Fer 56 2 000 2010 PDF

Whole Number 229
ISSN 0429-8284
Power Semiconductor contributing in

energy and environment region
Power Semiconductor contributing

in energy and environment region
CONTENTS
The Current Status and Future Outlook for Power Semiconductors
47
3.3 kV IGBT Modules
51
New Lineup of V-Series IGBT Modules
56
V-Series Intelligent Power Modules
60
Superjunction MOSFET
65
High Reliability Power MOSFETs for Space Applications
69
Pressure Sensor for Exhaust Systems
74
Thermal Management Technology for IGBT Modules
79
Cover photo:
To protect the earth from global warming, there are many things that mankind
must do. While harnessing the synergistic
effect from power electronics and power
semiconductors to boost the power of each,
Fuji Electric is focusing on the energy
and environment fields, and is working to
make energy more environmentally-friendly and to help conserve the global environment.
Fuji Electrics power semiconductors
are also used in outer space, where higher
reliability and greater energy savings are
required than on land.
The cover image shows Fuji Electrics
space-use high-reliability MOSFETs (Metal-Oxide-Semiconductor Field-Effect Transistors) that are being used in the International Space Station. Fuji Electrics power
semiconductors help to achieve energy savings not only on the Space station, but also
on Spaceship Earth, and where they are
making a positive contribution to protecting the environment.
FUJI ELECTRIC REVIEW vol.56 no.2 2010

date of issue: May 20, 2010
editor-in-chief and publisher Hisao Shigekane
Corporate R & D Headquarters

Fuji Electric Holdings Co., Ltd.
Gate City Ohsaki, East Tower,
11-2, Osaki 1-chome, Shinagawa-ku,
Tokyo 141-0032, Japan
http://www.fujielectric.co.jp
editorial office
Fuji Electric Journal Editorial Office

c/o Fuji Office & Life Service Co., Ltd.
9-4, Asahigaoka 1-chome, Hino-shi, Tokyo 191-0065, Japan
Fuji Electric Holdings Co., Ltd. reserves all rights concerning

the republication and publication after translation into other
languages of articles appearing herein.
All brand names and product names in this journal might be
trademarks or registered trademarks of their respective companies.
The Current Status and Future Outlook for

Power Semiconductors
Yasukazu Seki Toru Hosen Masaru Yamazoe
ABSTRACT
With the increased focus on efforts to protect the global environment, power semiconductors, which are the main
power electronics products, are becoming increasingly important. Using 6th-generation IGBT V-Series technology,
Fuji Electric has developed and commercialized high voltage, large capacity power modules that are capable of
operating at high temperatures. As post-silicon technology for the next generation, we are pursuing the development
of devices that utilize wide band gap semiconductor material and the development of superjunction MOSFETs. Additionally, MOSFETs designed for applications in outer space and used in the Japanese experimental module known
as Kibo attain high reliability and low loss, while ICs designed for use in power supply control achieve low noise and
energy savings. Exhaust system pressure sensors for use in automobiles and control ICs for use in hybrid vehicles
have been newly commercialized.
1. Introduction
In 2009, the Obama administration established
and inaugurated the Green New Deal. In September
of the same year, at the United Nations Summit on
Climate Change, Japanese Prime Minister Hatoyama
announced a new mid-term goal of cutting Japans
greenhouse gas emissions by 25% from 1990 levels by
2020. Among the new policies set forth by leaders from
the various countries, projects related to energy and
environment attracted the most attention.
Fuji Electric is presently reforming its business
structure to focus on energy and the environment.
Also, Fuji Electric has long been involved in efforts to
develop innovative power electronics technology which
is crucial for protecting the global environment and reducing CO2 emissions. Power electronics technology is
a key technology for converting energy into power, and
as the main components, power semiconductors are becoming more and more important.
The performance characteristics requested of power semiconductors are low loss, high speed switching,
low noise and ease of use.
This paper discusses the current status and future
outlook for the energy-saving and environmentallyfriendly power semiconductors which are being developed by Fuji Electric and their representative products
of power modules, power discretes, power supply ICs
and automotive devices.
2. Power Modules
With a commitment to energy savings and ecofriendliness, IGBT (insulated gate bipolar transistor)
Semiconductors Group, Fuji Electric Systems Co., Ltd.
Fuji Electric Systems Co., Ltd.
47
Fig.1 Requested characteristics of IGBT modules
Low on-voltage
(on-resistance)
Tradeoff relation
High-speed
switching
High resistance to
damage
power modules are being deployed in various fields.

The IGBT chip at the core of these modules is a 6th
generation IGBT chip, which has begun to be deployed
in Fuji Electrics V-Series. As can be seen in Fig. 1,
many items are requested of the IGBT modules but a
tradeoff relation exists between the requested items
and performance characteristics, and numerous technological breakthroughs would be needed to satisfy all
the requests.
With the V-Series 6th generation IGBT module,
the attained characteristics are close to their theoretical limits and dissipation loss was reduced. Also, the
design was implemented with good awareness of the
environment. For example, the V-Series complies
with the RoHS directive*1 through the use of lead-free
materials, the package structure significantly reduces
generated noise, and a small size and light weight are
realized simultaneously. Additionally, high temperature operation up to 175 C is possible.
A series of 600 V, 1,200 V and 1,700 V IGBT modules and IPMs have been commercialized using these
*1: RoHS: EU Directive on the restriction of the use of certain hazardous substances in electrical and electronic
equipment
3. Next Generation Power Modules

The characteristics of the 6th generation IGBT
V-Series are approaching the theoretical limits of
silicon. Silicon carbide (SiC) and gallium nitride (GaN)
are the leading candidates for materials to be used
instead of silicon in next-generation power devices.
Due to material properties well suited for use in power
devices, research and development of SiC material has
been ongoing for a long time, and recently, practical
applications have begun to be considered. SiC material is very expensive due to the difficulty of growing its
crystals and the process technology involved, but a way
toward a technical solution is gradually becoming visible, and sooner or later the cost issue is also expected
to be resolved.
Moreover, GaN, which became prominent with
blue diodes, has also been found to be a promising material for power devices, and is a candidate for the material used in next-generation power devices. Despite
having only a short history of use as a power device
material, the decisive factor for GaN, which can be
fabricated on a silicon wafer, will be whether or not the
advantage of its inexpensive cost can be leveraged.
Both SiC and GaN are promising candidates as the

material of future power devices. Fuji Electric and the
National Institute of Advanced Industrial Science and
Technology (AIST) are conducting joint research of SiC
material. Also, Fuji Electric and Furukawa Electric
Co., Ltd. have jointly established a next-generation
power device technical research association to research
GaN, and plan to incorporate the results into the nextgeneration power modules.
4. Power Discretes
As a high-voltage MOSFET (Metal-OxideSemiconductor Field-Effect-Transistor), Fuji Electric
has newly developed the Superjunction MOSFET
(SJ-MOSFET). In 2008, Fuji Electric developed the
SuperFAP-E3S 600 V Series having the industrys
lowest level of Ron A (on-resistance normalized to a
unit area) as a planer MOSFET, which realized low
noise and an improved tradeoff relation between low
loss and switching performance, and contributed to
higher efficiency in devices. Recently, in the field
of switching power supplies, which is the main application of MOSFETs, efforts to realize higher efficiency have been accelerated in order to comply with
international energy-saving regulations such as the
International Energy Star Program. In particular, a
power supply efficiency of 92% or above (at 50% load)
is requested of the large-capacity servers and the
like used at internet data centers, which are essential for IT companies. To realize this high efficiency,
MOSFETs must also have lower loss characteristics,
and therefore Fuji Electric developed the new low
on-resistance SJ-MOSFET.
The SJ-MOSFET has
the industrys lowest level of Ron A, and realizes approximately 1 / 4th the on-resistance of the previous
SuperFAP-E3S Series. When installed in the power
factor improvement circuit of a power supply, as shown
in Fig. 2, loss is reduced by approximately 14% and
development is accelerating toward the goal of early
product commercialization.
Fig.2 Comparison of power MOSFET dissipation loss (input
AC100 V / output 400 W)
1.2
1.0
Dissipation loss (a.u.)
superior V-Series IGBTs.

Fuji Electric has previously developed IGBT modules and IPMs (intelligent power modules) for use in
hybrid vehicles. As key components for hybrid vehicles, Fuji Electric has also previously developed plated
IGBT and FWD (free wheeling diode) chips housed in a
package having a double-sided cooling structure. This
IGBT achieves approximately twice the current density of a general-purpose IGBT. Now, with the additional application of 6th generation IGBT V-Series technology and microfabrication technology, and by improving
the FS (field stop) structure, a dramatic improvement
in characteristics has been achieved. Moreover, temperature sensing and current sensing functions have
been built-in to realize a chip that is even easier to use.
As an example application to the energy and environment field, high-voltage high-power IGBT modules
have begun to be used in wind power generators. As
a result of recent trends of applications involving
wind power generation, there is strong demand for
high-voltage and high-power IGBT modules, and in
response, Fuji Electric has prepared 1,200 V, 1,700 V
and 3,300 V high-power modules. A wide variety of
standard package groups including the HPM (high
power module), EconoPACK+ and the like, have been
assembled and are being deployed by our customers.
As new types of IGBTs, the RB-IGBT (reverse
blocking IGBT) and RC-IGBT (reverse conduction
IGBT) are being developed by Fuji Electric. In particular, as next-generation applications for the RB-IGBT,
application to the highly-anticipated matrix converters
and application to a new 3-level inverter are promising.
0.8
14 %reduction
Turn-off
loss
0.6
0.4
0.2
0.0
Turn-on
loss
Conduction
loss
FMH23N60ES
SJ-MOSFET
48
Fuji Electric has also commercialized highly reliable MOSFETs for space applications such as in artificial satellites. In 1994, Fuji Electrics initial spaceuse devices were installed in and contributed to the
successful launch of the first rocket built exclusively
in Japan. Leveraging that experience and incorporating subsequent research and development results,
the present lineup of space-use MOSFETs overcame
prior obstacles to achieve low on-resistance and the
required tolerance to ionization radiation and highenergy charged particles in outer space. Additionally,
these MOSFETs are installed in Kibo, the Japanese
Experimental Module of the International Space
Station, and have continued to operate properly since
being launched into orbit in 2008. In the future, Fuji
Electric intends to expand the range of applications to
overseas space industries as well.
Meanwhile, Fuji Electric has developed a Schottky
barrier diode (SBD) having an ultra low IR and suitable for use in a high temperature environment such
as for photovoltaic power generation. The reverse
leakage current IR has been reduced to 1 / 10th that of
a conventional SBD, enabling guaranteed operation at
junction temperatures of up to 175 C.
Through joint development with third parties, as
described above, Fuji Electric is accelerating the development of next-generation devices that use SiC and
GaN to realize dramatically lower loss compared to the
present generation of silicon-based devices.
oped, and realize both high performance and small size

in a SOP8 pin package. In particular, the FA5610N
realizes a high power factor and low noise with an oscillation frequency distribution function, enabling the
input filter to be simplified.
In the power supply control IC field, to satisfy increasingly severe requirements for higher efficiency,
energy savings, smaller size, lower device cost and the
like, Fuji Electric continues to research and develop a
distinctive, proprietary control method.
5. Power Supply Control ICs
6. Automotive Devices
To help achieve energy savings and lower noise

in switching power supplies and to lower the cost of
devices, Fuji Electric has independently developed and
applied a control method for power supply control ICs.
Devices such as televisions, PCs and printers that are
plugged-in continuously an electrical outlet remain in
a standby state for extended periods of time. Their
consumption of power during standby must be reduced
in order to save energy. The EPA 5.0 standard issued
by the U.S. Environmental Protection Agency prescribes the average efficiency even during light-load
operation, and in response, Fuji Electric developed the
FA5592 Series of current mode PWM control ICs that
comply with the EPA 5.0 standard. The FA5592 significantly improves efficiency during light load operation,
as shown in Fig. 3, by reducing the switching frequency
when the load rate is below 60%. Moreover, with a
built-in switching frequency distribution function that
realizes lower noise and an internal startup circuit
that ensures operation at up to 750 V, the FA5592 can
be used in countries or regions in which there are large
fluctuations in power supply voltage.
For relatively large-capacity power supplies,
FA5604 / FA5605 multifunction voltage mode PMW
control ICs and FA5610 / FA5611 low-noise continuous current mode PFC control ICs have been devel-
As automotive devices, IPSs (intelligent power

switches), exhaust system pressure sensors, and the
Fi009 hybrid vehicle IGBT drive IC have been developed by Fuji Electric.
(1) Exhaust system sensor
Automotive exhaust gas regulations are becoming
more stringent year after year, and in 2009, the Post
New Long-term Regulations in Japan and the Euro 5
emissions standards in Europe came into force. As a
result, exhaust gas recirculation systems, in which a
portion of exhaust gas is recirculated to the intake side
so as to control combustion in an engine, have begun
to be used. Newly developed exhaust system pressure
sensors, based on the CMOS (Complementary MetalOxide-Semiconductor) single-chip technology that has
been applied to conventional intake system pressure
sensors, utilize a new structure to realize dramatically
improved corrosion resistance. On the basis of this
technology, product development continues with the
aim of application to heavy machinery and the like.
(2) Fi009 driving IC for hybrid vehicle IGBTs
The demand for highly fuel-efficient hybrid vehicles, which are provided with both a gasoline engine
and an electric motor so as to optimize load-sharing
according to the running conditions of the automobile,
is increasing rapidly. IGBTs are the main devices used
49
Fig.3 Dependence of power supply efficiency on load current

for FA5592
Output 65 W/ 19 V
90
Efficiency (%)
86
82
78
FA5592 (for AC115 V input)
74
FA5592 (for AC230 V input)

Conventional model (for AC115 V input)
Conventional model (for AC230 V input)
70
25
50
75
100
Load factor (%)
Vol. 56 No. 2 FUJI ELECTRIC REVIEW
in the inverter systems that drive the motor in hybrid

vehicles. The control IC that drives this IGBT has the
important function of protecting the IGBT in order
to ensure high reliability of the inverter system. The
newly developed control IC has a function that protects
against overheating and overcurrent, and receives signals from temperature and current sensors inside the
IGBT chip made by Fuji Electric.
In addition to the products introduced herein, Fuji
Electrics automotive devices also include single-chip
igniters and other products that use Fuji Electrics proprietary and distinctive technology. Leveraging this
technology, Fuji Electric intends to continue to develop
highly reliable and high performance products that
satisfy customer needs.
7. Postscript
important topic of today, the implementation of environmental measures such as reducing CO2 emissions
and the development of new energy sources that do
not depend on fossil fuels are urgently needed. With
a commitment to energy savings and eco-friendliness,
Fuji Electric is actively addressing these problems and
aims to contribute their solutions. For this purpose,
technical innovation in power semiconductors, the key
components of power electronics technology, is absolutely necessary.
As has been discussed in this paper, Fuji Electric
is working to develop distinctive power semiconductor
products, and will apply innovative technology to realize lower loss, more advanced functionality, smaller
size, higher reliability and lower noise. Fuji Electric
intends to continue to advance technical development
with the aim of developing products from the customers perspective.
With protection of the global environment being an
50
3.3 kV IGBT Modules

