FROM THE TRANSISTOR AND p-n-p-n SWITCH TO

amountdollΗλεκτρονική - Συσκευές

2 Νοε 2013 (πριν από 3 χρόνια και 9 μήνες)

150 εμφανίσεις

FROM THE TRANSISTOR AND p
-
n
-
p
-
n SWITCH TO

THE LASER AND LIGHT EMITTING DIODE

Nick Holonyak, Jr.

Electrical Engineering Research Laboratory and

Center for Compound Semiconductor Microelectronics

University of Illinois at Urbana
-
Champaign, Urbana, Illinoi
s 61801


ABSTRACT


Following Bardeen and Brattain’s discovery of the transistor (16 December 1947) and
carrier injection, making the hole indeed equal to the electron, the semiconductor took on new
importance and became the basic substance of electronics.

The element, the crystal Ge ushered
in the transistor, and Si with its higher gap and unique oxide technology yielded first p
-
n
-
p
-
n
switches and transistors, and then the integrated circuit. In fact, largely because of its oxide, and
its relatively good
hole mobility, Si has become the dominant material of electronics (maybe
forever). Beyond Si and its energy
-
gap, indirect
-
gap, and heterojunction limitations, the direct
-
gap III
-
V materials, particularly III
-
V alloys, made possible light emitting diodes (
LEDs) and
lasers, and thus optoelectronics. The first practical visible
-
spectrum LED, not to mention the
first III
-
V alloy device (a first laser), began (1960
-
62) with the direct
-
gap III
-
V alloy GaAs
1
-
x
P
x
,
which is also the beginning of III
-
V epitaxy. Of

special importance, GaAs
1
-
x
P
x

established the
viability of III
-
V alloys and, with its energy
-
gap and wavelength “tunability,” set the basis, the
direction for construction of heterojunctions. The progression over four decades from the direct
-
gap alloy Ga
As
1
-
x
P
x
, the prototype, to later Al
x
Ga
1
-
x
As and next In
1
-
x
Ga
x
P, to then the shorter
wavelength direct
-
gap alloy In(Al
x
Ga
1
-
x
)P, and to more recently In(Al
x
Ga
1
-
x
)N, has led to

2

heterojunction LEDs that cover all of the visible spectrum. In addition, various
III
-
V alloys, with
increasingly sophisticated crystal growth and processing technology, have led to high
performance quantum well lasers, as well as quantum well LEDs, over a broad range of
wavelength and power. The direct
-
gap III
-
V LED after almost four

decades exceeds the
incandescent lamp (as well as other forms of lamps) in performance in much of the visible range,
and, beyond growing display applications, has put conventional lighting under long range threat
with a general purpose semiconductor lamp
-
-
an ultimate lamp that promises unusual
performance and energy saving.


3

I. INTRODUCTION


Following Bardeen (Fig.

1) and Brattain’s discovery of the transistor (16 December,
1947),
1,2

and identification of carrier injection,
3
-
5

making the hole (positive,
“p”) indeed equal to
the electron (negative, “n”), the semiconductor took on new importance. With the sudden
emergence of the transistor and the identification of carrier injection,
3
-
5

the semiconductor began
its climb to primacy, to the fundamental posit
ion in electronics, including also optoelectronics.
With the transistor
--
a new idea, a new principle, a new device, a new name
--
a new electronics
emerged that could not be based on or be matched by the capabilities of the vacuum tube, even if
we could or
would be willing to cover the earth with tubes. The semiconductor permitted a
NEW and vaster electronics, not merely a substitute for the old electronics, including now a p
-
n
junction (heterojunction) ultimate lamp
6
---
a light emitting diode (LED) lamp tha
t promises to
save energy.

In spite of what we see now, it is interesting that at first Bardeen was not sure whether the
transistor, based on cost, could compete with the vacuum tube except in special applications.
This is now merely amusing. Beyond the
Bardeen and Brattain research devices,
1,2

the first
transistors Bardeen demonstrated to many of us (Urbana, 1951), our first “look” as it happened,
are the two oscillator
-
amplifier point
-
contact transistors in the historic box of Fig.

2.
5

The
transparent
“instant turn
-
on” box of Fig.

2 was one of three built in 1949 (one for Bardeen, one
for Brattain, and one for Bell Labs) and is the World’s first, and now oldest, portable transistor
circuit and apparatus. The transistor box of Fig.

