Thursday, October 18, 2007

The Blue LED

A small present I had been offered for my birthday a few weeks ago is a nice little gadget, a keyring pendant with a small lamp which delivers an intense blue light. Not that this little torch light is something completely indispensable, but I like it, just because...

I like it because the blue light conveys so glaringly a development that has taken place over the last 10 years or so, and that has brought a new electronic device from the alchemist-like labs of experimental semiconductor research to the mass-fabrication for give-aways: the blue light-emitting diode.

A light-emitting diode, or LED for short, is a special kind of semiconductor device. And the intense blue light of the small torch is not created by an incandescent lamp, but in such a solid-state semiconductor device.

In chunks of matter that consist of large amounts of atoms, the energy levels of the electrons are not arranged in discrete steps, as in an isolated atom, but are merged into bands. The most important bands are the conduction band, which can contain mobile electrons which can sustain an electrical current, and the valence band, which is usually completely filled with electrons and does not contribute to the electrical current. A semiconductor is a material with a gap between the valence and conduction bands and where at zero temperature, the valence band is completely filled, while the the conduction band is completely empty. This means that there are no electrons available which can carry a current, and that the material is an insulator. However, if the temperature rises, some electrons can be excited across the gap into the conduction band, and those electrons can transport a current. Thus, a semiconductor is a material whose electric resistivity drops with increasing temperature.

The electrical conductivity of a semiconductor can be increased by inserting impurities in the material: atoms that have more, or less, electrons than the atoms of the semiconducting material, and thus offer extra electrons which can populated the conduction band, or "suck" electrons out of the valence band, creating holes in the valence band. This way to insert extra electrons or of holes is called n- or p-type doping, respectively.

Now, if a n-doped and a p-doped semiconductor are brought into contact and an electrical current is driven across the junction by an externally applied voltage, the surplus electrons and the holes run into each other and can recombine, setting free an amount of energy roughly corresponding to the band gap. If the electron and the hole have the same momentum, this energy can be carried away by a photon, and the device emits light - it's a LED. The requirement of equal momentum is both crucial and restrictive and has as a consequence that only some n-p junctions can be used as LEDs. Typically, a maximum of the valence band has to occur at the same momentum as a minimum in the conduction band. Such a feature of the band structure is called a direct gap, and the width of the direct gap determines the colour of the light emitted by the LED.

The band structure of Gallium nitride (GaN) has a direct gap with a width of roughly 3.4 eV, corresponding to a wavelength of 365 nm in the near ultraviolet. (Source: S. Bloom et al, Phys. Status Solidi (b) 66 (1974) 161; via Landolt-Börnstein III/41A1b).

While semiconducting materials with direct band gaps corresponding to red light have been easy to handle and produce since quite a while, creating devices for the production of blue light turned out to be much more complicated. A material with a suitable band gap is Gallium nitride (GaN). The figure shows the band structure of GaN, i.e. the energy of electrons as a function of momentum. The different labels along the momentum axis denote special points in the Brillouin zone of GaN. There is a direct band gap at the so-called Gamma point, which corresponds to electrons (and holes) with zero momentum. Such a direct gap at the Gamma point is ideally suited for the use in LEDs, and the width of the gap 3.4 eV means that the light emitted will be ultraviolet, with a wavelength of 365 nm. So, this looks like an ideal material to build blue LEDs, and indeed, GaN is at the base of today's LEDs, such as those in my little lamp.

Shuji Nakamura at his desk
at UCSB Santa Barbara
However, there are some caveats: techniques to grow suitable films of GaN had to be found, as had methods to dope the material. The number of defects had to be reduced as much as possible, since defects trigger the recombination of electrons and holes without the production of light. The first LED emitting blue light with high efficiency over a longer time was finally developed in 1993 by a small team lead by Shuji Nakamura at the Japanese company Nichia Chemical Industries. They had grown a quite complicated, layered structure of GaN, AlGaN, and InGaN, and thus laid the foundation for a tremendously growing business:

Blue LEDs based on InGaAs are now very reliable and comparably cheap to produce. They are used in displays, beamers, for lighting, indicator lights, and advertisements. Used in combination with red, amber and green LEDs, they produce white light and are more energy efficient and reliable than incandescent lamps, which they may replace in the near future. Devices optimised for the stimulated emission of coherent blue light are used as blue Laser diodes in the Blue Ray optical data storage format.

