Gallium Nitride: How GaN Transformed LEDs, RF and Power

Gallium Nitride (GaN) is a remarkable material that has profoundly impacted the fields of light, radio, and power, despite its legendary manufacturing challenges. Dr. Umesh Mishra, a pioneer in GaN transistors and LEDs, once described it as "God's gift to mankind" due to its surprising cooperativeness despite its inherent difficulties.

Beginnings

The element Gallium was discovered in 1875 by French chemist Francois Lecoq de Boisbaudran, who named it after France. In 1932, scientists synthesized Gallium Nitride by reacting pure gallium with ammonia gas at 1000 degrees Celsius. This resulted in a hard, chemically stable material with a direct bandgap two to three times wider than silicon's, a high electric breakdown field, and electron mobility comparable to silicon and silicon carbide. Its main drawback has always been its manufacturability, as it cannot be easily produced in large, pure boules like silicon.

The Blue Light Breakthrough Part 1

GaN's first major breakthrough was the blue LED. In the 1960s, RCA, under chairman David Sarnoff, aimed to create an LED to replace TV picture tubes. They successfully produced red and green LEDs using gallium arsenic and gallium phosphide but needed blue to complete the RGB spectrum.

Around 1968 or 1969, RCA scientist Herbert Maruska, at the suggestion of his boss James Tietjen, explored GaN for blue LEDs due to its UV light corresponding bandgap energy. The team pioneered Hydride Vapor Phase Epitaxy (HVPE) to grow GaN on sapphire substrates. This process involved running hydrochloric acid gas over liquid gallium, then mixing it with ammonia in a high-temperature chamber. Although the resulting GaN had many defects, it was a significant step forward, and other companies like Philips adopted the technique.

The next challenge was creating a PN junction, essential for LEDs. A PN junction requires one material doped with excess electrons (N-type) and another with electron holes (P-type). When voltage is applied, electrons and holes recombine, emitting light. However, nobody knew how to make P-type GaN. RCA experimented with a metal-insulator-semiconductor (MIS) structure, using zinc-doped GaN and indium surrounding N-type GaN to produce a green LED, and later magnesium to produce a blue one. While a remarkable achievement, the MIS approach was inefficient and not scalable for lasers, making it a commercial dead end.

RCA's GaN project faced further misfortune after David Sarnoff's death in 1971 and significant financial losses from their computer development efforts, leading to the project's cancellation. Other major research institutions like Bell Labs, Philips, Hitachi, and Toshiba also abandoned GaN-based blue LED research, shifting to seemingly more promising materials like zinc selenide. GaN research continued only through the efforts of a few dedicated individual researchers.

The Blue Light Breakthrough Part 2

Two main problems hindered the development of a PN-junction GaN-based blue LED: growing defect-free GaN layers and creating P-type GaN.

The first issue, growing high-quality GaN, was addressed by Hiroshi Amano. He applied the concept of "buffer layers," previously used for gallium arsenide, to GaN. Using metal organic chemical vapor deposition (MOCVD), a new tool unavailable in the 1960s, Amano deposited an aluminum nitride buffer layer on sapphire at 600 degrees Celsius, resulting in a smooth GaN layer despite the crystal structure mismatch.

The second issue, P-type GaN, was resolved by realizing that magnesium was indeed the correct dopant, but it needed activation. The ammonia used in Maruska's HVPE method contained water impurities, leading to hydrogen atoms trapping magnesium and preventing electrical activation. In 1989, Amano and his colleague Isamu Akasaki addressed this by running a low-energy electron beam over the magnesium-doped GaN, creating a weak P-type GaN. This led to the first PN-junction-based blue LED in 1989, though it was not very bright and primarily emitted UV light.

Shuji Nakamura of Nichia Chemical Industries further improved this in the late 1990s by replacing the e-beam annealing with a more uniform thermal annealing method, producing a better-performing P-type layer and a brighter blue LED. By alloying GaN with Indium Nitride, they eventually developed the practical, high-brightness Candela-class double-heterostructure (DH) blue LED, which is widely used today.

