FUJI ELECTRIC REVIEW Vol.58

FUJI ELECTRIC REVIEW Vol.58
Whole Number 237
ISSN 0429-8284
Power Semiconductor
contributing in energy and
environment region
Power Semiconductor
contributing in energy and
environment region
CONTENTS
Power Semiconductor contributing in energy and environment region
IGBT Module Series for Advanced-NPC Circuits
50
Direct Liquid Cooling IGBT Module for Automotive Applications
55
Expanded Lineup of High-Power 6th Generation IGBT Module Families
60
New Lineup of Large-Capacity “V-Series” Intelligent Power Modules
65
Hybrid Si-IGBT and SiC-SBD Modules
70
Packaging Technologies for SiC Power Modules
75
“Super J-MOS” Low Power Loss Superjunction MOSFETs
79
P-Channel Power MOSFETs for Space Applications
83
Cover photo:
To help prevent global warming, the
popularization of clean energy such as solar
power and wind power, as well as the application of power electronics technology
that efficiently utilizes this energy, are
highly anticipated worldwide.
To response to this heightened anticipation, Fuji Electric has developed various
types of power semiconductors having high
energy conversion efficiencies and that are
environmentally friendly. These devices
are utilized in many products in a wide
range of fields, including the environment,
power generation, automotive, industrial
machine, public infrastructure and consumer electronics fields, and are making
positive contributions worldwide.
The cover photo is of automotive-use
insulated gate bipolar transistor (IGBT)
modules that are installed in eco-friendly
vehicles such as electric vehicles or hybrid
vehicles. Featuring a direct water-cooling
structure, these IGBT modules achieve significant reductions in thermal resistance
and size in comparison to previous models.
Supplemental Explanation
3-level inverter technology, Miller period
87
FUJI ELECTRIC REVIEW vol.58 no.2 2012
date of issue: May 20, 2012
editor-in-chief and publisher Naoya Eguchi
Corporate R & D Headquarters
Fuji Electric 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.
1, Fujimachi, Hino-shi, Tokyo 191-8502, Japan
Fuji Electric 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.
IGBT Module Series for Advanced-NPC Circuits
Kosuke Komatsu † Takahito Harada † Yoshiyuki Kusunoki †
ABSTRACT
A series of insulated gate bipolar transistor (IGBT) modules has been developed to enable advanced neutralpoint-clamped (A-NPC) inverters. Modules in this series integrate A-NPC circuits for three phases with thermistors in
a single package. Loss is minimized by the adoption of 6th-generation IGBT, free wheeling diode (FWD) and reverse
blocking IGBT (RB-IGBT) devices. Power dissipation is reduced by 51% compared to conventional two-level inverters and by 33% compared to conventional NPC three-level inverters. Two types of pin configuration are available,
and selectable according to customer requirements.
1. Introduction
In recent years, initiatives to reduce CO2 emissions
in order to protect the environment have been implemented in countries throughout the world. The shift to
clean energy, such as to wind power and solar power,
which does not rely on conventional fossil fuels, is becoming increasingly prominent.
The use of power electronics devices to conserve
energy can be found in a wide variety of applications,
from consumer electronics to electric railways, FA systems and the like. Moreover, power electronics are
used not only in power-consuming applications, but
their use has also spread to the fields of power generation, transmission and supply such as in uninterruptible power supplies (UPS), wind power generators
and solar power generators. In particular, multi-level
inverters have been proposed as an efficient way to
increase the power conversion efficiency of a UPS or
power generation system(1), and neutral-point-clamped
(NPC) inverters have been put into practical use. A
3-level inverter*1 having a simpler circuit configuration
than this NPC inverter has also been proposed, but
when configured with typical insulated gate bipolar
transistor (IGBT) and diode, an increase in conduction
loss and a high surge voltage due to the wiring inductance were problems.
Fuji Electric has developed circuit systems for inverters and converters, which are power electronics
devices, and has contributed to energy conservation
mainly in devices in the industrial field. Additionally,
by adopting a custom low inductance package using
*1: 3-level inverter technology; See supplemental explanation 1 on page 87.
†
50
Fuji Electric Co., Ltd.
a reverse-blocking IGBT(2) (RB-IGBT), a proprietarily
developed power semiconductor, Fuji Electric has developed an IGBT module for use in advanced NPC
(A-NPC) circuits that solves the aforementioned problems(3). A UPS that utilizes this module has been introduced to the market.
Presently, Fuji Electric is aiming to expand its
series of IGBT module for A-NPC circuits, and is developing an IGBT module for A-NPC circuits that integrates a three-phase A-NPC 3-level inverter circuit
and a thermistor into a single package. This paper
presents an overview of these efforts.
