Why GaN Powers Modern LEDs: Gallium Nitride Explained

On October 7, 2014, the Royal Swedish Academy of Sciences awarded the Nobel Prize in Physics to Isamu Akasaki, Hiroshi Amano, and Shuji Nakamura "for the invention of efficient blue light-emitting diodes which has enabled bright and energy-saving white light sources." Behind that terse citation lies one of the most consequential materials science breakthroughs of the 20th century: figuring out how to make gallium nitride work.

GaN had been a curiosity for decades before that. Researchers knew it had the right bandgap for blue and UV emission. They also knew it was miserably difficult to grow with decent crystal quality. Most of the semiconductor community had given up on it by the 1980s, chasing zinc selenide (ZnSe) instead as the pathway to blue LEDs. Nakamura, working at a small chemical company called Nichia with minimal resources, proved them all wrong.

What Is Gallium Nitride?

Gallium nitride is a binary III-V semiconductor: gallium from group III, nitrogen from group V of the periodic table. It crystallizes in the wurtzite structure (hexagonal) under normal growth conditions. Its key property for optoelectronics is a direct bandgap of 3.4 electron volts (eV), corresponding to a wavelength of about 365nm in the near-ultraviolet.

"Direct bandgap" means the minimum of the conduction band and the maximum of the valence band occur at the same point in momentum space (the gamma point). When an electron drops from the conduction band to the valence band, it can emit a photon directly, without needing to simultaneously emit or absorb a phonon (crystal lattice vibration) to conserve momentum. This makes radiative recombination efficient. Indirect bandgap materials like silicon can emit light, but the process is so inefficient that silicon LEDs are not practical.

GaN is also extremely hard, chemically stable, and tolerant of high temperatures. Its melting point exceeds 2,500 degrees C. These properties make it durable in demanding applications, but they also made it extremely difficult to process during early development. You cannot easily etch GaN with standard wet chemicals. You cannot melt it and pull crystals from a melt (like silicon). Growing high-quality GaN requires metal-organic chemical vapor deposition (MOCVD) or molecular beam epitaxy (MBE), both of which are complex and expensive.

The Alloy System: InGaN and AlGaN

Pure GaN emits at 365nm. That is useful for some UV applications, but not for the blue light needed for white LEDs, and not for the deep UV needed for germicidal applications. The trick is alloying.

InGaN: Blue and Green

Adding indium to GaN creates indium gallium nitride (InGaN). Indium nitride (InN) has a bandgap of about 0.7 eV (1770nm, infrared). By mixing indium and gallium, you can theoretically tune the bandgap anywhere from 0.7 eV to 3.4 eV, covering the entire visible spectrum and into the infrared.

In practice, InGaN LEDs work well in the blue (440-470nm, around 18-20% indium content) and green (520-530nm, around 25-30% indium) ranges. Going further into the yellow and red requires higher indium content, which introduces severe lattice strain and compositional instability (indium clustering). This is the "green gap" problem: LED efficiency drops dramatically between green and yellow, exactly where the human eye is most sensitive.

The blue InGaN LED, specifically the one Nakamura developed at Nichia in 1993, is the foundation of modern solid-state lighting. Every white LED in your home, phone, car, and office starts with a blue InGaN chip. A phosphor coating converts some of the blue light to yellow and red, and the combination appears white to the human eye. The blue LED literally made white LED lighting possible.

AlGaN: Ultraviolet

Adding aluminum to GaN creates aluminum gallium nitride (AlGaN). Aluminum nitride (AlN) has a bandgap of 6.2 eV (200nm, deep UV). AlGaN alloys cover the range from 3.4 eV (365nm) to 6.2 eV (200nm), spanning the entire UV-A, UV-B, and UV-C spectrum.

This is the material system behind UV-C LEDs for germicidal applications. A UV-C LED emitting at 265nm (peak DNA absorption) uses an AlGaN active layer with roughly 50% aluminum content. The efficiency challenges are severe: high aluminum content degrades crystal quality, reduces hole mobility in the p-type layer, and creates light extraction difficulties. But the physics works, and efficiency is improving steadily.

