Semiconductor Devices

Chapter 9
Power Semiconductor Devices

Post Graduate Program in Electronics and Communication Engineering
Semiconductor Devices • Post Graduate

9. Power Semiconductor Devices

Introduction and Context

Power semiconductor devices are the controlled valves through which electrical energy is converted, regulated, and delivered in modern power-electronic systems. Unlike small-signal devices, they must withstand substantial voltage and current while dissipating acceptable power during conduction and switching. Their behaviour therefore depends not only on semiconductor physics, but also on thermal design, packaging, gate-drive circuitry, electromagnetic compatibility, protection, and system-level reliability.

The principal device families are power diodes, thyristors, bipolar junction transistors (BJTs), power MOSFETs, insulated-gate bipolar transistors (IGBTs), and wide-bandgap devices based mainly on silicon carbide (SiC) and gallium nitride (GaN). No single device is universally superior. A high-frequency converter may favour a low-charge MOSFET, whereas a utility-controlled rectifier may require the voltage capability and ruggedness of a thyristor. Device selection is consequently a multi-objective engineering decision involving switching loss, conduction loss, blocking voltage, thermal behaviour, safe operating area (SOA), reliability, cost, and application requirements.

Learning Objectives

After studying this chapter, the student should be able to:

1. Explain the operating principles and characteristic parameters of major power semiconductor devices.

2. Distinguish conduction and switching losses and relate them to converter efficiency.

3. Compare the voltage, current, frequency, thermal, SOA, reliability, and cost characteristics of different device families.

4. Select an appropriate device for rectifiers, motor drives, switched-mode power supplies, traction systems, and high-frequency converters.

5. Interpret datasheet ratings and identify common reliability and protection issues.

6. Evaluate the role of SiC and GaN in emerging power-conversion systems.

From Signal Devices to Power Valves

A power device is designed to operate in at least one of three states: forward conduction, reverse or forward blocking, and transition between these states. Ideally, a conducting device has zero voltage drop and a blocking device has zero current. Real devices depart from this ideal, producing conduction loss and switching loss.

For a device carrying current I with on-state voltage V_on, the approximate conduction loss is

P_cond = V_on I_avg

for a device with nearly constant voltage drop. For a resistive device such as a MOSFET, it is more appropriately written as

P_cond = I_rms² R_DS(on).

Switching loss arises because voltage and current overlap during turn-on and turn-off. A first-order estimate is

P_sw ≈ 1/2 V I (t_on + t_off) f_s,

where V is the switched voltage, I is the switched current, t_on and t_off are transition intervals, and f_s is switching frequency. Actual loss also includes gate-drive energy, output-capacitance charging and discharging, reverse-recovery loss, and parasitic-inductance effects.

The basic device ratings must be interpreted together. Breakdown voltage specifies the blocking capability under stated temperature and test conditions. Current ratings may refer to continuous, pulsed, average, or RMS current and are strongly dependent on case temperature and cooling. Junction temperature is limited by the semiconductor and package. A useful thermal relation is

T_j = T_a + P_loss R_θJA,

or, when a heat sink is specified,

T_j = T_c + P_loss R_θJC,

where T_j, T_a, and T_c are junction, ambient, and case temperatures, respectively. Transient thermal impedance is needed for pulsed operation.

Device Families and Their Operating Character

Power Diodes

A power diode is an uncontrolled, two-terminal device that conducts when forward biased and blocks when reverse biased. The p-i-n diode is common at medium and high voltage because its lightly doped intrinsic region supports blocking voltage. Schottky diodes use a metal-semiconductor junction and have low forward voltage and negligible minority-carrier storage, giving very fast switching. Their reverse-voltage capability and leakage performance have traditionally been more limited than those of p-i-n devices, although SiC Schottky diodes provide high voltage and low reverse-recovery charge.

The reverse-recovery current of a p-i-n diode can produce voltage overshoot, electromagnetic interference, and additional switching loss in the companion transistor. Important diode specifications therefore include forward voltage, reverse-recovery time, reverse-recovery charge, repetitive peak reverse voltage, surge-current capability, leakage current, and thermal resistance. Diodes are used in uncontrolled rectifiers, freewheeling paths, clamp networks, boost converters, and commutation circuits.