Takeharu Koga Yasuhiko Arita Takatoshi Kobayashi
ABSTRACT
Fuji Electric has developed a 3.3 kV-1.2 kA IGBT module in response to market needs for inverters suitable
for industrial and vehicle applications. The package of the newly developed module incorporates IGBT high-power
module technology. Compared to previous modules, internal inductance has been reduced by 33% and the current
flow to chips on each insulating substrates shows good uniformity. Power cycle tests were implemented with this
module, and sufficient durability was verified. 3.3 kV-0.8 kA IGBT modules has been developed. 3.3 kV-1.5 kA and
3.3 kV-0.8 kA IGBT modules are also being developed to expand the product lineup.
1. Introduction
2. Specifications of 3.3 kV IGBT Module
IGBT (Insulated Gate Bipolar Transistor) modules
have gained widespread popularity due to their low
loss characteristics, suitability for configuring simple
driving circuits, wide safety operation area and high
reliability. The high-voltage high-power sector in
which GTO (Gate Turn-OFF) thyristors have been
used widely is also transitioning to the use of IGBT
modules. IGBT modules are being applied widely in
high power inverter and high voltage inverter systems
and the like. Especially as a result of recent efforts to
prevent global warming, the market for new energy
(wind power generation, solar power generation) has
grown rapidly. The inverter systems used in this sector have been advanced to achieve higher power capacities, and demand for IGBT modules is growing.
Fuji Electric has previously developed high-power
IGBT modules for application to the high-power sector. In 2007, Fuji Electric enhanced the chip, package
design and manufacturing technology of the previous
high-power 1.2 kV and 1.7 kV IGBT modules to develop a 3.3 kV 1.2 A high-power IGBT module. In 2008,
Fuji Electric improved the low-loss, thermal characteristics and environmental durability of the 1.2 kV and
1.7 kV modules to develop new high-power IGBT modules(1) and is expanding its product lineup. The package of these modules has low inductance and achieves
excellent current balance.
This package technology is also applied to the newly developed 3.3 kV 1.2 kA IGBT module. The 3.3 kV
IGBT modules are designed for use in vehicle applications as well as industrial applications. This paper
presents an overview and describes the performance of
the new 3.3 kV IGBT modules.
51
Figure 1 shows an external view of 3.3 kV 12 kA

IGBT module. With a package size of 190 mm
140 mm, the module is compatible with modules made
by other companies. Table 1 lists the target specifications of the 3.3 kV modules.
3. Electrical Characteristics
3.1 IGBT and FWD characteristics
(1) IGBT chip characteristics

IGBT chips incorporate a trench structure and a
field stop (FS) structure (U-Series IGBT) that provide
an excellent trade-off in the relation between the collector-emitter saturation voltage VCE(sat) and turn-off
loss Eoff, and utilize a cell-pitch optimized for 3.3 kV
to achieve lower loss. Also, in the high-power sector, it
is essential that an IGBT chip has high-power switching capability (wide reverse bias safety operation area
(RBSOA) and short-circuit capability) in order to realize high reliability. To provide a wide RBSOA, the
Fig.1 External view of 3.3 kV 1.2 kA IGBT module
190 mm
140 mm
Table 1 Target specifications of 3.3 kV IGBT module (1.5 kA modules are under development)
Item
Symbol
1MBI1200UE-330
1MBI1500UE-330
1MBI800UG-330
Collector current
IC
1,200
1,500
800
Package size
190 140
190 140
130 140
mm
Collector-Emitter
saturation voltage (chip)
V CEsat
3.15 V (typical value)

(at 150 C and 1,200 A)

(at 150 C and 1,500A)

(at 150 C and 800 A)
Forward voltage (chip)
VF

(at 150 C and 1,200 A)

(at 150 C and 1,500 A)

(at 150 C and 800 A)
IGBT
Thermal
resistance
R thj-c
FWD
V iso
Isolation voltage
8.5
8.0
13.0
K/kW
17.0
15.0
25.0
K/kW
6.0
6.0
6.0
kV
Fig.2 VCE(sat)-I C characteristics

3,000
Fig.3 VF-IF characteristics

3,000
V GE = 15 V
2,500
T j =25 C
Forward current I(A)

F
Collector current I(A)

C
2,500
2,000
T j = 150 C
1,500
1,000
500
0
Units
1,500
500
0
Collector-emitter saturation voltage

VCE(sat)(V)
3.2 VCE(sat)-IC characteristics and VF-I F characteristics
Figure 2 shows the VCE(sat)-IC characteristics.

Similar to Fuji Electrics low-voltage class of trench
IGBTs, a positive temperature coefficient is exhibited.
With a reduced current imbalance for parallel connections, these modules allows for easy parallel connections necessary for achieving larger currents.
Figure 3 shows the VF-IF characteristics. The
FWD, at a forward voltage, also exhibits a positive
temperature coefficient at currents of at least one-half
of the rated value, and allows for easy parallel connections.
T j =150 C
1,000
chip was designed with a structure that suppresses the

concentration of current at the edge of the active area.
In addition, carrier injection on the collector side is adjusted to ensure sufficient short-circuit capability.
(2) FWD (Free Wheeling Diode) chip characteristics
In order to achieve (1) lower loss and (2) suppression of oscillation and surge voltage during reverse
recovery at low current levels, FWD chips are designed
with an optimized crystalline wafer structure and a
deep collector-side n+ layer concentration profile. Also,
to provide high reverse recovery capability (high di / dt
capability) the cathode-side is designed with a structure that suppresses the concentration of current at
the edge of the active area.
T j =25 C
2,000
Forward voltage V
F (V)
Fig.4 Switching waveforms (inductive load)

(VCC = 1,800 V, IC = 1,200 A, Tj = 150 C)
0V
0V
VGE
VCE
IC
VGE
IC
VCE
0V
0A
0V
0A
Turn-on waveform
Turn-off waveform
VAK (V CE )
VCE : 500 V /div

IC
I F : 500 A /div
VGE : 20 V /div
t
: 1 s /div
0V
0A
IF
Reverse recovery waveform
3.3 Switching characteristics
Figure 4 shows the turn-on, turn-off and reverse

recovery waveforms. These waveforms, exhibiting neither noise nor the generation of a large surge voltage,
are free of problems.
4. Package Structure
High reliability and high thermal conductivity (low
thermal resistance) are required for the high-power
3.3 kV IGBT Modules
52
modules used in high-power inverter system. The

reduction of current imbalances among chips inside
a module and the reduction of heat generation in a
package are also important issues for achieving large
currents.
4.1 General internal structure of package
Figure 5 shows the general internal structure of a

3.3 kV IGBT module.
To enhance the heat dissipation capability of the insulating substrate in a 3.3 kV module, instead of the alumina and silicon nitride substrates typically used with
low-voltage modules, an AlN substrate having 2.5 to 8
times the thermal conductivity was used. The result
realized the low thermal resistance shown in Table 1.
A copper (Cu) base is typically used as the base
material in low-voltage modules. With the 3.3 kV
IGBT modules, an AlSiC base is used to ensure high
reliability. AlSiC is a composite of Al and SiC, and
because its coefficient of thermal expansion is close to
that of the AlN substrate, the heat cycle lifetime and
power cycle lifetime are several times higher than in
the case of a Cu base.
4.2 Improved structure of main terminal
In the design of the main terminal, the following

three items are very important.
(a) Reduction of internal inductance
(b) Reduction of current imbalance of chips on each
insulating substrates
(c) Reduction of stress at contact areas with insulating substrates (leading to higher heat cycle
and power cycle lifetimes)
The main terminal was changed to the same new

structure as used with high-power IGBT modules.
Figure 6 shows the main terminal before and after the
improvement.
(1) Reduction of internal inductance
The internal inductance was reduced by shortening
the length of the main terminal lead as much as possible, arranging the conducting areas of the collector
and the emitter leads vertically, and by actively using
the mutual interaction of the magnetic field.
Measured results show that the internal inductance has been reduced from 30 nH for a conventional
terminal to 20 nH. According to the value of the internal inductance L and the current turn di/ dt during switching, a voltage DV = L (di/ dt) is generated
between the chip and the contact area of the terminal.
Therefore a reduction in internal inductance reduces
the overvoltage applied to the chip.
(2) Reduction of current imbalance between insulating substrates
Because of the main terminal arrangement of the
module, the insulating substrate is separated into a
substrate directly below the emitter terminal and a
substrate directly below the collector terminal. These
are connected in parallel with the shortest possible
lead, thereby forming a structure in which a current
imbalance between insulating substrates occurs easily.
Fig.7 Measured current distribution of chips on each insulating
substrates
Emitter Termial
Collector terminal
Insulating
substrate 2
Chip current on
insulating substrate 1
Fig.5 General internal structure of 3.3 kV IGBT module

Main terminal
Silicon gel
Solder
Chip current on
Insulating
substrate 1
Wire bonding
Chip
Solder
Metal
AlN substrate (insulating substrate)
Insulating
substrate
Metal
Base
Solder
AlSiC base
Fig.6 Main terminal before and after improvement
3.3 kV 1.2 kA IGBT module

*Chip is mounted on insulating substrate
Chip current on
Chip current on
Inductance
30 nH
(per terminal)
(a) Before improvement
53
Inductance
20 nH
(per terminal)
VCE : 500 V/ div

I C : 100 A/ div
t : 5 s/div
(b) After improvement
Fig.8 Tj power cycle test results
Fig.9 Tc power cycle test results

107
108
106
Lifetime (cycles)
109
Lifetime (cycles)
107
106
2 s on/18 s off
105
More than 20,000

cycles
(without damage)
105
Lifetime of module having
Al2O3 substrate and Cu base
(where F(t)=20%)
104
103
T j
Tj
104
Tc
2s
18 s
102
F (t)= 1% line
3.3 kV1.2 kA
(AlN substrate and AlSiC base)
10
20
103
102
10
20
50
100
50
100
200
T C (K)
200
T j (K)
The path of current flow in the emitter terminal and

the collector terminal were analyzed. The structure
was designed such that the currents would be equal.
Figure 7 shows waveforms of the current distribution measured of chips on each insulating substrates.
The measured results and the distribution of current of
chips on each insulating substrates were good.
(3) Reduction of stress at contact areas with insulating substrate
To reduce the stress at contact areas with the insulating substrate, an optimized heat treatment is applied to soften the material of the main terminal. Also,
the main terminal is integrally formed with the terminal case so as to lessen the generation of stress on the
main terminal.
cycle test of an IGBT module, Fuji Electric has verified

that the DTj power cycle capability is determined by
the lifespan of the solder underneath the chip and aluminum wire junction of the chip.
Figure 8 shows the DTj power cycle test results for
3.3 kV modules. The 3.3 kV modules, as in the case
of the 1.2 kV and 1.7 kV high-power IGBT modules,
use highly rigid Sn-Ag solder underneath the chip. By
equalizing the current of chips on each the insulating
substrates, DTj power cycle test results that are equivalent to those of the 1.2 kV and 1.7 kV high-power
IGBT modules could be verified.
5.2 Tc power cycle test results
As shown in Fig. 9, a power cycle capability of more

than 20,000 cycles was verified under the condition of
DTc = 80 K. The 3.3 kV modules use an AlSiC base and
thus have at least 3 times the DTc power cycle capability as in the case of a copper base.
5. Ensuring Power Cycle Capability
6. Product Lineup
High-voltage modules are required to provide high

reliability for their intended applications. In the
marketplace, the reliability considered to be important is the power cycle capability. A power cycle test
(intermittent power-on test) is performed by applying energized and isolated electrical loads to an IGBT
module with its heat dissipating fin temperature held
constant, and causing the junction temperature Tj of
the IGBT chip to rise and fall so as to generate thermal
stress, and the test is carried out until the thermal
stress damages the chip. Types of power cycle tests include the DTj power cycle test that causes the junction
temperature to rise and fall with a relatively short period and the DTc power cycle test that causes the case
temperature Tc to rise up and fall to a certain temperature with a long period.
At present, a 3.3 kV 1.5 kA IGBT module having the same package size as the 3.3 kV 1.2 kA IGBT
(190 mm 140 mm) but with larger IGBT and FWD
chip sizees are being developed. A 3.3 kV 800 A IGBT
module having a 130 mm 140 m package size has
been developed. The target specifications for each
3.3 kV IGBT module are listed in Table 1.
5.1 Tj power cycle evaluation results
From module analysis performed after a power
3.3 kV IGBT Modules
7. Postscript
Fuji Electric has recently developed a 3.3 kV
1.2 kA IGBT module as a high-power IGBT module in
a package. Compared to the module characteristics
prior to improvement, internal inductance has been
reduced by 33% and the current flow of chips on each
insulating substrates showed good uniformity. Power
cycle tests were carried out with this module and sufficient durability was verified. 3.3 kV 1.2 kA IGBT module is planned for commercial release in 2010.
54
References
(1) Nishimura, T. et al. High-power IGBT Modules. Fuji
Electric Review. 2008, vol.55, no.2, p.51-55.
55
(2) Morozumi, A. et al. Reliability of Power Cycling for

IGBT Power Semiconductor Module. Conf. Rec. IEEE
Ind. Appl. Conf. 36th. 2001, p.1912-1918.

Kouta Takahashi Shinichi Yoshiwatari Yusuke Sekino
ABSTRACT
Fuji Electric is developing a series of products that use the latest generation V-Series IGBTs. V-Series IGBT
modules realize lower chip loss and improved package heat dissipation to achieve a smaller IGBT module size and
higher power density. The chip and package characteristics have also been improved to enhance reliability and
guarantee the maximum temperature of 175 C. For these high power density and highly reliable V-Series IGBTs,
Fuji Electric has developed new packages, such as a large capacity 2-in-1 package and a small sized PIM (Power
Integrated Module) package. The product lineup will be expanded up to 1,700 V.
Energy savings and the reduction of CO2 emissions

have been seen as important challenges for the industrial sector in recent years. As a result, the demand
for inverters is continuing to increase. In consideration of power dissipation, switching capability, drive
circuit design and the like, the power semiconductors
used most commonly in inverters are IGBTs (insulated
gate bipolar transistors).
Fuji Electric developed a line of commercial IGBTs
in 1988, and since then has responded to market
needs by steadily releasing successive generations of
new models that realize low loss and a compact size.
Recently, Fuji Electric has realized even lower levels of loss and a more compact size with its lineup of
V-Series IGBT modules that incorporate the latest
generation of IGBTs. Also, the V-Series of IGBT modules feature enhanced chip and package characteristics
that improve reliability and enable the guarantee of
(non-continuous) operation at chip junction temperatures of Tj = 175 C when the inverter is in an emergency operation.
This paper discusses the reliability and product
lineup of Fuji Electrics V-Series of IGBT modules.
2. IGBT V-Series Characteristics
(1)
Figure 1 shows the changes in chip size for successive generations of the IGBT chips installed in commercially available IGBT modules. Fuji Electric has
reduced the chip size by 50% over 10 years, and the
V-Series further shrinks the die size by an additional
30% compared to the U-Series. With the V-Series,
the use of a silicon nitride substrate (SiN) having good
thermal conductivity as the insulating substrate, in

addition to chip improvements, improves the heat dissipation performance and increases the power density
of the IGBT module. These improvements contribute
significantly to reducing the size and lowering the cost
of the entire inverter system. Also, with the V-Series,
the optimization of the MOS (metal-oxide-semiconductor) gate structure facilitates control of the abrupt
behavior of the IGBT current and FWD (free-wheeling
diode) voltage during turn-on switching, which is a
cause of EM (electromagnetic) noise(2). Accordingly,
the V-Series of IGBT modules facilitate EM noise reduction and inverter design.
3. Reliability of V-Series of IGBT Modules