Nakamura's Breakthrough: What He Actually Did

To appreciate why the Nobel committee waited until 2014 to award this prize (for work done in the early 1990s), you need to understand what Nakamura was up against.

GaN crystals grow on foreign substrates (usually sapphire, Al2O3) because bulk GaN crystals large enough for device fabrication did not exist. Sapphire has a lattice constant of 4.758 angstroms. GaN's lattice constant is 3.189 angstroms. That is a 16% mismatch, which is enormous by semiconductor standards. For comparison, silicon and germanium differ by only 4%, and even that causes significant defect issues.

When you grow GaN on sapphire, the lattice mismatch creates threading dislocations at densities of 10^8 to 10^10 per square centimeter. Each dislocation is a non-radiative recombination center: a place where electrons and holes recombine by generating heat instead of light. With defect densities that high, GaN should not be able to emit light efficiently. In GaAs, dislocation densities above 10^4 per square centimeter kill LED performance.

Nakamura solved three critical problems in sequence. First, he developed a "two-flow" MOCVD reactor that improved gas flow dynamics and enabled high-quality GaN growth. Second, he refined the GaN buffer layer technique (building on Amano and Akasaki's earlier work with AlN buffer layers) to reduce dislocation density. Third, and most importantly, he figured out how to make p-type GaN.

P-type doping was the showstopper. GaN is naturally n-type (electron-rich) due to nitrogen vacancies. Making it p-type requires substituting magnesium atoms for gallium. Researchers had been trying this for years, but the magnesium-doped GaN always remained highly resistive. Akasaki and Amano discovered in 1989 that electron beam irradiation activated the p-type conductivity, but the mechanism was mysterious and the process impractical for manufacturing.

Nakamura discovered in 1992 that a simple thermal anneal in nitrogen atmosphere accomplished the same activation. The mechanism: during MOCVD growth, hydrogen passivates the magnesium acceptors by forming Mg-H complexes. Annealing above 700 degrees C in nitrogen drives out the hydrogen, freeing the magnesium to act as an acceptor and creating genuine p-type conductivity. This was a manufacturing-compatible process, and it unlocked mass production of GaN p-n junctions.

Why Not Other Materials?

GaN was not the obvious choice for blue LEDs. Other materials were explored first and found wanting.

GaAs and GaP: Wrong Bandgap

Gallium arsenide (GaAs, 1.42 eV, 870nm) and gallium phosphide (GaP, 2.26 eV, 549nm) were the first LED materials, commercialized in the 1960s. Their alloys (AlGaAs for red, GaAsP for orange-yellow, GaP:N for green) covered the infrared through green spectrum. But they cannot reach blue. AlGaAs maxes out at about 620nm (red). GaP has an indirect bandgap, making it inherently inefficient. There is no III-V arsenide or phosphide material with both a direct bandgap and the right bandgap energy for blue emission.

ZnSe: Close But Fragile

Zinc selenide (ZnSe, 2.7 eV, 460nm) was the leading blue LED candidate through the 1980s and early 1990s. Several groups, including teams at 3M and Sony, demonstrated ZnSe blue LEDs and even laser diodes. The problem was lifetime. ZnSe devices degraded rapidly during operation, with dark-line defects propagating through the active region. No one could make a ZnSe blue LED that lasted more than a few hundred hours. The material is simply too soft and too prone to defect multiplication under the stress of carrier injection.

When Nakamura's GaN blue LED hit the market with thousands of hours of lifetime, ZnSe research largely stopped. The materials science community shifted overwhelmingly to GaN.

SiC: Indirect Bandgap Problem

Silicon carbide (SiC, 3.0 eV for 6H polytype) was used for early blue LEDs by Cree in the late 1980s. But SiC has an indirect bandgap, which means photon emission requires simultaneous phonon emission. The resulting LEDs were extremely dim, suitable for indicator lights only. Once direct-bandgap GaN LEDs arrived, SiC as an LED emitter became obsolete. (SiC remains important as a substrate for GaN growth and as a power semiconductor material.)