2 was built with poin
t contact transistors of
necessity; the junction transistor did not yet exist, had not yet been realized. The demonstration
transistor box was used by Bardeen in many of his seminars and talks. Who could have
imagined what was to begin with the transisto
r, let us say with this box, and that this would lead

4

to a new form of electronics, a new scale in electronics, indeed, a revolution, a revolution that
extends beyond Si to the III
-
V semiconductor
---

and with it lasers and light emitting diodes
(LEDs), LE
Ds that have become high brightness energy
-
saving lamps!


II. THE TRANSISTOR


From his involvement with Gerald Pearson and Walter Brattain in a shared office when
he went to Bell Labs at the end of World War II, Bardeen began studying the surface behavior

of
semiconductors and explained (1947) why, because of surface states, field effect amplifier
experiments did not work.
7

In further studies and experiments, Bardeen and Brattain devised a
totally new device, a device that could later be named, because of

how it functioned, "the
transistor".
1
-
4

A totally new idea was at issue! Who would have believed that it could be something as
far
-
fetched as minority carrier injection (hole injection into an n
-
type crystal) at low impedance
at one electrode, which Bard
een called the “emitter” (later comment to Holonyak
5
), and carrier
collection at high impedance at the other electrode (a reverse
-
biased remote “collector”)
--
not to
mention provide gain (amplification)? At the time (1947), semiconductors such as Ge and Si
,
which were well known after World War II radar use, were thought to be direct gap, not indirect
gap with the capability of possessing long electron
-
hole lifetimes (thus making possible carrier
injection and collection over a significant distance).


There

is, of course, much more to this story, which, in the interest of briefness, cannot be
repeated here. Nevertheless, by the Christmas of 1947 there was a transistor, a new and
revolutionary bipolar amplifier, a device that inherently employed both the ele
ctron and the hole.
It is interesting that in Urbana in the Spring of 1952 John Bardeen drew the metal
-
semiconductor

5

diagram of Fig.

3 on the blackboard and pointed at the holes located at the inversion layer near
the metal
-
semiconductor boundary (at the
“p
-
n” transition), smiled slightly, and said, “If
Schottky in the 30’s had looked to see what the holes were doing, the transistor would have been
invented.”
5

(That is the extent to which Bardeen would point at himself!) In effect, Fig.

3 is a

p
-
n

ju
nction. There is enough evidence to make clear that Bardeen recognized, via the Bardeen
and Brattain transistor (the first transistor, the original bipolar device), carrier injection with a
current. Besides the transistor itself, at last the basis existe
d to explain current
-
driven light
emission in semiconductor crystals. Besides the fact that injected carriers (e.g., holes injected
into n
-
type Ge) could be collected, thus yielding the bipolar transistor, they also could
recombine, i.e., annihilate one a
nother and produce heat or light (photons). The latter,
recombination radiation (in a proper material), is the key to the semiconductor laser, light
emitting diode, optoelectronics, and the high
-
brightness LED
-
lamp.


III. TRANSISTOR DEVELOPMENT


The poin
t contact transistor of Bardeen and Brattain was quickly superseded by the
junction transistor (in itself a long story), and over a period of years one form of junction
transistor replaced the next, and the next, and the next. We show in Fig.

4 an alloyed

transistor,
an n
-
p
-
n Si alloyed transistor (1954) with an “alloyed” n
-
type collector on top and an “alloyed”
n
-
type emitter on bottom. Billions of transistors, mainly in Ge, were made by R. N. Hall’s alloy
process (~1950),
8

which we recognize is merely l
ocal liquid phase epitaxy. The alloyed
transistor, which carried the field from more or less 1950 to 1960, was a successful high
-
volume
transistor, but automatically limited. Any transistor built from two references, i.e., from two
surfaces (top and bott
om) as in Fig.

4, is limited, which, incidentally, was not the failing of

6

Bardeen and Brattains’s original point contact device with nearby emitter and collector contacts
on one surface. To make higher performance, higher speed junction transistors, it is

necessary to
reduce the distance from the emitter to collector, i.e., reduce the spacing from the lower junction
to the upper junction in Fig.

4.


