. c.ou.tc. ft a ~ r l O l

Figure 8: Schematic diagram of the double-surface transistor firstdemonstrated by John Shive on 8 February 1948. Holes must flow by

diffusion through the bulk of the germ nium nd not along the surf ce in

order for this device to work. (Reprinted from Shive, The Double-Surface

Transistor. )

The most likely (but still not certain) interpretation of

this sequenceof

events is that Shockley felt he needed toestablish that minority carriers could indeed diffuse throughthe bulk of the semiconductor, something that was by nomeans obv,ious in late January 1948. Shive s serendipitous

2


21/34


experiment, done without any knowledge of Shockley sJanuary 23 disclosure, gave him the evidence he required.The experiment also may have injected a heady dose ofurgency into the solid state physics group. On February 26, e ll a bs applied for a total of four patents onsemiconductor amplifiers, including Bardeen and Brattain soriginal patent application on the point-contact transistor.Their two landmark papers, titled The Transistor, a SemiConductor Triode and Nature of the Forward Current inGermanium Point Contacts, went of f to the hysic lReview four months later, on 25 June 1948. One day afterthat, Bell applied for Shockley s patent on the junct iontransistor. And then on June 30 it announced the inventionof the transistor in a famous press conference.

In July, Shockley proved that hole injection (as hedubbed the flow of minority carriers in transistor action) wasindeed occurring in n type germanium. Working withHaynes, he showed that the charge carriers traveling fromemitter to collector were in fact positive particles with amobility of about 1.2 x 10 3 cm 2/volt-sec. 54 Their paperwas published in early 1949 immediately after Shive s paperon the two-sided transistor.

Shockley had another blind spot to overcome in histhinking about: the minority carriers before it finally becamepossible to fabricate functioning junction transistors usinghis ideas. One of the problems behind the failure of hisfield-effect transistor had been the slow rate at which chargecarriers diffused through the polycrystall ine sil icon andgermanium films used in the experiments. Gordon Teal, oneof Shockley s colleagues at Bell Labs, realized the merits ofusing single crystals of germanium and eventually silicon.Teal knew by this time that in polycrystalline films these


22/34

iordan oddeson

carriers do not survive long enough to make it from theemitter to the collector in sufficient numbers and thatminority carriers had lifetimes 20 to 100 times longer insingle crystals. He tried to convince Shockley of this criticaladvantage of s ingle crystals, but Shockley ignored thesuggestion and was satisfied to use polycrystalline films ofgermanium. 55

Fortunate ly Jack Morton, who headed Bell Lab sefforts to develop the point-contact transistor into acommercially viable product, took Teal s suggestionseriously and late in 1949 gave him a modicum of support topursue this avenue of research. Working with MorganSparks, Teal modified the crystal-making machine that heand a colleague John Little had recently developed forpul ling single crystals out of molten germanium. Themodification allowed doping the germanium in a controlledmanner and made it possible for Sparks and Teal to fabricatethe first practical p junction transistor in April 1950. 56

On the date of its demonstration to Bell Labs executives, 20April 1950, Shockley penned a note in the margin of his

January 1948 disclosure see Figure 7), stating, An punit was demonstrated today to Bown, Fisk, Wilson,Morton. 57

on lus ion

From this examination of the different ways in whichthe inventors of the first two transistors interpreted the flowof charge carriers in these devices, it seems clear thatBardeen made the fundamental breakthrough in establishingthat minority carriers-in this g r m n i u ~case, holes-areresponsible for transistor action. One might quibble that

22


23/34


until Shive s experiment he had an imperfect understandingof the detailed geometry of the flow. And one might arguethat holes trickling through a p-type inversion layer, asBardeen at first conceived their flow, are not really minoritycarriers in that layer where the dominant charge carriersare in fact holes. Still, Bardeen s realization of the crucialrole the holes were playing was the great leap forward thatmade semiconductor amplifiers a reality. It was no smallleap of the imagination in those postwar years to realize thatholes- carriers of sign opposite to those normally.in excessin the body of the block 59--could serve this function. Hadthe Purdue physicists made the same interpretation of thespreading resistance they had observed almost a year earlier,they might well have beaten Bell Labs to the transistor.