Thyristors

The silicon-controlled rectifier (SCR) is a four-layer p-n-p-n device with anode, cathode, and gate terminals. A forward-biased SCR remains blocking until a suitable gate pulse is applied or a breakover condition occurs. Once latched, it continues conducting while forward current exceeds the holding current; removal of the gate signal alone does not turn it off. Turn-off requires current commutation or natural current zero crossing.

Thyristors offer very high voltage and current ratings, low on-state loss, strong surge capability, and comparatively low cost. Their principal limitation is poor turn-off controllability and relatively low switching frequency. They are widely used in line-frequency controlled rectifiers, soft starters, AC voltage controllers, high-power motor drives, HVDC converters, and static bypass systems. Gate-turn-off thyristors and integrated gate-commutated thyristors extend controllability, but require powerful gate circuits and careful management of switching stresses.

Power BJTs

A power BJT is a current-controlled minority-carrier device. Its collector current depends on base drive, and saturation can produce a low on-state voltage. However, stored charge in the base causes storage time during turn-off. The required continuous base current creates drive loss, and the device is vulnerable to secondary breakdown and thermal runaway because current gain and current distribution vary with temperature.

Power BJTs were important in early switching converters and inverters, but they have largely been replaced by MOSFETs and IGBTs. They remain useful for understanding minority-carrier transport, charge storage, and the trade-off between low conduction voltage and switching speed.

Power MOSFETs

The power MOSFET is a voltage-controlled majority-carrier device. A gate voltage creates an inversion channel, allowing current to flow between source and drain. Since the gate is insulated, steady-state gate current is very small, although the gate must be charged and discharged during switching.

Its principal on-state parameter is R_DS(on), which generally increases with junction temperature. A MOSFET has no minority-carrier storage, permitting very fast switching. This makes it attractive in low- and medium-voltage converters, synchronous rectifiers, battery systems, data-centre power supplies, automotive electronics, and high-frequency motor drives. At high blocking voltage, the drift region must be thicker and more lightly doped, causing a substantial increase in resistance. This silicon trade-off limits the economic use of silicon MOSFETs at high voltage.

The gate threshold voltage is not a recommended operating voltage; it merely indicates the onset of conduction at a specified small current. Designers must use the R_DS(on) specification at the actual gate voltage and temperature. Gate charge, Miller charge, output capacitance, body-diode behaviour, avalanche rating, and reverse-recovery characteristics are also important.

IGBTs

The insulated-gate bipolar transistor combines a MOS gate with bipolar conduction. Its MOS input requires little steady-state current, while minority carriers reduce the drift-region resistance. Consequently, an IGBT offers high voltage and current capability with a relatively low conduction drop compared with a high-voltage silicon MOSFET.

The main penalty is the current tail during turn-off, caused by stored minority carriers. This increases switching loss and limits practical frequency compared with MOSFETs. IGBTs are widely used in industrial motor drives, photovoltaic inverters, uninterruptible power supplies, induction heating, traction, welding, and medium-frequency high-power converters. Modern field-stop and trench IGBTs have improved the trade-off among saturation voltage, tail current, short-circuit capability, and switching speed.

An IGBT normally requires an antiparallel diode when reverse current is needed. The electrical and thermal characteristics of the IGBT and diode must be coordinated; a fast but electrically mismatched diode can increase overshoot and losses.

Comparing the Principal Trade-offs

The correct comparison is application-dependent rather than based on one headline rating.

|---|---|---|---|---|

The voltage range alone does not determine selection. A 650 V GaN device may outperform a silicon MOSFET in a high-frequency, hard-switched converter, while a 1.2 kV SiC MOSFET may replace an IGBT where switching frequency, passive-component size, or efficiency justifies its cost. At several kilovolts and very high current, thyristor technology may remain more economical and robust.

