(a) Some commonly used III-V or II-VI elements are shown above, with the roman numerals denoting the number of valence electrons for elements in the column. Note that III + V equals 8; so does II + VI. Valence electrons for elements at the top of the column are more tightly bound. Therefore, the bandgap energy is greater for elements higher up the column.
Figure 1 continued. (b) The number 8 denotes the number of electrons needed to complete covalent bonds between atoms to form a molecule or crystal. GaP and GaN are stable crystals with valence electrons filling the outer shells. (c) To make GaN conductive to electrons, i.e., n-type, you dope it with silicon, which has four valence electrons. That is, you replace a small percentage of the gallium atoms with silicon. This process fills up the outer shells and leaves extra electrons to move about freely, as in a conductor. (d) To make GaN conductive to holes, i.e., p-type, you dope it with magnesium, which has two valence electrons. So, you replace some of the gallium atoms with magnesium. This process creates holes (i.e., you don't fill up the 8 covalent bonds). Thus, an electron from an adjacent bond can then jump into a hole, ad continuum, making the process a migration of holes. Under forward bias, holes and electrons recombine in the active region to produce light. Adding Al and In to GaN, which forms the quaternary compound semiconductor AlInGaN, adjusts bandgap (wavelength) and more closely matches the lattice constant of the substrate (for better efficiency).
In LEDs, holes and electrons recombine in a double heterostructure or quantum well active region when passing through a p-n junction of doped semiconductors under forward bias (figures 1 and 2). The active region is made of small-bandgap material and the barrier or confinement regions are made of large-bandgap material. This kind of structure can be used for laser diodes, too, as an LED is just a laser diode without an optical cavity. This "technology of light" is the product of a synergistic integration of technologies that has made possible the Information Age (see Nobel laureates aided Information Age).
LEDs, ELDs (electroluminescent diodes), and OLEDs (organic LEDs) are coming into their own and competing against their more straight-shooting cousin, the laser diode. LEDs may supplant laser diodes as the light source in fiber optic communications, and high-power LEDs are being used for illumination, pioneering the drive to replace incandescent bulbs and fluorescent lamps. ELDs and OLEDs may replace CRTs and liquid crystal displays. And OLEDs, which can be deposited on flexible substrates, may be useful for new applications such as "lighted" wallpaper.
Many of the low-powered red emitters we're familiar with, which are based on GaAsP, are not bright. Grown on GaAs substrate, they are not lattice matched, and hence are not efficient emitters. "The closer the lattice constant of the substrate is to the lattice constant of the III-V crystal to be grown, the better the lattice match is," said Paul Martin of LumiLeds (San Jose, CA).
The breakthrough was the introduction of the AlGaInP quaternary compound semiconductors for the red and yellow and GaN materials for the blue and green. AlGaInP, as the active region in either a double heterostructure or quantum well device structure and grown on GaAs substrate, is highly luminous. The range of bandgaps, 1.9 to 2.2 eV, corresponds to emission at 650 to 570 nm, and can be varied by mixing the Al to Ga ratio in AlGaInP. Usually the Al goes from 0 to 50 percent and the Ga from 50 to 100 percent.
For GaN materials, one can mix GaN, which emits in the UV, with InN, which emits in the red, to form InGaN, which emits any color from UV to red, depending on the ratio of In to Ga.
Blue and green LEDs were developed and commercialized by Shuji Nakamura (see OE Reports, August 1996), formerly of Nichia Chemical Industries (Japan) and now at the Univ. of California/Santa Barbara. At that time (about 1988), most people were pursuing blue LEDs and diode lasers using II-VI-based materials such as ZnSe. Such materials, though easy to grow, had short lifetimes. Nakamura instead chose a III-V-based material, GaN, even though it was more difficult to grow. The lattice mismatch of GaN grown on sapphire was about 13 percent, but they developed a two-flow MOCVD (metal-organic chemical vapor deposition) system and a GaN buffer layer technique to grow high-quality GaN films. The bandgap of AlInGaN can be changed from 2.0 to 6.2eV, corresponding to emissions from 620 to 200 nm.