To reduce the transistor base thickness with some control, preferably with great precision,
requires abandon
ing some of the early methods of making transistors in favor of processes such
as impurity diffusion. It is interesting that Bardeen mentioned this in his Urbana notebook in
early 1952. Also he was aware that Si, with its lower leakage, was a better choi
ce of material for
transistors than Ge. At Bell Labs in 1954
-
55 the technology, mainly at John Moll’s urging,
turned from Ge to Si, to impurity diffusion, oxide masking, and metal evaporation for shallow
junctions and contacts. At last, transistor techno
logy had turned to an invariant form that would
evolve and evolve, but not be superseded. For almost 50 years we have witnessed the further
and further development and refinement of this technology. In fact, this is where the integrated
circuit (IC) had
its origin.


Figure

5 shows a comparison of a 1955 Si crystal with a present
-
day Si ingot. Sliced,
polished, and etched wafers of a similar small ingot were processed by diffusion (1954
-
55) from
one side into transistors such as shown in Fig

6. Figure

7
shows N.H. (by the accident of
military service) describing this ring
-
base Si transistor (Fig.

6) to Makoto Kikuchi (later Sony’s
research director) in Tokyo (Nov, 1956), but avoiding any mention of Bell Labs’ secret of oxide
masking. Besides the fact tha
t the Si transistor of Fig.

6 does not have thermo
-
compression
bonded lead connections on the metallized electrodes, it also has an overly thick collector region
(the Si wafer thickness) and thus too much series resistance. This problem persisted until ab
out
1960 when vapor phase epitaxial processes were introduced to grow lightly doped Si on heavily

7

doped substrate wafers, thus yielding in effect a thin high quality layer (for top
-
side emitter and
collector junction fabrication) on a thick “slab” of heav
ily doped low resistance substrate crystal.


We note that if the bottom side diffused n
-
type layer of the crystal of Fig.

6 is not
removed, the results is a p
-
n
-
p
-
n thyristor switch,
9,10

not the p
-
n
-
p transistor of Fig.

6. This is a
large and interesting
story in its own right,
10

and is a vital part of the history of how Si
technology was transferred from the East Coast to the the West Coast of the U.S., leading to
Silicon Valley. We remark also that the Si p
-
n
-
p
-
n switch became the thyristor and yielded
perhaps the power industry's most important control device,
10

single devices that can handle
megawatts.


IV. CRYSTAL EPITAXY AND III
-
V ALLOYS


We see that as transistor development proceeded, crystal growth and materials processing
grew in scale and sophi
stication. The transistor engendered, actually drove, a new era of
intensive materials study and development and led to new processing methods such as vapor
phase epitaxy (VPE), which, of course, could be extended beyond just Ge and Si transistors and
swi
tches. The III
-
V materials, which were interesting to some of us in the late fifties and early
sixties (1959
-
62) for tunnel diodes
11

and parametric diodes, not to mention light emitters, were
technologically more mysterious than the elemental materials Ge

and Si, and more intractable.
In spite of considerable skepticism in some quarters, some of us felt, nevertheless, that III
-
V
materials offered important opportunities and should be explored also. For example, a III
-
V such
as GaAs made possible higher v
oltage tunnel diodes than Ge or Si (Fig.

8)
11

but then introduced
other problems and the need for further studies. Fortunately, some agencies, the Air Force and
Army among others, agreed and supported exploratory III
-
V materials and device work, and by

8

19
60 we could demonstrate III
-
V alloy and heterojunction work employing vapor phase epitaxy
that was the basis for an early publication
12

and was sufficiently basic for issuance of a patent
(Fig. 9).


Figure 9 shows a sketch of a simple closed tube halide t
ransport III
-
V epitaxial crystal
growth process, which, as shown, was used as early as 1960 to grow GaAs on GaAs, GaAs on
Ge, GaP on GaP, GaP on GaAs, and GaAs
1
-
y
P
y

on GaAs
1
-
x
P
x.
. The last is of special interest
because it shows that co
-
transport of GaAs
and GaP could be used to grow the alloy GaAs
1
-
x
P
x
,
12

either as uniform crystal or as a heterojunction (GaAs
1
-
x
P
x

on GaAs
1
-
y
P
y
). Until the III
-
V vapor
phase epitaxial (VPE) crystal growth of Fig.

9 (1960) some expert opinion held that a III
-
V alloy
such as

GaAsP could be synthesized by simply diffusing phosphorus (P) into GaAs, which, of
course, would have required an inordinate amount of time (years, decades, longer) and never was
heard of again (except perhaps as a laugh).


