Power MOSFETs & IGBTs

The two workhorse power switches. Learn power MOSFET structures (VDMOS, trench, shielded-gate, superjunction), R_DS(on) components and temperature coefficient, body diode, gate charge, avalanche and SOA, paralleling; then the IGBT: MOS-gated bipolar structure (p⁺ collector injection), conductivity modulation, V_CE(sat), punch-through and field-stop designs, tail current, latch-up of the parasitic thyristor, short-circuit withstand, co-packaged diodes, and the MOSFET-vs-IGBT decision — with Python comparing conduction loss vs current, the loss crossover point and IGBT tail loss vs frequency.

Power MOSFETs are fast, easy to drive and resistive when on — ideal at lower voltages and higher frequencies. IGBTs add a bipolar injector to a MOSFET, giving a low, nearly constant on-state voltage at high voltages and currents — at the cost of slower turn-off.

Choosing between them (or SiC/GaN) is one of the most common power-electronics decisions.

An analogy

A MOSFET is a pipe whose resistance is fixed once open (losses grow with current squared); an IGBT is like a pipe with a small fixed toll (knee voltage) that then flows very freely — expensive for small flows, cheap for large ones, but slow to close because water keeps trickling (tail current).

Power MOSFET structures

Evolution:

MOSFET behaviour

R_DS(on) has a positive temperature coefficient (mobility falls) → natural current sharing when paralleled (with matched gate drive and layout). Avalanche ruggedness allows unclamped inductive energy up to E_AS. Linear-mode SOA must be checked for hot-swap and e-fuse applications. Q_g and Q_oss set switching and gate losses; C_oss nonlinearity (especially in superjunctions) affects ZVS design.

IGBT behaviour

Punch-through and field-stop designs use thinner drift layers with an n buffer to stop the field, reducing V_CE(sat) and tail. Trade-off curve: lower V_CE(sat) ⇔ more stored charge ⇔ higher E_off (tuned by collector injection and lifetime). The parasitic thyristor must never latch (short n⁺ emitters, heavily doped p body). IGBTs typically withstand short circuits for 5–10 µs, giving time for desaturation protection. V_CE(sat) may have negative TC at low currents (bipolar) and positive at high currents.

Choosing

MOSFET: < ~600–900 V, high frequency, light-load efficiency, synchronous rectification. IGBT: ≥ 600 V, high current, 2–30 kHz, short-circuit robustness, cost. SiC MOSFET: 650–1700 V where both efficiency and frequency matter. Evaluate at the real load profile (average and RMS current, temperature), not just at rated current.

Formulas

Worked example: IGBT loss

Given: V_CE0 1.0 V, r 20 mΩ, 40 A DC

Find: P.

Answer: 72 W

Worked example: MOSFET paralleling

Given: 2 × 30 mΩ, 40 A total

Find: Loss.

Answer: 24 W

Worked example: Crossover

Given: V_CE0 0.8 V, r_CE 30 mΩ, R_DS(on) 70 mΩ

Find: I_x.

Answer: 20 A

Trench MOSFET vs field-stop IGBT

  Trench MOSFET                         Field-stop trench IGBT
   S ─ n⁺ ─ p body ─┐gate│                E ─ n⁺ ─ p body ─┐gate│
        channel ↓    │trench              electrons ↓ (MOS channel)
   n⁻ drift (electrons only)             n⁻ drift  ◄═ holes injected from below (conductivity modulation)
   n⁺ substrate ── D                      n field-stop │ p⁺ collector ── C
   body diode: p body → n drift           parasitic n⁺pn⁻p⁺ thyristor (latch-up must be suppressed)

The IGBT is a MOSFET whose n⁺ drain is replaced by a p⁺ collector. When on, the MOS channel supplies electrons that act as base current of a wide-base PNP; holes injected from the collector flood the drift region. At turn-off, those stored carriers must recombine — the tail current. IGBTs have no body diode, so a co-packed diode is used.

Conduction loss crossover and tail loss

# Conduction loss: MOSFET (resistive) vs IGBT (knee voltage + slope), and IGBT tail-current loss
Rds = 0.060                       # 650 V MOSFET at Tj 100 C (ohm)
Vce0, rce = 0.9, 0.025            # 650 V IGBT: knee + slope
for I in (5, 15, 30, 50):
    print(f"I = {I:2d} A: MOSFET {I*I*Rds:6.1f} W, IGBT {I*(Vce0+rce*I):6.1f} W")
Ix = Vce0 / (Rds - rce)
print(f"crossover current where losses are equal: {Ix:.1f} A (MOSFET better below)")
Vbus, Itail, ttail = 400, 5.0, 300e-9
for f in (10e3, 50e3):
    print(f"tail-current loss (triangular, 5 A for 300 ns) at {f/1e3:.0f} kHz: {0.5*Vbus*Itail*ttail*f:.1f} W")
# I =  5 A: MOSFET    1.5 W, IGBT    5.1 W
# I = 15 A: MOSFET   13.5 W, IGBT   19.1 W
# I = 30 A: MOSFET   54.0 W, IGBT   49.5 W
# I = 50 A: MOSFET  150.0 W, IGBT  107.5 W
# crossover current where losses are equal: 25.7 A (MOSFET better below)
# tail-current loss (triangular, 5 A for 300 ns) at 10 kHz: 3.0 W
# tail-current loss (triangular, 5 A for 300 ns) at 50 kHz: 15.0 W

A 60 mΩ (hot) 650 V MOSFET beats a comparable IGBT below ~26 A because its loss grows with I² while the IGBT's grows almost linearly after the 0.9 V knee. At light load (typical of many converters) the MOSFET wins by far. The IGBT's tail loss scales with frequency: 3 W at 10 kHz but 15 W at 50 kHz — why IGBTs stay below ~20–50 kHz.

Common mistakes

Applications

Key takeaways

Going further