IF THE 20th is the century of electronics, the 21st will be that of light.
More and more, computers and communications are relying on light to send, store
and process information. Devices that work with light are much faster, and can
store more data, than those in which electrons do the legwork.
But these developments have been hampered by the difficulties in generating
the full spectrum of colours. The miniaturised lasers and light-emitting diodes
(LEDs) needed to push the optoelectronics revolution forward have long been able
to offer red, yellow and, at a push, green light. But the real gold lies at the
blue end of the rainbow, and for decades it has remained out of reach.
Now all of that has changed. Blue LEDs are already lighting up the streets of
Japan in advertising display screens. And blue microlasers are opening up the
prospect of video CDs and capacious computer discs. But no giant of the
electronics industry has made this breakthrough. It has come instead from a
small company called Nichia Chemical Industries, based on the sleepy island of
Shikoku in southern Japan.
Advertisement
Blue light opens up several opportunities. Blue LEDs are essential for
full-colour flat-screen displays. And blue lasers will change the face of CD
technology. Lasers are key components of CD players, and they are also used in
magneto-optic storage systems for computer hard discs. The shorter the
wavelength of the laser鈥檚 light, the smaller the width of the focused beam, and
so the smaller the size of a data bit that can be written or read optically.
Older CD players used infrared lasers to read the information on disc, but most
new machines are now switching to red light. Because blue light has a shorter
wavelength, it can afford higher data storage densities than either of these:
going from infrared light to blue quadruples the amount of data that can be
stored in a given area.
One major application will be the digital video discs that are set to replace
video tapes, just as audio CDs have displaced audio cassettes and vinyl discs.
You need high storage densities to make this practical, because it takes a vast
quantity of data to represent a video image. Blue lasers should also usher in
higher-resolution laser printing and projection television, where TV pictures
are projected through space. Because lasers can be miniaturised and their beams
do not spread as they travel through air, it may be possible to project sharp TV
pictures onto large screens from devices the size of a pen.
These are exciting ideas, but they have proved hard to turn into reality.
With both LEDs and lasers, physicists have been stumped by the problem of how to
produce the high-energy photons that make up blue light. It鈥檚 all down to the
way lasers and LEDs work. Both are made from a double layer of semiconducting
material, in which the two layers have had different impurities added to
change the way they behave. One layer, called 鈥渘-doped鈥�, has impurities that
provide extra electrons. Conversely, 鈥減-doped鈥� material has impurities that soak
up electrons, leaving behind positive 鈥渉oles鈥� that act as charge carriers. These
are not real particles at all, but empty spaces in the 鈥渟ea鈥� of electrons which
act like positively charged electrons.
Interesting things happen when you put thin films of p and n-doped
semiconductor back to back. The electrons in the n-doped material can cross the
junction between the layers and fall into the holes in the p-doped material.
When this happens, the electron loses some energy, just as a ball on a pool
table loses energy when it drops into a pocket. The lost energy is given off as
a photon of light, the colour of which depends on the energy of the photon: the
higher the energy, the shorter the wavelength, and the further the colour is
towards the blue end of the spectrum.
The energy of the photons depends, in turn, on the depth of the energy pocket
that the electrons fall into when they recombine with holes. This is called the
band gap. The larger it is, the bluer the photon.
There are plenty of semiconductors with band gaps that are suitable for
making light at the red end of the spectrum. But for many years it looked as
though only two were strong candidates for making blue and green light: silicon
carbide and zinc selenide. Both have large band gaps, and so throw out
high-energy photons. But neither seemed to have quite what it takes for commercial
applications. Silicon carbide has an 鈥渋ndirect鈥� band gap鈥攍ike a crooked
pool-table pocket鈥攕o it cannot emit light efficiently. Zinc selenide
doesn鈥檛 have this problem, but suffers from others. For one thing, researchers
struggled for years to make a p-doped version, and though this problem has now
been solved, the p-n junction does not survive long enough for most commercial
needs. As the material glows, it accumulates crack-like defects that ruin its
light-emitting ability.