Figure 2. (a) Heterojunction under forward bias. Today, most high-efficiency LEDs use the double heterostructure design where the carriers are confined to the active region, WDH. Ec = energy of conduction band, Ev = energy of the valence band, EF = Fermi energy, the energy at which the probability that a quantum state is occupied by an electron is 1/2 . t is the spontaneous recombination lifetime. Typically, for a double heterostructure, WDH > 500 Angstroms, while for a quantum well, WDH = 50-200 Angstroms. (b) Structure of a double heterostructure LED. For example, the active region can be AlGaInP with the top and bottom confinement layers AlInP and the current spreading layer AlGaAs. The substrate is GaAs. (From Light-Emitting Diodes: Device Physics, Fabrication, and Applications, SPIE Short Course Notes, Photonics West 2000, E. Fred Schubert, Boston Univ.).
These high-brightness LEDs promise to be a new source of white light illumination. Traditional LEDs had very small output powers, about a few milliwatts, but today's AlInGaP-based LEDs have record output powers as high as 100 lumens in the amber. Usually, blue and yellow are combined to form the white light. "The advantages include reliability, less heat generation, and higher efficiency than comparably sized incandescents," said Bob Karlicek of GELcore (Valley View, OH).
In the best quantum well devices, the internal efficiency (the conversion of electron-hole pairs to light) approaches 90 percent. If an extraction efficiency (going from a quantum well device to a finished LED) of 45 to 55 percent can be achieved, then the overall efficiency of an LED is about 50 percent. "If we think about a regular light bulb, it's 10 times worse than that," said E. Fred Schubert of Boston Univ.
Martin said LEDs today have an input power of 1 watt, which will increase to about 4 watts in a year and 10 watts in a couple of years. These 10-w LEDs will operate in the white at about 20 lumens per watt, which means an amazing 200 lumens out of a single LED, compared to 900 lumens out of a 60-w incandescent bulb.
Schubert and his colleagues have developed what is called a photon recycling semiconductor light emitting diode. The active region (InGaN) of the primary LED emits in blue. Bonded on top of that is a photon recycling wafer (Al0.3Ga0.7InP) that emits yellow, the complementary color. The LED is dicromatic. The human eye sees the result as white light.
Figure 3. Simple model of the GaN:RE excitation process, where RE is a rare earth element. The Schottky diode is the rectifying metal-semiconductor junction, the ITO-GaN contact, where ITO is indium tin oxide. A bright ELD uses a large concentration of rare earth in order to maximize the probability of the radiative events 3a and 3b while maintaining adequate crystallinity so as to decrease nonradiative scattering events 1 and 2. In 3a, a hot (accelerated) electron in the conduction band of GaN transfers its energy (broken arrow) to a rare earth atom, raising it to an excited state. In 3b, the electron at the bottom of the conduction band of the GaN has lost its energy. The electron in the excited state ES of the rare earth returns to its ground state after emitting a photon. Image courtesy of Jason Heikenfeld and Andrew Steckl, Univ. of Cincinnati.
Electroluminescent diodes (ELDs) are another promising solid state emitter. The main difference between an LED and ELD is that the LED is a p-n junction operating at low voltage and high current, while an ELD is a Schottky diode (Figure 3), which is basically a metal to semiconductor that operates at high voltage and low current. In an ELD, rare earth atoms, such as erbium, are uniformly spread throughout the host material (GaN in this case) using molecular beam epitaxy. "Inside each rare earth, there are certain electronic transitions, certain energy levels, which, if excited, will then de-excite through emission of light," said Andrew Steckl of the Univ. of Cincinnati. Unlike LEDs, these excited energy levels are on the inner shell and, to get them excited, you have to bombard this rare earth using high voltage. As the electron or hole goes through the material under high field it picks up energy, collides with the rare earth, and, in the process, excites the inner shell electron.
Steckl, citing the advantage of ELDs, said, "The colors they emit are very monochromatic. You can incorporate two or more rare earths in the same layer and generate a mixed color, and the colors themselves are very pure."
ELDs would be used for flat panel displays, supplanting liquid crystal displays, which are slow, limited in temperature range, have a limited viewing angle and lifetime, and which cannot operate at a TV frequency. ELDs would be an all-solid-state solution to flat panel displays, replacing the liquid crystal encapsulated between two glass plates. They could also find possible uses in optical fiber communication, lighting, optical memory, and as instrument indicators.
