What Is GaN (Gallium Nitride)? Why GaN Chargers Are Small

14 min read · Updated: October 23, 2026

GaN (gallium nitride) is a semiconductor material made of gallium and nitrogen. It can withstand a stronger electric push than silicon, and it can form a path where electrons move very easily. So it can switch electricity on and off faster than silicon. That lets a charger use a smaller transformer and smaller capacitors.

What you'll learn

  • Where GaN is used (chargers, LEDs, cell phone base stations, data centers)
  • How it differs from silicon (a bigger energy step, and an electron path at a boundary)
  • Why fast on-off switching makes a charger smaller
  • When to use GaN, SiC or silicon, and what is still hard about GaN

Prerequisites: it helps to know that a power semiconductor is "a part that switches electricity on and off fast to change its form" ("What Is a Power Semiconductor?").

In everyday life: from chargers to base stations

More and more chargers in electronics stores are labeled "GaN." In Japan, for example, Anker Japan launched a 30-watt USB charger using GaN in February 2019. It was one of the early examples.

Many USB-C chargers follow a standard called USB Power Delivery. The charger talks with the phone or laptop to agree on the voltage and current to send. A revision in May 2021 raised the maximum from 100 watts to 240 watts (USB-IF). GaN helps handle that much power in a small box.

Besides chargers, GaN works in these places.

  • Blue and white LEDs: GaN is the material of blue LEDs. Their inventors, Isamu Akasaki, Hiroshi Amano and Shuji Nakamura, received the 2014 Nobel Prize in Physics ("How Do LEDs Work?")
  • Cell phone base stations: GaN is used in amplifiers that boost radio waves before they are sent. In June 2024, Mitsubishi Electric began shipping samples of a GaN amplifier part for 5G base stations
  • Data centers: GaN is used in the power supplies that feed servers. In May 2024, Infineon announced plans for large power supplies for AI that combine silicon, SiC and GaN
  • Electric cars: GaN is still under development for the "on-board charger," which charges the battery from a home outlet (as of 2026)

What kind of material is GaN?

GaN is a crystal of gallium (Ga) and nitrogen (N) atoms in a regular pattern. A semiconductor is a material that can easily be switched between passing and blocking electricity ("What Is a Semiconductor?").

One key number sets a material's character. It is the height of the "step" an electron must climb before it can move freely. This step is called the band gap. It is measured in eV (electron volts). Silicon's is 1.12 eV, and GaN's is about 3.4 eV. So GaN's step is about three times as high as silicon's.

Materials with a larger step than silicon are called wide-bandgap semiconductors. SiC (silicon carbide) belongs to this group too. This article mainly looks at GaN as a part that switches electricity.

How it works ①: It withstands a strong electric push, so the chip can be small

A voltage creates an "electric push" on the electrons inside a material. If the push is too strong, fast electrons hit atoms and knock out new electrons. If this grows like an avalanche, the current can no longer be stopped. In a material with a large step, knocking out an electron takes more energy. So this happens less easily (this is a simplified picture).

The strength of the push is the voltage divided by the thickness of the layer that holds it. The limit is about 300,000 volts per centimeter of thickness in silicon. In GaN it is about 3.3 million volts per centimeter (Toshiba's table of material values). That is about 10 times silicon. With 10 times the strength, the same voltage fits in about one-tenth the thickness. In GaN transistors, current flows sideways (see the next section). There, this "thickness" is the distance between two electrodes.

A thinner layer also means a shorter path for the current. As a result, the resistance left when the switch is on can be smaller. This is the on-resistance. If the same resistance is good enough, the chip area can be smaller instead.

A small chip also helps it switch fast. Each time a transistor turns on or off, a little charge must flow into or out of it. A smaller chip needs less of this charge. That is why GaN transistors can switch quickly.

How it works ②: An "electron path" at a boundary

GaN's other strength is how it makes a path for electrons. A thin layer containing aluminum (AlGaN) is placed on top of the GaN. Then electrons gather in a thin sheet just below the boundary between the two layers. This sheet of electrons is called a two-dimensional electron gas (2DEG).

