Elemental semiconductors are made from a single chemical element like silicon, while compound semiconductors are synthesized from two or more elements, like gallium nitride, to achieve specific electronic properties. This fundamental material choice dictates a component's bandgap energy, which directly changes maximum operating temperature, breakdown voltage, and switching speed in your power circuits. A common pitfall for hobbyists and junior engineers is confusing the doping of a semiconductor (adding trace impurities like boron to silicon to create P-N junctions) with creating a compound semiconductor (alloying entirely different base elements like gallium and nitrogen to form a new crystal lattice).
The Physics: Bandgap and Electron Mobility
To understand why a circuit designer chooses one material over another, you have to look at the bandgap—the energy required (measured in electron-volts, eV) to excite an electron from the valence band to the conduction band. Silicon has a relatively narrow bandgap of 1.12 eV. At high temperatures, ambient thermal energy is enough to push electrons across this gap, causing severe leakage currents and eventual thermal runaway.
Compound semiconductors like Silicon Carbide (SiC) and Gallium Nitride (GaN) are "wide-bandgap" (WBG) materials. Because their bandgaps are roughly three times larger than silicon, they remain stable at much higher temperatures and can sustain vastly higher electric fields before breaking down. This allows manufacturers to build thinner drift regions, which drastically reduces the on-resistance ($R_{DS(on)}$) for a given voltage rating.
| Material | Type | Bandgap (eV) | Electron Mobility (cm²/V·s) | Critical Breakdown Field (MV/cm) |
|---|---|---|---|---|
| Silicon (Si) | Elemental | 1.12 | 1,400 | 0.3 |
| Germanium (Ge) | Elemental | 0.67 | 3,900 | 0.1 |
| Gallium Arsenide (GaAs) | Compound | 1.42 | 8,500 | 0.4 |
| Silicon Carbide (SiC) | Compound | 3.26 | 900 | 2.5 |
| Gallium Nitride (GaN) | Compound | 3.40 | 1,500 (2DEG) | 3.3 |
Notice the electron mobility of GaN. While bulk GaN mobility is moderate, GaN High Electron Mobility Transistors (HEMTs) utilize a two-dimensional electron gas (2DEG) at the AlGaN/GaN heterojunction, yielding exceptionally fast switching characteristics that silicon cannot match. For a deeper look at how these heterojunctions are manufactured, refer to EPC's technical guide on GaN fundamentals.
Worked Example: Switching Loss in a 400V DC-DC Converter
Theoretical physics is useful, but bench engineers care about thermal management and magnetics sizing. Let's calculate the hard-switching turn-on and turn-off losses for a 400V DC bus, 10A load, operating at a 200 kHz switching frequency ($f_{sw}$).
The formula for switching power loss is:
P_sw = 0.5 × V_DS × I_D × (t_r + t_f) × f_sw
Using a standard Si device (e.g., Infineon IPW60R099CP), the combined rise and fall times ($t_r + t_f$) under hard switching are approximately 40 ns.
P_sw = 0.5 × 400V × 10A × 40ns × 200,000 Hz = 16.0 WResult: 16W of pure switching loss per device. You will need a substantial heatsink and forced air cooling.
Using a GaN device (e.g., EPC2016C), the absence of reverse recovery charge and lower gate charge drops the combined transition time to roughly 5 ns.
P_sw = 0.5 × 400V × 10A × 5ns × 200,000 Hz = 2.0 WResult: 2.0 W of switching loss. This 8x reduction allows you to use a much smaller heatsink, or push the switching frequency to 1 MHz to drastically shrink your inductors and capacitors.
This math illustrates exactly what the material changes in a real installation: compound semiconductors shift the bottleneck from the silicon die's thermal limits to the PCB's parasitic inductance and the magnetic core losses of your passive components.
Where You Meet This in Practice
You are already using both elemental and compound semiconductors daily, often without realizing the material science driving the form factor.
- Elemental (Silicon): The backbone of digital logic and low-frequency power. Your Arduino's ATmega328P, standard 74HC logic gates, and the 60Hz/100kHz MOSFETs in basic LED drivers are all silicon. It is cheap, highly reliable, and benefits from decades of manufacturing perfection.
- Compound (Silicon Carbide - SiC): Dominates high-voltage, high-power applications. If you are building or repairing an EV traction inverter handling an 800V battery bus, or a 50kW commercial solar string inverter, you will find SiC MOSFETs (like the Wolfspeed C3M0060120D). SiC's high thermal conductivity allows it to survive the brutal heat of heavy industrial loads. Read more about SiC's role in modern power grids via Wolfspeed's SiC technology overview.
- Compound (Gallium Nitride - GaN): The king of high-frequency, medium-voltage consumer and RF electronics. Those ultra-compact 100W USB-C PD wall chargers that barely get warm? They rely on GaN switching at 500kHz to 1MHz, allowing the internal transformer to shrink to the size of a coin. GaN is also the standard for 5G cellular base station RF power amplifiers and automotive LiDAR pulser circuits.
Frequently Asked Questions
Why don't we just use compound semiconductors for everything?
Cost, wafer size, and defect density. Silicon wafers are grown at massive scales (300mm diameter) with near-zero crystal defects, driving the cost per die down to fractions of a cent. Compound semiconductors like SiC and GaN are notoriously difficult to grow; SiC sublimation requires temperatures over 2,000°C, and GaN is typically grown heteroepitaxially on silicon or sapphire substrates, introducing lattice mismatch defects. A GaN power IC might cost $2.50, while an equivalently rated Si MOSFET costs $0.40. You only pay the compound premium when the system-level savings (smaller magnetics, eliminated cooling) justify the BOM cost.
Is silicon carbide (SiC) considered a compound semiconductor?
Yes. Even though both silicon and carbon are Group IV elements (making it a IV-IV compound), SiC is definitively a compound semiconductor because its crystal lattice consists of two distinct elements bonded together. This alternating Si-C bond is what gives the material its extreme mechanical hardness and wide 3.26 eV bandgap.
How does the choice of semiconductor affect my PCB layout?
Drastically. Because GaN and SiC switch so fast, the rate of current change ($di/dt$) can easily exceed 100 A/ns. According to $V = L(di/dt)$, even 1 nanohenry (nH) of parasitic trace inductance will generate 100V of ringing across your drain-source terminals. When laying out a PCB for compound semiconductors, you must minimize the high-frequency power loop area, use multiple vias for ground returns, and place high-frequency ceramic decoupling capacitors within millimeters of the device pins. Standard TO-220 through-hole layouts used for silicon will cause catastrophic voltage overshoot and device failure with GaN.
Can I replace a silicon MOSFET with a GaN FET in an existing circuit?
Almost never as a direct drop-in replacement. The gate drive requirements are fundamentally different. A standard Si MOSFET requires 10V to 15V on the gate to fully enhance the channel. A GaN HEMT typically requires only 5V to 6V, and its absolute maximum gate voltage is often around 7V. If your existing silicon gate driver outputs 12V, it will instantly punch through the gate oxide of a GaN device and destroy it. Furthermore, SiC MOSFETs often require a negative turn-off bias (e.g., -3V to 0V) to prevent false Miller turn-on, requiring a completely different isolated gate driver topology.