V. III
-
V ALLOYS AND LIGHT EMIT
TING DIODES (LEDs)


The transistor and carrier injection, with a current, made it apparent why a p
-
n junction
could emit light. III
-
V compounds gave, moreover, energy band properties and energy gaps
(wavelengths) that made it possible to seek and to make
light emitting diodes (LEDs), and even
semiconductor lasers. In fact, at least two of us (Holonyak at GE and Rediker at Lincoln
Laboratory
-
MIT) felt that the energy band property that made possible a laser (1962), i.e., a
direct gap and conserved electron

and hole momentum in carrier recombination, was what was
required also for the best LEDs. Thus visible
-
red direct
-
gap GaAsP was a better LED candidate
than, let us say, shorter wavelength but indirect
-
gap GaP (and its offset electron and hole
momentum).



9


In a debate (not always friendly) that started ~1962 and continued for decades it became
clearer and clearer that lasers and LEDs required III
-
V alloys and not just the usual binary
crystals such as GaAs or GaP, which, in fact, are now being relegated
to use as substrates. The
prototype III
-
V alloy system proved to be GaAsP, which, indeed, operated as one of the first
semiconductor lasers (1962) and was introduced also as the first practical LED. The first GaAsP
laser is shown in Fig.

10.
13

Figure

11

shows another (similar) GaAsP laser operating in the
visible red with its output beam photographed directly on colored film without the benefit of any
kind of infra
-
red converter or detector. This is the first semiconductor laser photographed with
its ow
n light. The same GaAsP p
-
n junction material that was used for the lasers of Figs.

10 and
11, except shifted in composition to shorter wavelength (more P), could be packaged in regular
pig
-
tail diode packages (glass packages) as shown in Fig.

12 and be u
sed as LEDs. At the first
semiconductor laser conference in Schenectady, November, 1962, which was held by GE for
invited U.S. Defense Department guests, LEDs as shown in Fig.

12 were given to some of the
participants. The first practical LEDs and the fi
rst offered for sale (GE, 1962) were made from
the III
-
V alloy GaAs
1
-
x
P
x

and were offered for sale by GE directly and a little later through the
Allied Radio Catalogue. In fact, in an interview with Harlan Manchester of Readers Digest (Feb,
1963), we even

claimed that visible
-
red GaAsP LEDs would exceed lasers in importance. As it
happens, to this day LED sales exceed semiconductor laser sales and, as the LED now becomes a
full
-
fledged lamp, LED use and sales promise to become enormous
---
as well as energy

saving in
comparison to less efficient forms of conventional lighting.


The GaAsP laser and LED proved at once (1962), directly by their device performance,
that III
-
V alloys, although in a sense stochastic, were sufficiently “smooth”, uniform, and defect
-
free so as to be viable systems, useful systems. Now, the III
-
V alloy could be exploited, and, of

10

course, being an alloy, could be used in heterojunctions. Ultimately heterojunctions would
displace homojunctions, and would prove to be indispensable also

for quantum well
heterostructures and superlattices. Without III
-
V alloys, optoelectronics as known today could
not exist. Figure

13 shows (Sept 1967) Zhores I. Alferov and N. Holonyak, Jr. at the Physico
-
Technical Institute (Leningrad) at the beginning

of their many years of discussions on III
-
V
heterojunction devices and optoelectronics. As is well known, Alferov and his group pioneered
in the introduction of the Al
x
Ga
1
-
x
As
-
GaAs double heterjunction and the demonstration of
continuous room temperature

laser operation.
14



VI. III
-
V ALLOYS AND HIGH BRIGHTNESS LEDs


The III
-
V semiconductor laser and LED research of the early 1960’s, which occurred
primarily in industrial laboratories, just as did early transistor research, could equally well be
perform
ed in university laboratories. Massive laboratories were not yet needed for
semiconductor materials and device research, and III
-
V materials and device research could
move into university laboratories, as, indeed, happened with III
-
V alloy semiconductor r
esearch.
Based on an invitation from John Bardeen, we moved (1963) GaAsP research, as well as the
study of other III
-
V alloys, from General Electric (Syracuse) to Urbana. In Fig.

