GaN FETs: The gallium nitride power transistor

A GaN FET (gallium nitride field-effect transistor) is a wide-bandgap power semiconductor switch. It uses a heterojunction between gallium nitride (GaN) and aluminium gallium nitride (AlGaN) to form a high-mobility conduction channel. Prized in modern power electronics, gallium nitride FETs deliver faster switching, higher power density, and lower on-resistance than conventional silicon transistors.

Many modern GaN devices are manufactured on a standard silicon wafer, combining the performance advantages of GaN with the cost benefits and scalability of established silicon manufacturing processes.

Types of gallium nitride transistors

Did you know that terms like GaN HEMTs, GaN FETs, GaN transistors, and GaN power semiconductors are all terms used to refer to GaN transistor products?
 
  • GaN HEMTs: The technical description for a type of GaN transistor using a high-electron-mobility-transistor (HEMT) structure. GaN HEMT technology forms the basis of most commercial GaN power transistors today.
  • GaN FETs: Casual reference to GaN transistors by those who have a long history with traditional silicon Field Effect Transistor technology.
  • GaN transistors: A broader term encompassing GaN transistors which encapsulates HEMT and other, lesser used, transistor structures.
  • GaN power semiconductors: A very broad term encompassing GaN transistors as discrete devices or those that also integrate additional functions such as sensing, drivers and more.
  • CoolGaN™ Transistors: A specific type of GaN HEMT transistor developed by Infineon, known for high device performance, quality and reliability.

Benefits of GaN FET technology

The key benefits of GaN FET technology include:

  • higher switching speeds
  • improved power efficiency
  • lower losses
  • excellent thermal performance

These advantages allow designers to achieve greater power density at virtually every power level, particularly in demanding high frequency power-conversion systems.

Compared with traditional silicon devices, GaN transistors require smaller passive components and cooling systems, resulting in more compact and efficient designs.

Gallium nitride transistors compared to conventional Silicon MOSFETs


Besides silicon MOSFETs, designers often compare GaN technology with silicon carbide (SiC) devices. While silicon carbide is frequently preferred for very high-voltage applications, GaN FETs typically deliver superior switching performance in high-frequency power-conversion systems.

Operating principle & the 2DEG channel

The GaN band gap of roughly 3.4 eV — about three times silicon's 1.1 eV — gives the material a far higher critical electric field. This is what enables the high breakdown voltage and low on-resistance that define GaN FETs.

Power GaN FETs are typically designed as High-Electron-Mobility Transistors (HEMTs). A GaN HEMT device achieves its superior conductivity without requiring traditional chemical doping, relying instead on the atomic properties of the crystal materials.

The two-dimensional electron gas (2DEG)

  • The heterojunction: A thin barrier layer of AlGaN is grown over a thicker buffer layer of GaN. There is a strong spontaneous and piezoelectric polarization inherent to GaN materials.
  • Creating the 2DEG: This polarization naturally pulls electrons out of the crystal and draws them to the boundary between the GaN and AlGaN. There, they collect in an incredibly thin layer — only about 1 nanometre thick. This creates a two-dimensional electron gas (2DEG).
  • Electron mobility: The 2DEG forms a highly conductive channel. The electrons float freely in this plane and aren't tied to any atom's lattice. As a result, they barely scatter off atoms and can move at very high speed.

Learn more about GaN HEMT structure, 2DEG physics, and enhancement-mode architectures

Switching behavior

Most commercial power GaN FETs are enhancement-mode (e-mode) devices, meaning they are normally-off.

  1. OFF state (VGS = 0 V): When no voltage is applied to the gate, the 2DEG layer under the gate is depleted (missing), preventing electrical current from flowing between the source and drain.
  2. Turn-ON (VGS = Vth): When a positive threshold voltage is applied to the gate, it creates a field effect that pulls electrons back into the 2DEG channel beneath it. This completes the electrical path from the drain to the source, and the transistor conducts current.
  3. Turn-OFF (VGS = 0 V): When the gate voltage drops to zero, the electrons immediately disperse away from the gate area, breaking the 2DEG channel and turning the device off.

How switching differences impact circuits

  • Zero reverse recovery: Silicon MOSFETs contain a parasitic P-N junction that forms an internal body diode. When this diode conducts in reverse, it requires time and energy to sweep the trapped carriers away (reverse recovery). Because GaN FETs rely on the 2DEG and do not have a physical P-N junction, they have zero reverse-recovery charge. This drastically reduces power loss and electromagnetic interference (EMI) during high-frequency switching.
  • Low parasitic capacitance: Because GaN FETs feature a horizontal (lateral) structure, their parasitic capacitances (input/output/reverse) are significantly lower than those of bulky, vertical silicon MOSFETs. Less charge is required to turn the device on or off, making transitions occur much faster.
  • Reverse conduction: A GaN FET doesn't rely on a slow body diode to conduct in reverse. It carries current backwards through its 2DEG channel instead. This lets it act as a synchronous rectifier with zero reverse recovery.