Not only are the V-Series of IGBT modules compact in size and have good noise controllability, but
they are also highly reliable. With the V-Series, noncontinuous operation is guaranteed during instantaneous emergency operation at temperatures of up to
Fig.1 Reduction of IGBT chip size
Chip area (a.u.)
(when the 1,200 V L-Series is set as 100)
1. Introduction
120
L-Series
1,200 V/50 A IGBT
100
N
80
60
40
20
0
1985
L-Series
600 V/50 A IGBT
1990
1995
2000
U
V
2005
2010
(Year)
56
Fig.2 High-temperature switching waveform of 1,200 V

V-Series IGBT
VGE
IGBT turn-off waveform
tures of up to 150 C, long-term reliability of the chips

breakdown voltage, and reliability of the package,
including solder and wire bonding, are essential. The
long-term reliability of a chips breakdown voltage is
usually verified with an accelerated aging test in which
a DC voltage is applied to a chip while being heated
Fig.4 1,200 V V-Series power cycle test results
Tj
T j
Tc
t off = 18 s
t on = 2 s
Ic
Tj Power cycle conduction pattern and temperature change

(Tj : Chip junction temperature, Tc : Case temperature, Tf : Cooling fin temperature)
1010
*Dashed lines indicate
estimated lifetimes.
109
108
Tjmin=25 C
107
106
105
104
Tj max =150 C
0
20
RG = 1.1 , VG =+ 15/ 15 V
VCE
Breakdown voltage BVr (V)
1,550
EP2
UU4 series
EP2
V series
EP2XT
EP3
EP3
50 A
EP3XT
100 A
150 A
1,530
New Dual series

1,200 V
1,520
150 A
1,510
200 A
S series
Standard
UU4 series
New Dual
V series
New Dual XT
300 A
1,500
No. of samples : 10
0
1,000
2,000
Time (h)
57
35 A
75 A
1,540
1,490
100
S series
25 A
FWD Tj = 175 C Vr = 960 V
80
PIN series
1,200 V
15 A
Fig.3 Breakdown voltage reliability test results for 1,200 V

V-Series FWD
60
Fig.5 1,200 V V-Series PIM series and New Dual series
10 A
1 s
40
Tj (C)
IC
VCC =800 V, IC = 300 A (twice of rated current)
Power chip
T c
Tf
Lifetime (cycles)
175 C, and continuous operation is guaranteed during

normal operation conditions at temperatures of up to
150 C. Compared to the U-Series, this is an increase
of 25 C for each operating state. To make this possible, Fuji Electric improved the reliability and switching capability of the IGBT chip during high-temperature operation, and also improved the package reliability, including the solder and Al bond wires. Details of
these accomplishments are described below.
To guarantee operation during instantaneous
emergency conditions at temperatures of up to 175 C,
reverse blocking capacity must be maintained even
at high temperatures and the chip must be capable
of stable switching even at large currents. We have
verified that with the V-Series, an IGBT with no lifetime killer and a FWD that uses electron irradiation
as carrier lifetime control enables a stable breakdown
voltage to be maintained, without thermal runaway, at
temperatures of up to 200 C. Also, to verify the hightemperature switching capability, we have verified
IGBT switching and FWD switching at 200 C. Figure
2 shows IGBT turn-off waveforms for a chip junction
temperature Tj = 200 C at twice the rated current.
Based on these results, the V-Series is able to guarantee operation during instantaneous emergency conditions at temperatures of up to 175 C.
To guarantee during normal operation at tempera-
450 A
3,000
600 A
Fig.6 Future expansion of the V-Series lineup

Max. rated current
Main
(A)
terminal
10
15
25
35
50
75
100
150
PIM (EP2XT)
200
225
300
400
450
EP2/ PC2
900
1,200
1,400
2,400
3,600
EP3/ PC3
PIM (EP3XT)
Pins
550
600
6-in-1 (PC2XT)
6-in-1 (PC3XT)
Conventional
terminals
2-in-1 (DualXT)
Dual
New model 2-in-1
Screws
Under development
Solder-free
terminals
Control terminal
spring
Screws
2-in-1 (DualXT)
Spring Dual
(a) 1,200 V series

Max. rated current
Main
(A)
terminal
10
15
25
35
50
75
100
150
PIM (EP2XT)
200
225
300
EP2/ PC2
400
450
600
900
1,200
EP3/ PC3
PIM (EP3XT)
Pins
6-in-1 (PC2XT)
Conventional
terminals
Dual
2-in-1 (DualXT)
Screws
(b) 600 V series
in an electronic oven. Figure 3 shows the results of a

FWD reliability test. For a Tj (chip temperature) of
175 C, we verified that sufficient reliability is exhibited, without degradation of the breakdown voltage,
even after 80% of the rated voltage had been applied
for 3,000 hours. An IGBT having a similar junction
termination as the FWD was found to have the equivalent reliability.
Typically, in a power module, the package reliability is affected by repeated thermal cycles, wherein
a chip becomes hot while in the on-state and then
becomes cool during the off-state, and as a result, the
solder and wire bonding contact underneath the chip
are stressed by thermal expansion and contraction.
Sufficient capacity to withstand this stress is necessary. Figure 4 shows the results of a V-Series power
cycle test. The Tjmin = 25 C line indicates the chips
lifetime (number of cycles) when the cooling fin temperature is held constant and the maximum chip junction temperature is varied during the load cycle. For
example, DTj = 30 C represents a load cycle in which
the cooling fin temperature is 25 C and the maximum
chip temperature is 55 C. The Tjmax = 150 C line indicates the chips lifetime when the maximum chip
temperature is held constant and the cooling fin temperature is varied during the load cycle. For example,
when DTj = 30 C, the cooling fin temperature is 125 C
and the maximum chip temperature is 150 C during

the load cycle.
As shown in the graph, even if DTj is the same,
higher cooling fin and chip temperatures result in
shorter lifetimes. In other words, the Tjmax =150 C
line represents the worst case operating conditions, but
even under these conditions, it was still possible to verify that the V-Series provides sufficient reliability with
a lifetime of more than 50,000 cycles at DTj =100 C.
Based on these results, the improvements to the chip
and package reliability guarantee the normal operation of the V-Series at temperatures of up to 150 C.
Thus, the reliability of the V-Series during hightemperature operation has been improved significantly, and this improved reliability contributes greatly
to enhancing the inverter reliability and the degree of
freedom in the thermal design.
4. Lineup of V-Series IGBT Modules

At present, the V-Series is available in PIM (power
integrated module) and New Dual models as shown in
Fig. 5.
The PIM models integrate upper and lower arm
IGBTs and FWDs for 3 phases, a converter diode and
a brake IGBT (7-in-1), enabling a 3-phase AC inverter
to be configured from a single module. As a result, an
58
inverter system can be made more compact in size and

the design can be streamlined. For the V-Series, the
lineup of PIM models was expanded up to a rated current of 150 A.
The New Dual models are provided for use with
rated currents of 200 A and above. These modules
integrate upper and lower arm IGBTs and FWDs for 1
phase (2-in-1). Three chips are used in parallel inside
the module. When power semiconductors are used in
parallel, there is a concern that a current imbalance
will occur in which current is concentrated in some
chips. The New Dual models feature an innovative internal interconnect pattern that results in an extremely easy-to-use device with almost no current imbalance
among chips. For the V-Series, the lineup of New Dual
models was expanded up to a rated current of 600 A.
5. Future Outlook
Fuji Electric plans to continue to expand the
V-Series lineup. Future plans for expansion of the
lineup are shown in Fig. 6. Development of the following new packages is being advanced.
(1) High-power modules (Econo PACK+*1, new 2-in-1)
(2) Compact module (MiniSKiiP*2)
Presently, the market for high-power inverters for
electric power transformer applications such as wind
power generators, traction applications and the like
is expanding, and Fuji Electric plans to develop highpower modules to meet the needs of these markets.
We plan to develop a line of EconoPACK+ models up
to rated values of 550 A and 1,200 V, and a new 2-in-1
model up to rated values of 1,400 A and 1,200 V. Also,
*1: Econo PACK is a trademark or registered trademark of
Infineon Technologies AG.
2: MiniSKiiP
is a trademark or a registered trademark of
*
SEMIKRON.
59
for low-power applications, Fuji Electric is developing

the MiniSKiip, which is solder-free, is easy to assemble
and has equivalent functions as the PIM models but
with a significantly smaller size. We plan to develop
a line of MiniSKiip models rated at 8 to 100 A and
1,200 V.
Additionally, Fuji Electric is also developing a
1,700 V chip, and plans to deploy this chip in successive high-power applications such as the new 2-in-1
and New Dual models.
6. Postscript
This paper has described the features and lineup
of Fuji Electrics V-Series IGBT modules that use the
latest generation IGBT chips. Through improved chip
technology and package heat dissipation performance,
the V-Series achieves a smaller size and will significantly aid in the production of more compact inverters.
Moreover, improved reliability of the chip and package
enables the operation to be guaranteed at temperatures up to 175 C. Consequently, the V-Series makes
an important contribution to enhancing the inverter
reliability and the degree of freedom in the thermal
design.
In the future, Fuji Electric plans to deploy the
V-Series technology in 1,700 V high-power modules
and compact modules, and will to continue to support
the needs of our customers.
References
(1) Onozawa, Y., et al. U-Series IGBT Modules (1,200 V),
Fuji Electric Journal. 2002, vol.75, no.10, p.563-566.
(2) Igarashi, S., et al. Analysis and Reduction Methods of
EMI Radiation Noise from Converter Systems, IEEJ
Transactions on Industry Applications. 1998, vol.118D, no.6, p.757-766. (Japanese).

Naoki Shimizu Hideaki Takahashi Keishirou Kumada
ABSTRACT
Fuji Electric has developed a series of intelligent power modules for industrial applications, known as V-Series
IPMs. This product combines high-performance 6th-generation V-chip technology for the IPMs with a new control
IC to realize lower loss and a smaller package size. The short-circuit protection function was made to operate at
faster speeds and the trade-off relation between conduction loss and short-circuit withstand capability was improved
to reduce switching loss. Additionally, the new control IC and the package were optimized to reduce turn-on loss
and improve radiation noise characteristics. In addition to the conventional protection functions, a new function
that outputs different alarm pulse widths for each alarm factor is also provided. Ground-fault protection can also be
provided even in a small package.
1. Introduction
An IPM (Intelligent Power Module) is an intelligent power device that combines a standard IGBT
module containing an IGBT (Insulated Gate Bipolar
Transistor) chip and a FWD (Free Wheeling Diode)
with a drive IC equipped with drive and protective
functions. IPMs are used in machines in a wide
range of fields, but mostly in motor-driven equipment (such as NC (Numerical Control) machine tools,
general-purpose inverters, servos, air conditioners and
elevators) and power supply devices (such as UPS
(Uninterruptible Power Supplies), and PCS (Power
Conditioning Systems) for solar energy generation),
and are required to have compact size, high efficiency,
low noise, long life and high reliability.
Fuji Electric, in 1997, devised the industrys first
internal IGBT chip over-temperature protection function, and developed the R-IPM Series that aimed
to achieve higher reliability with an all-silicon design that reduced the number of parts. In 2002, Fuji
Electric developed the R-IPM3 Series in which the
chip structure was changed from PT (punch through)
to NPT (non-punch through). In 2004, Fuji Electric
integrated a newly developed NPT-type trench-gate
structure IGBT and a new structure FWD into the
U-IPM Series which improves the tradeoff relation
between lower switching loss and radiation noise.
Recently, Fuji Electric has developed an FS (field
stop) type V-Series IGBT chip (V-IGBT) having a
trench gate structure that achieves even lower loss and
lower input capacitance. Fuji Electric has also developed a drive IC that uses finer line widths to realize a
more compact size, and features improved temperature
characteristics with less variation among chips. These
technologies and packages have been optimized to develop the V-Series IPM (V-IPM) that is housed in a
new compact package and that features an improved
tradeoff relation between total dissipation loss and radiation noise. This paper describes Fuji Electrics new
V-IPMs.
2. Product Concept and Product Lineup

Development concepts for the V-IPM are listed
below.
(1) Reduction of total dissipation loss
(2) Improvement of tradeoff relation between switching loss and radiation noise
(3) Shorter dead time
(4) Individual outputs for each cause of alarm
(5) More compact and thinner packages
(6) Provision of ground fault protection for small capacity packages
(7) Compliance with RoHS*1 directive
Details of these concepts are explained in chapters
3 to 6 below.
Table 1 shows the V-IPM product lineup. For
V-IPM devices, the current capacity has been increased
compared to the previous devices, and 600 V-rated
devices have rated currents of 20 to 400 A, and
1,200 V-rated devices have rated currents of 10 to
200 A. The four types of newly developed packages are
all RoHS compliant.
Table 2 lists characteristics and internal functions
of the V-IPM devices. A small capacity package (P629)
newly realizes short-circuit protection (ground fault
protection) function in its upper arm. Also, throughout
*1: RoHS: EU Directive on the restriction of the use of certain hazardous substances in electrical and electronic
equipment
60
Fig.1 Comparison of IGBT chip cross-sectional structures

(660 V-IPM)
Table 1 V-IPM product lineup

(a) Product lineup
Rated Number of
20 A 30 A 50 A 75 A 100 A 150 A 200 A 300 A 400 A
voltage elements
P626
P630
6 in 1
7 in 1
Gate
P631
Rated Number of
10 A 15 A 25 A 35 A 50 A
voltage elements
P629
6 in 1
1,200 V
75 A 100 A 150 A 200 A
P626
P630
6 in 1
7 in 1
P631
(b) Package types
P629
package
4
P626
package
10 15
1
P
PU
P630
package
9 1319
n+
n drift layer
p+ collector layer
p+ collector layer
9 1319
LWH
(mm)
49.570 12
50 87 12
84 128.5 14
110 142 27
Collector
Collector
U-IPM (5th generation)
NPT, trench
IGBT
technology
Table 2 V-IPM characteristics and internal functions
P626
P630
P631
Chip
Low
Over Short- over
Drive current circuit temper- voltage Alarm
circuit protec- protec- ature protec- output
tion protection
tion
tion
Lower
Upper
Lower
Upper
Lower
Upper
Lower
: Improvement from existing product
the lineup, a pulse output width that changes according to the alarm cause makes it possible to specify the
cause of the alarm.
3. Characteristics of V-IPM Power Chips

V-Series IPMs are utilizing 6 th generation
V-IGBT chips. The V-IGBT is an FS-type IGBT having
a trench gate structure formed by using thin wafer process technology with a FZ (floating zone) wafer. Figure
1 compares the cross-sectional structure of various
chips, and Table 3 shows the history of IGBT technology.
The surface structure has been optimized to make
61
V-IPM (6th generation)

FS, trench
Generation
3rd
generation
4rd
generation
IPM
R-IPM
R-IPM3
U-IPM
V-IPM
N-IGBT
T-IGBT
U-IGBT
V-IGBT
IGBT
Wafer
Epitaxial
Structure
PT
Gate structure
Upper
Collector
R-IPM3 (4th generation)

NPT, planer
LWH
(mm)
P629
n+ field stop
layer
p+ collector layer
Table 3 History of IGBT technology (600 V-IPM)
1 4 7 10 16
P
LWH
(mm)
Arm
n drift layer
P631
package
LW H
(mm)
Package
Gate
Gate
n drift layer
B
N
n+
p+
: Upgrade of capacity compared to existing products
Emitter
Emitter
P629
6 in 1
600 V
Emitter
5rd
6rd
generation generation
FZ
NPT
FS
Planar
Trench
Lifetime control
Yes
No
Carrier injection
High
injection
Low injection
Transport efficiency
Low
efficiency
High efficiency
a drift layer have lower resistance and be thinner. As the result, the V-IGBT has the advantage of
lower on-voltage VCE(sat) and improved switching loss.
Additionally, the optimization of the surface structure
and the lower resistance of the drift layer also improve
the controllability of the turn-on di/ dt. The radiation
noise is less than compared to a conventional device.
The IGBT chip for the V-IPM, compared to a chip
for an IGBT module, realizes an improved tradeoff relation between VCE(sat) and turn-off loss Eoff as a result
of a finer surface structure. When VCE(sat) is reduced,
the short-circuit current increases and the short-circuit
withstand capability decreases, and short-circuit protection must engage rapidly in practical applications.
The FWD with improved lifetime control realizes
lower recovery current and soft recovery.
4. Characteristics
4.1 Total dissipation loss
The marketplace requests IPMs to have lower loss.