In 1980 Bardeen reflected in an interview on thedifference between Shockley s view and his in theinterpretation of transistor action:

The difference between ourselves [Bardeen and Brattain] and

Shockley came in the picture of how the holes flow from the

emitter to the collector. They could flow predominantly

through the inversion layer at the surface, which does contain

holes. And the collector would be draining out the holes fromthe inversion layer. They could also flow through the bulk of

the semiconductor, with their charge compensated by the

increased number of electrons in the bulk 60

After acknowledging Shive s definitiveexperiment, which showed definitely that the holes wereflowing through the bulk of the germanium, 61 he noted:

. what you were modulating was the conductance ofminority carriers in the inversion layer. And we probably had

23


24/34

iordan oddeson

tha t too much on our minds, when we were interpre ting the

first results of the point-contact transistor, in that we knew

that there was an inversion layer there which would provide a

channel for holes going from the emitter to the collector. The

question is, how important that aspect was, compared with the

flow of the holes through the bulk, which Shockley cal led

injection to differentiate it from the word emitter, which

we'd lre dy introduced.

Shive's experiment and that of Haynes and Shockley laterthat year resolved this question. Note Bardeen's emphasisabove on the semantic difference between hole injection andhole emission. According to Bardeen, Shockley introducedthe word injection because he wanted a different word todescribe the flow of minority carriers in the bulk of thesemiconductor.

This is not to diminish Shockley's important work indiscerning a new physical process necessary to designsemiconductor amplifiers that could be easily mass-producedwith relatively stable and predictable properties. And hecame up with the junction transistor idea almost a month for Shive s pivotal exper iment demonstra ted thatminority carriers did indeed diffuse through the bulkmaterial. Bardeen and Brattain made the fundamentalscientific breakthrough by recognizing that holes, notelectrons, were the crucial charge carriers involved in theirtransistor. Shockley took their work a major step further,providing a more detailed understanding of the flow ofminority carriers which could be either holes or electrons.

Their combined breakthroughs in semiconductor physics arewhat opened the road to Silicon Valley.

24


25/34

2

3


An early version of this paper was presented by one ofus (Riordan) at the 1994 Pittsburgh meeting of theAmerican Physical Society. It is based on researchdone in preparation for writing our book Crystal FireThe Birth the Information Age (New York: W.W.Norton Co., 1997).

W.J. Pietenpol, Bell System and CommercialApplications of Transistors, Manuscript, 7 June1958, AT T Archives.

W. Shockley, The Path to the Conception of theJunction Transistor, IEEE Transactions on ElectronDevices ED-23, no. 7 (July 1976): 597-620, hereaftercited as Path. Quote on p 597.

Other than a few papers written' by Shockley himself(which hardly qualify as objective scholarship), there isnothing to compare with the extensive works on theorigins of the point-contact transistor. W. Shockley, Path, 597-620; The Invention of the Transistor: AnExample of Creative-Failure Methodology,Proceedings the Second European Solid StateDevice Research Conference (London: Institute ofPhysics,1973): 55-75; and The Invention of theTransistor n Example of Creative-FailureMethodology, Proceedings the Conference on thePublic Need and the Role the Inventor Monterey,California, 11-14 June 1973 (Washington, NationalBureau of Standards Special Publication No. 388, May1974),47-89. Andrew Goldstein describes the role ofGordon Teal and his techniques for growing singlegermanium crystals in fabricating the first grown-

25


26/34

Riordan Hoddeson

5

6

7

junction transistor in Finding the Right Material:Gordon Teal as Inventor and Manager, in Sparks Genius Portraits Electrical EngineeringExcellence ed. Frederick Nebeker (New York: IEEEPress, 1993),93-126. See also Ernest Braun, Selected Topics from the History o SemiconductorPhysics and its Applications, in Out the CrystalMaze eds. L. Hoddeson, e t al. (New York: OxfordUniversity Press, 1992), 443-488.

The meeting was devoted to the work o the late JohnBardeen. N. Holonyak, John Bardeen and the PointContact Transistor, Physics Today 45, no. 4 (April1992): 36-43. The detailed microscopic understanding

o the flow o minority carriers in semiconductorsbecame an important factor allowing the subsequentminiaturization o transistors as components o

integrated circuits.

Shockley, Path, 615.

