Nobel Prize in Physics recognizes research leading to high-brightness blue LEDs

At the heart of an LED is a p–n junction between a semiconductor with a surplus of electrons (the n-type layer) and another semiconductor with a surplus of holes (the p-type layer), as diagrammed in figure 2. A positive voltage applied to the p-type layer (left) drives holes from that layer and electrons from the n-type layer (right) toward the junction, where they combine to generate light.

In materials with a direct bandgap—that is, materials in which the momentum is the same for electrons in the conduction band as for the holes in the valence band—the charge carriers combine radiatively and the energy carried by the photon (blue) equals the energy gap of the material. Recombination in materials with an indirect bandgap requires some interaction with the lattice, so the radiative recombination is less efficient and the photon energy is less than the bandgap.

The search for short-wavelength, blue-light emission has concentrated on wide, direct-bandgap materials such as GaN, whose bandgap is 3.4 eV at 300 K. Zinc selenide is another direct-bandgap material with an appropriate bandgap. In the early 1990s, researchers demonstrated a blue-green laser based on ZnSe, but technical challenges such as problems in doping and short operating lifetimes limited further progress toward a practical device. (See the article by Gertrude Neumark, Robert Park, and James DePuydt, Physics Today, June 1994, page 26.)

Yet a third material of interest for blue LEDs is the indirect-bandgap compound silicon carbide. Unfortunately, p–n junctions made from SiC have had extremely low efficiencies.

Electroluminescence, or electrically stimulated light emission, was first seen in 1907 as a yellow glow emanating from a silicon carbide crystal upon the application of a voltage. The phenomenon was explained only decades later^1,2 in terms of radiative recombination of electrons and holes, thanks to the insight gained from research on semiconductors and p–n junctions.

The first LEDs, emitting in the IR, were made in the 1950s from efficient gallium arsenide p–n junctions. To get emissions in the shorter-wavelength visible region, researchers at industrial labs in Germany, the UK, and the US explored gallium phosphide, which has a wider, indirect bandgap. (As one goes from GaAs to GaP and then to GaN in the III–V group, the semiconductors have progressively wider, direct bandgaps because the elements paired with the Ga atoms have progressively smaller radii and therefore tighter binding.) In 1962 Nick Holonyak Jr and a coworker at the General Electric Laboratory in the US reported red light emission³ from an alloy of gallium, phosphorus, and arsenic, a combination that is used in red LEDs today. Subsequent developments added yellow and then green to the list.

Interest in blue emitters drew some researchers to look at gallium nitride. But the work ran into difficult challenges posed by the material. One is the problem of growing a defect-free crystal because of the large lattice mismatch with the available substrates (most commonly, sapphire). The second challenge was to dope the material with acceptors (p-type dopants) at sufficiently high concentrations.

In the late 1950s, a team at Philips Research Laboratory in Germany saw electroluminescence over a wide spectrum from GaN infused with a variety of dopants.⁴ Their observations, however, were limited to powders consisting of very small GaN crystals, as they were unable to grow large enough crystals to form a p–n junction.

In the 1960s Herbert Maruska at RCA Laboratories in Princeton, New Jersey, succeeded in growing single crystals of GaN films on sapphire using hydride vapor phase epitaxy (HVPE).⁵ He found that the material was n-doped by nature, but p-doping remained a challenge. When RCA sent Maruska for graduate work at Stanford University, Jacques Pankove and others in the group doped the GaN crystal with Zn to form a diode that emitted green light.⁶ Maruska then tried doping with magnesium and in 1972 produced blue-violet emission.⁷ 

The Mg-doped layers formed by the RCA team were insulating, which indicated that the holes were not free to move. Thus the structure was not a p–n junction, and the team proposed that the blue luminescence was caused by impact excitation of the filled Mg acceptors by hot electrons, followed by recombination. The emission efficiency was low. Before the researchers could make further progress, RCA canceled the GaN work.

Maruska comments now that growing GaN crystals with the HVPE technology of that time would never have worked because it contributed oxygen atoms that may have been combining with the Mg ions and rendering them ineffective as p dopants; he had just built a setup for the newer and more promising technique of metalorganic vapor phase epitaxy (MOVPE) when RCA pulled the plug.

Despite the availability of such tools as MOVPE and molecular-beam epitaxy, by the mid 1970s, most work on GaN had stopped. But in 1973, Akasaki, then at Matsushita Research Institute in Tokyo, took up the challenge. After he moved to Nagoya University in 1981, he was joined by Amano and other coworkers. Persistently pursuing the work through many experiments, the team succeeded in growing a high-quality crystal⁸ in 1986.

The key was first to deposit a layer of aluminum nitride on the substrate at low temperature. When later heated to the temperature used for GaN crystal growth, the AlN layer develops small crystallites on which the GaN can more easily nucleate. Close to the AlN buffer layer, the defect density is still high, but the growth becomes more uniform as additional layers are added.