There are two objectives for reducing the loss, one is
to increase the carrier frequency in order to improve
equipment controllability, and the other is to increase
the output current at the same carrier frequency.
Also, the reduction in loss will lead to reduced cost of

the equipment since less cooling capability will be required than before.
Figure 2 compares examples of dissipation loss during PWM inverter operation. The loss in the V-IPM is
approximately 27% lower than that of the R-IPM, approximately 17% lower than that of the R-IPM3, and
approximately 10% lower than that of the U-IPM.
This reduction in loss is due to an improved tradeoff relation between VCE(sat) and Eoff, and a lower
turn-on loss Eon. These techniques for reducing loss
are described below.
Also, optimization to improve the VCE(sat) tradeoff

is being implemented with the goal of maintaining the
same level of surge voltage as before, but with lower
loss.
4.3 Lower turn-on loss
As a result of lower IGBT input capacitance, a new

drive method for the control IC and improved temperature characteristics, as shown in Fig. 4, the tail of the
collector-emitter voltage VCE becomes shorter and Eon
is reduced by approximately 35%.
4.4 Comparison of radiation noise
4.2 Improvement of VCE(sat) and Eoff tradeoff relation
Static loss and Eoff account for more than 50% of

the total loss in an IGBT chip. However, as shown in
Fig. 3, tradeoff relations exist among VCE(sat), which determines the static loss of the IGBT chip, Eoff and the
short-circuit withstand capability, This tradeoff must
be optimized. Since the IPM has a short-circuit protection function, by increasing the speed of the short-circuit protection, the IGBT chip short-circuit withstand
capability can be diverted to reduce loss. With the
V-IPM, the achievement of higher speed short-circuit
protection resulted in an improved tradeoff relation
between VCE(sat) and Eoff, and lower loss.
A tradeoff relation exists between radiation noise

and Eon, whereby loss can be reduced when di/dt is
increased and radiation noise can be reduced when di /
dt is decreased. Figure 5 shows an example of relative
comparison test results as measured by Fuji Electric.
Using a dummy load testing device, measurement was
made during acceleration and deceleration operations.
By employing the method for reducing turn-on loss
described in section 4.3 and by optimizing the internal
circuit wiring pattern in the package to reduce the
radiating area, the peak value of radiation noise was
approximately 3db lower than that of a conventional
product. With this V-IPM, both lower Eon and lower
radiation noise were realized.
Fig.2 Comparison of total dissipation loss (simulation results)

4.5 Shorter dead time interval
(In the case of 150 A)
Total dissipation loss (W)
100
Recovery loss (FWD)

Static loss (FWD)
Turn-on loss (IGBT)
Turn-off loss (IGBT)
Static loss (IGBT)
80
T j = 125, E d = 300 V,
V CC = 15 V, I o = 50 Arms
cos =0.8, = 1
88
75
70
60
60
52
46
42
40
In an inverter circuit, a dead time interval is

Fig.4 Turn-on waveform
70
Conditions : Ed =300 V, VCC =15 V, IC =150 A, Tj =125 C
62
56
50
U-IPM
V-IPM
38
Improved
VCE tail
20
R-IPM R-IPM3 U-IPM V-IPM R-IPM R-IPM3 U-IPM V-IPM R-IPM R-IPM3 U-IPM V-IPM
f c =5 kHz
f c = 10 kHz
E on = 10.3 mJ
f c = 15 kHz
E on = 6.7 mJ
Channel 1 : VCE (100 V/ div), Channel 2 : Collector current IC (50 A/ div)
Fig.3 IGBT tradeoff relationships

Fig.5 Comparison of radiation noise (results of relative comparison test at Fuji Electric)
VCE(sat)
IGBT
short-circuit
withstand
capability
Improved
tradeoff
Improvement due to
optimization of 6th
generation V-IGBT
Key point of
improvement
Turn-off loss
Reduced turn-off surge as

a result of package optimization
90
Noise level (dB)
High-speed short-circuit
protection
(IPM-specific
improvementeffect)
U-IPMP611 package
80
70
60
50
40
30
30
V-IPMP630 package
50
70
90
Frequency (MHz)
110
62
provided in order to prevent overlapping of the onintervals of the upper and lower arms of the IPM.
Shortening of the dead time interval is important for
reducing waveform distortion and rotational unevenness. With the V-IPM, the switching time interval has
been optimized by improving the temperature characteristics and reducing fluctuation during switching
by the control IC. As a result of these techniques, the
V-IPM realizes a minimum dead time interval of 1 s
at its IPM input section.
5. Protection Functions
5.1 Short-circuit protection
As described in section 4.2, the tradeoff relation between VCE(sat) and Eoff in the IPM can be improved by
diverting the short-circuit withstand capability of the
IGBT chip to loss reduction. To realize this improvement, the speed at which the short-circuit protection
operates must be increased. To speed up the protection circuit, the design of the V-IPM must ensure that
the short-circuit protection function operates correctly
with a shorter filter time. For this purpose, the sense

voltage for the collector current monitor, which performs recognition sensing for the protection circuit,
must be optimized.
With the IGBT, the sense ratio is adjusted to
reduce the sense current and to stabilize the sense
voltage during switching. Also, with the control IC,
by changing to a new drive method from a gate resistance method based on the highly temperature
dependent on-resistance of a MOSFET (Metal-OxideSemiconductor Field-Effect Transistor), the sense voltage is limited especially at high temperature conditions. As a result of these improvements, we successfully improved the tradeoff relation between loss and
Fig.6 External appearance of package
P630
Table 4 Alarm pulse width for each alarm cause
Alarm cause
P626
Alarm pulse width
Over current protection

(including short-circuit protection)
2 ms (standard value)
Low voltage protection
IGBT Chip over temperature

protection
P629
Fig.7 V-IPM block diagram

P
VccU
Vin
GNDU
VccV
VinV
GNDV
VccW
VinW
GNDW
Vcc
Driver
IC1
Driver
IC1
Driver
IC1
ALMV
VinV
GNDV
VccW
ALMW
VinW
GNDW
VinX
VinY
Driver
IC2
VinZ
VinZ
GND
Driver
IC
Driver
IC
Driver
IC
Driver
IC
Driver
IC
Driver
IC
GND
ALM
(a) Small capacity IPM P629 (6 in 1)
63
ALMU
VinU
GNDU
VccV
Vcc
VinX
VinY
VccU
ALM
(b) Medium and large capacity IPM P626P630P631 (6 in 1)
short-circuit withstand capability.
packages have RoHS compliant structures.
5.2 Separate output for each alarm cause
7. Block Diagram of IPM
A conventional IPM only has a single alarm pulse

width of 2 ms, from which the cause of an alarm cannot be identified, but with the V-IPM, the alarm pulse
width changes for each alarm cause, enabling the
cause analysis to be performed more quickly when
troubleshooting.
The alarm pulse widths for each V-IPM alarm
cause are listed in Table 4.
6. Package
Figure 6 shows the external appearance of several
packages. P629 is a small capacity package, P626 is a
medium capacity and small size package, P630 is a medium capacity, thin package, and P631 (not shown) is a
large capacity package.
The newly developed IPM has an external package shape designed in response to requests for thinner modules. A conventional package uses overhead
crossing type bar wiring, and the wiring distance was
shortened. With the new package, the internal circuit
wiring is fabricated only from aluminum wire and the
copper pattern on the insulating substrate to realize
a thin package. Additionally, the internal inductance
was reduced due to the mutual inductance effect from
parallel PN lines. The resultant effect is that radiation
noise is decreased, and turn-off surge does not become
excessively large.
An additional advantage is that since a 50 A / 600 V
capacity is realized with the thin package of the P629,
the height on the device is uniform, and the P629 can
be used together with the P626. Moreover, the package design ensures sufficient insulation distance to
ground and inter-phase insulating distance, and therefore, the insulating distance can be ensured without
any special insulation design on the device side. All
Figure 7 shows block diagrams of IPMs. With a

conventional small capacity type, current is detected
with a shunt resistance that has been inserted into the
N line, and therefore short-circuit protection could be
implemented in the lower arm only. With the V-IPM,
since all IGBTs use the sense current for detection,
short-circuit protection of the upper arm device can
be implemented. The P626, P630 and P631 packages
are also provided with an alarm terminal on the upper arm, and alarm signals can be transmitted to the
device side. Because P629 is designed to be compatible
with the installation of previous models, an alarm signal terminal is not provided for the upper arm.
8. Postscript
Fuji Electrics FS-type V-Series IGBT chips
(V-IGBT) having a trench gate structure and the
V-Series IPM (V-IPM) that incorporates a new control IC into a new package have been introduced. The
V-IPM realizes a compact size and high efficiency, and
is able to meet market expectations. In the future,
Fuji Electric intends to continue to concentrate on
product development to satisfy market needs.
References
(1) Onozawa, Y. et al. Development of the next generation 1,200 V trench-gate FS-IGBT featuring lower EMI
noise and lower switching loss. Proc. of ISPSD 07.
p.13-16.
(2) Onozawa, Y. et al. Development of the 1,200 V FZdiode with Soft Recovery Characteristics by the New
Local Lifetime Control Technique. Proc. of ISPSD 08.
p.80.
64
Yasuhiko Oonishi Akihiko Ooi Takayuki Shimatou
ABSTRACT
600 V-class superjunction (SJ) MOSFETs (package : TO-220) with a maximum on-resistance of 0.16 have
been fabricated by using multi-epitaxial growth technology which has an excellent capability for controlling the doping
concentration. By optimizing the doping concentration in the SJ structure, the fabricated SJ-MOSFET achieves an
approximate 70% reduction in specific on-resistance compared to that of a conventional MOSFET SuperFAP-E 3.
This is the industrys highest level of specific on-resistance, and its value exceeds the theoretical limit for conventional
MOSFETs. The avalanche withstand capability of the fabricated SJ-MOSFET has been also improved over the rated
current by optimizing the doping profile of the SJ structure in the depth direction and the thickness and resistivity of
the n-buffer layer.
Advanced Technology Laboratory, Fuji Electric Holdings

Co., Ltd.
65
2. Technical Trends of Power MOSFETs

Figure 1 shows the trends of Ron A and Ron QGD
for 600 V-class power MOSFETs. As described above,
a theoretical limit exists for the Ron A of the power
MOSFET, and therefore development efforts until now
have focused on how to make the Ron A approach the
silicon limit as closely as possible. In order to reduce
the Ron A, the drift resistance, which forms the silicon
limit, was reduced and other resistance components
were also reduced. The latter entails reducing the
channel resistance and JFET (Junction Field-Effect
Transistor) resistance by improving the cell density
Fig.1 Power MOSFET RonA and RonQGD trends
1.2
1.2
Conventional MOSFET
Quasi-plane junction
technology
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.0
1985
SJ-MOSFET
1990
1995
(Year)
2000
2005
0.2
0.0
2010
Normalized Ron QGD (a.u.)
In recent years, as concern for protecting the

global environmental has heightened, in the IT (information technology) sector, green IT has been attracting attention as a way to achieve power savings.
In order to reduce the power loss of IT equipment,
the power converters used with the IT equipment
must be made more efficient. Consequently there is
strong demand for low-loss power MOSFETs (MetalOxide-Semiconductor Field-Effect Transistors). The
power MOSFETs used in power converters operate as
switching devices and their associated dissipation loss
consists of conduction loss while the power MOSFET
is ON, and switching loss when the power MOSFET
turns on and turns off. Generally, the conduction
loss is dominant in applications having a low switching frequency, and the switching loss is dominant in
applications having a high switching frequency. The
on-resistance normalized to a unit area (Ron A) is used
as a figure of merit for the conduction loss, and the
gate-to-drain charge normalized to the on-resistance
(Ron QGD) is used as a figure of merit for the switching
loss. Thus, in order to reduce the dissipation loss of a
power MOSFET, minimization of these figures of merit
is strongly requested.
However, the breakdown voltage and Ron A are associated with a theoretical limit that is determined by
the material (in the case of silicon, this theoretical limit is known as the silicon limit), and it is not thought
to be possible to obtain a Ron A value that exceeds this
theoretical limit. The super junction (SJ) structure is
an innovative breakthrough that overcomes this limi-
tation, and is attracting attention due to its ability to

realize dramatically lower Ron A(1)(2).
This paper describes the fabrication method and
characteristics of Fuji Electrics newly developed
600 V-class SJ-MOSFET that realizes the lowest onresistance in the industry and a high level of inductive
load avalanche withstand capability.
Normalized Ron A (a.u.)
1. Introduction
of the surface MOSFET area and optimizing the cell

structure, and the former entails reducing the drift
resistance by optimizing the resistivity and thickness of the drift layer that ensure the inductive load
avalanche withstand capability and the breakdown
voltage. With a structure that uses a p-well region
to ensure the inductive load avalanche withstand
capability, there is a limit to extent with which the
drift resistance can be reduced, but the application of
quasi-plane junction technology enables this limit to
be lowered(3). With quasi-plane junction technology,
instead of using a p-well structure, the drift resistance
is reduced by narrowing the distance between p-base
regions without increasing the JFET resistance, and
by optimizing the p-base shape, an inductive load
avalanche withstand capability equivalent to that of
a p-well structure is guaranteed. By applying quasiplane junction technology, the Ron A of a conventional
MOSFET has been improved to 110% of the silicon
limit, and the Ron QGD has also been improved significantly with the reduction in Ron A and the narrower
distance between p-base regions. Fuji Electric has
applied this quasi-plane junction technology to commercialize the SuperFAP-G Series which is an easyto-use successor of the SuperFAP-E3 Series(4).
Recently, the SJ-MOSFET, which breaks through
the silicon limit, has been attracting attention. As
shown in Fig. 2, the SJ-MOSFET replaces the p-type
and n-type regions in the drift layer of a conventional
MOSFET with alternating regions of p-pillars and
n-pillars, and because the impurity concentration of
the n-type regions can be increased, the Ron A can be
reduced dramatically. Moreover, since the Ron A can
be reduced, the active area can be made smaller for the
same on-resistance, thereby enabling the Ron QGD to be
reduced as well.
3. SJ-MOSFET Development
3.1 Multi-epitaxial growth technology
The SJ structure enables a dramatically lower

value of Ron A, but because it is a charge compensation
structure, it has the disadvantage of being unable to
maintain its breakdown voltage unless a charge balance can be maintained between n-type and p-type
regions. If the charge balance collapses, the breakdown voltage drops significantly and the rated voltage
will not be maintainable. When fabricating an SJMOSFET, the impurity concentrations of the n-type
and p-type regions must be controlled precisely. To reduce deviation of the breakdown voltage, the impurity
concentration in the depth direction of the p-type region must be distributed. Therefore, the SJ structure
is fabricated with a method of multi-epitaxial growth
in which the processes of introducing impurities into
a certain area by ion implantation, which has excellent impurity concentration control performance, and
epitaxial growth are performed repeatedly(5). Figure 3
shows the process flow of SJ structure fabrication by
the multi-epitaxial growth method. First, an n-type
layer having a low impurity concentration is grown
epitaxially on an n-type substrate with consideration
given to the thickness of the buffer layer (step 1). Next
after completion of the fabrication of that layer, phosphorous (P) and boron (B) ions are injected into regions
Fig.3 SJ structure fabrication process flow using multi-epitaxial growth method
n-type epitaxial layer
Step 1
n-type substrate
P+
B+
P+
B+
P+