Holonyak s argument is based largely on (1) a statementBardeen made about his role in 1990, in a filmedinterview for Japanese television; (2) Holonyak s vividmemory and informed interpretation o a commentBardeen made in a graduate seminar Holonyak attendedin the spring o 1952 (that Walter Schottky had askedwhat the holes were doing in his rectification studies,the transistor would have been invented then); (3) acredit Shockley gave to Bardeen and Brattain in thepreface o his widely used textbook, Electrons andHoles in Semiconductors with Applications toTransistor Electronics (Princeton: D. Van Nostrand,


27/34


1950), and a handwritten inscription by Shockley inBardeen's copy the book, acknowledging Bardeen'simportant role in the developments treated; and (4)Holonyak's personal experience Bardeen's generousnature and modesty in presenting his contributions.

8 All interviews cited here are (or will shortly be) ondeposit at the American Institute Physics Center for

History Physics, College Park, Maryland (hereafter AlP .

Historical accounts the point-contact transistorinclude: L. Hoddeson, The Discovery the PointContact Transistor, Historical Studies in thePhysical Sciences 12, no. 1 (1981): 41-76, hereafter

cited as Point-Contact ;C.

Weiner, How theTransistor Emerged, IEEE Spectrum 10 (1973): 2433; E. Braun and S. MacDonald, Revolution inMiniature: The History and Impact Semiconductor Electronics (Cambridge: CambridgeUniversity Press, 1978); John Bardeen, Semiconductor Research Leading to the Point-Contact

Transistor, The Nobel Lectures 1942 1962(Amsterdam: 1964),313-386; and Walter Brattain, Genesis the Transistor, The Physics Teacher

(1968): 108-114. A much fuller treatment can be foundin Michael Riordan and Lillian Hoddeson, Crystal Fire:The Birth the Information Age (New York: W.W.Norton Co., 1997).

10 The tests were conducted in May-June 1945 by J.Hayes, H. McSkimin, W. Yager, and Russell Ohl.

o r r m r ~s ~ Point-Con1at, 62-63.

27


28/34

Riordan Hoddeson

2 J Bardeen, Bell Labs Notebook (BLNB) No. 20780, 8 March-23 April 1946, 38-57, AT T Archives; Surface States and Rectification at a Metal toSemiconductor Contact, Physical Review 7 (1947):717-727.

13 W. Shockley, Path, 605.

4 The full program is described in the laboratorynotebooks of the members of the group and summarizedin Point-Contact, 64-66. The sequence of steps to thepoint-contact transistor detailed in this article follow, forthe most part, the account in Point-Contact. It was tobecome a lifelong pattern for Bardeen to work closelywith experimentalists and only on rare occasions such

as in his work on the theory of superconductivity-withother theorists.

15 W. Brattain, Bell Labs Notebook (BLNB) No. 18194,17-29 November 1947,142-151, AT T Archives; Genesis of the Transistor, Physical Review(1948): 345; interview with Brattain by A.N. Holdenand W.J. King (AlP).


17 Bardeen, Semiconductor Research, 318-341.

18 In addition, using such a layer circumvented theconsiderable practical difficulty of depositing a goodthin layer of semiconductor material. Bardeen interviewby L. Hoddeson, 3 February 1980 (AlP).

19 Bardeen, BLNB No. 20780, entries of 22 November1947,61-67 and 23 November, 68-70.

28


29/34


The best treatment of this Purdue program is in PaulHenriksen, Solid State Physics Research at Purdue,Osiris 3, Sere 2 1987): 237-260. For a vivid accountof how Lark-Horovitz worked with students in thisgroup see Helmut Fritzsche interview by L. Hoddeson,3 March 1987 AlP).

21 Brattain, BLNB No. 18194, 175-76. This is the first

recorded instance we can find in which Bardeen andBrattain recognized the possibility that holes were actingas charge carriers. Note that Bardeen still proposed thatthe flow occurred within a narrow inversion layer at thesemiconductor surface.

L. Hoddeson, Point Contact, 71-72.

23 L. Hoddeson, Point-Contact, 72, which indicates thatthis event occurred on Thursday, December. Acloser examination of Brattain, BLNB No. 18194,183-92 indicates that there was a period of confusionfollowed by the actual breakthrough on 5 December.See Riordan and Hoddeson, Crystal Fire Chapter 7,for a more complete discussion of this sequence of

events. J Bardeen, Semiconductor Research Leading to the

Point Contact Transistor, 338.

25 W. Brattain, BLNB No. 18194, 16 December 1947,192-193; Brattain interview by Holden and King AlP).

6 L. Hoddeson, Point-Contact, 74.

7 Andrew Pickering, Constructing Quarks ASociological History Particle Physics Edinburgh:

29


30/34

Riordan oddeson

University Press, 1984),7.