14 Bardeen and I
are looking, in the early 70’s, at red
-
orange
-
yellow
-
green

(ROYG) LED’s in the higher energy
InGaP alloy system. In 1970 we were able to show, with graduate student labor, laser operation
in this III
-
V alloy, and then began a long period of studying and extending this system to shorter
wavelengths (red



green).



With the continued development of vapor phase epitaxial (VPE) crystal growth
processes, and then the demonstration of AlGaAs
-
GaAs heterojunction lasers grown by

11

metalorganic chemical vapor deposition (MOCVD, a VPE process, see Dupuis, et al., 1977
15
),
it
was clear that Al
-
Ga substitution in the In(Al
x
Ga
1
-
x
)P system, which is a “take
-
off” (an
extension) on InGaP, would yield high performance ROYG heterojunction LEDs. It is worth
mentioning that, besides high
-
brightness red
-
orange
-
yellow LEDs, the red
-
or
ange lasers in
digital video disc (DVD) recording machines employ the In(Al
x
Ga
1
-
x
)P alloy, and are mainly
grown by Dupuis
-
style MOCVD.
15


For the best LEDs it is not sufficient merely to have a direct
-
gap visible
-
spectrum III
-
V
alloy. It is important also

to make the LED in the form of a double heterojunction, i.e., a wider
bandgap hole (p) and electron (n) emitter on either side of a narrower gap active region, a
concept championed very early by Zh. I. Alferov for lasers.
14

This is advantageous for impro
ved
carrier injection and also in allowing escape of photons from the active region (less absorption),
and is, of course, part of the design of high brightness ROYG In(AlGa)P LEDs. This does not
take care of all of the problems of absorption because photo
ns, even in a very efficient
quaternary double heterojunction (DH), tend to be contained in the dense medium of the crystal
(the problem of crystal and free space mismatch). Hence, the absorbing GaAs substrate on
which In(AlGa)P DHs are grown by VPE must
be removed and replaced with, for example, a
transparent “platform” such as GaP.


Kish in Craford’s group at Hewlett
-
Packard (both alumni of the Urbana III
-
V
semiconductor research) has accomplished this task, and introduced a new family of high
brightness

In(AlGa)P LEDs.
16

Figure

15 (after Kish, et al.) shows Kish’s scheme for GaAs
substrate removal and replacement with GaP on VPE (MOCVD) In(AlGa)P double
heterojunctions (DHs). Figure

15 (bottom) shows quite strikingly the difference in behavior of
an ab
sorbing substrate (AS) and transparent substrate (TS) In(AlGa)P LED. The improvement in

12

LED performance, depending upon the VPE crystal and wavelength, can be as large as 200% or
more. In addition, as shown in Fig.

16 (after Krames, et al.
17
), if the LED

crystal is properly
shaped to reduce multiple photon reflections in the crystal, the LED performance can be further
enhanced, e. g., reaching an external quantum efficiency (Fig.

16) as high as 50 to 60%. The
LED, based on performance, is indeed an ultim
ate lamp.
6


When an already successful product is improved on this scale, the effect is not just
evolutionary, it is revolutionary. Figure

17 (after Craford and co
-
workers) shows where III
-
V
alloy LEDs fit in performance in comparison with other well kno
wn light sources. Besides
ROYG In(AlGa)P LEDs, Fig.

17 shows also where blue
-
green In(AlGa)N LEDs (Nichia and
now others), which also are part of the III
-
V alloy family, fit in the same comparison. Figure

17
shows distinctly that visible
-
spectrum III
-
V a
lloy LEDs now exceed the standard incandescent
lamp and a number of other sources in performance (lumens per Watt) at many wavelengths.
With further improvement III
-
V alloy LEDs threaten to out
-
perform a number of other light
sources.


VII. CONCLUSIONS
AND FUTURE OUTLOOK


At this point we can look back over a considerable period (over 40 years, Fig.

18) and see
what has happened to the LED, which, of course, is itself a consequence of transistor era
developments. The GaAs
1
-
x
P
x

laser gave an unambiguous
start to the first practical LED, which
in Fig.