The story of Nakamura, who was awarded the 2006 Millennium Technology Prize as the inventor of new source of light, is not less exciting: He managed to convince his boss at Nichia to pursuit his search for the blue LED based on Gallium nitride, and, as the Chairman of the International Selection committee for the Millenium Prize explains,

Shuji Nakamura is a splendid example of perseverance and dedicated research work, and of making a major breakthrough. He has worked with great determination for decades, and even severe setbacks have not prevented him from achieving something that other workers in the field regarded as almost impossible: using a reactor system of his own design to develop a solid material, in this case gallium nitride, into a powerful light source producing blue, green and white light, and also creating a blue laser.

That's the story of the little blue lamp.

  • The first paper about the blue LED is Shuji Nakamura, Takashi Mukai, and Masayuki Senoh, Applied Physics Letters 64 (1994) 1687; doi 10.1063/1.111832. A good general introduction is High-Luminosity Blue and Blue-Green Gallium Nitride Light-Emitting Diodes by H. Morkoç and S. N. Mohammad, Science 267 (1995) 51; doi 10.1126/science.267.5194.51 (subscription required for both papers).

  • Blue Laser Diodes by Joachim Piprek is a freely available review about practical examples of GaN laser simulation, analysis, and optimisation, but gives also some general background on the GaN based blue LED technology.

  • Nichia 's Shuji Nakamura: Dream of the Blue Laser Diode is an interesting interview with Nakamura.

Update: via Cocktail Party Physics, here is a report covering a talk by Nakamura on the Current Status of Solid State Lighting at the American Institute of Physics Industrial Physics Forum on "The Energy Challenge" last Monday, October 15, in Seattle.

Solid-State Lighting, the use of LEDs for lighting, has huge prospects because it is more energy-efficient than using usual incandescent or fluorescent lamps, and because LEDs are becoming ever cheaper and more durable. More information can be found, e.g., in Physics Today, December 2001: The Promise and Challenge of Solid-State Lighting (doi 10.1063/1.1445547), and at this page on Solid-State Lighting of the U.S. Department of Energy - Energy Efficiency and Renewable Energy - check out the PDF files.



  1. Shuji Nakamura was treated like a dog. He was reviled and cheated by his employer. When he finally sought and eventually obtained financial redress he was thoroughly screwed by his legal counsel. He left (was de facto expelled) from Japan.

    Good lab doggy! Return to the CVD bench and do something wonderful for us again. Management needs a new pair of golf courses.

  2. Nice post Stefan, and thanks for the story of Shuji Nakamura.

    Uncle AL, you said that he was basically expelled from Japan? So he's now in UCSB.

    "Thus, a semiconductor is a material whose electric resistivity drops with increasing temperature."

    But isn't this true of metals as well? Oops, no, I now recall: for metals the resistivity increases linearly as a fn. of temp.


  3. hi all,

    can anyone tell the watt/lumen ratio of LEDs compared to incandecent lamps?

    I use the 4,5v MAGLITE LED on the job every day and am very pleased.

    I observe,that the color of light does not change as the battery wear out, only the amount of light decreases. nice feature!



  4. Hi Klaus,

    can anyone tell the watt/lumen ratio of LEDs compared to incandecent lamps?

    For a quick answer, have a look at this plot from an article in Physics Today, The Promise and Challenge of Solid-State Lighting, December 2001. The InGaN blue LED has reached the performance of incandescent lamps of about 20 lumen/Watt in 2000, and the efficiency of LEDs has continued to grow ever since.