28 It is interesting that some of the entries in Brattain snotebook during those critical weeks in December 1947are in Bardeen s hand. The two obviously wereworking side-by-side in the laboratory.

29 W. Brattain, BLNB No. 21780, 19 December 1947,3.

3 J Bardeen, BLNB No. 2 78 24 December 1947,72.31 J Bardeen and W. Brattain, Three-Electrode Circuit

Element Utilizing Semiconductive Materials, U.S.Patent No. 2,524,035, Application 17 June 1948, SerialNo. 33,466, Sheet 1 Figure 1 See column 10, lines64-70. (Emphasis added.)

32 J Bardeen and W.H. Brattain, The Transistor, ASemi-Conductor Triode, hysical Review 74 (1948):230-31, on 230.

33 Paul W. Henriksen, Solid State Physics Research atPurdue, 253-4.

34 It depended inversely on both the magnitude of the field

and the duration of the pulse. R Bray, interview by PHenriksen, 14 May 1982 (AlP).

35 Bray interview by Paul Henriksen, quoted inHenriksen, The Emergence of Solid State PhysicsResearch at Purdue University During World War II,Manuscript, University of Illinois, 1983.

36 J Bardeen interviews by L. Hoddeson, 1 Dec. 1977,14 April 1978, and 13 Feb. 1980 (AlP).

37 Holonyak interview by M Riordan and L. Hoddeson,

3


31/34


30 July 1993 (AlP). This exchange recounted in W. Brattain interview by C.

Weiner, 28 May 1974 (AlP), 28.


4 W. Shockley, Bell Labs Notebook (BLNB) No. 20455,71, 76-78, 80, 88-89, AT T Archives.

W. Shockley, Path, 608. (Emphasis in the original.)

W. Brattain, BLNB No. 18194, 169-70; G Pearson,Bell Labs Notebook (BLNB) No. 20912, 75, AT TArchives.

4 W. Shockley, BLNB No. 20455,91.

44

W. Shockley, BLNB No. 20455, 106-109; Bardeencredits him with this suggestion in his BLNB No.20780, 72.

45 Letter from W. Shockley to M.B. Shockley postmarked30 December 1947, Shockley Archives, Box 7, Folder3, Stanford University Special Collections.

46 W. Shockley, BLNB No. 20455, 110-113.

47 W. Shockley, Path, 613-14.

48 W. Shockley, BLNB No. 20455, 128-132.


5 W. Shockley, Path, 615. (Emphasis in the original.)

5 1Holonyak gives a reasonable argument that once thepoint-contact transistor and the notion of minoritycarrier flow were on hand, the p junction transistorwas bound to follow. He recalls a dinner in the mid-

3


32/34

Riordan Hoddeson

'80s in Urbana at which Bardeen stated that he andBrattain had planned to move on to that device as soonas they completed their taxing work preparing patentsfor the original transistor, only to find that Shockley hadalready tied up this area of work with the BellLaboratories patent attorney (who, in John's words, was in Shockley's pocket. ) Holonyak interview by

Hoddeson on 10 January 1992.5 J.N. Shive, The Double-Surface Transistor, Physical

Review 75 (1949): 689-90. See also Shive, Bell LabsNotebook (BLNB) No. 21869, 30-35, AT TArchives.


54 J.R. Haynes and W. Shockley, Investigation of HoleInjection in Transistor Action, Physical Review 7(1949): 691.

55 Goldstein, Finding the Right Material, 104. See alsoGordon Teal and John Little, Growth ofGermanium Single-Crystals, Physical Review 78(1950): 647 (reported at a meeting held 16-18 March1950).

56 Goldstein, Finding the Right Material, 104-5.

57 W. Shockley, Notebook No. 20455, 128. We tell thefull story of these developments in our book CrystalFire

58 In a paper about Bardeen delivered in 1992 atIndianapolis, Conyers Herring however stated that thecurrent in the point-contact transistor was primarily


33/34


carried by holes, that is, minority carriers moving intothe semiconductor rather than by electrons moving out emphasis added). See C Herring, Recollectionsfrom the Early Years of Solid State Physics, PhysicsToday 45, no. 5 April, 1992): 26-33, on 31.

59 J Bardeen and W. Brattain, The Transistor, 230.

6 Interview of Bardeen by L Hoddeson, 3 February1980, 2.

61 Interview of Bardeen, 4.

6 Interview of Bardeen, 2

33


34/34