In 1988 Akasaki and Amano discovered a solution to the vexing challenge of p-doping. They noticed that irradiating the Mg-doped GaN layer with low-energy electrons greatly improved its properties.⁹ That innovation opened the door for p–n junction formation.

In 1988 Shuji Nakamura was working at Nichia Chemical Corp, a small company in Tokushima, Japan, that specialized in phosphors. After working on GaAs, Nakamura grew interested in pursuing GaN. Like the team at Nagoya, he succeeded in growing high-quality crystals but with a different buffer layer. Instead of an AlN buffer layer, he grew a thin layer of GaN at low temperatures before laying down the rest of the crystal at higher temperatures.¹⁰ Nakamura had learned a similar two-step growth process for GaAs when he was a visiting researcher at the University of Florida. The technique is used in commercial production today.

In 1992 Nakamura was able to explain why the low-energy irradiation had improved the p-dopant behavior in the p–n junctions made by Akasaki and Amano. He proposed that hydrogen impurities introduced during growth were forming complexes with the Mg dopants that were rendering them passive and not effecive in creating holes. The electron radiation apparently drove off the hydrogen and activated the Mg acceptors. Nakamura devised another way to solve the hydrogen problem: postgrowth heating of the crystal in a hydrogen-free environment.¹¹ That’s the technique used commercially today.

Nakamura says that a third key step in getting a blue LED with high- luminescence efficiency was the introduction of a layer of indium gallium nitride as the emitting, or active, layer—that is, the layer where holes and electrons combine.¹² It is sandwiched between the n-type GaN and the p-type GaN layers. Like a GaN p–n junction, the InGaN structure has a high density of dislocations or crystal defects caused by the mismatch between the GaN and its sapphire substrate. Nevertheless, the structure with an InGaN active layer turns out to be far more luminescent than GaN alone.

An LED is a first step toward producing a laser, and both the Nagoya and Nichia groups continued their developments in that direction. Nakamura reported a long-lived blue GaN laser diode¹³ in 1996 (see Physics Today, April 1996, page 18).

At that time, one driving force for the development of a short-wavelength laser was its potential for use in high-density optical storage. Interestingly, however, the Blu-ray disk enabled by the blue LED is now competing with direct download of video information. The impact of blue (and white) LEDs for lighting has been far larger, comments Michael Kneissl of the Technical University of Berlin, because of the huge energy savings worldwide.

Steven DenBaars of UCSB recalls attending a conference on ZnSe in 1993. “Among the thousands of researchers who were studying ZnSe were just three notable ones who were working on GaN: Akasaki, Amano, and Nakamura.” After hearing their results, DenBaars himself switched to GaN materials. It appears he was not alone. The three researchers “were a spark that ignited a fire,” he says. The intense research that ensued has not abated.

The performance of LEDs only continues to improve, as researchers tackle various problems such as high initial cost, light quality, and loss of efficiency at higher electrical currents. (See Physics Today, July 2013, page 12.)

In addition to its applications in lighting, GaN is garnering increasing interest in other applications. According to Noble Johnson of PARC, a Xerox company, GaN switches are a major development that “could rival lighting as a killer app.” In particular, the strong bonding that gives rise to GaN’s wide bandgap also makes the material attractive for switching. As explained by UCSB’s Umesh Mishra, a switch is a device that can hold a voltage in its “off” state and pass current in its “on” state. Strong bonding enhances a material’s ability to hold the voltage and to switch rapidly.

Mishra, who has started a company called Transphorm to develop GaN transistors, notes that the energy lost in power conversion, such as switching from AC to DC power, consumes 10% of all electricity generated in the US. More efficient switches could cut that loss in half.

Isamu Akasaki was born in 1929 in Chiran, Japan. He graduated from the School of Science at Kyoto University in 1952 and started working at Kobe Kogyo Corp. In 1959 he went to Nagoya University, first as a research associate and then as a lecturer. After earning his PhD at Nagoya in 1964, he went to Matsushita Electric Industrial Co before returning to Nagoya in 1981. He retired from Nagoya in 1992 and was appointed a professor at Meijo University. He also holds the title of distinguished professor at Nagoya.

Born in 1960 in Hamamatsu, Japan, Hiroshi Amano earned his undergraduate degree and PhD in engineering from Nagoya University in 1989. After working there as a research associate, he went to Meijo University in 1992 and became a professor in 2002. In 2010, he returned to Nagoya as a professor.

Shuji Nakamura was born in 1954 in Ikata, Japan. After getting his master’s degree at the University of Tokushima, Nakamura worked for Nichia Chemical Corp in Tokushima. He spent 1988–89 at the University of Florida as a visiting researcher. He received a PhD in 1994 from the University of Tokushima and moved to UCSB in 2000.