Switching Loss, Soft Switching, and Commutation

Turn-on loss depends on the rate at which voltage falls and current rises, gate resistance, driver strength, stray inductance, and the opposing diode's reverse recovery. Turn-off loss depends on current fall time, voltage rise, stored charge, output capacitance, and circuit inductance. The Miller plateau reflects the interval in which gate voltage is used mainly to change drain or collector voltage.

Soft-switching methods reduce voltage-current overlap. In zero-voltage switching, the switch turns on when its voltage is near zero; in zero-current switching, it turns off or on near zero current. Resonant and quasi-resonant converters can therefore operate at higher frequency with reduced switching loss and electromagnetic interference, though they introduce additional circulating current, control complexity, and voltage stress.

The reverse-recovery behaviour of a diode can dominate the loss of a nominally fast transistor. In a half-bridge, for example, the incoming switch may first remove stored charge from the outgoing diode, producing a reverse-recovery current. SiC Schottky diodes nearly eliminate minority-carrier recovery, while appropriate dead time and synchronous rectification are important with MOSFET body diodes.

Thermal Design and Safe Operating Area

Power-device reliability is frequently determined by temperature cycling rather than by nominal electrical ratings alone. Heat flows from the junction through the die attach, package, interface material, heat sink, and cooling medium. Thermal resistance is not always constant: mounting pressure, interface thickness, airflow, coolant temperature, and aging influence it. In pulsed applications, Foster or Cauer thermal networks can represent transient heat flow.

The SOA defines combinations of voltage, current, pulse duration, and temperature for which operation is permitted. It includes forward-bias SOA, reverse-bias SOA, short-circuit withstand, avalanche capability, repetitive peak current, and sometimes unclamped inductive-switching limits. A device rated for a high continuous current may not tolerate the same current at high voltage during a switching transient.

MOSFETs generally have favourable current-sharing behaviour because their on-resistance rises with temperature, although localized parasitics can still cause failure. BJTs can suffer thermal runaway and secondary breakdown. IGBTs require attention to short-circuit duration and latch-up-related limits. Thyristors need adequate triggering, dv/dt control, and di/dt limitation. All devices need protection against overvoltage, overcurrent, excessive gate voltage, and excessive junction temperature.

Reliability, Packaging, and Protection

Failure mechanisms include bond-wire lift-off, solder fatigue, die cracking, gate-oxide damage, metallization degradation, cosmic-ray-induced failure at high voltage, and electrical overstress. Repeated thermal expansion mismatch produces mechanical fatigue. Reliability analysis should therefore consider mission profile: load cycle, switching frequency, ambient temperature, cooling conditions, overload frequency, and expected service life.

Effective protection commonly combines a gate-driver undervoltage lockout, controlled turn-on and turn-off, desaturation or overcurrent detection, active clamping, snubbers, fuses, current sensors, and thermal monitoring. Kelvin-source or Kelvin-emitter connections reduce common-source inductance and improve gate-drive accuracy. The layout must minimize commutation-loop inductance; otherwise even a correctly rated device may experience destructive voltage overshoot.

Datasheet comparison requires discipline. Ratings may be specified at different temperatures, pulse widths, gate voltages, or test circuits. Typical rather than guaranteed values should not be used as design limits. A design margin is necessary, but excessive oversizing increases cost, capacitance, and gate-drive requirements.

SiC and GaN: Wide-Bandgap Devices

Silicon carbide has a larger bandgap, higher critical electric field, and higher thermal conductivity than silicon. These properties permit thinner drift regions, lower specific on-resistance at high voltage, high-temperature operation, and fast switching. SiC MOSFETs are particularly attractive at 650 V, 1.2 kV, and higher voltage classes. SiC Schottky diodes exhibit negligible reverse-recovery charge, benefiting hard-switched boost and inverter circuits.

Gallium nitride devices, commonly implemented as high-electron-mobility transistors, exploit a high-mobility two-dimensional electron gas. They offer low capacitance, low charge, and very rapid switching, especially in the sub-650 V range. Their speed creates strict requirements for gate voltage control, common-source inductance, dead time, isolation, and measurement technique. Enhancement-mode GaN devices simplify use but may have narrower gate-voltage margins than conventional silicon MOSFETs.