The way atoms are arranged in GaN and AlGaN crystals leaves a built-in imbalance of plus and minus charge. The size of this imbalance differs between the two layers. So a plus charge is left at the boundary, and it pulls electrons toward it (this is a very simplified picture). The number of electrons is a few trillion to 20 trillion per square centimeter (Ambacher and colleagues, 1999).

To add electrons to silicon, a small amount of another element is usually mixed in. These added atoms also get in the way of moving electrons. A 2DEG forms without such additives, so electrons have fewer things to hit. How easily electrons move is called mobility. In a 2DEG, it is roughly 1,500 to 2,000 cm²/(V·s) at room temperature. Even with many electrons packed in, they move about as easily as electrons in lightly doped silicon (about 1,400).

Cross section of a GaN transistor (HEMT). GaN and AlGaN layers sit on a silicon substrate, and a sheet of electrons forms just below their boundary Source Gate Drain AlGaN (thin layer) Electron path (2DEG) GaN Buffer layers Silicon substrate Electrons flow sideways along the surface Gate voltage clears the sheet below it
A sketch of a GaN transistor (HEMT) in cross section. The arrow shows the direction electrons travel (current flows the opposite way). The buffer layers ease the mismatch between silicon and GaN (explained later). Not to scale: the AlGaN is tens of nm thick, and the electron sheet is a few nm. Left out: the protective surface film, the detailed shape of the electrodes, and the many sublayers of the buffer.

A transistor that uses this path is called a HEMT (high-electron-mobility transistor). Of the electrodes on top, the middle one is the gate. A voltage on the gate removes or restores the electron sheet just below it. Current flows sideways along the chip surface, not up and down as in many power semiconductors.

The HEMT was invented by Takashi Mimura of Fujitsu Laboratories. He came up with the idea in 1979, using a material called gallium arsenide (GaAs), not GaN. HEMTs spread into products such as satellite TV receivers. In December 2019, the IEEE (a professional society for electrical and electronics engineering) presented a Milestone plaque for it. For GaN HEMTs, the Sumitomo Electric group began volume production for base stations in the mid-2000s.

How it works ③: Why faster switching makes a charger smaller

A charger first turns the AC from the outlet into DC. It then chops that DC on and off in small pieces and sends it through a transformer. A transformer is a part that raises or lowers voltage with two coils ("AC vs DC Explained"). Finally, a capacitor stores charge and releases it bit by bit, giving smooth DC ("What Is a Capacitor?").

During each on-off cycle, the transformer and capacitors hold the electrical energy for a moment. Suppose the amount of energy sent each second stays the same. Then switching 10 times as often means each cycle holds only one-tenth as much. The less each cycle holds, the smaller the parts can be.

Switching more often means less energy is held each time, so the transformer can be smaller Slow switching Transformer Fast switching Transformer Time → Less energy per cycle, so it is smaller
A sketch of switching on 2 times (top) and 8 times (bottom) in the same time. The box sizes only show the idea; they are not the real size ratio. Left out: capacitors, the rectifier circuit, the control chip, and the waveform during each transition.

This is where GaN comes in. Each time a transistor switches, part of the electricity is lost as heat. This is called switching loss, and the total grows with the number of switching events. If a silicon transistor switches too often, the heat builds up and efficiency drops.

A GaN transistor loses less each time it switches, so more frequent switching adds less heat. As a rough guide, many silicon designs switch tens of thousands to about 100,000 times a second (tens of kilohertz to 100 kHz). Some GaN designs run at hundreds of thousands to a million times a second (up to 1 megahertz).

Lower loss means less heat. Then the metal plate that carries heat away (the heat sink) can be smaller too. A charger gets smaller thanks to both fast switching and low heat.

GaN vs SiC vs silicon: which one where

Here is a rough comparison of the three main materials for power semiconductors.

