IGBT-basic know-how

IGBT: how does an Insulated Gate Bipolar Transistor work?

Dr. Martin Schulz

Principal Application Engineering
Infineon Technologies

www.infineon.com v1.0
IGBT-basic know-how

About the author

Dr. Martin Schulz has been with Infineon since 2005. In 2011, he joined the
application-engineering group. His responsibilities are in the area of electrified
commercial and agricultural vehicles (CAV). Furthermore, thermal management
and interconnection technologies are his field of expertise. Martin Schulz holds
a Dr.-degree in electrical engineering, gained at the University of Siegen at the
department of Power Electronics and Electrical Drives. He holds several patents
in the field of power electronics and is a Senior IEEE-Member.
Dr. Martin Schulz

Today, the IGBT comes very close to being considered an ideal switch. Then again – what is the difference between
an IGBT and a MOSFET? What advantages does an IGBT offer and how does it work?

In technical communities focusing on power electronics, questions like this one regularly arise: “I have to design
an H-bridge to control a motor. Voltage is 320 V, the current is 2 A, switching frequency is 30 kHz. To remain on the
safe side, I’m searching for a switch with 600 V blocking capability. I have no idea whether to use MOSFET or IGBT.
Any hints as to criteria to do a proper selection?”

In this case, there is no simple decision, as some important parameters are missing. For the target set, solutions
with 600 V MOSFET as well as using 600 V IGBT come to mind. Important criteria not mentioned in the query, re-
late to size, efficiency and cost targets.

The IGBT, or Insulated Gate Bipolar Transistor, became the most used power electronic component in industrial
applications. In the meantime it has become a central component in inverters for all types of electric drives, bat-
tery chargers, and solar and wind power plants. But why? What is so special about this component? What are the
strengths and what challenges have to be handled when using this technology? The answer to these questions
lies within the technology itself.

The wide portfolio of available parts

From speed-controlled, low-power compressor drives in refrigerators to traction drives in railways: the IGBT has
taken over a dominant position within the last decades.

From a multitude of packages, the user can choose devices from 300 V in discrete designs to power modules
supporting 6500 V. Current-carrying capability of a single transistor spans a range from a few amps to several
kilo-amps.

Besides the well-established TO-package series, SMD-components are available, accompanied by power mod-
ules for the highest power demands.

www.infineon.com v1.0
IGBT-basic know-how

Depending on the power to be handled, soldering or press-in connectors are in use, while currents exceeding
200 A typically demand screw-type terminals. Figure 1 includes a minute section of packages available on the
market.

10000

1000
Forward current [A]

100

10
100 1000 10000

Blocking voltage [V]

Figure 1: IGBT-Overview on dimensions

IGBT: a simple technology

The most basic function of an IGBT is the fastest possible switching of electric currents, thus achieving the lowest
possible switching losses. As the name “Insulated Gate Bipolar Transistor” reveals, an IGBT is a bipolar transistor
with an isolated gate structure; the gate itself is basically a MOSFET. Therefore, the IGBT combines the advan-
tages of high current-carrying capabilities and high blocking voltages of a bipolar transistor with the capacitive,
almost zero-power based control of a MOSFET. Figure 2 depicts how a MOSFET and a Bipolar Transistor combined
lead to the IGBT.

C
B
C
G UCE
G
UGE
E
E
IGBT Insulated Bipolar
gate transistor

Figure 2: IGBT in detail – the MOSFET represents the gate, the bipolar transistor the output stage

www.infineon.com 3
11-2019
IGBT-basic know-how

The schematic only depicts the general structure. In reality, the technical design is not based on two independent
devices. The overall function is the consequence of integrating the structure at the chip level.

With the MOSFET acting as a gate structure, the Base of the bipolar transistor is no longer available. The device is
now connected via Collector, Gate and Emitter.