Step 2
n-type substrate
Fig.2 Structure of conventional MOSFET and SJ-MOSFET

Gate
Source
Gate
Source
Step 3

n-type substrate
n+
p+
n+
p+
n p
n
n
Step 4
n+
n+
Drain
Drain
(a) Conventional MOSFET
(b) SJ-MOSFET
n
n-type substrate
66
that will become n-type and p-type regions (step 2),

and then an n-type layer having a low impurity concentration is grown epitaxially (step 3). Because the
impurity concentrations in the fabricated n-type and
p-type regions are determined by the dimensions of the
resist for ion injection and by the accuracy of the ion
injection itself, the impurity concentrations are easy to
control. Steps 2 and 3 are repeated until a certain drift
layer thickness is achieved, and then lastly, thermal
diffusion is used to fabricate consecutive n-type and
p-type regions. Thereafter, an ordinary DMOSFET
(Double Diffused MOSFET) process is used to fabricate
a DMOSFET on the surface of the SJ structure and
create the SJ-MOSFET. Figure 4 shows an SCM (scanning capacitance microscopy) image of a cross-section
Fig.4 SCM image of SJ-MOSFET cross-section
SiO2
p+
of an SJ-MOSFET fabricated by the multi-epitaxial

growth method. From the figure, it can be verified
that an SJ structure, connected to both the p-type and
n-type regions in the depth direction, has been fabricated.
3.2 Improvement of RonA
Using the multi-epitaxial growth method, an SJMOSFET rated at 600 V and 0.16 was developed.
As shown in Fig. 5(a), the Ron A of the developed SJMOSFET is approximately 73% lower than that of the
SuperFAP-E3, which is a conventional MOSFET. This
is the industrys lowest level of Ron A. A comparison of
Ron QGD is shown in Fig. 5(b). The Ron QGD of the SJMOSFET is also approximately 69% lower than that
of the SuperFAP-E3. Moreover, the surface MOSFET
structure of the SJ-MOSFET is based on the same
concept as the SuperFAP-E3 and inherits its ease-ofuse. Figure 6 shows the output characteristics of the
SJ-MOSFET for VGS(th) = 3.0 V.
Fig.6 Output characteristics of SJ-MOSFET
60
VGS =10.0 V
Ratings : 600 V/ 0.16

Room temperature
n
50
8.0 V
7.0 V
I D (A)
40
n-type buffer layer
30
6.0 V
20
Fig.5 Comparison of figures of merit for conventional

MOSFET and SJ-MOSFET
10
5.0 V
0
Normalized Ron A (a.u.)
1.2
10
15
20
V DS (V)
1.0
0.8
73%
reduction
0.6
0.4
Fig.7 Inductive load avalanche waveform of SJ-MOSFET
0.2
0.0
SuperFAP-E3
(Conventional MOSFET)
Ratings : 600 V/0.16

Room temperature
SJ-MOSFET
ID : 10 A/div
VDS : 200 V/div
Time scale : 5 s/ div
(a) Ron A
VDS
Normalized Ron QGD (a.u.)
1.2
1.0
0.8
69%
reduction
0.6
ID
0.4
0.2
0.0
SuperFAP-E3
(Conventional MOSFET)
SJ-MOSFET
(b) Ron QGD
67
Fig.8 Power supply circuit for application evaluation

Power factor improvement circuit
Fig.9 Comparison of power MOSFET dissipation loss

(AC100 V input/400W output)
1.2
Dissipation loss (a.u.)
1.0
0.8
Turn-off
loss
14%
reduction
0.6
0.4
0.2
0.0
Turn-on
loss
Conduction
loss
FMH23N60ES
SJ-MOSFET
3.3 Improved inductive load avalanche withstand capability
It has been noted that the inductive load avalanche

withstand current of an SJ-MOSFET is less than that
of a conventional MOSFET. As a result of an optimized impurity concentration profile and n-type buffer
layer in the depth direction of the SJ structure, the
newly developed SJ-MOSFET guarantees an inductive
load avalanche withstand capability of 150 A /cm2 even
under a balanced charge condition. Figure 7 shows a
typical inductive load avalanche waveform of an SJMOSFET.
4. Application Results
An SJ-MOSFET (rated at 600 V / 0.16 ) fabricated
with the multi-epitaxial growth method and assembled
in a TO-220 package was installed in the power factor
improvement circuit (Fig. 8) of a 400 W-class ATX power supply (server power supply that conforms to the
ATX standard), and the loss and temperature rise were
evaluated. Figure 9 compares the power MOSFET area
loss for an SJ-MOSFET and for the FMH23N60ES
(rated at 600 V and 0.16 ), which has the lowest onresistance among SuperFAP-E3 series products and is
housed in a TO-3P package. Compared to the conventional FMH23N60ES MOSFET, the SJ-MOSFET realizes 16% lower conduction loss and an approximately
14% reduction in total power loss. Moreover, the temperature rise has been confirmed to be approximately
5 C lower, and an approximate 0.5% improvement
in the total power conversion efficiency has also been
verified. Although differences exist between the TO220 and TO-3P packages, simple replacement with an
SJ-MOSFET can expect to yield a lower dissipation
loss and improved efficiency.
5. Postscript
An SJ-MOSFET (in a TO-220 package) that is
fabricated by multi-epitaxial growth technology and
rated at 600 V and 0.16 has been developed. The
SJ-MOSFET features an optimized impurity concentration in the SJ structure and an optimized n-type
buffer layer to achieve the industrys lowest level of
low Ron A, and a high inductive load avalanche withstand capability. In the future, Fuji Electric intends
to develop a product lineup of low-loss SJ-MOSFETs
and to contribute to efforts for protecting the global
environment.
References
(1) Fujihira, T. Theory of Semiconductor Superjunction
Devices. Jpn. J. Appl. Phys. Oct. 1997, vol.36, p.62546262.
(2) Deboy, G. et al. A New Generation of High Voltage
MOSFETs Breaks the Limit Line of Silicon. Proc.
IEDM. 1998, p.683-685.
(3) Kobayashi, T., et al, High-Voltage Power MOSFETs
Reached Almost to the Si limit. Proc. ISPSD 2001, p.
99-102.
(4) Niimura, Y., et al. A Low Loss, Low Noise and Robust
500 to 900 V Class Power MOSFET with Multiple RESURF Guardring Edge Structure. Proc. PCIM. China,
June 2009, p.150-155.
(5) Onishi, Y. et al. 24 mcm2 680 V Silicon Superjunction. MOSFET. Proc. ISPSD. June 2002, p.241-244.
68
High Reliability Power MOSFETs for

Space Applications
Masanori Inoue Takashi Kobayashi Atsushi Maruyama
ABSTRACT
We have developed highly reliable and radiation-hardened power MOSFETs for use in outer space applications
in satellites and space stations. The largest difference between these newly developed Rad-Hard Power MOSFETs
and general-use MOSFETs is that they have excellent durability against high energy charged particles and ionizing
radiation. To provide increased durability, electrical characteristics had been sacrificed in the past. With this device,
however, to provide durability against high energy charged particles, a drift diffusion model was modified so as to
enable simulation of the mechanism. A power MOSFET designed for use in outer space applications and having
the worlds top level performance is realized by providing a thick epitaxial layer with low specific resistance as a
countermeasure to ensure durability against SEB (single event burn-out).
1. Introduction
It is well known that the benefits derived from outer space applications such as communication satellites,
weather satellites, GPS (Global Positioning System)
and earth observation are pervasive in society today.
The electronic devices and switching power supplies installed in artificial satellites are required to be
both highly efficient so that the limited electric power
in outer space can be utilized effectively and to have
a reduced part count so as to ensure reliability as a
system. Accordingly, power MOSFETs (Metal-OxideSemiconductor Field-Effect Transistors) are required
to have low loss and, in a space environment, high
reliability with tolerance(2) to ionizing radiation(1), highenergy charged particles and the like. Particularly, in
outer space, most semiconductor components exhibit
a significant degradation in their electrical characteristics, and products that are commonly used in terrestrial applications are unable to provide guaranteed
reliability in space.
At the time when Fuji Electric began to develop
power MOSFETs for space applications, there already
existed much research(1),(2) into the degradation of power MOSFET characteristics caused by ionization radiation and countermeasures to prevent that degradation.
However, the mechanism of the SEB (single event
burnout) phenomenon, in which high-energy charged
particles cause instantaneous burnout, was unknown.
Consequently, Fuji Electrics 1st generation of spaceuse power MOSFETs sacrificed electrical characteristics for SEB tolerance, but by employing an estimated
SEB mechanism as described below, Fuji Electric has
recently developed a 2nd generation of high-reliability
space-use power MOSFETs having electric characteris Semiconductors Group, Fuji Electric Systems Co., Ltd.
69
tics that are equivalent to those of consumer-use power

MOSFETs. Figure 1 shows the external appearance of
these power MOSFET products, and Table 1 lists their
main characteristics.
The results of this achievement are described
herein.
2. Fuji Electrics Contribution to Space

Development
Fuji Electrics contribution to space development
began in the 1980s with involvement in the development of the H-II rocket which was developed and
built exclusively in Japan by combining the technical
expertise of various Japanese manufacturers under
the guidance of the former NASDA (National Space
Development Agency of Japan). Fuji Electric developed and supplied high-reliability BJTs (bipolar junction transistors) for this effort, and contributed to the
successful launch of the first H-II rocket in 1994.
Fuji Electrics 1st generation space-use power
MOSFETs are also installed in the International Space
Fig.1 External appearance of power MOSFET products
Table 1 Product list
V DSS
(V)
Model
ID
(A)
R DS(on)1
max.
()
P D2
(W
V GS
(V)
Qg
max.
(nC)
V GS(th)
(V)
Radiation
level (krad)
Package type
JAXA R 2SK4217
100
42
0.013
250
20
2.5 - 4.5
220
100
SMD-2
JAXA R 2SK4218
100
42
0.028
150
20
2.5 - 4.5
100
100
SMD-1
JAXA R 2SK4219
100
15
0.064
70
20
2.5 - 4.5
50
100
SMD-0.5
JAXA R 2SK4152
130
42
0.017
250
20
2.5 - 4.5
220
100
SMD-2
JAXA R 2SK4153
130
39
0.039
150
20
2.5 - 4.5
100
100
SMD-1
JAXA R 2SK4154
130
15
0.089
70
20
2.5 - 4.5
50
100
SMD-0.5
JAXA R 2SK4155
200
42
0.026
250
20
2.5 - 4.5
220
100
SMD-2
JAXA R 2SK4156
200
32
0.062
150
20
2.5 - 4.5
100
100
SMD-1
JAXA R 2SK4157
200
14
0.148
70
20
2.5 - 4.5
50
100
SMD-0.5
JAXA R 2SK4158
250
42
0.038
250
20
2.5 - 4.5
220
100
SMD-2
JAXA R 2SK4159
250
26
0.091
150
20
2.5 - 4.5
100
100
SMD-1
JAXA R 2SK4160
250
12
0.223
70
20
2.5 - 4.5
50
100
SMD-0.5
JAXA R 2SK4188
500
23
0.18
250
20
2.5 - 4.5
300
100
SMD-2
JAXA R 2SK4189
500
10
0.48
150
20
2.5 - 4.5
120
100
SMD-1
JAXA R 2SK4190
500
4.5
1.15
70
20
2.5 - 4.5
48
100
SMD-0.5
R DS(on) : V GS = 12 V, 2 P D : T C = 25 C
Table 2 Requirements for space-use power MOSFETs
Requirement
Electrical
characteristic
Actual or targeted performance

Space-use MOSFETs
Actual perTargeted
Actual
formance of
consumer-use performance performance
MOSFETs of 1st gen. of 2nd gen.
MOSFETs MOSFETs
Break down
voltage (V
250 V
500 V
250 V
On-resistance
Long-term reliability
Tolerance to radiation (TID)
Tolerance to high-energy
charged particles
(SEB tolerance)
Fig.2 Internal structure of power MOSFETs for space applications

TO-254
SMD-2
SMD-1
SMD-0.5
: Satisfied requirements
: Slightly below requirements
: Did not meet requirements
Station that is being assembled in orbit through cooperation from the United States, Russia, Japan, Canada
and Europe. Approximately 3,000 of these MOSFETs
are installed in the Japanese Experimental Module
known as Kibo, and these MOSFETs have continued
to operate properly ever since their insertion into orbit
in March 2008.
3. Development of 2nd Generation Space-use

Power MOSFETs
The aforementioned 1st generation space-use power MOSFETs have sufficient tolerance to ionization
radiation. However, in order to ensure adequate SEB
tolerance, the element breakdown voltage was increased. But, increasing the element breakdown voltage leads to increased on-resistance and greater loss.
An objective of the 2nd generation of space-use
power MOSFETs was to enhance the SEB toler-
ance and to realize electrical characteristics that are

equivalent to those of consumer-use power MOSFETs.
Table 2 lists the requirements for space-use power
MOSFETs.
3.1 Technical challenges of power MOSFETs for space
applications
Consumer-use power MOSFETs have excellent

electrical characteristics but are not compatible with
ionizing radiation and high-energy charged particles. Because the 1st generation of space-use power
MOSFETs had insufficient SEB tolerance, the breakdown voltage had to be raised to 500 V, and the onresistance was sacrificed.
The reduction of on-resistance while ensuring SEB
tolerance and TID (total ionizing dose) tolerance, as
well as ensuring long-term reliability, were challenges
for the 2nd generation of space-use power MOSFETs.
With the measures described in paragraphs (1) and
(2) below, requirements of the 1st generation space-use
70
power MOSFET can be satisfied. The SEB measure

described in paragraph (3) is a special feature of the
2nd generation space-use power MOSFETs.
(1) Use of hermetically sealed package to realize longterm reliability
Reliability is improved with the use of a metal
hermetically sealed package. A sintered compact of
copper-tungsten (Cu-W), having a coefficient of thermal expansion that is extremely close to that of silicon
(the raw material of the MOSFET chip), is used in
the package frame (area in which the MOSFET chip
is installed) to improve the temperature cycling tolerance. Moreover, as shown in Fig. 2, the interior of the
hermetically sealed package is hollow, and this void
is filled with dry nitrogen gas to protect the power
MOSFET chip from extrinsic degradation.
(2) Use of low-temperature process to ensure TID
Generally, when a power MOSFET for terrestrial
applications is used in an environment of ionizing radiation, the breakdown voltage decreases and a shift
occurs in the threshold voltage Vth of the gate that controls the on-off switching of the power MOSFET. The
degradation of the power MOSFETs characteristics
Fig.3 Distribution of electric field intensity and electron-hole
pair generation
due to TID is a phenomenon caused by electric charge

trapped in the oxide film.
In the manufacture of power MOSFETs for space
applications, the heat treatment after fabrication of
the oxide film is performed at a low temperature in order to decrease the amount of electric charges trapped
in the oxide film and to provide a TID tolerance of
1,000 Gy which is equivalent to the ionizing radiation
exposure for 10 years in a geostationary orbit.
(3) Use of 2-step epitaxial layer structure to ensure
SEB
Since about 1986, the phenomena of semiconductor
device malfunction and sudden damage due to highenergy charged particles (such as nickel (Ni) ions, for
example) have been reported. Such phenomena can be
caused by a single high-energy charged particle, and
are collectively known as single event effects (SEEs).
One type of a SEE that affects power MOSFETs is
SEB, which has been reported as the instantaneous
burnout of a device.
At the time when Fuji Electric began development
of 2nd generation space-use power MOSFETs in 1992,
the mechanism behind the SEB phenomenon was not
understood. Fuji Electric began to analyze the mechanism using simulation technology.
3.2 Estimated SEB mechanism
2.4
2.8
3.2
3.6
4.4
4.8
5.2
5.6
High-energy charged particle
Ni+
Electric field (V /cm)
n epitaxial layer
The SEB phenomenon could not be reproduced

with a conventional simulation model, but this issue
was overcome by performing the simulation with a corrected drift-diffusion model. From our analysis of the
simulation, the following can be understood.
Figure 3 shows the distribution of electric field
intensity and the distribution of electron-hole pair genFig.4 Distribution of electric potential at incidence of highenergy charged particles
Source
46.49
100
200
264.2
n+ substrate
(m)
Gate
(GND)
(a) Distribution of electric field intensity