Researchers are also developing OLEDs, mainly for displays. The advantages of OLEDs are (1) they are easier and cheaper to manufacture; (2) they can be deposited on almost any substrate, including flexible ones such as plastic; and (3) they can be made very large (luminescing sheets).
These LEDs are built out of polymers, which are normally insulating. In the early 1970s, however, Hideki Shirakawa of the Univ. of Tsukuba discovered that some common polymers can be made to conduct. The polymer has to have alternating single and double bonds, called conjugated double bonds. An example of this is polyacetylene (Figure 4a). To conduct, the polymer either has to be oxidized (remove electrons), which makes it p-type, or reduced (add electrons), which makes it n-type (Figure 4b). Under an electric field, the π electrons or holes of the conjugated bonds jump from one double bond to another (π orbital electrons are not as strongly held as the σ orbital electrons of a double bond). A polaron or radical cation (positively charged molecular entity with an upaired electron) is produced and migrates along the chain of the polymer molecule. What is amazing is that this doping of polyacetylene films by halogen oxidation made it 109 times more conductive, giving it a conductivity of 105 Siemens/meter. In comparison, teflon has a conductivity of 10-16 S/m and copper and silver 108 S/m.
Figure 4. (a) The conjugated (i.e., alternating single and double) bonds of the polyacetylene polymer. (b) The reaction equations for oxidation (remove electrons) and reduction (add electrons) of polyacetylene. [CH]n is the polymer and x is an integer number representing the total number of positive or negative charges.
OLEDs are also easier to deposit, as spin casting can be used to lay down the layers. "Depending on the speed with which you spin and the viscosity of the polymer sample that one starts out with, you'll be able to control the thickness and quality of the film that you get out of it," said Ram Sivaraman, Radiant Photonics (Austin, TX). The process takes about two to three hours as opposed to an inorganic material growth, which can take a day or even more.
The p-n junction is built by sandwiching a confinement layer, poly(phenylene vinylene), between a hole-rich polymer layer such as poly(vinyl- carbazole) and an electron-rich polymer layer such as polyaniline (Figure 5). Sivaraman's structure was built on a glass substrate with indium tin oxide as the bottom contact layer and a gold electrode on top. Electrons and holes are injected toward the middle layer where they would recombine and emit light.
Figure 5. A schematic showing the structure of an organic LED. PANI=polyaniline, PPV=poly(phenylene vinylene), PVK=poly(vinyl-carbazole). The active region or confinement layer, PPV, is sandwiched between a hole-rich layer, PVK, and an electron-rich layer, PANI. The ITO (indium tin oxide), bottom electrode, is the hole injection layer. The PANI is doped with sulfonic acid to make it electron rich and is the electron transport layer. The gold electrode on top of it is used as the electron injection layer. Image courtesy of Ram Sivaraman, Radiant Photonics.
The emission peak for poly(vinyl-carbazole) is in the blue. Other researchers have developed OLEDs emitting in the green and red, but most of those involve using inorganic materials as the electron-injection layer.
William Gillin of Queen Mary College, Univ. of London, is concentrating on developing OLED infrared emitters based on rare-earth-containing molecules, such as lanthanide hydroxyquinoline chelates. His OLEDs may supplant laser diodes in fiber optic communication because, he said, they "can be deposited directly on a silicon substrate, which can provide all of the signal processing and drive electronics. Waveguides for 1.5-µm light can easily be formed in silicon and integrated with our OLEDs." Once the drive electronics circuit is made, it only takes a couple of low temperature evaporations to make the OLED, so integration is simpler and cheaper. "For conventional LEDs, which are grown onto InP or GaAs substrates, these devices must be grown, diced, and then individually connected to some separate driving electronics."
As the vacuum tube gave way to the transistor, LEDs may one day supplant the incandescent bulb and fluorescent tubes. They are already used for stoplights, automobile lighting, signs, and directed illumination. OLEDs promise to revolutionize the flat panel display market and are already in use in some cell phones.
In technology, what is dreams today, is tomorrow, reality.