18 is the point (arrow, 1962) way over at the lower left. This figure, which was supplied
by my former student and colleague of many years, George Craford (LumiLeds), shows how
LEDs have evolved in performan
ce over the years. It is not necessary to describe all the steps
and plateaus in LED performance in this figure except to state that the figure begins (1962) on

13

the lower left with the direct
-
gap III
-
V alloy GaAs
1
-
x
P
x
, the prototype red
-
spectrum alloy, an
d
then at the far right (~

2000) climbs, because of the direct
-
gap alloy In
0.6
(Al
x
Ga
1
-
x
)
0.5
P grown
lattice matched to GaAs, to well above the performance of conventional lamps. The III
-
V alloys
have prevailed as LEDs. They have eliminated everything else
, and still have room for
considerable improvement (cf., Fig.

17). It is clear what this means: On a logarithmic scale of
decades (10, 20, 50, 100 years) the semiconductor will become the ultimate lamp. This is
possible in theory
--
it is what p
-
n junction

theory allows.
6

And it is possible now based on what
semiconductor technology will permit. Every form of display will be possible, as well as every
form of lamp. In short, more than the transistor has come from the transistor, and from the
uniqueness o
f the semiconductor, which is a universal substance employing and equating in
importance the electron, hole, and photon.


Wherever my teacher John Bardeen is, I am sure he is pleased with what has happened.
But maybe we should take another look at Fig.

18 and see how much time and effort are needed
to research, to learn, to build, to accomplish. Obviously not everything should be done and
measured with a short
-
term perspective and the expectation of immediate gain or yield. That is
not how we got here,

to the high
-
brightness LED
-
lamp, and the performance and energy savings
it promises.


ACKNOWLEDGMENTS


In closing, I want to mention that our early work, the first successful III
-
V alloy laser and
LED work, was supported by the Air Force, and more recen
tly, in the era of quantum wells, by
the Army Research Office and the National Science Foundation. I want to thank my students
and colleagues for all their contributions, and for their efforts in making the LED into the high
-

14

brightness LED lamp. I don’t
think any of us can repay our debt to John Bardeen. Finally, I
want to thank Zhores I. Alferov, the Global Energy Foundation, and the Russian People for the
Global Energy International Prize
---
for founding this Prize and focusing attention on energy (its
availability, use, conservation, etc.) and its critical role in human life and progress.



15

REFERENCES

1.

J. Bardeen and W. H. Brattain, "The Transistor, Semi
-
Conductor Triode," Phys. Rev.,
Vol. 74, pp. 230
-
231 (June, 1948).

2.

J. Bardeen and W. H. Brattain, "Th
ree
-
Electrode Element Utilizing Semiconductive
Materials," U.S. Patent 2,524,035, Oct. 3, 1950 (Filed June 17, 1948).

3.

J. Bardeen, "Pre
-
History of the Semiconductor Laser," Optoelectronics
-
Devices Technol.,
Vol. 2, pp. 124
-
126 (June, 1987).

4.

J. Bardeen, “Th
e Early Days of the Transistor,” Abstracts, New Materials Conference,
Osaka, 1990. (Reference supplied by M. Kikuchi, Tokyo, Nov 5, 1999.)

5.

N. Holonyak, Jr., "John Bardeen and the Point
-
Contact Transistor," Phys. Today, Vol. 45
(#4), pp. 36
-
43 (April, 1992
).

6.

N. Holonyak, Jr., “Is the Light Emitting Diode (LED) an Ultimate Lamp?," Am. J. Phys.,
Vol. 68, pp
.

864
-
866 (Sept, 2000).

7.

J. Bardeen, "Surface States and Rectification at a Metal Semi
-
Conductor Contact," Phys.
Rev., Vol. 71, pp. 717
-
727 (May 15, 1947).

8.

R. N. Hall, "Power Rectifiers and Transistors," Proc. IRE, Vol. 40, pp. 1512
-
1518 (Nov,
1952). See also, R. N. Hall and W. C. Dunlap, "P
-
N Junctions Prepared by Impurity
Diffusion," Phys. Rev., Vol. 80, pp. 467
-
468 (Nov 1, 1950).

9.

J. L. Moll, M. Tanenbaum
, J. M. Goldey, and N. Holonyak, "P
-
N
-
P
-
N Transistor
Switches," Proc. I.R.E., Vol. 44, pp. 1174
-
1182 (Sept, 1956).

10.

N. Holonyak, Jr., "The Silicon p
-
n
-
p
-
n Switch and Controlled Rectifier (Thyristor),"
IEEE Trans. Power Electronics, Vol. 16, pp. 8
-
16 (Jan 20
01).

11.