Aluminum electrode
Insulating layer
21.2
21.6
22
22.4
22.8
23.2
23.6
24
24.4
24.8
Electron-hole pair generation (/ cm3s)
Polysilicon
n+
p
n epitaxial layer
39.96
100
n+ substrate
200
R j fet
R epi
n epitaxial layer
262.6
R ch
R sub
n+ substrate
(m)
(b) Distribution of electron-hole pair generation
71
Drain (+)
eration. Electron-hole pairs are generated along the

trajectory of a high-energy charged particle. In the vicinity of a collision of a high-energy charged particle to
the n+ substrate, a high electric field region is formed
and a large quantity of electron-hole pairs is generated. An excess of holes is formed up until the boundary
between the epitaxial layer and the n+ substrate, and
at this tip, the electric field is strong, and electron-hole
pairs are generated actively.
From these findings, the SEB mechanism can be
estimated as follows.
(1) As a result of the incident high-energy charged
particles, holes from the generated electron-hole
pairs are supplied as base current to operate a
parasitic npn transistor.
(2) The excess of holes increase along the trajectory
of the incident high-energy charged particles (Fig.
4), and the base of the parasitic npn transistor is
pushed-out. As for the effective edge of the base,
a high electric field region is formed to block the
path with the n+ substrate.
In this region, a super high-density current
flows and, as a result, even at voltages less than
the breakdown voltage, dynamic avalanching occurs easily and electron-hole pairs are generated.
(3) The generated holes are supplied again as base
current, which facilitates the operation of the
parasitic npn transistor.
(4) Thus, the generation of holes due to dynamic avalanching in the high electric field region in the vicinity of the boundary between the substrate and
epitaxial layer causes a positive feedback action
similar to that of a thyristor, and leads to damage.
4. Proposed Structure to Prevent SEB
With this structure, it is thought that even if the

base is pushed-out, as long as sufficient distance to the
n+ substrate is ensured, an electric field will not reach
a high enough value to cause dynamic avalanching,
and the generation of SEB can be reduced. To verify
the effectiveness of the proposed structure, actual
prototypes of MOSFETs with different epitaxial layer
thicknesses were fabricated and tested. As the test
results, Fig. 5 shows the dependency of epitaxial layer
thickness on SEB-generating voltage.
To obtain a breakdown voltage of 250 V with this
structure, an epitaxial layer thickness of approximately 29 m is needed, but with this design, SEB occurred
at about 200 V, which is approximately 60% of the
breakdown voltage. On the other hand, with a device
having a thicker epitaxial layer, we found that the
SEB-generating voltage is higher and that the desired
effect could be obtained. Based on these test results,
the targeted SEB-generating voltage could be attained
by setting the epitaxial layer thickness to 32 m.
On the other hand, this measure sacrifices (increases) the on-resistance, which is a critical characteristic of power MOSFETs. Figure 4 also shows the
on-resistance configuration of a power MOSFET. The
n- epitaxial layer is the current path when the power
MOSFET turns on and operates, and is directly linked
to an increase in on-resistance. The on-resistance Repi
of the n- epitaxial layer accounts for a large percentage
of the total on-resistance, and in the case of a 250 V
Fig.6 Cross-section of active area of 2nd generation spaceuse power MOSFET that uses a 2-step epitaxial layer
structure
Source
Gate
(GND)
In consideration of the aforementioned mechanism,

the following structure to prevent SEB was proposed
and implemented.
Aluminum electrode
Insulating layer
Fig.5 Dependency of SEB tolerance on n epitaxial layer thickness
Polysilicon
p
SEB-generating voltage (V)
400
300
Target value :
250 V or higher
R j fet
R epi1
n epitaxial layer
200
R epi2
Low resistivity n epitaxial layer
Epitaxial layer thickness of

consumer-use MOSFET
100
R sub
n+ substrate
0
10
R ch
20
30
40
50
Epitaxial layer thickness (m)
60
Drain (+)
72
Fig.7 Dependence of SEB capacity on electrical resistance of

epitaxial layer
SEB-generating voltage (V)
400
Single epitaxial layer structure

2-step epitaxial layer structure
300
250 V target value
200
Design of consumer-use
MOSFET
100
0
10
20
30
40
Fig.8 Breakdown voltage dependency of on-resistance

10
9
1st generation space-use power MOSFET
8
7
6
5
4
Consumer-use MOSFET and

2nd generation space-use
power MOSFET
Margin reserved to
ensure SEB tolerance
2
1
0
100
200
300
400
Breakdown voltage (V)
73
500
tity of electrons injected from the parasitic npn transistor and the relative difference in impurity concentrations (corresponding to the resistivity) of the epitaxial
layers, the base is pushed out and a high electric field
is formed, and therefore, increasing the impurity concentrations of the epitaxial layers (lowering the resistivity) is also expected to have the effect of suppressing
the generation of a high electric field.
Figure 7 shows the relationship between the electrical resistance of the epitaxial layer and the voltage
at which SEB is generated. With the innovative 2-step
epitaxial layer structure, the targeted SEB-generating
voltage can be achieved even if the resistance of the entire epitaxial layer is lowered by about 50%.
The application of this 2-step epitaxial layer structure enables the SEB tolerance to be ensured and the
increase in on-resistance to be minimized (3% or less).
Figure 8 shows the tradeoff between breakdown voltage
and Ron. The breakdown voltage margin that had been
reserved in order to ensure SEB tolerance is eliminated, and a 2nd generation high-reliability space-use
power MOSFET having excellent on-resistance characteristics was developed.
5. Postscript
Electrical resistance of epitaxial layer (mcm2)

(epitaxial layer resistivity epitaxial layer thickness)
Indexed Ron
(consumer-use MOSFET is set as 1)
R on
rated device, for example, accounts for approximately

80% of the total.
To avoid this increase in on-resistance, instead of
simply increasing the thickness of the epitaxial layer,
we proposed a 2-step epitaxial layer structure (Fig. 6)
that is provided with an n- epitaxial layer of low resistivity Repi2.
With this structure, a resistivity higher than that
of an n+ substrate can be obtained even with an epitaxial layer of low-resistivity, and an effect equivalent
to increasing the epitaxial layer thickness is thought to
be obtainable. Moreover, as a result of the large quan-
600
By successfully estimating the SEB mechanism,

we have minimized the increase in on-resistance
and commercialized a space-use high-reliability power MOSFET product having electrical characteristics equivalent to those of a consumer-use power
MOSFET. This paper has described a 250 V-class
product, but a lineup that includes 100 V, 130 V, 200 V
and 500 V-class products using this technology is available.
The 2nd generation space-use high-reliability power MOSFET realizes the top level of performance in
the world. In the future, Fuji Electric intends to continue to make contributions to space development.
References
(1) Gover, J. E. Basic Radiation Effects in Electronics
Technology. Colorado Springs, CO, Proc. 1984 IEEE
NSREC Tutorial Short Course on Radiation Effects.
July 22, 1984.
(2) Waskiewicz, A. E. et al. Burnout of Power MOSFET with
Heavy Ions of CaliFornium-252. IEEE Trans. Nucl. Sci.
Dec. 1986, vol.NS-33, no.6, p.1710-1713.

Katsuyuki Uematsu Hiroko Tanaka Hirofumi Kato
ABSTRACT
Exhaust gas regulations for motor vehicles are becoming stricter year-after-year, and as internal combustion
engines are being made more efficient and their exhaust gas reduced, there is increased demand for pressure
sensing in the exhaust systems of diesel engine vehicles. Applying a single-chip semiconductor pressure sensor
fabricated by a CMOS process, which has been successfully used for manifold pressure measurement, Fuji Electric
has developed an exhaust system pressure sensor capable of withstanding an exhaust gas environment containing
corrosive substances. The newly developed sensor, in a DIN standard SO2 gas test, exhibited the ability to withstand
corrosion that is more than 2.5 times greater than that of conventional sensors, and is suitable for use in absolute
pressure sensing and relative pressure sensing applications.
1. Introduction
Industrial development and expanded distribution in the BRIC countries (Brazil, Russia, India and
China) and elsewhere are ongoing. The associated
increase in economic activity on a global scale is acerbating the problems of global warming caused by the
emission of greenhouse gases such as carbon dioxide
(CO2) and acid rain and air pollution caused by nitrogen oxides (NOx) and sulfur oxides (SOx).
As exemplified by the Green New Deal policies in
the US, and Japans announced intent to reduce CO2
exhaust emissions by 25% by the year 2020, environmental issues are being addressed as national and
global problems. With the establishment of strict targets for emission regulation, the development of new
environmental technologies and the accompanying economic activity are expected to provide opportunities for
environmental businesses.
In particular, environmental measures implemented by the automobile industry, which account for a
major portion of the economic activities relating to industry and distribution, are being watched closely. In
order to meet the exhaust emission regulations which
are being strengthened from year to year, automobile
manufacturers have concentrated their efforts on increasing the efficiency and realizing cleaner emissions
of combustion engines, and on developing hybrid vehicles and electric vehicles.
mobiles, engine information such as the load status,

aspiration and temperature acquired from sensors
installed at various locations in the engine is used to
control combustion electronically to realize an optimal
air-fuel ratio.
Automotive pressure sensors attached to the intake manifold of an engine and used primarily for measuring intake pressure have helped to make engine
control more efficient. Also, in addition to intake pressure measurement sensors, multiple other pressure
sensors, such as boost pressure sensors for turbo chargers and super chargers, clogging detection sensors for
air filters and barometric pressure sensors(1) for compensating the intake pressure at higher altitudes at
which the air becomes thinner, are being used.
Fuji Electric began mass-producing manifold absolute pressure (MAP) sensors for automobile engines
in 1984. The first sensor had many parts, including a
circuit for adjusting the characteristics of a piezo-resistive gauge and a circuit board on which SMDs (surface
mounted devices) for dealing with EMC (electromagnetic compatibility) such as chip capacitors, chip resisFig.1 Sensor unit of pressure sensor
Diffused piezo resister
Diaphragm
2. Fuji Electrics Pressure Sensors

To achieve cleaner emissions and more efficient
fuel consumption of automobiles, the engine control
must be made more accurate. In present-day auto-
Glass base
Pressure sensor chip (silicon)
74
tors and the like are mounted, and the parts had many
electrical contacts(2).
To enhance the reliability by reducing the number
of parts, Fuji Electric developed a pressure sensor in
which the sensing unit, amplifier circuit, temperature
characteristic compensation circuit and EMC protection devices are integrated into a single chip using a
standard semiconductor CMOS (complementary metaloxide-semiconductor) process. This single-chip type
of pressure sensor has been deployed in the market
mainly as a MAP sensor since 2002.
Figure 1 shows an overview of the sensor unit of
this pressure sensor, and Fig. 2 shows the cross-sectional structure of this pressure sensor.
Diffused piezo-resistors are fabricated on a diaphragm at the same time as the IC process is performed, and a Wheatstone bridge is configured with
four diffused piezo-resistors. The diaphragm ensures a
high burst pressure and is formed, using Fuji Electrics
proprietary silicon etching process, in a precise round
shape.
When the application of pressure causes the diaphragm to flex, the resistance values of the diffused
piezo-resistors come to change due to the piezo-resistive effect, and the output of the Wheatstone bridge
Fig.2 Cross-sectional structure of pressure sensor unit
changes. As a result of this principle, the pressure

quantity can be converted into a voltage. Moreover,
a high precision amplifier for amplifying the signal
output from the Wheatstone bridge and an adjustment
circuit for compensating the sensor characteristics are
built on a single chip. Additionally, protection devices
for protecting internal circuits from surges generated
by the automobile engine control system, and from
static electricity during the assembly process and electromagnetic noise from external sources, and so on, are
all provided on a single chip(3).
As shown in Fig. 3, the sensor unit is embedded in
a resin package, and electrical and mechanical connections are made to the exterior. The sensor cell package
is made from PPS (polyphenylene sulfide), which is
used for automotive parts and has excellent resistance
to acid, and the sensor chip unit, lead frame and wires
are coated with fluoro silicone gel, which has excellent
chemical resistance and oil resistance.
3. Application to Pressure Sensors for Exhaust

System
Exhaust emissions regulations for automobiles
have been strengthened year after year since the
Muskie Act in the 1970s, and in Japan, Post-New
Long-Term exhaust emission regulations came into
effect in 2009. In Europe, the EURO5 emission regula-
Diffused piezo resister
Sensor chip
(silicon)
Vacuum
chamber
Glass base
Fig.4 Changes in NOx and PM exhaust emission regulations

for diesel passenger cars

(Prepared by Fuji Electric and based on documents
from the Japanese Ministry of Land, Infrastructure and
Transport, the Japanese Ministry of Economy, Trade
and Industry, and Japan Automobile Manufacturers Association)
Pressure
0.5
Emission (g/ km)
Bending stress
of diaphragm
Japan
US
EU
0.4
0.3
EURO4
long-term
EURO5
Tier2Bin9 New
exhaust emission
regulation
Post-new long-term
exhaust emission
regulation
Tier2Bin5
0.2
0.1
0
2004
2006
2008
2010
2012
EURO6
2014
(Year)
(a) Nitrogen oxides
Fig.3 Structure of pressure sensor cell

Pressure P
Fluoro silicon gel
Lead frame
Sensor chip
Resin package
75
Adhesive
Glass
Emission (g /km)
Aluminum wire bonding
0.05
Japan
US
EU
Tier2Bin9
0.04
0.03
EURO4
New long-term
exhaust emission
regulation
0.02
EURO5
0.01
0
Post-new long-term
exhaust emission
regulation
EURO6
Tier2Bin5
2004
2006
2008
2010
2012
2014
(Year)
(b) Particulate matter
tions came into effect in 2009, and the EURO6 regulations are slated to come into effect in 2014. Figure 4
shows the changes in exhaust emission regulations for
diesel passenger cars.
Aiming to improve fuel economy and to reduce
CO2 greenhouse gas emissions, diesel automobiles
are in high demand chiefly in Europe and account for
more than half of the total number of automobiles
sold in Europe. Diesel automobiles have much better
fuel economy and lower CO2 emissions than gasolinepowered automobiles, but the NOx emissions that accompany their higher combustion temperature and PM
(particulate matter) emissions due to imperfect combustion are problems.
Recently, a DPF (diesel particulate filter) used to
remove PM contained in exhaust gas has begun to be
installed mainly in diesel automobiles. A pressure sensor is used to sense clogging of this filter. Additionally,
an EGR (exhaust gas recirculation) system in which a
portion of exhaust gas is recirculated to the intake side
so as to control combustion has begun to be utilized,
and pressure sensors have come to be used to measure
the pressure of exhaust gas (Fig. 5).
The sensors used in these exhaust systems must
be protected from corrosive matter such as acid caused
by the NOx and SOx contained in the exhaust gas, and
incompletely combusted fuels and oils.
Fuji Electrics newly developed pressure sensor for
exhaust systems uses a conventional intake pressure

sensor and implements the corrosion free design required for exhaust system pressure sensors (Fig. 6).
4. Corrosion Resistant Design to Protect Against