Wide-bandgap adoption is not automatically beneficial. The system must exploit higher frequency or reduced losses sufficiently to offset device price, gate-driver requirements, insulation stress, electromagnetic interference, and design effort. At system level, smaller magnetics and heat sinks can compensate for a higher semiconductor cost.

Application-Based Selection Framework

A practical selection process begins with the electrical mission rather than a preferred device family.

1. Define maximum blocking voltage, continuous and peak current, overload, switching frequency, duty ratio, isolation, and ambient temperature.

2. Identify the switching topology and whether the device must conduct reverse current or withstand hard commutation.

3. Estimate conduction, switching, gate-drive, reverse-recovery, and magnetic losses.

4. Check SOA, short-circuit capability, avalanche or unclamped inductive behaviour, thermal impedance, and transient voltage margin.

5. Evaluate gate-driver complexity, layout sensitivity, protection, electromagnetic compatibility, and control requirements.

6. Compare total cost of ownership, including cooling, magnetics, assembly, qualification, and field reliability.

For a low-voltage synchronous buck converter, a silicon MOSFET is often economical because its low R_DS(on) dominates and switching frequency may be high. For a 400–800 V photovoltaic inverter, an IGBT may provide a cost-effective compromise at moderate frequency, while SiC MOSFETs become attractive when higher efficiency, smaller filters, or high switching frequency is valuable. For a line-commutated HVDC link, thyristors remain compelling because frequency is low and voltage/current capability is paramount. For a compact USB-C adapter, GaN can reduce magnetic size and improve power density when its high-frequency capability is properly used.

Illustrative Case Study: Selecting a Switch for a 20 kW Solar Inverter

Background. An engineering team is redesigning a 20 kW, 800 V dc-link photovoltaic inverter. The target switching frequency is 20 kHz, and the system must operate outdoors for a long service life. Efficiency, thermal size, cost, and field reliability are all important.

Problem. The existing silicon IGBT stage meets the power requirement but has significant turn-off loss and requires a large heat sink. The team must choose between an improved IGBT module and a 1.2 kV SiC MOSFET module.

Analysis. The IGBT offers a lower initial semiconductor cost, mature gate-drive technology, and strong short-circuit capability. At 20 kHz, however, its tail current contributes appreciably to switching loss. The SiC module has lower switching energy and permits a reduction in heat-sink size. Its higher dv/dt may increase common-mode current and electromagnetic interference, while the gate driver must limit negative transients and avoid excessive voltage. The layout must have a low-inductance commutation path, and the antiparallel diode characteristics must be included in the loss calculation.

Approach. The team evaluates both devices using the actual mission profile rather than nominal data-sheet current. It calculates conduction and switching losses at minimum, nominal, and maximum temperature; models junction-to-case and case-to-ambient thermal paths; tests double-pulse switching; and verifies overcurrent protection. The SiC option is selected only after the added device cost is compared with savings in cooling hardware and projected energy losses.

Outcome. In the illustrative design, SiC is preferred when the energy-saving and thermal benefits offset the purchase premium. The IGBT remains preferable for a cost-sensitive product with lower switching frequency and less demanding power density.

Learning points. Device selection is a system decision. Switching frequency changes the balance between conduction and switching loss; packaging and layout can determine whether a theoretical device advantage is realized; and reliability qualification must include real thermal and electrical transients.

Challenges, Professional Practice, and Emerging Directions

Common mistakes include selecting by maximum current alone, ignoring temperature dependence, using threshold voltage as a drive voltage, omitting reverse-recovery loss, neglecting dead-time effects, and treating typical datasheet curves as guaranteed limits. Another frequent error is to replace an IGBT with a faster device without reconsidering insulation, EMI filtering, bearing currents in motors, and control-loop sampling.