The fundamental function of the IGBT is rather simple. A positive voltage UGE from gate to emitter turns on the
MOSFET. Then, the voltage connected to the collector can drive the base current through the bipolar transistor
and the MOSFET; the bipolar transistor turns on and the load current can flow.

Vice versa, a voltage UGE ≤ 0 V turns of the MOSFET, the base current is interrupted, and the bipolar transistor turns
off as well.

Due to the capacitive nature of the MOSFET, the gate current only needs to charge the gate capacity. The RMS
value of the gate current sums up to almost zero. Therefore, you hear very often that the power to control an IGBT
is zero.

This simplification often is a root cause for troubles in designing the application. Developing hardware to control
an IGBT – a gate driver – is a task that may keep a small development team busy for a while.

However, this much effort is most likely unnecessary. Some semiconductor manufacturers offer suitable hard-
ware with a wide variety of functionalities as integrated solutions. A suitable gate driver can be designed by uti-
lizing dedicated gate driver ICs and sticking to the proposals given in datasheets and application notes.

As a power electronic device, the IGBT is optimized for high switching speeds. Operating it in linear mode similar
to MOSFETs in former audio amplifiers is highly undesirable. This mode of operation would lead to massively
increased losses.

With the output characteristics of the bipolar transistor, further features of the device result. An IGBT can carry
current in one direction only, and during operation there is always a forward voltage correlated to a PN junction.
IGBTs are well suited for a switching frequency range up to 30 kHz. Using special techniques, so-called resonant
topologies, the switching losses can be reduced, and higher switching frequencies can be achieved.

In contradiction to a MOSFET, IGBTs can be built to withstand very high voltages. With an overlap between 300 V
and 600 V, the low-voltage domain is covered by MOSFETs, while voltages exceeding 600 V today are dominated
by IGBTs.

Other than with a MOSFET, an IGBT does not inherently contain a freewheeling, or body, diode by design. This
diode, however, is a part needed to protect the switch providing a freewheeling path to prevent reverse current.
When choosing a component, care has to be taken to either add a suitable diode or go for a component with an
integrated diode die.

www.infineon.com 4
11-2019
IGBT-basic know-how

The schematics of the components given in figure 3 suggest devices containing an IGBT only or a so-called
co-pack including IGBT and diode.

C C
G G

E E

IGBT IGBT + diode

Figure 3: Only IGBT, and IGBT co-packed with freewheeling diode

To improve the switching performance of an IGBT, auxiliary emitters are commonly used. This connection reduces
the influence of stray inductances within the gate circuit. What has become a de-facto standard in power mod-
ules, has also recently reached discrete components in the newly designed TO247-4. Figure 4 depicts the details
in the differences between the two packages.

IGBT + diode IGBT + diode

TO247 TO247-4

C C
G G

E E’
C C
E E E’
G E G

Figure 4: Difference between TO247 and TO247-4

The auxiliary Emitter E’ does not contribute to carrying the load current. This reduces the distortion resulting
from inductive coupling. A current change di/dt within the gate circuit is eliminated, so is an induced voltage. As a
result, the switching gets cleaner, easing the work to be done to achieve EMI targets.

IGBT and more – cleverly integrated in power modules

In most cases, a single switch is not enough for a developer to build the design. Within a frequency converter,
typically two units are necessary. First, the supplying voltage taken from the grid has to be rectified, and the
DC-voltage level sometimes needs to be adapted or stabilized. Afterwards, an inverter is used to convert the
DC-link voltage into the AC system desired, which can be different from the supplying grid in frequency, ampli-
tude and even number of phases.

In case the application does not demand regenerative operation, a simple diode rectifier can be chosen. The
energy from the application leads to an increase in the DC-link voltage. Here, a break chopper is installed, and in
the case of excess energy, it provides a path for handling energy safely by converting it into heat.

www.infineon.com 5
11-2019
IGBT-basic know-how

To support building common designs and to minimize the effects of unwanted influences, semiconductor man-
ufacturers combine the necessary components into power modules. The schematic given in figure 5 includes a
rectifier, a break chopper and the inverter needed to form a converter. The colors denote groups, which are avail-
able as individual components; table 1 summarizes names and contents.