Nichia鈥檚 breakthroughs have come thanks to a light-emitting wonder material:
the semiconductor gallium nitride. This material has long been an outsider in
the race for blue light. Its large band gap made it a contender back in the
1960s, when semiconductor lasers had just been invented, and things looked
hopeful in 1969 when researchers at the RCA Laboratories in Princeton, New
Jersey, made crystalline thin films of the stuff. But then problems
appeared.
As with all thin films of semiconductors, you need a good surface to grow it
on鈥攐ne where the spacing between the atoms in the film matches that in the
material underneath. Otherwise, the first few atomic layers of the film get
stretched or compressed to fit with the lattice below鈥攁nd this leads to
flaws, like the cracks in paint as it dries and shrinks. But gallium nitride can
only grow at temperatures of around 1000 掳C, and there are only two substrates
with the right atom spacing that can survive these high temperatures: silicon
carbide and sapphire. Silicon carbide is very expensive. Sapphire is cheaper,
but the match between the atomic spacings is not very close, so gallium nitride
films grown on sapphire are full of defects. It also turns out to be very hard
to make the p-doped gallium nitride: for years, no one had a clue how to do
it.
Then around ten years ago Isamu Akasaki, a scientist at Nagoya University in
Japan, took the first steps towards overcoming both these problems. In spite of
general scepticism at the time, Akasaki was convinced that gallium nitride was
worth pursuing, and in 1986 he found a partial solution to the problem of
lattice matching. Akasaki and his colleagues found that a buffer layer of
aluminium nitride laid down between the sapphire substrate and the gallium
nitride film would enable them to grow extremely smooth films. But even so the
film still had lots of defects.
Akasaki鈥檚 second breakthrough involved a generous helping of luck. Various
workers had believed that magnesium might be a good impurity to use to make
p-type films of gallium nitride, but whenever anyone tried it, they found that
it didn鈥檛 work鈥攆ilms containing magnesium did not conduct electricity at
all. In 1989 Akasaki and his student Hiroshi Amano were investigating just such
a film when they noticed it glowing as they zapped it with a beam of electrons.
What鈥檚 more, as time went on the glow became brighter and brighter. Somehow they
had managed to convert the films into p-doped semiconductors.
Sacrificial layers
But Akasaki鈥檚 team was not quite alone in the field. Shuji Nakamura, a young
scientist who had been enticed to Nichia by the gleam of blue light, knew about
Akasaki鈥檚 discoveries and set out to see how they might be taken further. First,
he looked at the question of the buffer layer. Simply copying the idea of using
aluminium nitride would bring him up against Akasaki鈥檚 patents, so Nakamura
decided to use gallium nitride itself as the buffer. Though this meant
sacrificing a thin layer of gallium nitride to the mercy of defects, the nitride
layer on top was much smoother. He then discovered that adding indium to the
buffer layer made the top layer smoother still.
Then in 1992 Nakamura took another big step forward when he found a better
way to create a p-doped material. Instead of using an electron beam to convert
the insulating magnesium-doped material to a semiconductor, he simply heated it.
This, he had concluded, was all that Akasaki鈥檚 electron beam was doing anyway.
Nakamura鈥檚 heating process is not just cheaper than electron-beam irradiation;
it also gives a higher concentration of positive holes.
Nakamura wasted no time in using these building blocks to seize the
initiative in blue lasers. In December 1995 he unveiled a blue-purple laser,
based on indium gallium nitride. Akasaki, who had moved to Meijo University in
Nagoya, was hot on Nichia鈥檚 heels. By June of 1996, he too announced an indium
gallium nitride laser, which gave off light with even shorter
wavelengths鈥攊n the ultraviolet. And there were signs that the big guns
were eventually catching up when, last October, researchers at Toshiba also said
they had made a violet nitride laser.