Corrosive Matter in Exhaust Gas
The corrosion resistance required of exhaust system pressure sensors has been achieved by implementing the following.
(1) Gold plating of electrode pads on the chip
(2) Gold wire bonding between the chip and lead
frame
(3) Gold plating of lead frame for the resin case
Details of these techniques and their evaluation
results are described below.
4.1 Corrosion resistant design of sensor chip
The electrode pads of the semiconductor chip,

which is fabricated by Fuji Electrics standard CMOS
process, are made from an aluminum alloy. If SOx
gas and steam contained in the exhaust gas permeate
into the fluoro silicone gel on the chip, sulfuric acid is
generated by the mechanism listed below, and the aluminum corrodes.
2SO2 + O2 / 2SO3
SO3 + H2O / 2H+ + (SO4)22Al + 6H+ / 2Al3+ + 3H2
Fig.5 Engine intake/exhaust system and pressure sensor

MAP sensor
Turbo charger
Boost pressure Inter-cooler

sensor
Air filter
Pressure sensor for DPF
Throttle
DPF
Air
EGR
valve
Muffler
Catalyst
Engine block
Intake air
Air duct
Pressure sensor for EGR
EGR cooler
Exhaust gas
Therefore, the electrode pads of the chip are coated

with a gold plating layer to form a corrosion resistant
structure. Figure 7 shows the cross-sectional structure
of the electrode pads and the gold plating on the chip.
In order to prevent aluminum and gold from diffusing
between the aluminum alloy layer and the gold plating
layer, a barrier layer of a titanium-tungsten (Ti-W) alloy is provided (for the purpose of explanation, the proportion of horizontal and vertical length in the figure
differs from that of an actual chip).
Moreover, conventionally, the chip and lead frame
were bonded together by aluminum wire, but because
the same chemical reaction would cause the wire to
corrode and possibly break, gold wire was used for the
Fig.6 Exhaust system pressure sensor

Fig.7 Corrosion resistant structure of sensor chip electrode
pads
Gold plating layer
Oxidation layer
Electrode pads
(aluminum alloy)
Barrier layer (Ti-W)
Passivation layer
76
Fig.8 Structure of exhaust system pressure sensor

Gold wire bonding
Pressure P
Fluoro silicon gel
Fig.9 SO2 gas test results
Sensor for
exhaust gas
system
No wire breakage
after 50 cycles
Conventional
product
Gold plated
lead frame
Resin package
Adhesive
Sensor chip
with gold
plated electrode
Glass base
Table 1 SO2 gas test conditions (DIN50018-SFW1.0S)
Theoretical SO2 concentration at start

of cycle (vol. %)
0.33
Condensate climate
DIN50018-SFW 1.0S
1st test process (h)

Cycles
2nd test process (h)

Total (h)
Temperature (C)
Test room
conditions
1st
test
process
Relative humidity
(%)
2nd
test
process
Temperature (C)
Relative humidity
(%)
8 (including heating)
16 (including cooling)
(test box is open or
ventilated)
24
40 3
Approximately 100
(forming condensation on the test
material)
18 to 28
75 (max.)
he theoretical SO2 concentration, in the case of a test box having a

:T
300 L volume, corresponds to 1.0L of SO2 added for each cycle.
new sensor.
4.2 Corrosion resistant design of the package
In an existing pressure sensor cell package, the

lead frame is plated with nickel, but depending on the
surface condition of the plating, the leads may become
corroded by acid that has formed from the exhaust gas,
resulting in a break in the electrical connections to the
exterior. Therefore, a gold layer is plated on top of the
nickel plating to create a structure that prevents corrosion.
Since acid formed from corrosive gas reaches the
interface between the lead frame and resin, the side
surfaces, i.e., the cut surfaces, of the lead frame that
are implanted into the resin must be corrosion-resistant. After press-cutting of the lead frame, a gold plating process is performed, and the cut edge surfaces are
gold-plated (Fig. 8).
4.3 Corrosion resistance test results
To evaluate the corrosion resistance, an SO2

gas test was performed based on the DIN standard
DIN50018-SFW1.0S. The test conditions are listed
in Table 1. This test exposes the product surface, on
which condensation is formed, to SO2 gas and is more
severe than the exhaust gas environment of an actual automobile, but is used as an accelerated test for
77
Wire breakage
occurs at 20th cycle
0
10
20
30
40
50
60
Number of cycles
Table 2 Basic specifications of the exhaust system

pressure sensor
Item
Units
Specification
Comments
Over-voltage
16.5 V
1 min
Proof Pressure
kPa abs.
600
Storage
temperature
- 40 to 150
Usage temperature
- 40 to 135
Power supply
voltage
5.00 0.25
Pressure range
(absolute pressure)
kPa abs.
50 to 400
Type of pressure
used
(gauge pressure)
kPa gauge
50 to 400
Output range
0.5 to 4.5
Interface
Diagnostic output
Accuracy
300
Pull-up
100
Pull-down
<0.2, >4.8
%F.S.
< 1.2
10 to 85 C
%F.S.
< 2.0
- 40 / 135 C
Standards that
have passed EMC
verification
JASO D00-87CISPR 25
ISO11452-2ISO7637
1 : Full-scale pressure can be changed freely

2 : Detect open mode at Vcc and / or Vout
verifying the lifetime until reaching the failure mode of

wire breakage due to corrosion of the product.
Figure 9 compares broken wire failure modes of
the chip electrode pads and case lead frame caused by
corrosion in the SO2 gas test for a conventional product and the newly developed exhaust system pressure
sensor. We verified that the newly developed exhaust
system pressure sensor has more than 2.5 times that
corrosion resistance of a conventional product (where 1
cycle is defined as the time from product condensation
until exposure to SO2 gas).
5. Basic Specifications
Basic specifications of the newly developed exhaust
system pressure sensor are as listed in Table 2. The
sensor chip circuitry directly reuses a configuration
having a history of past success as a MAP sensor, and

features a diagnostic function for detecting breakage
of the wiring and an over-voltage protection function,
and is resistant to electromagnetic noise. Either absolute pressure or gauge pressure can be selected for the
pressure measurement.
6. Postscript
Fuji Electrics new exhaust system pressure sensor was developed mainly for automotive applications,
but at present, efforts to reduce the exhaust gas from
automobiles are targeting not only the exhaust from
automobiles, but the entire process from the production to scrapping of automobiles. For example, since
heavy machinery equipped with diesel engines that
emit large quantities of exhaust are used in mining
the iron or raw material from which automobile bodies
are fabricated, and in the mining of rare metals used
to make reduction catalysts and batteries, there is a
movement to establish regulations for this exhaust gas
and to reduce the amount of these emissions.
If the newly developed pressure sensor is to be applied to non-automotive engines, such as engines in
heavy machinery and the like, then according to such
conditions as the type of engine, location of sensor installation and type of fuel used, significant variations
will exist in the quantity of corrosive matter contained
in the exhaust gas, and in the exhaust pressure and
ambient temperature, and it is important that product development be advanced with familiarity of the
upper-level applications.
Fuji Electric will continue to develop world-class
technology, aiming to develop products that will please
our customers and contribute to environmental protection measures.
The authors wish to take this opportunity to express their gratitude to Hitachi Automotive Systems,
Ltd. for their cooperation in conducting the SO2 gas
test.
References
(1) K. Saitou, Latest Trends of Automotive Sensors. CMC
Publishing Co., Ltd. 2009, p.38-51.
(2) T. Takahama, et al. Semiconductor Pressure Sensors.
Fuji Electric Journal. 1986, vol.59, no.11, p.707-710.
(3) K. Uematsu. About Automobile Pressure Sensors.
Material Stage. Technical Information Institute, Ltd.
2009, vol.9, no.1, p.26-30.
78

Yoshitaka Nishimura Mitsukane Oonota Fumihiko Momose
ABSTRACT
In power conversion systems that use IGBT modules, the thermal conductivity of the thermal grease utilized is
designed as low as possible. As a result, the thermal grease thickness affects the IGBT chip temperature. This paper describes the effects of the hardness of components and the ingredients of the thermal grease on the spreading
ability of the compound and the distribution of stress when tightening with screws. Additionally, in consideration of
the stress distribution, we have proposed a metal mask pattern for applying the thermal grease as thinly as possible.
As a result, the thermal grease can be applied at approximately 1/3rd the thickness of the conventional application
technique.
1. Introduction
Recently, in response to global warming, new types
of energy, such as wind and solar power, are being
utilized and efforts to popularize hybrid cars are being promoted with the goal of reducing carbon dioxide
emissions resulting from the use of fossil fuels. Also,
as conventional electric and electronic devices are
requested to provide ever higher levels of energy savings, IGBT (Insulated Gate Bipolar Transistor) modules required for power conversion and motor control
are becoming more and more important. Moreover,
as society becomes increasingly information-oriented,
digitized data is becoming more and more prevalent,
and there is renewed need for uninterruptable power
supplies (UPS) and the like.
The required IGBT module characteristics differ
according to the particular market, but higher efficiency and down-sizing are common requirements.
For wind power generation, IGBT modules are used in
the devices that convert the generated power. Power
converters are often installed in limited spaces, such as
inside a tower, and the IGBT modules are water-cooled
to increase their mounting density and achieve a more
compact size(1).
In response to requests for high density mounting,
Fuji Electric has increased the performance(2) of IGBT
chips. The IGBT modules have been developed with
thermal management to achieve a more compact size
and greater capacity simultaneously(3),(4). Fuji Electric
also supplies a simulator for simulating the IGBT module loss and temperature under the operating conditions, and this simulator is necessary when designing
an apparatus requested by a customer.
This paper introduces technology for attaining low
79
thermal resistance by optimizing the usage of the thermal grease that thermally connects the IGBT module
and cooling fin.
2. Background
The structure of a product that uses an IGBT
module and the thermal conductivities of constituent
components are shown in Fig. 1. So that heat generated from the die can escape, the IGBT module is used
with a cooling fin attached. The cooling fin typically
used with an inverter may have a surface roughness of
up to about 100 m maximum. When an IGBT module
is mounted on a cooling fin, a gap is created, and this is
a factor that degrades the thermal contact resistance.
Generally, in order to reduce the thermal contact resistance, thermal grease is applied (printed) between the
cooling fin and IGBT module.
The thermal conductivity of a typical thermal
grease is approximately 1 W / (m K), and as can be
seen in Fig. 1, thermal conductivity is lowest in the
path of thermal dissipation away from the IGBT modFig.1 IGBT module structure and main thermal conductivities
Case
IGBT die
Solder:
50 W/(mK)
Copper circuit:
390 W/(mK)
Ceramic:
20 to 170 W /(mK)
Solder:
50 W/(mK)
Copper base:
390 W/ (mK)
Thermal grease:
1 W/ (mK)
Cooling fin
ule. To increase the efficiency of the entire product

that uses the IGBT module, reduction of the thickness
of the thermal grease is requested. However, if the
thickness of the thermal grease is reduced, spreading
the thermal grease over the entire interface between
the IGBT module and the fin will be difficult.
In the chapters below, we report on (1) factors affecting the spreadability of thermal grease and (2) the
thermal grease printing method, and propose reducing
the thermal contact resistance in a product that uses
an IGBT module.
3. Factors Affecting the Spreadability of Thermal

Grease
The thermal grease is printed onto the IGBT
module or the cooling fin, and the force exerted by
tightening the screws that secure the IGBT module to
the cooling fin causes the thermal grease to spread out
and fill the gap between the IGBT module and cooling
fin. At this time, the IGBT module and cooling fin will
be thermally connected. To observe the spreading of
the thermal grease, grease was printed at a thickness
of 30 m in an area of 81 mm2 between a glass block
(having surface roughness of 5 m or less) and a fin
(having surface roughness of 5 m or less), and screws
to secure the glass block to the fin were tightened with
3.5 N m of torque. The results are shown in Fig. 2. As
can be seen in the figure, the thermal grease spread
Fig.2 Spreading of thermal grease
Glass block
Thermal grease
t =30 m
Fin
out around the screw locations, but despite using a

flat fin, the spreading of thermal grease in the center
area was poor. To investigate the cause, a photoelastic
camera was used to measure the distribution of stress
generated between the IGBT module and fin when the
IGBT module and fin mounting screws are tightened.
An FEM (finite element method) analysis was also performed at the same time.
Figure 3 shows the results of measurement with
the photoelastic camera of the distribution of stress
generated between the IGBT module and fin. A large
stress was found to be generated in the area surrounding the screws, and this finding exhibits a similar
tendency as the grease spreading test results shown in
Fig. 2.
Figure 4 shows the FEM analysis model and the
analysis results. The analysis was carried out for the
conditions of an IGBT module size of 119 mm 59 mm
and the screws being tightened with 3.5 N m of torque.
The results of the FEM analysis showed that the stress
generated between the IGBT module and fin is highest
at the areas around the screws, but decreases when approaching the center area. These results clearly show
that generated stress between the IGBT module and
fin varies according to the distance from a screw location.
Next, the relationship between stress and the
thickness of the thermal grease was investigated using grease manufactured by several different companies. The following types of thermal grease were used
: Electrolubes HTC, Dow Cornings SH 340, American
Oil & Supplys AOS 340 and Shin-Etsu Silicones G747.
A uniform amount of each type of grease was printed
onto a flat plate, and then a glass block was laid on
Fig.4 FEM analysis model and analysis results
(a) Diagram of test configuration
IGBT module
Applied pressure
(b) Thermal grease printing
(c) After screw tightening
(a) FEM analysis model

Y1
Fig.3 Results of stress distribution measurement using photoelastic camera

X1
X2
X1
Stress
Stress
Y2
X2
Y1
Y2
(b) FEM analysis results (stress distribution)
80
top, and pressure was applied. From the spread-out

area of thermal grease at this time, the corresponding
thickness of the thermal grease thickness was computed for each applied pressure.
Figure 5 shows the relationship between applied
pressure and thermal grease thickness. It can be seen
that the thermal grease thickness decreases as the applied pressure increases up to 0.1 MPa. However, at
pressures above 0.1 MPa, the thickness did not fall below a certain level even if the applied pressure was increased. The reason for this behavior is thought to be
attributable to the effect of the material composition of
the thermal grease. The main components of ordinary
thermal grease are ceramic powder and oil(6).
Figure 6 shows a SEM (scanning electron microscope) photograph of the ceramic particles used in
thermal grease. It can be seen that the ceramic particles are several microns in size. The ceramic grains
are believed to be the reason that the thermal grease
thickness does not decrease below a certain level even
if additional pressure is applied.
Thus, we examined factors that affect spreading
of the thermal grease. As shown in Fig. 7 (a), in the
Thermal grease thickness (m)
Fig.5 Relationship between applied pressure and grease

thickness
case where thermal grease is present at a location of

high stress in the vicinity of a screw, the gap between
the module and fin will be large, and there is a concern
that force may not be transmitted to the center area.
In other words, the printing pattern of the thermal
grease is thought to affect its spreading. Therefore, we
investigated the relationship between the initial pattern and the spreading of the thermal grease. Figure
8 shows how the thermal grease spreads according to
whether grease is present in the areas surrounding
the screws. In the case of sample A, in which thermal
grease has been printed at the screw locations, the center area does not contact the fin. However, with sample B, in which thermal grease has not been printed at
the screw locations, the thermal grease spreads over
the entire surface.
If there exists areas to which the thermal grease
has not spread, an air layer will be formed and result in a dramatic increase in thermal resistance.
Therefore, until now, poor spreading of the thermal
grease was counteracted by increasing the quantity of
thermal grease applied.
From the above results, it is thought that the application of thermal grease at appropriate locations
will enable a smaller quantity of thermal grease to be
used, resulting in a lesser thickness of the thermal
grease.
120
Fig.7 Relationship between grease location and pressure