MaterialBest voltage rangeMain uses
SiliconFrom low voltages up to several thousand voltsLow cost. Most power supplies, home appliances, trains, electric cars
SiCFrom the 600-volt class to several thousand voltsElectric car motor drives, railways, solar power
GaNMainly up to a few hundred voltsSmall chargers, server power supplies, base station amplifiers

SiC carries heat away well, so it suits high voltages and high power. GaN suits jobs up to a few hundred volts where fast switching pays off. Silicon can be made cheaply in huge volumes. Where speed matters less, it is still the most economical choice. For more detail, see the tables in "SiC vs GaN" and "Si vs SiC."

Base stations use the same HEMT. There, it does not switch power on and off. It works as an amplifier that makes the radio signal stronger. It withstands high voltage and moves many electrons fast, so a small chip can put out strong radio power. Mitsubishi Electric's 2024 part works with radio waves that oscillate 3.3 to 3.8 billion times a second (3.3–3.8 gigahertz). It puts out 16 watts on average ("What Is 5G?").

In electric cars, the inverter that drives the motor (a device that turns DC into AC) mainly uses SiC or silicon. GaN is just starting to be researched and adopted, for example in the on-board charger that charges the battery at home. The European research project "HiPower 5.0" aims to fit a 22-kilowatt on-board charger into 4 liters using GaN. That is one-third of the market average of 12 liters (Fraunhofer IZM, April 2026).

What is still hard about GaN

Large GaN wafers are hard to make

For silicon, a large crystal is pulled from a melt, then sliced into round plates 300 mm across. These plates are called wafers. GaN, heated at normal pressure, loses its nitrogen before it melts. So large crystals cannot be made the same way. In most cases, a thin film of GaN is grown on a wafer of another material, such as silicon.

Growing a crystal on a different material is called heteroepitaxy. Silicon and GaN have different spacing between their atoms. This difference is called lattice mismatch. They also shrink by different amounts as they cool. Grown directly, the crystal gets defects, and the wafer may crack or warp. To ease this, many "buffer layers" are placed in between.

Silicon wafers are cheap and large, and they can use the equipment of silicon factories. In September 2024, Infineon announced that it had developed a way to grow GaN on 300 mm silicon wafers. It says each wafer yields 2.3 times as many chips as a 200 mm wafer. LEDs, by contrast, mainly use sapphire wafers.

Left alone, it stays "on"

The 2DEG is there from the start, even with no voltage on the gate. So a basic HEMT lets current flow when nothing is done to it. This is called "normally-on." In a power circuit, that is dangerous. If the control circuit stops, the switch may stay on, and a large current may keep flowing.

So designers use tricks to make it normally-off, meaning it is off when no voltage is applied. A common method puts a layer of "p-type" GaN under the gate. A p-type material has many holes (gaps left where electrons are missing). This layer removes the electron sheet below it in advance. With this method, the allowed range of gate voltage is narrow. So the circuit that drives the gate, the gate driver, needs care. Another method connects a low-voltage silicon transistor in series, so the pair as a whole is normally-off.

Long-term reliability, and price

GaN power semiconductors have a short history. Ways to test how they behave over long use are still being set up. For example, after high voltage is applied, electrons can get trapped in crystal defects. This can raise the resistance for a while. JEDEC, a body that writes semiconductor standards, formed a committee (JC-70) in 2017. Its job is to set test methods and other standards for wide-bandgap semiconductors.

Price is another issue. Part for part, a silicon transistor is usually still cheaper. But the transformer and heat sink get smaller, so the whole product can come out ahead. The industry is moving fast, with a run of acquisitions. Infineon bought GaN Systems in October 2023, and Renesas bought Transphorm in June 2024. Meanwhile, the major contract manufacturer TSMC was reported to be ending its GaN foundry production by the end of July 2027 (July 2025; "What Is a Semiconductor Foundry?").

Common myths

Myth: Every GaN charger is safe and never gets hot

In fact, GaN reduces losses but does not remove them. Take a charger that sends out 95% of the power it takes in (95% efficiency). To send out 100 watts, it takes in about 105 watts, and about 5 watts becomes heat (our own calculation). A small charger also has little surface area to shed heat, so it is common for it to feel warm. Safety depends on design and testing, not on the material. In Japan, for example, AC adapters are "specified electrical appliances" under the Electrical Appliances and Materials Safety Act. They must carry the diamond-shaped PSE mark (Ministry of Economy, Trade and Industry). The myth spreads easily because ads stress "less heat."