Manufacturing Challenges

The Substrate Problem

Ideal LED manufacturing would use a native substrate: growing GaN on GaN. This eliminates lattice mismatch entirely, reducing dislocation density to below 10^4 per square centimeter. Native GaN substrates exist, but they are small (2-inch diameter is standard, 4-inch is emerging) and extremely expensive ($1,000-5,000 per wafer). For comparison, 6-inch sapphire wafers cost $20-50.

Most commercial GaN LEDs are grown on sapphire (for visible LEDs) or on AlN templates on sapphire (for UV LEDs). Silicon substrates (150mm, 200mm) offer a cost advantage for GaN power electronics but introduce additional challenges for LED applications due to thermal expansion mismatch and light absorption.

Defect Management

Since we cannot eliminate dislocations entirely on foreign substrates, the strategy is to manage them. Epitaxial lateral overgrowth (ELOG) uses a patterned mask to filter dislocations, reducing density by 100-1000x in the overgrown regions. Pendeo-epitaxy, nano-patterned substrates, and various buffer layer engineering approaches all aim at the same goal: getting dislocation density low enough for acceptable device performance.

For visible InGaN LEDs, the carrier localization effect mentioned earlier makes the material surprisingly tolerant of defects. For AlGaN UV-C LEDs, localization is weaker and defects are more damaging, which is a major reason UV-C LED efficiency lags visible LED efficiency by such a wide margin.

Beyond LEDs: GaN in Power Electronics

GaN's wide bandgap, high breakdown field, and high electron mobility make it valuable far beyond LEDs. GaN power transistors (high electron mobility transistors, or HEMTs) are displacing silicon in:

• EV chargers and on-board chargers. GaN enables smaller, lighter, more efficient power converters. Tesla, Apple, and Anker all use GaN-based chargers.
• 5G base station amplifiers. GaN RF transistors deliver higher power density and efficiency than silicon LDMOS or GaAs at 5G frequencies.
• Data center power supplies. Google, Microsoft, and AWS are adopting GaN power stages for server power supplies, targeting 99%+ conversion efficiency.
• Automotive power conversion. GaN is entering EV drivetrain inverters, competing with silicon carbide (SiC) for the high-voltage power conversion market.

The GaN power electronics market is growing at over 35% CAGR, faster even than the GaN LED market grew at its peak. The same material that lights your room is increasingly managing the electrons that charge your car.

Current Research Frontiers

Three active research areas will shape GaN's next decade.

Micro-LEDs. Arrays of GaN micro-LEDs (each pixel smaller than 10 micrometers) are being developed for next-generation displays. Apple, Samsung, and Meta are investing heavily. The challenge is achieving uniform emission and high yield across millions of tiny GaN devices on a single panel.

Deep-UV efficiency. Pushing AlGaN UV-C LED efficiency from 5% toward 20% requires solving the p-type doping problem for high-aluminum-content alloys, improving light extraction from high-refractive-index AlGaN, and reducing defect density through better substrate and buffer technology. This is directly relevant to the ongoing work in GaN materials science for short wavelength applications.

Vertical GaN power devices. Current GaN power transistors are lateral devices grown on foreign substrates. Vertical devices on native GaN substrates would handle higher voltages and currents, competing directly with SiC for EV and grid applications. Native substrate cost and size remain the primary barriers.

Why GaN Matters

Gallium nitride is one of perhaps five semiconductor materials that have genuinely changed the world. Silicon gave us computing. Germanium started the transistor revolution. GaAs enabled fiber optic communications. GaN gave us efficient lighting and is now transforming power electronics.

The numbers tell the story. LED lighting has reduced global electricity consumption for lighting by an estimated 40% compared to incandescent technology. That translates to roughly 1,400 terawatt-hours of annual electricity savings, equivalent to the total electricity output of about 150 nuclear power plants. All because three stubborn researchers figured out how to grow a difficult crystal and make it conduct holes.

For short wavelength sources specifically, GaN's alloy system (AlGaN) is the only proven path to efficient solid-state UV emitters. Every advance in UV-C LED efficiency builds on the GaN materials science foundation laid in the 1990s. The material that solved blue LEDs is now being pushed to solve germicidal UV, and the progress, while harder, is following the same trajectory.