(schematic diagram)
100
80
40
IGBT module
HTC
20
0
Compound grain
Screw
Screw
60
SH340
0
0.1
0.2
IGBT module
ASO340
G747
0.3
0.4
Fin
Fin
Pressure (MPa)
Applied pressure
Applied pressure
Large
Fig.6 SEM photograph of ceramic material used in thermal

grease
Small
(a) Grease in vicinity of screw
Large
Small
(b) No grease in vicinity of screw
Fig.8 Relationship between thermal grease printing pattern

and spreading (test results)
Thermal grease
printing pattern
(Module side)
10
mm
Sample A
10 mm
Sample B
Thermal grease
spreading after
tightening of
screws
(Fin side)
1 m
81
Sample A
Sample B
(a) Grease in vicinity

of screw
(b) No grease in
vicinity of screw
4. Development of Recommended Printing

Method for Thermal Grease
4.1 Metal mask design
Based on the results of investigations into the

causes of thermal grease spreading, we considered
which method of thermal grease printing Fuji Electric
should recommend. Typical methods of thermal grease
printing involve the use of : (1) a dispenser, (2) a roller
or spatula, and (3) a metal mask. When a dispenser
is used for thermal grease printing, there is a large
distance throughout which the grease can spread, and
the spreading is easily affected by the viscosity of the
thermal grease. In the case of thermal grease printing
using a roller or spatula, there is significant variation
in the application quantity and the quality will be inconsistent. For these reasons, Fuji Electric adopted
the following method for controlling thickness of the
thermal grease.
Fig.9 Thermal grease printing method using a metal mask
Metal squeegee
IGBT set in jig
Printing by metal squeegee

(begin)
Printing by metal squeegee
With metal mask removed
(1) Thermal grease printing shall be performed using

a metal mask.
(2) Thermal grease shall be applied at appropriate locations (and no grease shall be applied in the area
surrounding a screw).
(3) The quantity of grease applied shall be changed
according to the level of stress generated between
the IGBT module and the fin.
Figure 9 shows the method of thermal grease
printing using a metal mask and Fig. 10 shows Fuji
Electrics recommended metal mask design.
Fuji
Electrics recommended metal mask was designed so
that thermal grease is not applied in the surrounding
areas of screws and that the required quantity of thermal grease at each location is printed according to the
stress level computed from the FEM analysis.
Next, we investigated the relationship between the
printing method and the minimum thickness (printing
quantity) of thermal grease required for grease spreading. For testing, a single IGBT module (having dimensions 119 mm 59 mm) and fin, and HTC thermal
grease were used. Figure 11 shows the relationship
between the thermal grease printing method and minimum required thickness (printing quantity). When
using a roller, a mass of thermal grease corresponding
to a thickness of approximately 100 m is printed to
spread over the entire surface between the IGBT module and the fin. On the other hand, it can be seen that
when using Fuji Electrics recommended metal mask, a
mass of thermal grease corresponding to a thickness of
approximately 50 m to spread over the entire surface
between the IGBT module and the fin.
The above result demonstrates that printing using
an optimal metal mask is a valid method, and that the
use of Fuji Electrics recommended metal mask design
Fig.11 Relationship between thermal grease printing method
and rate of spreading
New stencil mask design
General stencil mask design
Grease is not printed in the surrounding area of screw.

Since the grease will not compress to less than its grain size,
there should be no grease in the area surrounding a screw that
applies force.
Thermal grease
spreading (%)
Fig.10 Recommend metal mask design
100
90
80
70
60
30
50
70
90
110
130
150

(Average value computed from quantity of printed grease)
Grease density is low in the areas surrounding screws that

apply large pressures, but is high in the center area.
An optimal quantity of thermal grease is applied in proportion
to the applied pressure.
Roller
Dispenser
82
Fig.12 Relationship between thermal grease printing method

and die temperature
utilize IGBT modules.
5. Postscript
Dispenser
Tj (deg)
116
Roller
114
General stencil mask design
112
New stencil mask design
110
0
50
100
150
200

(Average value calculated from quantity of printed grease)
is highly effective in optimizing the quantity of thermal grease to be printed.

4.2 Affect on die temperature
We investigated the effect of the printing method

on the die temperature of the IGBT module.
In the testing, using the same IGBT module and
fin, a current of 80 A was applied to the IGBT die
and after 5 minutes had elapsed, the die temperature
was measured with an IR camera. Figure 12 shows
the required minimum thermal grease thickness for
each printing method, and the measured results of the
IGBT die temperature for each thickness. It can be
seen that the IGBT die temperatures are different according to the thermal grease printing method.
It was verified that when using Fuji Electrics
recommended metal mask, the maximum IGBT die
temperature was 6 C lower than in the case of a conventional printing method. The IGBT die temperature
is the largest factor affecting the lifespan and efficiency
of products that utilize IGBT modules. The use of Fuji
Electrics recommended metal mask enables an appropriate quantity of thermal grease to be printed and realizes a decrease in thermal resistance in products that
83
The following items were clarified in the results

of our study of thermal grease printing as a thermal
management method in product.
(a) The stress generated between the IGBT module
and fin is affected by the printed pattern of the
grease.
(b) An optimal pattern and quantity exist for the
thermal grease to be printed.
From the above results, an optimal metal mask
design was proposed for thermal grease printing. By
disclosing the thermal grease printing method to our
customers and applying Fuji Electrics IGBT modules,
we aim to contribute to higher efficiency and energy
savings in equipment.
References
(1) Ambo, T., et al. Power Electronics for Large Scale Wind
Power Generation. Journal of the Institute of Electrical
Engineers of Japan. 2009, vol.129, no.5, p.291-294.
(2) Onozawa, Y. et al. Development of the Next Generation
1,200 V Trench-gate FS-IGBT Featuring Lower EMI
Noise and Lower Switching Loss. ISPSD 2007, p.13-16.
(3) Kobayashi, Y. et al. New Concept IGBT-PIM Using the
Latest Technology. Fuji Electric Journal. 2006, vol.79,
no.5, p.358-361.
(4) Nishimura Y. et al. Development of a New-Generation
RoHS IGBT Module Structure for Power Management.
Transaction of the Japan Institute of Electronics Packaging. 2008, vol.1, no.1.
(5) Takaku, T. et al. Power Loss and Temperature Simulator for IGBT Module. Fuji Electric Journal. 2008,
vol.81, no.6, p.438-442.
(6) Mita, K. Development of Thermal Interface Materials
for IT-Related Applications (Silicone type TIM). Journal of the Society of Rubber Industry, Japan. 2005,
vol.78, no.4, p.153-157.
Overseas Subsidiaries
North America
Fuji Electric Corp. of America
Sales and marketing of inverters, power distributors & control
equipment, power supplies, and ring blowers
Headquarters Office
Phone +1-510-440-1060
New Jersey Office
Phone +1-201-712-0555
Ohio Office
Phone +1-513-326-1280
Illinois Office
Phone +1-847-397-8030
Virginia Office
Phone +1-540-491-9625
Texas Office
Phone +1-713-789-8322
Los Angeles Office
Phone +1-714-540-3854
Fax +1-513-326-1288
Manufacture and marketing of photoconductive drums for copiers and

printers
Fax +1-847-925-9632
Fuji Electric FA (Asia) Co., Ltd.
Fax +1-540-491-9629
Marketing of inverters, power distributors, control equipment and

semiconductors
Fax +1-713-789-8358
Fax +1-714-540-5731
Fax +1-732-457-0042
Fax +33-4-73-98-26-99
Phone +852-2311-8282
Fax +852-2312-0566
Fuji Electric Taiwan Co., Ltd.

Fax +886-2-2515-1860
Fuji Electric FA Taiwan Co., Ltd.

Marketing of power distributors, control equipment and drive system
products in Taiwan
Phone +886-2-2370-2390
Fax +886-2-2370-2389
Manufacture of small- to medium-capacity industrial motors
Fax +886-3-321-7890
Fuji Electric FA Korea Co., Ltd.

Materials procurement and marketing of electrical and electronic
machinery and components
Phone +82-2-780-5011
Fax +82-2-783-1707
Korea FA Systems Co., Ltd.

Development and marketing of software for computers
Fax +86-21-5496-0189
Fax +86-411-762-2030
Wuxi Fuji Electric FA Co., Ltd.

Manufacture of inverters
Phone +86-510-8815-2088 Fax +86-510-8815-9159
Shanghai Fuji Electric Switchgear Co., Ltd.

Fax +86-21-5718-1448
Shanghai Fuji Electric Transformer Co., Ltd.

Manufacture and marketing of molded transformers
Fax +86-21-5718-5745
Shanghai General Fuji Refrigeration

Equipment Co., Ltd.
Manufacture of refrigerated showcases
Fax +86-21-6921-1066
Fuji Electric (Shanghai) Co., Ltd.

Marketing of inverters, switchgear and transformers
Fax +86-21-6422-4650
Suzhou Lanlian-Fuji Instruments Co., Ltd.

Marketing of measuring instruments
Marketing of semiconductor devices
Phone +886-3-321-3030
Manufacture of low-voltage circuit breakers and motors
Phone +86-21-5496-1177
Fax +852-2312-0566
Fuji Electric Device Technology Hong Kong

Co., Limited.
Fax +49-9131-28831
Fuji Electric Dalian Co., Ltd.
Phone +86-21-6921-1088
Fax +852-2664-8040
Atai Fuji Electric Co., Ltd.
Fuji Electric Holdings (Shanghai) Co., Ltd.
Phone +86-21-5718-5747
Phone +852-2311-8282
Fax +86-571-796-8198
Fax +49-69-6661020
East Asia
Phone +86-21-5718-5740
Phone +852-2664-8699
Phone +886-2-2515-1850
Manufacture of measuring instruments
Manufacture of switchgear
Phone +86-571-796-8118
Marketing of semiconductors
Fuji Electric France S.A.
Phone +86-411-762-2000
Hangzhou Fuji Refrigerating Machine Co., Ltd.
Hong Kong Fujidenki Co., Ltd.
Marketing of Drive & Automation equipment, semi-conductors,

photoconductive drums for copiers and printers
Phone +86-21-5496-3311
Fax +86-755-220-5812
Fax +1-201-368-8258
Fuji Electric Europe GmbH
Phone +33-4-73-98-26-98
Phone +86-755-220-2745
Development of software for vending machines and other businesses
EU
Head Office
Phone +49-69-6690290
Erlangen Office
Phone +49-9131-729613
Marketing and after-sales service of inverters
Fax +1-510-440-1063
Marketing of semiconductor devices and photoconductive drums
Phone +1-732-560-9410
Fuji Electric Technology and Service

(Shenzhen) Co., Ltd.
Phone +86-512-8881-2966 Fax +86-512-8881-2971
Phone +82-2-598-0651
Fax +82-2-573-1904
Southeast and South Asia

Mahajak International Electric Co., Ltd.
Manufacture and marketing of watt-hour meters
Phone +66-2-253-2350
Fax +66-2-253-2354
Fuji Electric (Malaysia) Sdn. Bhd.

Manufacture of magnetic recording media
Phone +60-4-403-1111
Fax +60-4-403-1491
Fuji Electric Semiconductor (Malaysia)

Sdn. Bhd.
Manufacture of power semiconductors
Phone +60-4-403-1111
Fax +60-4-403-5993
Fuji Electric Philippines, Inc.

Manufacture of power semiconductors
Phone +63-2-844-6183
Fax +63-2-844-6196
Fuji-Haya Electric Corp. of the Philippines

Manufacture and marketing of switch boards and electrical control
equipment
Phone +63-2-892-8886
Fax +63-2-893-5645
P.T. Fuji Dharma Electric.

Manufacture of watt-hour meters
Phone. +62-21-4606247
Fax +62-21-4610338
Fuji Electric Asia Pacific Pte. Ltd.

Marketing of Drive & Automation equipment and semiconductors,
in Southeast Asia
Phone +65-6533-0010
Fax +65-6533-0021
Printed on recycled paper

Fer 56 2 000 2010 PDF

Uploaded by

Copyright:

Available Formats

Fer 56 2 000 2010 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Fer 56 2 000 2010 PDF

Uploaded by

Copyright:

Available Formats

Whole Number 229

Power Semiconductor contributing in

Power Semiconductor contributing

3.3 kV IGBT Modules

New Lineup of V-Series IGBT Modules

V-Series Intelligent Power Modules

High Reliability Power MOSFETs for Space Applications

Pressure Sensor for Exhaust Systems

Thermal Management Technology for IGBT Modules

FUJI ELECTRIC REVIEW vol.56 no.2 2010

editor-in-chief and publisher Hisao Shigekane

Corporate R & D Headquarters

Fuji Electric Journal Editorial Office

Fuji Electric Holdings Co., Ltd. reserves all rights concerning

The Current Status and Future Outlook for

Fig.1 Requested characteristics of IGBT modules

power modules are being deployed in various fields.

3. Next Generation Power Modules

The Current Status and Future Outlook for Power Semiconductors

Both SiC and GaN are promising candidates as the

superior V-Series IGBTs.

oped, and realize both high performance and small size

5. Power Supply Control ICs

To help achieve energy savings and lower noise

As automotive devices, IPSs (intelligent power

Fig.3 Dependence of power supply efficiency on load current

FA5592 (for AC230 V input)

Load factor (%)

Vol. 56 No. 2 FUJI ELECTRIC REVIEW

in the inverter systems that drive the motor in hybrid

With protection of the global environment being an

The Current Status and Future Outlook for Power Semiconductors

3.3 kV IGBT Modules

Figure 1 shows an external view of 3.3 kV 12 kA

(1) IGBT chip characteristics

3.15 V (typical value)

3.15 V (typical value)

3.15 V (typical value)

Forward voltage (chip)

2.75 V (typical value)

2.75 V (typical value)

2.75 V (typical value)

Fig.2 VCE(sat)-I C characteristics

Fig.3 VF-IF characteristics

Forward current I(A)

Collector current I(A)

Collector-emitter saturation voltage

3.2 VCE(sat)-IC characteristics and VF-I F characteristics

Figure 2 shows the VCE(sat)-IC characteristics.

chip was designed with a structure that suppresses the

Fig.4 Switching waveforms (inductive load)

VCE : 500 V /div

Reverse recovery waveform

3.3 Switching characteristics

Figure 4 shows the turn-on, turn-off and reverse

3.3 kV IGBT Modules

modules used in high-power inverter system. The

Figure 5 shows the general internal structure of a

In the design of the main terminal, the following