    Here is a link to a more recent table, which shows also that the lifetime of LEDs for lighting is much higher than for other lamps, and that the total cost after 50000 hours of usage is much lower.

    You can find more data - albeit a bit outdated - at this page of the U.S. Department of Energy - Energy Efficiency and Renewable Energy on Solid-State Lighting - check out the PDF files.

    (These links come thanks to Jennifer Ouellette's post LED-ing the Way)

    I observe,that the color of light does not change as the battery wear out, only the amount of light decreases. nice feature!

    Yeah, it's a good thing in general that the quantum mechanics of the solid state - band structure and all that - is not influenced by applying moderate voltages ;-)

    Best, Stefan

  5. Hi Stefan, great post about the source of light.

    The source of light:

    is really interesting of course, I wonder how long before the solid-state era is envoloped into "liquid-state" physics?

    The process of LED as your links details show, is very like the process of "sonoluminessence", wherby the bubble replaces the "Electron"?

    Neat torch, best paul.

  6. Hi Paul,

    The mechanism of light emission in sonoluminescence isn't well understood. But it's from continuously oscillating bubbles driven by self-organizing spherical waves. Can't see much similarity to electrons in a semiconductor.

    (Hmm. Then again, that's a little like the waves in the Wheeler-Feynman absorber theory of radiation. Also, Dirac had a semi-classical model of the electron as a bubble in the electromagnetic field.)

  7. Nakamura was not "expelled" from Japan. He left voluntarily when he received a good offer in the U.S.

    The Japanese didn't want him to leave. His departure was broadcast live on TV as an "event". He is often cited as an example that Japan's corporate research system needs reform. He himself has written books on the issue that are widely read and cited in Japan.

  8. Hi Kris, I aint heard of Wheeler-Feynman absorber theory? I dowloaded a very interesting paper from your linked site, its 110 pages long, but it will really absorb me for a while, especially as it is very relevant to my recent forays of De-Broglie, very interesting and thanks, I think you may have something there!, paul_v.

  9. Hi Paul,

    Here's the reference:

    Wheeler, John Archibald and Feynman, Richard Phillips (1945). "Interaction with the Absorber as the Mechanism of Radiation” Reviews of Modern Physics, 17, 157-181.

    Note Wheeler's hand-drawn pictures of incoming and outgoing waves. Feel free to email me with questions about the 110 page paper.

  10. Hi Paul,

    I cannot add more about sonoluminesce than Kris has already said - but from the little I know about it, it seems to me that it is way less efficient in the production of light than are LEDs.

    Best, Stefan

  11. There's an interesting episode in Nakamura's story relating to patents.

    Nakamura was the first inventor to sue his former employer for royalties on the blue LED and succeed in a trial court in Tokyo. His verdict was upheld by the first appelate court, but then overturned by the Supreme Court.

    Japan, like Germany, but unlike the default rule in the U.S., has a law (Article 35 of the Japanese patent law) that requires employers to pay employee-inventors a reasonable royalty on their inventions.

    The U.S. actually does have a similar provision in its law in the form of the Bayh-Dole Act, which nominally requires universities to share royalties with academic inventors.

    Determining what constitutes a reasonable royalty is extremely difficult because it takes so long for inventions to be commercialized. The default rule in most of the U.S. is basically that the salary that a corporation and inventor negotiate in advance of employment is basically fair compensation for inventive work, and that inventors shouldn't expect more than that (unless they do the work on their own time, with their own equipment and money).

    Although it may seem counterintuitive, this is probably part of the reason that so much more R&D was done in the U.S. for a long time. Corporations were more likely to fund R&D when they knew they would profit from anything that came out of it.

    The Bayh-Dole Act, the Japanese rule, the German rule, and a recent ruling in Australia all make it harder for inventors to negotiate with corporations that would otherwise be interested in commercializing inventions.

    Nakamura is a living example of the difficulties.

  12. This comment has been removed by the author.


Comment moderation on this blog is turned on.
Submitted comments will only appear after manual approval, which can take up to 24 hours.