Electrical Engineering › Section P · Semiconductor theory, diodes, and transistors › Topic P.13
Insulated-gate FETs (MOSFETs) and IGBTs
Insulate the gate of a field-effect transistor and it draws no steady current at all. That one change made the MOSFET the building block of almost every chip and most power switches, and pairing it with a bipolar output gave the IGBT, the workhorse of high-voltage motor drives.
An insulated gate
A JFET's gate meets the channel at a PN junction that must stay reverse biased. An insulated-gate FET (IGFET) separates gate and channel with a thin insulating layer, so the gate is one plate of a tiny capacitor. Its field still controls the channel, but no steady current flows into the gate, and either gate polarity is safe. The usual build is a metal or polysilicon gate on silicon dioxide over silicon, hence the common name MOSFET, for metal-oxide-semiconductor FET. Like JFETs they are unipolar and voltage-controlled, with gate, source and drain.
Four terminals and a body diode
A MOSFET sits on a body, or substrate, doped opposite to its channel: a fourth terminal. Discrete MOSFETs tie the body to the source inside, leaving three leads; the symbol's body arrow points in for N-channel, out for P-channel. That link leaves a PN junction between body and drain: in an N-channel power MOSFET, the body diode, anode at the source and cathode at the drain. It conducts if the drain is pulled below the source, useful in motor bridges, but it means a MOSFET blocks only one polarity. P-channel parts mirror every polarity.
Depletion-type MOSFETs
A depletion-type MOSFET is made with its channel already in place, so it conducts with \(V_{GS} = 0\), like a JFET. For an N-channel part, a negative gate repels electrons and depletes the channel, down to cutoff at \(V_{GS(off)}\); a positive gate attracts more and enhances it, which the insulated gate makes safe. Shockley's equation from "Junction field-effect transistors: switching and active mode" holds on both sides of zero: with \(I_{DSS} = 4.0\,\text{mA}\) and \(V_{GS(off)} = -2.0\,\text{V}\), a device passes 1.0 mA at −1 V, 4.0 mA at 0 V and 9.0 mA at +1 V. Depletion parts are now uncommon.
Enhancement-type MOSFETs
An enhancement-type MOSFET has no channel at \(V_{GS} = 0\): it is normally off, like a BJT, which is why it dominates. Raise an N-channel gate above the threshold \(V_{GS(th)}\) and the field gathers enough electrons under the oxide to form an N-type channel, an inversion layer. In the current-regulating saturation region the current follows a square law,
where \(K\) depends on the device: with \(K = 1.0\,\text{mA/V}^2\) and a 2.0 V threshold, 4 V on the gate gives 4.0 mA. Data sheets quote \(V_{GS(th)}\) at a tiny drain current, often 250 µA, so the threshold is where conduction begins, not where the MOSFET is fully on. Logic-level parts have thresholds of about 1 to 2 V.
Fully on: R_DS(on)
A MOSFET used as a switch is driven far above threshold, deep into the ohmic region, where the channel is simply a small resistance, \(R_{DS(on)}\). Data sheets give it at stated gate voltages such as 10 V, 4.5 V or 2.5 V; a part may not be fully on at lower drive. The conduction loss is
Unlike a BJT there is no fixed saturation voltage: the drop is \(I_D R_{DS(on)}\), and power MOSFETs reach milliohms. \(R_{DS(on)}\) rises with temperature, which helps paralleled MOSFETs share, and rises steeply with voltage rating, so high-voltage silicon MOSFETs conduct less well.
Worked example: switching a 2 A load from 3.3 V
A 3.3 V microcontroller pin must switch a 12 V, 2.0 A LED strip. A logic-level N-channel MOSFET is rated \(R_{DS(on)} = 50\,\text{m}\Omega\) at \(V_{GS} = 2.5\,\text{V}\), an assumed figure. 1. Low-side switch: strip from 12 V to drain, source to ground, pin to gate. 2. Drop: \(2.0 \times 0.050 = 0.10\,\text{V}\). 3. Loss: \(2.0^2 \times 0.050 = 0.20\,\text{W}\), needing no large heatsink. 4. Once charged, the gate draws nothing; a 100 Ω series resistor limits the charging pulse to 33 mA, and 100 kΩ from gate to source holds it off if the pin floats. 5. Contrast: a BJT with β = 50 would need 40 mA of base current, too much for the pin, and would lose about \(0.3 \times 2.0 = 0.6\,\text{W}\).