Myth: GaN is better than silicon, so CPUs will be made of GaN someday

In fact, computing chips like CPUs are built with CMOS, which pairs two kinds of transistors with different properties. One kind carries current with electrons, the other with holes. In GaN, holes move very poorly: their mobility is tens of times lower than that of electrons, or worse. Making good p-type layers is also hard, so a good pair is still at the research stage (this is simplified). Also, what matters in computing chips is packing billions of transistors that run at low voltage. GaN's strength of withstanding high voltage does not help there.

Myth: A "GaN charger" is made entirely of GaN

In fact, GaN is mainly in the transistors that switch the power, a small share of the parts. The control chip, transformer, capacitors and rectifier diodes are parts made of other materials.

By the numbers

ItemValueNote
Step height (band gap)Si 1.12 eV / GaN about 3.4 eVGaN is about 3 times silicon
Electric push it can withstandSi about 0.3 MV/cm / GaN about 3.3 MV/cmAbout 10 times. A thin layer holds the voltage
Electrons in the 2DEGA few trillion to 20 trillion per cm²Gather without additives
Electron mobility in the 2DEGAbout 1,500–2,000 cm²/(V·s)Electrons in silicon: about 1,400
GaN amplifier for base stations16 W average3.3–3.8 GHz, Mitsubishi Electric (2024)

MV/cm means "how many million volts per centimeter of thickness." The mobility unit cm²/(V·s) means "how many centimeters per second an electron moves under a push of 1 volt per centimeter."

Learn more

Read next

Articles in the series “Chips in everyday life”

  1. Memory vs Storage: What's the Difference? (RAM vs SSD)
  2. What Chips Are Inside a Smartphone? The Main Chips Explained
  3. What Is a Power Semiconductor? SiC, GaN and IGBTs Explained
  4. How Do Image Sensors Work? Inside a Phone Camera Sensor
  5. What Is a Microcontroller? Tiny Computers in Everyday Things
  6. How Do Solar Cells Work? Light to Electricity and Its Limits
  7. How Do LEDs Work? Turning Electricity Into Light in a Chip
  8. How Do Touchscreens Work? Capacitive Touch Explained
  9. How Do SSDs Work? Inside NAND Flash Memory, Explained
  10. What Is an Analog Chip? Analog vs Digital ICs Explained
  11. How Does GPS Work? Satellites, Timing and Your Phone
  12. LCD vs OLED: How Phone and TV Screens Make Color
  13. How DRAM Works: Cells, Refresh and Sense Amplifiers
  14. What Is MEMS? How Accelerometers and Gyroscopes Work
  15. How Lithium-Ion Batteries Work: Charging, Aging and Safety
  16. What Is a Laser Diode? How It Differs From an LED
  17. What Is GaN (Gallium Nitride)? Why GaN Chargers Are Small
  18. What Is Silicon Carbide (SiC)? Why EV Inverters Use It
  19. How Wireless Charging Works: Induction, Qi and Qi2
  20. What Is an Inverter? DC to AC for Air Conditioners and EVs
  21. What Is USB? USB-C vs USB 3, USB4 and Power Delivery
  22. How Noise Cancelling Works: Using Sound to Erase Sound
  23. How Are Chips Cooled? Heat Sinks, Heat Pipes and Limits
  24. How AC Adapters and Chargers Work: From Wall Outlet to 5 V
  25. What Chips Are Inside an Electric Car? EV Chips Explained
  26. How Do Smartwatches Measure Heart Rate? Reading Blood Flow
  27. How Power Banks Work: How Many Charges From 10,000 mAh?
  28. What Is NFC? How Contactless Cards and Tap to Pay Work

Questions on this topic: Compound semiconductors

Sources

All explainers