The blue LED enabled the creation of blue lasers for disc readers, projectors, and interferometry. More significantly, it led to the development of white LEDs. This was achieved by placing a blue LED inside a semi-transparent chamber coated with a special phosphor made from yttrium aluminum garnet (YAG) doped with cerium. The blue light from the LED partially passes through, while the rest is absorbed by the YAG phosphor, which then emits a greenish-yellow light. The combination of these lights produces white light.

Amano, Akasaki, and Nakamura were awarded the Nobel Prize in 2014 for their groundbreaking work, a testament to their belief and technical skill.

The LED Blows Up

In 1995, Wired Magazine called the blue laser the "Holy Grail" of the semiconductor industry. The article predicted that efficient, durable, and long-lasting white LEDs would replace traditional tungsten-filament lamps, a view considered extreme at the time but which has since become reality.

Initially, blue LEDs cost about $15 per unit. Competition from companies like Cree, which developed GaN-on-silicon carbide chips, helped reduce prices. Japanese, Taiwanese, and American companies rapidly entered the white LED market. Early white LEDs produced a cool, bluish light suitable for flashlights, car headlights, and signs. However, for general home lighting, consumers desired warmer light, which became available around 2002. This gradually opened up the massive home lighting market as LED brightness and cost structures improved.

Another significant market for LEDs emerged in mobile devices, where they served as backlights for displays, keypad lights, and camera lighting. By 2005, mobile devices accounted for 52% of the white LED market. Today, white LEDs dominate 80-85% of the general lighting market, saving immense amounts of electricity globally. This LED boom significantly expanded the GaN fabrication and research ecosystem, paving the way for GaN's second revolution: Radio Frequency (RF).

RF

RF chips are crucial for wireless communication, transmitting and receiving signals. They upconvert digital signals, boost power for transmission, and perform the reverse for reception. For a long time, RF transistors were niche, primarily serving defense applications like radars and radios, operating in the gigahertz range. The rise of satellite TV in the 1980s created the first substantial civilian RF markets, followed by the massive growth of mobile phones, which scaled up the RF chip industry.

The GaN HEMTs

Unlike blue LEDs, RF applications did not require P-type GaN. In the late 1970s, Bell Labs, building on work by Leo Esaki and Raphael Tsu at IBM, demonstrated two-dimensional electron gas (2DEG), a state where electrons move freely in two dimensions but are constrained in the third. This led to the concept of the High-Electron-Mobility Transistor (HEMT), a field-effect transistor with a source, drain, and gate made from a heterojunction (a junction between two different semiconductor materials). When sufficient voltage is applied, the heterojunction forms a 2DEG layer, allowing electrons to travel at high speeds and enabling fast transistor switching.

The first HEMTs were produced by Fujitsu in 1979, using heterojunctions of undoped Gallium Arsenide and n-doped Aluminium Gallium Arsenide. In the early 1990s, it was discovered that HEMTs could be made with N-type GaN and Aluminium Gallium Nitride heterojunctions, attracting the attention of the US defense industry.

The DARPA Challenge

In the 1990s, the US Navy sought more powerful radars to track distant targets and extend missile ranges, exceeding the capabilities of existing gallium arsenide chips. The Office of Naval Research (ONR) had been funding fundamental research into wide bandgap semiconductors since the 1970s and 1980s, with GaN emerging as a leading candidate.

In 2002, DARPA launched the Wide Bandgap Semiconductors for Radio Frequency Applications (WBGS-RF) program. At the time, early GaN chips were plagued by defects. The program aimed to improve GaN quality and the entire industrial ecosystem for RF GaN. Over more than a decade and three phases, the WBGS-RF program successfully improved GaN epitaxial quality, developed longer-lasting high-performance HEMTs, and produced monolithic microwave ICs with record performance. This program is considered one of DARPA's most successful commercialization projects.