The gate is a capacitor
A MOSFET draws gate current only while its gate capacitance charges or discharges. That is negligible in a slow switch, but in a power converter switching at tens of kilohertz the gate charge must move in nanoseconds, so gate-driver chips supply amps briefly. The oxide is very thin: a gate-source voltage beyond the rated maximum, often ±20 V for power parts, punctures it, and so can static from a person, so MOSFETs need anti-static handling. A floating gate can charge to any voltage and turn the device partly on; a pull-down resistor prevents this, like a JFET's bleed resistor.
CMOS: complementary pairs
Complementary MOS (CMOS) logic pairs an N-channel and a P-channel enhancement MOSFET. In an inverter, the P-channel part connects the output to the supply and the N-channel part connects it to ground, with both gates driven by the input. A high input turns the N-channel on and the P-channel off, so the output goes low; a low input does the reverse. In either steady state one transistor is off, so there is no dc path from supply to ground, and current flows almost only while switching. That made CMOS the technology of nearly all processors and memories, and shrinking MOSFETs drove Moore's law, transistor counts doubling roughly every two years. "CMOS circuits" builds on this.
The IGBT
A power BJT with β = 25 carrying 60 A needs 2.4 A of base current; a Darlington of two still needs about 90 mA. High-voltage MOSFETs need no drive current but have high \(R_{DS(on)}\). The insulated-gate bipolar transistor (IGBT) combines them. It behaves like an N-channel MOSFET supplying the base current of a PNP transistor, whose injected carriers make the thick silicon conduct well: an insulated gate with a low on-state drop at high current. Its terminals are gate, collector and emitter, its \(V_{CE(sat)}\) typically about 1.5 to 2.5 V, and it turns off more slowly than a MOSFET, with a tail of current. IGBTs dominate drives and inverters from hundreds of volts to kilovolts, though silicon carbide MOSFETs now compete at 600 to 1200 V.
Worked example: MOSFET or IGBT?
Compare the conduction losses of two 600 V devices, using assumed but typical figures and a simple model: a MOSFET with \(R_{DS(on)} = 0.10\,\Omega\), and an IGBT with a constant \(V_{CE(sat)} = 1.8\,\text{V}\). 1. At 5 A: MOSFET \(5^2 \times 0.10 = 2.5\,\text{W}\); IGBT \(1.8 \times 5 = 9.0\,\text{W}\). 2. At 30 A: MOSFET \(30^2 \times 0.10 = 90\,\text{W}\); IGBT \(1.8 \times 30 = 54\,\text{W}\). 3. Crossover where \(I^2 R = V I\): \(I = 1.8/0.10 = 18\,\text{A}\), where both lose 32.4 W. 4. So the MOSFET wins at light load and the IGBT at heavy load, since one loss grows with \(I^2\) and the other with \(I\). 5. A real choice also weighs switching losses, where the IGBT's tail current counts against it.
Common mistakes to avoid
First, treating \(V_{GS(th)}\) as the voltage for full conduction; use the voltage at which \(R_{DS(on)}\) is specified. Second, assuming an enhancement MOSFET conducts at \(V_{GS} = 0\). Third, leaving a gate floating. Fourth, exceeding the gate voltage rating or ignoring static. Fifth, forgetting the body diode. Sixth, calling an IGBT's terminals drain and source; they are collector and emitter.
Key takeaways
MOSFETs insulate the gate with oxide, so there is no steady gate current. The body is tied to the source, leaving a body diode. Depletion types conduct at \(V_{GS} = 0\); enhancement types are normally off and conduct above \(V_{GS(th)}\), with \(I_D = K(V_{GS} - V_{GS(th)})^2\) in saturation. A switching MOSFET loses \(I^2 R_{DS(on)}\). The gate is a capacitor: drive it hard, never let it float, guard it from static. CMOS pairs N- and P-channel MOSFETs for near-zero static power. The IGBT joins a MOSFET input to a bipolar output for high-voltage, high-current switching.
Flashcards
Flashcard
FET whose gate is insulated from the channel by oxide
Test your understanding
Question 1 of 5
An enhancement-type N-channel MOSFET has V_GS(th) = 2.0 V. What does it do with 1.5 V between gate and source?
Question 2 of 5
A MOSFET switch with R_DS(on) = 25 mΩ carries 4.0 A. What is its conduction loss?
Question 3 of 5
A depletion-type N-channel MOSFET has I_DSS = 6.0 mA and V_GS(off) = −3.0 V. What drain current flows in saturation at V_GS = +1.5 V?
Question 4 of 5
Using a simple model, an IGBT has V_CE(sat) = 2.0 V and a MOSFET has R_DS(on) = 80 mΩ. At 40 A, which has the lower conduction loss?
Question 5 of 5
Why does a CMOS inverter draw almost no current while its input is steady?