Explore more about GaN

Frequently asked questions about GaN FETs

What are GaN FETs used for?

GaN FETs are used in high-frequency, high-density power conversion, where both efficiency and size matter.

Typical applications include:

GaN technology is also playing an increasingly important role in electric vehicles, supporting efficient onboard chargers, DC-DC converters, traction-related subsystems, and auxiliary power electronics.

They excel in applications where higher switching frequencies allow smaller magnetics, reduced system size, and improved overall efficiency.

The main difference is how each transistor conducts current. A silicon MOSFET uses a vertical, doped channel and has a built-in body diode. A GaN FET (a HEMT) uses a lateral 2DEG channel at the AlGaN/GaN heterojunction and has no body diode. In contrast to older technologies such as bipolar transistors, GaN FETs offer voltage-controlled operation with significantly lower switching losses and much higher switching frequencies.

GaN also has a wider band gap than silicon — about 3.4 eV versus 1.1 eV — and a higher critical field. This gives GaN FETs lower on-resistance, lower gate charge, and no reverse-recovery charge, so they switch faster and with lower losses.

Silicon MOSFETs still have their place: they are cheaper and more rugged, with true avalanche capability.

No. A lateral GaN HEMT has no p-n body diode like a silicon MOSFET. It can still conduct in reverse (third-quadrant conduction) through the 2DEG channel, behaving diode-like but with no stored charge and therefore no reverse-recovery losses. The trade-off is a higher reverse voltage drop, which makes dead-time management important.

Because standard enhancement-mode (e-mode) GaN FETs are lateral devices without a standard body diode, impact ionization does not cause a controlled clamping effect, hence no avalanche effect.

However, the 650 V-rated GaN HEMT can tolerate a much higher reverse voltage; it reaches 910 V at the same drain current value as the 650 V silicon MOSFET.

Nevertheless, similar to the silicon device, it can tolerate this condition only for a period of time. The ability of Infineon CoolGaN™ devices to temporarily withstand higher-than-rated voltage spikes for a limited time enables robust operation during extreme operating events, such as a current or voltage surge condition.

Accordingly, several useful transient voltage ratings are provided on CoolGaN™datasheets. With these parameters, the device’s transient behavior is well described.

HEMT stands for High Electron Mobility Transistor. It conducts through a two-dimensional electron gas (2DEG). This 2DEG forms at the junction between two semiconductors with different band gaps, such as AlGaN and GaN.

Because the electrons move in an undoped channel, they avoid impurity scattering and reach very high mobility.

A 2DEG (two-dimensional electron gas) is a thin, highly conductive sheet of electrons confined at the AlGaN/GaN heterojunction interface. It forms from spontaneous and piezoelectric polarisation rather than from doping, creating a high-density, high-mobility channel.

This 2DEG is the current-carrying path of a GaN HEMT.

Not automatically. Faster switching has two sides. On the upside, it lowers the energy lost per transition and lets you use smaller passive components. On the downside, a higher frequency means more switching events per second — which can raise parasitic inductance and capacitance creating new design challenges.

So, instead of indiscriminately maximizing speed, engineers use GaN to optimize the balance between switching losses and conduction losses. To achieve peak efficiency, designers must tailor the switching speed and frequency to the specific topology (e.g., resonant converters vs. hard-switching) and optimize the board layout to minimize parasitic losses. 

In GaN transistors, normally-off (e-mode) devices remain turned off when zero voltage is applied to the gate, requiring a positive voltage to conduct.

In contrast, normally-on (d-mode) devices naturally conduct current at zero volts and require a negative voltage applied to the gate to turn them off. In RF amplifiers and specialized defense/satellite electronics, normally-on is preferred.

In power electronics for commercial power supplies as an example, normally-off is preferred. Infineon’s lateral eHEMT is inherently normally-off. The only way to make d-mode GaN normally-off is to add another transistor in series.

The result is a cascode transistor combination consisting of a common-source silicon MOSFET driving a common-base depletion-mode GaN HEMT. Translating this to performance means that e-mode uses less drive, switches faster, and typically needs fewer circuit components whereas the d-mode cascode often requires extra damping and has reverse-current/di/dt limits that limit practical operating frequency.