Figure 5: Semiconductors, grouped in power modules

Table 1: Choice of devices and basic parameters

Name Content Voltage Current Example
[V] [A]
PIM or CIB Rectifier, 600–1700 6–150
Power Integrated break
Module chopper,
Converter, Inverter, inverter
Break
Sixpack or full bridge Inverter 600–1700 6–600
DC to 3~

Break Chopper Diode/IGBT 600–1700 25–1400

with
freewheeling
diode

Half bridge Inverter 600–6500 200–1800

DC to 1~

Single switch IGBT with 1200–6500 200–3600

freewheeling
diode

www.infineon.com 6
11-2019
IGBT-basic know-how

About 90% of the industrial applications are pumps, fans and other systems that do not demand a bidirectional
flow of energy. The solution based on a diode rectifier therefore serves the majority of systems. Besides the most
common topologies, advanced combinations of semiconductors are available. Among other things, this includes
various multi-level designs as well as Vienna rectifiers or matrix-converter structures.

Controllable thermal properties

Today’s IGBTs are quite similar to what is considered the ideal switch. An ideal switch is characterized by zero
current when turned off, and by zero voltage across the device when turned on. As a consequence, an ideal switch
does not generate any losses, and therefore does not create heat. Modern IGBTs reach efficiency levels of > 99%
for the individual switch, however, thermal management is an important topic that should not be neglected.

Considering power levels where discrete components are sufficient, assembling the power semiconductors is
business as usual. As known from TO-packages, for example, the component is mounted to a heat sink, which
in turn is fixed to the PCB by means of soldering or screwing. Attention has to be paid here too, as most discrete
packages feature an electrically active back side, typically the collector of a switch.

On a common heat sink, only parts that share a common voltage can be grouped. Failing this, to avoid short
circuit, isolation barriers between the switch and the heat sink become mandatory. These in turn reduce the
thermal transfer and therefore the performance of the semiconductor.

In higher power levels, IGBTs grouped in power modules are used. The thermally active backside of these mod-
ules is galvanically isolated. This way, all dies involved share a common surface that can be mounted to a single
heat sink. To improve thermal performance, applying a thermal interface material is mandatory. For this, care
has to be taken to choose a material that can support the thermal situation as well as the lifetime demands of
the application.

Despite the efficiency levels reached today, power losses can be serious. In a windmill application with 5 MW
throughput, power electronics generate, to put it simply, 100 kW of losses if the overall efficiency is 98%. This is
heat in other words, which needs to be handled and dissipated in a safe manner.

Though the voluminous high-power modules appear to be massive, solid entities, mechanical handling and
mounting deserves some caution. Despite the solid design, mechanical issues can later be distinguished as the
root cause of the damage, be it done during mounting or failure in operation.

Here too, the manufacturer’s application notes give an insight into tried-and-tested procedures. Respecting the
given limits and instructions is a first step towards a reliable, long-lasting design.

www.infineon.com 7
11-2019
IGBT-basic know-how

Easy-to-estimate dimensioning
Back to the posting found on the Internet about which component to use – IGBT or MOSFET, from the few details
given, a first estimate can still be made.

The voltage is rated 320 V, a typical value when rectifying a 240 V line voltage. This does not take into account
that line voltages include some tolerances, which also reflect in the DC voltage after rectifying. With a 10% adder,
the maximum DC-link voltage within the system can easily reach 350 V. This eliminates devices rated 400 V or less
from the list. The reason behind this is a transient phenomenon, creating a voltage spike during turn-off. As a
rule of thumb, a device should not be driven with voltages higher than about 2/3rd of the rated blocking, or break
through, voltage. The remaining 33% are kept as a margin for transient overvoltage spikes.