GaN is now a cornerstone defense technology. In the 2010s, Raytheon and other defense contractors began integrating GaN-based RF electronics to replace older Gallium Arsenide in applications like ballistic missile tracking radars and jamming systems. In the consumer sector, GaN has supported the mobile phone revolution by powering cellular infrastructure like base stations for millimeter wave, though its high cost generally precludes its use in handsets.

The Gates of Power

Power semiconductors, found in devices like phone chargers and electric vehicles (EVs), control the flow of energy. They regulate power to motors, adjust LED brightness, and manage battery charging. While there is some overlap, RF and power semiconductors have different requirements. RF chips prioritize high frequency and efficiency in converting DC to RF power. Power transistors, on the other hand, are high-voltage switches. Frequency is less critical; they must handle high voltages without breakdown, allow high current flow with minimal resistance and heat, and ideally be "default OFF" (enhancement-mode) for safety.

GaN and Power

Historically, power transistors were made from silicon. However, silicon's relatively low avalanche breakdown field (around 0.3 megavolts per centimeter) limits its high-voltage capabilities. GaN's avalanche breakdown field is 10 to 11 times higher than silicon's, allowing GaN power switches to handle higher voltages or operate more efficiently at the same voltages, leading to smaller and lighter devices.

GaN-based HEMTs have replaced silicon in certain power applications. While not used in phone handsets due to cost, GaN is prevalent in phone chargers. GaN-enabled adapters, though existing since around 2010, gained popularity later in the decade with mobile phone OEMs like Xiaomi, Samsung, and Oppo. These adapters, along with audio amplifiers and computer power supplies, showcase GaN's strengths in the low to medium power semiconductor range (30 to 650 volts).

GaNakin, I have the High Ground

Despite its success in the low to medium voltage range, GaN has struggled to compete with silicon carbide (SiC) in high-voltage applications like EVs, data centers, HVDC, and trains. This is primarily because SiC excels with vertical MOSFETs, while GaN's HEMTs are lateral devices.

A lateral HEMT's source, gate, and drain are on the same plane, with current flowing horizontally through the 2DEG layer. While effective for high frequencies, scaling lateral structures to higher voltages requires proportionally increasing the distance between the source and drain, making the chip impractically long. For example, a 1,200V GaN lateral HEMT would need a channel length of 16+ micrometers.

Vertical transistors, where sources and drains are on opposite sides of the wafer and current flows through the bulk of the device, are better suited for high voltages. Increasing the breakdown voltage in vertical transistors simply involves thickening the distance between the source and drain.

Vertical Limit

Developing vertical GaN transistors presents several challenges:

1. Substrate: The most economical way to grow GaN is on other substrates like sapphire, silicon carbide, or silicon. However, a true vertical GaN transistor requires a bulk GaN crystal. While bulk GaN crystals are becoming possible, often grown via hydride vapor phase epitaxy (the same method used by RCA decades ago), issues remain.
2. P-type GaN: Vertical MOSFETs require robust P-type GaN. While P-type GaN is sufficient for LEDs, its magnesium dopants require higher energies to activate, leading to a low activation percentage. This is not adequate for high-performance vertical MOSFETs like those made from SiC. Although a niche vertical structure called "current aperture vertical electron transistors" (CAVET) uses P-type GaN superficially, a fully capable vertical GaN MOSFET has been elusive.

However, progress is being made. In October 2025, Onsemi debuted a vertical GaN transistor, built on a GaN-on-GaN substrate, capable of handling 1,200 volts. This development could enable GaN to compete with silicon carbide in high-voltage applications.

Conclusion

Gallium Nitride is a versatile material, excelling in three distinct fields: light, radio, and power. While it is unlikely to surpass silicon in digital logic, GaN is increasingly vital in its niche spaces within RF and power. These sectors are growing rapidly, worth tens of billions of dollars, and are crucial for the performance of AI-enabled high-tech applications such as robots, EVs, and renewable energy systems.