N. Holonyak, Jr. and I. A. Lesk, "Gallium Arsenide Tunnel Diodes," Proc. IRE, Vol. 48,
pp. 1405
-
1409 (Aug, 1960).

12.

N. Holonyak, Jr., D. C. Jillson, and S. F. Bevacqua, “Halogen Vapor Transport and
Growth of Epitaxial Layers of Intermetallic Compounds a
nd Compound Mixtures,”
AIME Conf. (August 1961), Los Angeles, in J. B. Schroeder, ed., Vol. 15,

16

METALLURGY OF SEMICONDUCTOR MATERIALS (Interscience Publishers, New
York 1962), pp. 49
-
59.

13.

N. Holonyak, Jr. and S. F. Bevacqua, “Coherent (Visible) Light Emissi
on from
Ga(As
1
-
x
P
x
) Junctions,” Appl. Phys. Lett., Vol. 1, pp. 82
-
83 (Dec 1, 1962).

14.

Zh. I. Alferov, V. M. Andreev, D. Z. Garbuzov, Y. U. Zhilyaev, E. P. Morozov, E. L.
Portnoy, and V. G. Trofim, "Effect of AlAs
-
GaAs Heterostructure Parameters on
Threshold

Current of Lasers and the Achievement of cw Generation at Room
Temperature," Fiz. Tekh. Poluprovodn.,Vol. 4, pp. 1826
-
1829 (Sept 1970).

15.


R. D. Dupuis, "III
-
V Semiconductor Heterojunction Devices Grown by Metalorganic
Chemical Vapor Deposition," IEEE J. Se
lected Topics in Quantum Electronics, Vol. 6,
pp. 1040
-
1050 (Nov/Dec, 2000).

16.


F. A. Kish, F. M. Steranka, D. C. DeFevere, D. A. Vanderwater, K. G. Park, C. P. Kuo,
T. D. Osentowski, M. J. Peanasky, J. G. Yu, R. M. Fletcher, D. A. Steigerwald, M. G.
Craford
, and V. M. Robbins, "Very High
-
Efficiency Semiconductor Wafer
-
Bonded
Transparent
-
Substrate (Al
x
Ga
1
-
x
)
0.5
In
0.5
P/GaP Light
-
Emitting Diodes," Appl. Phys. Lett.,
Vol. 64, pp. 2839
-
2841 (May 23, 1994).

17.


M. R. Krames, M. Ochiai
-
Holcomb, G. E. Hofler, C. Carter
-
Coman, E. I. Chen, I.
-
H.
Tan, P. Grillot, N. F. Gardner, H. C. Chui, J.
-
W. Huang, S. A. Stockman, F. A. Kish, M.
G. Craford, T. S. Tan, C. P. Kocot, M. Hueschen, J. Posselt. B. Loh, G. Sasser, and D.
Collins, "High
-
Power Truncated
-
Inverted
-
Pyramid (Al
x
Ga
1
-
x
)
0.5
In
0.5
P/GaP Light
-
Emitting
Diodes Exhibiting >50% External Quantum Efficiency," Appl. Phys. Lett., Vol. 75, pp.
2365
-
2367 (Oct 18, 1999).


17


FIGURE CAPTIONS


Fig. 1

John Bardeen at the age of 80. Before Bardeen there was no transistor. After
Bardeen t
here was a transistor, and the basis to understand carrier injection and
thus light emiting diodes (lasers and LEDs).


Fig. 2

Bardeen's point
-
contact transistor oscillator
-
amplifier box made in 1949. This is
the first portable transistor circuit apparatus

and, inside the transparent box,
reveals the first two transistors seen in Urbana (1951).


Fig. 3

Energy band diagram of metal
-
semiconductor contact used by Bardeen (Urbana
class, Jan, 1952) to illustrate hole injection in n
-
type Ge (see Ref. 5).


Fig. 4

Silicon n
-
p
-
n alloyed transistor (1954) made with Au (+Sb) alloyed emitter and
collector regions. The n
-
type emitter and collector regions regrown (local liquid
phase epitaxy) from the Au+Si (+Sb) alloy are marked with arrows. Note the
large base (p
-
type
) thickness, which is hard to control from two sides (two
references).


Fig. 5

Comparison of (a) 1954
-
55 Si crystal used for the first diffused
-
impurity
transistor devices with (b) a modern Si ingot used for epitaxial growth of high
quality thin layers for

use in integrated circuits.