To carry a motor current of 2 A, a device rated 2 A is not sufficient either. The reason is to be found in the thermal
budget of the switch. This includes the forward losses when the switch is turned on, as well as the switching loss-
es and thermal resistances of the device itself. At the same time, over-dimensioning by factor ten is undoubtedly
too conservative.

Comparing switching losses, a MOSFET outperforms an IGBT when both have the same current rating, making
it the predestined device for higher switching frequencies. However, the static losses grow by I2 in the MOSFET
according to PVMOS,Stat = I2 • RDS.

Due to its bipolar character, losses in an IGBT only grow linearly with current according to PVIGBT,Stat = I • UCE.
Typically, this makes it the better choice in high-current applications.

With a few fundamental values taken from a datasheet and a little mathematical effort, the thermal correlations
can be evaluated. This allows for a first estimation whether or not a considered component could be a suitable
choice. An important prerequisite for this thermal estimation is an anchor point to start from, given by a tem-
perature and the thermal connection. Using the chain of thermal resistances RthJH from the chip’s junction to the
heat sink, and the maximum heat sink temperature THS presents a suitable approach. A further alternative would
be to choose the ambient temperature Tamb as a reference, accompanied by the thermal resistance RthJA from chip
to ambient. In this case, the heat sink’s value RthHA is included.

Table 2: Basic math for thermal evaluation

Parameter Switching losses Static losses Chip temperature
Eon, Eoff Pdyn = (Eon + Eoff) • fsw
RDS PVMOS,Stat = I² • RDS
UCE PVIGBT,Stat = IC • UCE
Tamb TChip = Tamb + (Pdyn+Pstat) • RthJA
THS TChip = THS + (Pdyn + Pstat) • RthJH

www.infineon.com 8
11-2019
IGBT-basic know-how

In case the maximum chip temperature calculated remains within safe operation limits, the component general-
ly can be considered as suitable.

Whether or not the component provides a certain lifetime demands a more accurate examination including that
of the application’s load profile. A lab’s equipment, an experiment or another setup rarely in use has totally dif-
ferent requirements as compared to an industrial drive in permanent operation.

But then again, the question “How long does my setup last?” is rather asked by professionals than by someone
searching for help in an Internet forum.

Overview figures and tables

Figure 1: IGBT-Overview on dimensions
Figure 2: IGBT in detail – the MOSFET represents the gate, the bipolar transistor the output stage
Figure 3: Only IGBT, and IGBT co-packed with freewheeling diode
Figure 4: Difference between TO247 and TO247-4
Figure 5: Semiconductors, grouped in power modules

Table 1: Choice of devices and basic parameters

Table 2: Basic math for thermal evaluation

Published by Please note! Additional information

Infineon Technologies AG This Document is for information purposes only and any For further information on technologies, our products, the
81726 Munich, Germany information given herein shall in no event be regarded as a application of our products, delivery terms and conditions
warranty, guarantee or description of any functionality, and/or prices, please contact your nearest Infineon Technologies
© 2019 Infineon Technologies AG. conditions and/or quality of our products or any suitability for a office (www.infineon.com).
All Rights Reserved. particular purpose. With regard to the technical specifications of
our products, we kindly ask you to refer to the relevant product Warnings
data sheets provided by us. Our customers and their technical Due to technical requirements, our products may contain
departments are required to evaluate the suitability of our dangerous substances. For information on the types in question,
products for the intended application. please contact your nearest Infineon Technologies office.

We reserve the right to change this document and/or the Except as otherwise explicitly approved by us in a written
information given herein at any time. document signed by authorized representatives of Infineon
Technologies, our products may not be used in any life-­
endangering applications, including but not limited to medical,
nuclear, military, life-critical or any other applications where a
Order Number: B114-I0931-V1-7600-EU-EC failure of the product or any consequences of the use thereof
Date: 11 / 2019 can result in personal injury.