Professional practice requires traceable design calculations, qualified components, documented derating, production test limits, and compliance with applicable safety and electromagnetic-compatibility standards. High-voltage converters require adequate creepage and clearance, insulation coordination, discharge paths, and protection against hazardous stored energy. Environmental and lifecycle considerations also matter: efficiency reduces operating energy, while component durability, repairability, and material use affect total environmental impact.

Current development is moving toward higher-voltage SiC modules, integrated power stages, advanced sintered and double-sided packages, intelligent gate drivers, and condition monitoring based on electrical signatures. Digital twins and data analytics can estimate junction temperature, detect parameter drift, and support predictive maintenance, but their outputs must be validated against physical measurements. Hybrid modules combining silicon and wide-bandgap switches, advanced cooling, and higher-frequency magnetic materials are also being investigated.

Best Practices for Device Selection and Design

Compare devices at the actual junction temperature, gate voltage, current waveform, and switching frequency.
Separate conduction loss, switching loss, gate-drive loss, diode loss, and magnetic loss.
Use double-pulse tests and low-inductance measurement methods to validate switching behaviour.
Design for SOA and transient stress, not merely steady-state ratings.
Coordinate the switch, diode, driver, snubber, protection circuit, and PCB layout as one commutation system.
Include thermal cycling, overload, short-circuit, and electromagnetic-compatibility tests in qualification.
Evaluate total system cost and reliability rather than device purchase price alone.

Chapter Summary

Power semiconductor devices control the conversion of electrical energy by alternating between conduction and blocking. Their selection requires a balance between conduction loss, switching loss, voltage and current capability, thermal resistance, SOA, reliability, cost, and system requirements. Diodes provide economical rectification; thyristors excel in high-power, low-frequency control; BJTs illustrate minority-carrier limitations; MOSFETs provide fast switching and low-voltage efficiency; IGBTs offer a useful high-voltage compromise; and SiC and GaN extend switching frequency, temperature, and power-density capability. Thermal design, reverse recovery, parasitic inductance, gate driving, protection, packaging, and mission profile are as important as the semiconductor die. A successful design compares complete system consequences rather than isolated datasheet figures.

Key Terms

Power diode: An uncontrolled two-terminal device that conducts in forward bias and blocks in reverse bias.

Reverse recovery: The transient removal of stored charge when a conducting diode is forced into reverse bias.

Thyristor: A latching four-layer semiconductor device that is triggered into conduction by a gate signal.

Holding current: The minimum current required to keep a thyristor conducting.

Power BJT: A current-controlled bipolar transistor designed to handle substantial voltage and current.

Power MOSFET: A voltage-controlled majority-carrier transistor used for fast power switching.

R_DS(on): The drain-to-source resistance of a conducting MOSFET under specified conditions.

IGBT: An insulated-gate bipolar transistor combining MOS gate control with bipolar conduction.

Tail current: The decaying turn-off current of an IGBT caused by stored minority carriers.

Conduction loss: Power dissipated while a device carries current in its on-state.

Switching loss: Power dissipated during transitions because voltage and current overlap.

Safe operating area: The permitted combinations of voltage, current, time, and temperature for reliable operation.

Thermal resistance: A measure of temperature rise per unit power transferred through a thermal path.

Miller plateau: The gate-voltage interval associated primarily with changing a transistor's terminal voltage.

Soft switching: Switching performed near zero voltage or zero current to reduce transition loss.

SiC: Silicon carbide, a wide-bandgap semiconductor suited to high-voltage and high-temperature power conversion.

GaN HEMT: A gallium-nitride high-electron-mobility transistor offering very rapid switching.

Mission profile: The time-dependent electrical, thermal, and environmental conditions expected during service.

Derating: Operating a component below its maximum rating to improve reliability and lifetime.