Fig. 6

Diffused
-
base Si p
-
n
-
p transistor with evaporated metal contacts made at BTL
(1955) by N.H. The hand lettering (N.H.) is simply for the convenience of
discussion. Note that thermo
-
compression bonding did not yet exist

and the
leads are spring contacts (bent out of place for device viewing). If the n
-
diffusion on the bottom of the crystal, which is the same as that on the top of the
p
-
type crystal, is not removed (etched off), the device becomes a p
-
n
-
p
-
n switch
(thyri
stor) and not p
-
n
-
p transistor (see Ref. 10).


Fig. 7

M. Kikuchi and N. Holonyak, Jr. examining the diffused
-
base Si transistor of
Fig. 6 at Denki Shikenjo (Tokyo) in 1956. (The note was written on the back of
the picture.)


Fig. 8

Comparison of the I
-
V c
haracteristics of Ge, Si, and GaAs tunnel diodes,
showing the obviously higher injection voltage at larger and larger energy gap in
the progression from Ge to Si to GaAs.


Fig. 9

Schematic diagram (from Holonyak, U. S. Patent 3,249,473, 1965) of the close
d
ampoule halide vapor phase epitaxial (VPE) growth (1960
-
62) of GaAs on
GaAs, GaAs on Ge, GaP on GaP, GaP on GaAs, and Ga1
-
yPy on Ga1
-
xPx, the
basis also for the co
-
transport of GaAs and GaP and growth of the Ga1
-
xPx used
for the first III
-
V alloy lasers
and LEDs (see Refs. 12 and 13).



18

Fig. 10

The first GaAsP p
-
n junction operated as a laser (1962). The GaAsP III
-
V alloy
crystal, with polished Fabry
-
Perot resonator sides, is attached to a TO
-
18 header.


Fig. 11

The first semiconductor laser, a GaAsP p
-
n

junction, photographed (near field)
by its own light (visible red) on ordinary film. The p
-
n junction and source of
light is just below the Ni lead. Directly below the laser spot, red light from the
vertically spreading laser beam is reflected from the
TO
-
18 header.


Fig. 12

GaAsP diffused junction inserted into a glass diode package in 1962 to form a
visible
-
spectrum red light emitting diode (LED)
---
the first practical LED, a III
-
V
alloy.


Fig. 13

Zh. I. Alferov (right) and N. Holonyak, Jr. (left) in fr
ont of the Ioffe Physico
-
Technical Institute (Leningrad, September 1967) on the occasion of first meeting
and discussing III
-
V alloy and heterojunction lasers and LED's.


Fig. 14

John Bardeen and Nick Holonyak, Jr. in 214 EERL (Urbana) in ~ 1971 viewing
re
d
-
orange
-
yellow
-
green InGaP light emitting diodes (LEDs), which with
MOCVD Al substitution later became the In(AlGa)P LED and then high
-
brightness LED
-
lamp.


Fig. 15

Kish's schematic scheme (Craford's group, Hewlett
-
Packard) for removing the
absorbing GaAs

substrate and substituting transparent GaP on red
-
orange
-
yellow
In(AlGa)P LEDs. Bottom shows the improvement from "AS" to "TS" in LED
performance (see Ref 16).


Fig. 16

Krames', et al. scheme (LumiLeds, Craford's group) for shaping the geometry of
an In(
AlGa)P transparent substrate LED to reduce multiple reflections and
promote photon escape from the crystal, yielding in the red the remarkable
external quantum efficiency of 50
-
60%, an ultimate lamp (see Refs. 17 and 6).


Fig. 17

Light emitting diode (LED,

III
-
V alloy heterojunction) performance at different
wavelengths shown plotted on the CIE eye performance curve, with LED data
shown in comparison to and exceeding many standard light sources (courtesy of
Craford and Kish).


Fig. 18

Evolution (history) of

the visible
-
spectrum light emitting diode (LED) from its
first introduction by Holonyak in 1962 (the alloy GaAsP) to the III
-
V alloy
heterojunction ultimate LED
-
lamps of In(AlGa)P and In(AlGa)N in ~ 2000 (and
beyond). Note that the LED begins with a III
-
V alloy and is (and remains) a III
-
V alloy as it becomes an ultimate lamp.