But the problem was only half solved. None of these devices gave a continuous
laser beam. Instead, the light came out in pulses lasting a few microseconds,
and that鈥檚 not much good for playing a CD. But by last October Nakamura had
managed to squeeze out beams lasting a few seconds and to run his lasers at room
temperature rather than having to cool them. By the end of the year he had a
device that could run continuously for up to 35 hours.
One hurdle that remains is that these devices consume much more power than
red-light lasers, so they run the risk that they will overheat and destroy
themselves. But Nakamura is confident that he can reduce the power requirements,
and Nichia expects to have a blue laser on the market by the end of this
year.
Meanwhile, Nichia has capitalised on its work with gallium nitride to make
blue LEDs, which it has been marketing commercially since 1995. LEDs are much
more efficient than electric light bulbs. Both use electricity to make light,
but the ordinary incandescent electric bulb does so indirectly and wastefully:
the electric current simply heats up a filament until it glows, and much of the
power is squandered as heat rather than visible light. But LEDs convert the
electrical energy directly to light. LEDs made from semiconductors have been
familiar for 30 years or so, since they started to be used in the numerical
displays of the first electronic calculators, for example. But it was only in
the early 1990s that they become bright enough to outshine incandescent
lamps.
Like lasers, LEDs are built around semiconductor p-n junctions. The materials
used in the present generation of bright LEDs have relatively small band gaps,
giving red, orange and yellow light. Yellow-green LEDs can be built from gallium
phosphide, but they are not very bright.
The market for visible-light LEDs is immense, worth around $1 billion
a year. For some purposes, red and yellow are enough: for instance, red LEDs
work well in displays on subway stations and bus stops. But there is an
increasing demand for full-colour LED displays, which would have the same colour
range as a television. This calls for green and blue light too: with a red,
green and blue source at each pixel, you can make any visible colour.
Billion-dollar blues
In November 1993, Nakamura unveiled a prototype blue LED made from gallium
nitride. The big companies were stunned by this announcement from an almost
unknown operation. 鈥淓verybody got caught with their pants down,鈥� commented one
Toshiba researcher. Nichia rapidly geared up for commercial production, and six
months later Nakamura and his team had made a blue-green LED so bright you could
hardly look at it. By September 1995 they had very bright green-light
devices too. These contained layers of gallium nitride just a few dozen atoms
thick, with large amounts of indium added to reduce the band gap slightly. They
far outshine the only commercial competitors, the gallium phosphide yellow-green
LEDs.
Nichia鈥檚 blue and green LEDs went onto the market in 1995. Until then, there
had been no green or blue LEDs bright enough to be used in outdoor
displays, and for a brief honeymoon period Nichia had this field to itself. Now
other manufacturers, including the American company Cree, are also beginning to
produce full-colour displays incorporating nitride devices.
Another important application for shorter-wavelength LEDs is traffic lights.
Everyone has seen traffic lights that wink from red through amber to nothing.
It鈥檚 not surprising: the incandescent bulbs used in most traffic lights have a
lifetime of less than a year. LEDs are both longer-lived and more efficient.
They can last up to 15 years, and take just a tenth of the power of conventional
bulbs of the same brightness. The problem had always been to find a bright and
long-lasting material for the green light.
Nichia has already developed a dazzling set of traffic lights that use a bank
of its green LEDs. 鈥淭he light was bouncing off the walls,鈥� says Kevin O鈥橠onnell,
a physicist from Strathclyde University in Glasgow, recalling the unveiling of
Nakamura鈥檚 lights in December 1995
Blue lasers are some way behind short-wavelength LEDs on the road to becoming
commercial items. When they arrive, most observers agree that their biggest
impact will be in data storage, helping the drive towards ever smaller disc
drives holding more and more data. O鈥橠onnell also looks forward to finding them,
alongside blue LEDs, in all sorts of less obvious niches. For example, nitride
ultraviolet sources could be used with ultraviolet-sensitive inks for security
checking systems, or for hardening light-sensitive substances used as dental
fillings. But whatever the details, the next century seems sure to kick off with
a blues revival.