Review Questions

Short Answer Questions

1. Distinguish conduction loss from switching loss.

2. Why does an SCR remain conducting after its gate signal is removed?

3. What is the significance of MOSFET R_DS(on) in thermal design?

4. Explain the origin of IGBT tail current.

5. Define safe operating area.

Long Answer Questions

1. Explain the construction, operation, merits, and limitations of power MOSFETs and IGBTs.

2. Compare silicon diodes, Schottky diodes, and SiC Schottky diodes in converter applications.

3. Discuss the factors that determine switching loss in a hard-switched half-bridge.

4. Explain thermal design, transient thermal impedance, and reliability concerns in power modules.

5. Compare thyristors, silicon MOSFETs, IGBTs, SiC MOSFETs, and GaN devices for power-conversion applications.

Analytical / Application Questions

1. A converter operates at high voltage and moderate frequency. Develop a decision process for choosing between an IGBT and a SiC MOSFET.

2. A MOSFET exhibits destructive voltage overshoot during turn-off despite having an adequate voltage rating. Identify likely causes and propose corrective measures.

3. For a compact high-frequency adapter, analyse when the higher cost of a GaN device can be justified at system level.

22. Multiple Choice Questions with Answers

Question 1

Which loss mechanism is most directly represented by I²R_DS(on) in a power MOSFET?

A. Conduction loss

B. Reverse-recovery loss

C. Gate-drive loss

D. Avalanche loss

Correct Answer: A. Conduction loss

Question 2

Why is an SCR unsuitable for many high-frequency switching applications?

A. It cannot block reverse voltage

B. Its gate requires continuous current during conduction

C. It normally cannot be turned off by removing the gate signal

D. It has no forward conduction state

Correct Answer: C. It normally cannot be turned off by removing the gate signal

Question 3

Which characteristic primarily limits high-voltage silicon MOSFET performance?

A. Increasing drift-region resistance

B. Excessive minority-carrier storage

C. Need for continuous base current

D. Absence of an insulated gate

Correct Answer: A. Increasing drift-region resistance

Question 4

The turn-off tail current of an IGBT is mainly caused by

A. Gate-oxide charging

B. Stored minority carriers

C. Reverse leakage of the gate

D. Avalanche multiplication in the body diode

Correct Answer: B. Stored minority carriers

Question 5

A converter designer wants to reduce diode-related hard-switching loss at 1.2 kV. Which choice is most appropriate?

A. A slow p-i-n diode with high stored charge

B. A SiC Schottky diode with negligible minority-carrier recovery

C. A low-current signal diode

D. An SCR operated without commutation

Correct Answer: B. A SiC Schottky diode with negligible minority-carrier recovery

Question 6

Which parameter is most important when estimating the junction temperature from steady power dissipation and case temperature?

A. Gate threshold voltage

B. Reverse-recovery time

C. Junction-to-case thermal resistance

D. Holding current

Correct Answer: C. Junction-to-case thermal resistance

Question 7

A 20 kW inverter increases its switching frequency substantially while retaining the same IGBT. Which effect is most likely?

A. Switching loss becomes less important

B. IGBT tail-current loss becomes more significant

C. Conduction loss becomes exactly zero

D. The device no longer requires thermal management

Correct Answer: B. IGBT tail-current loss becomes more significant

Question 8

Which feature most strongly favours GaN HEMTs in compact high-frequency adapters?

A. Very large minority-carrier storage charge

B. High gate-current requirement in steady state

C. Low capacitance and low switching charge

D. Line-frequency natural commutation

Correct Answer: C. Low capacitance and low switching charge

Question 9

Why must a device be evaluated using its SOA rather than its continuous current rating alone?

A. SOA applies only to unloaded operation

B. SOA accounts for permitted combinations of voltage, current, time, and temperature

C. Continuous current ratings always include switching transients

D. SOA eliminates the need for thermal calculations

Correct Answer: B. SOA accounts for permitted combinations of voltage, current, time, and temperature

Question 10

In selecting between an IGBT and a SiC MOSFET, which is a valid system-level consideration?

A. Only the purchase price of the semiconductor

B. Only the maximum rated current

C. The combined effects of switching loss, cooling, layout, EMI, reliability, and lifetime cost

D. The colour and physical appearance of the package

Correct Answer: C. The combined effects of switching loss, cooling, layout, EMI, reliability, and lifetime cost

Semiconductor Devices
Chapter 9 of 15

Chapter 9 Power Semiconductor Devices