SiC vs silicon: what is the difference?
Silicon is the default material of the semiconductor industry and is still used for most power transistors, from low-voltage MOSFETs to high-voltage IGBTs. SiC has spread quickly in electric vehicles and renewable-energy inverters because its physical properties suit high-voltage switching. Comparing the two explains both why SiC is gaining ground and why it is not replacing silicon wholesale.
The short answer
Silicon carbide (SiC) withstands roughly ten times the electric field of silicon thanks to a bandgap about three times wider, so a high-voltage device can use a much thinner, more heavily doped drift layer and lose far less power. SiC also conducts heat about three times better. Silicon, however, is far cheaper, available as large, nearly defect-free 300 mm wafers and supported by a mature process base, so it remains the material for logic, memory and most power devices below about 600 V.
Side-by-side comparison
| Aspect | SiC (4H-SiC) | Silicon |
|---|---|---|
| Bandgap | About 3.26 eV | About 1.12 eV |
| Critical electric field | About 2.5 to 3 MV/cm | About 0.3 MV/cm |
| Thermal conductivity | Roughly 3.7 to 4.9 W/(cm·K), depending on doping and direction | About 1.5 W/(cm·K) |
| Bulk electron mobility | About 900 to 1,000 cm²/(V·s) | About 1,400 cm²/(V·s) |
| Crystal growth | Sublimation (physical vapour transport) above 2,000 °C; slow, with dislocations to control | Czochralski pulling from a melt; fast, large and nearly dislocation-free |
| Mainstream wafer size | 150 mm, with 200 mm entering production | 300 mm for logic and memory; 200 mm common for power devices |
| Doping | Dopants barely diffuse; hot ion implantation and activation anneals above 1,600 °C are needed | Ion implantation with anneals around 1,000 °C; thermal diffusion also usable |
| Machining | Very hard; slicing, grinding and polishing are slow | Mature, high-throughput wafering |
| Main power devices | MOSFETs and Schottky barrier diodes, mainly 650 V to 3.3 kV | Low-voltage and superjunction MOSFETs, IGBTs up to several kV, diodes and thyristors |
Why a higher breakdown field means lower loss
When a power device is off, its voltage is held across a lightly doped region called the drift layer. The maximum field a material can stand sets how thick that layer must be and how lightly it must be doped. With a breakdown field about ten times that of silicon, SiC can hold the same voltage in a layer roughly one tenth as thick and much more heavily doped.
Both changes cut resistance, so the drift-layer resistance of an ideal SiC device is two to three orders of magnitude lower than that of a silicon device of the same voltage rating. In practice this lets SiC use a unipolar MOSFET at 1200 V and above, where silicon must switch to the bipolar IGBT to keep conduction loss acceptable.
Switching and temperature
A silicon IGBT lowers its on-state voltage by flooding the drift region with both electrons and holes, but those carriers must be removed at turn-off, producing a tail current and switching loss. A SiC MOSFET carries current with electrons alone, so it switches off cleanly and can run at higher frequencies, allowing smaller inductors, capacitors and filters.
The wide bandgap also keeps leakage current low at high temperature, and the high thermal conductivity spreads heat out of the chip. Together these allow higher power density and simpler cooling. SiC MOSFETs have their own weakness at the gate: the SiC-to-oxide interface traps charge and lowers channel mobility, which designers compensate for with trench gates and careful gate-oxide processing.
Why SiC wafers cost more
Silicon crystals are pulled from a melt by the Czochralski method, producing 300 mm ingots with almost no dislocations. SiC does not melt at normal pressure; it sublimes. Crystals are therefore grown by sublimating SiC source powder at over 2,000 °C and recondensing it on a seed crystal, a modified Lely process that is orders of magnitude slower. Defects such as basal plane dislocations must be controlled because they can extend into the epitaxial layer and cause bipolar degradation during operation.
SiC is also extremely hard, so slicing ingots and polishing wafers takes much longer. The result is that a SiC wafer costs far more per area than a silicon wafer, and that cost must be paid back through lower losses and smaller systems.
Choosing between them
For logic, memory and most power conversion below about 600 V, silicon is cheaper and good enough. SiC pays off where high voltage, high efficiency and compact size matter together: EV traction inverters and fast chargers, solar and storage inverters, industrial drives and railway traction. In Japan, ROHM, Mitsubishi Electric, Toshiba, Fuji Electric and Denso make SiC devices, while SUMCO and Shin-Etsu Handotai are major suppliers of the silicon wafers on which most other chips are built.
FAQ
Will SiC replace silicon?
Only in part of power electronics. Logic and memory remain silicon, and low- and medium-voltage power devices stay mostly silicon on cost. SiC is replacing silicon IGBTs and diodes where high voltage and low loss justify a higher device price.
What is 4H-SiC?
SiC exists in many polytypes that differ in how the atomic layers stack. 4H-SiC repeats every four layers and has the highest and least direction-dependent electron mobility among the common polytypes, which is why it is used for almost all power devices.
Is SiC a semiconductor or a ceramic?
Both. Polycrystalline SiC is a hard industrial ceramic and abrasive, also used for parts inside semiconductor equipment. Single-crystal SiC grown with controlled doping is a semiconductor used for power and RF devices.
Terms compared
- Silicon Carbide (SiC)A wide-bandgap compound of silicon and carbon whose high breakdown field and thermal conductivity make it the leading material for high-voltage power devices.
- Silicon (Si)Element 14, the base material of most semiconductor devices, refined from quartz into eleven-nines polysilicon and single-crystal wafers.
- SiC Wafer (Silicon Carbide Substrate)A single-crystal silicon carbide substrate for power devices, grown by sublimation rather than from a melt, which makes it slow and costly to produce.
- Silicon WaferA thin, mirror-polished disc of single-crystal silicon on which chips are built; 300 mm is the standard size for leading-edge production.
Japanese companies involved
- ROHM · SiC and silicon power devices, analog ICs
- Resonac · CMP slurries, packaging materials and SiC epitaxial wafers
- Mitsubishi Electric (power devices) · IGBT and SiC power modules, RF and optical devices
- Sumitomo Electric Industries (compound semiconductors) · SiC epitaxial wafers and GaAs and InP substrates
- Shin-Etsu Chemical · Silicon wafers, photoresists and photomask blanks
- SUMCO · Silicon wafers
- Tokuyama · Polysilicon and electronic chemicals
- DISCO · Dicing, grinding and polishing equipment
More power semiconductors comparisons
Sources
Semiconductor comparisons
- All semiconductor comparisons
- Gallium oxide vs SiC: what is the difference?
- SiC MOSFET vs IGBT: what is the difference?
- CoWoS vs SoIC: what is the difference?
- CoWoS vs InFO: what is the difference?
- Chiplet vs monolithic: what is the difference?
- Hybrid bonding vs flip chip: what is the difference?
- TSV vs wire bond: what is the difference?
Last updated: November 2, 2026