2. Characteristics of IGBT Modules for Advanced
NPC Circuits
2.1 Overview
An overview of the ratings, dimensions and the
like of Fuji Electric’s IGBT module series for A-NPC
circuits is shown in Table 1. The rated voltage of the
main switches is 1,200 V, the rated voltage of the intermediate bidirectional switches is 600 V, and the rated
current is 100 A. The modules have the following characteristics.
(a) Integration of a 3-phase A-NPC circuit and a
thermistor into a single package
(b) Selectable pin shape according to the inverter
production line
Figure 1(a) shows the appearance and Fig. 1(b)
shows the equivalent circuit of the IGBT modules.
2.2 Electrical characteristics of the device
(1) Main switches
For the main switches T1 and T2 (see Table 2), the
new “V Series” IGBT and free wheeling diode (FWD)
having a rated voltage of 1,200 V were used. The V
Series has the following characteristics.
Overview of IGBT modules for A-NPC circuits
Model
Package dimensions
12MBI100VN-120-50
(Solder pin type)
Rated voltage
1,200 V (M10Aain switch part)
600 V (Bidirectional switch part)
L122.5×W62.5×H17 (mm)
12MBI100VX-120-50
(Press-fit pin type)
Rated current
Table 2
100 A (Main switch part)
100 A (Bidirectional switch part)
NPC inverter and A-NPC inverter on-state voltage
comparison
issue: Power Semiconductor contributing in energy and environment region
Table 1
IC =100 A, VG E =+15 V, Tj = 25 °C
Current path
Mode 1
Mode 2
Mode 3
Mode 4
NPC inverter
3.20 V
3.20 V
3.20 V
3.20 V
A-NPC inverter
1.85 V
1.85 V
2.45 V
2.45 V
Solder pin type
Mode 1
P
Press-fit pin type
(a) Appearance
P
T1Gu
T3Gu
T3Eu
M
T3Gv
T3Ev
T3Gw
T3Ew
TH1
T1Gw
T1/T4Eu T1/T4Ev T1/T4Ew
T4Gu
TH2
U
V
D2
Mode 3
T1
Mode 4
T4
M
T4
U
U
T3
T3
Mode 3
T2
N
T1
N
Mode 2
(a) NPC inverter
T2
Mode 2
(b) A-NPC inverter
T4Gv
T2Gu
T4Gw
N
T1Gv
Mode 4
D1
M
Mode 1
P
T2Gv
T2Eu
W
T2Gw
T2Ev
T2Ew
(b) Equivalent circuit
Fig.1 Appearance and equivalent circuit of IGBT module for
A-NPC circuits
(a) Lower on-state voltage VCE (sat) and less switching loss due to optimized field stop (FS) and
trench gate structures
(b) Improved controllability of turn-on di / dt with
gate resistance Rg
(2) Bidirectional switches
For the bidirectional switches T3 and T4 (see Table
2), RB-IGBTs having a rated voltage of 600 V were
used. The RB-IGBT characteristics are as follows.
(a) RB-IGBT has reverse blocking voltage capability, and can therefore be connected in a anti
parallel configuration to enable bidirectional
switching.
(b) When a forward gate bias voltage is applied
to cause the chip to exhibit reverse recovery
switching as a FWD, the reverse recovery characteristics are the same as that of a conventional FWD.
(3) Conduction loss
An A-NPC inverter circuit, as compared to a conventional NPC inverter circuit, has half the number
IGBT Module Series for Advanced-NPC Circuits
Mode A
Mode B
Mode C
Fig.2 Example of current paths in each switching mode
of conducting elements throughout its entire current
path. As a result, conduction loss can be reduced by
approximately 30% compared to a conventional NPC
inverter. Table 2 compares the current paths and onstate voltages of the conventional NPC inverter and
the A-NPC inverter.
(4) Switching loss
An IGBT module for use in A-NPC circuit differs
from a conventional IGBT module in that it has the following three switching paths as shown in Fig. 2.
(a) Path in which the main IGBTs operate as
switches and the main FWDs operate in reverse
recovery (Mode A)
(b) Path in which the RB-IGBTs switch and the
main FWDs operate in reverse recovery (Mode
B)
51
VGE : 10 V/div
0V
VGE : 10 V/div
VAK: 100 V/div
0V
IF: 50 A/div
VCE: 100 V/div
IC: 50 A/div
0V
0A
VCE: 100 V/div
t: 200 ns/div
(a) Turn-on waveform
0V
0A
IC: 50 A/div
0V
0A
t: 200 ns/div
(b) Turn-off waveform
t: 200 ns/div
(c) Reverse recovery waveform
Fig.3 Switching waveform (mode B)
VCC =300 V, VGE = ± 15 V,
Rg = 24 Ω, Tj =125 °C
25
Eon
Switching loss (mJ)
Switching loss (mJ)
15
10
Eoff
5
Err
0
0
50
100
150
Current IC, IF (A)
200
10
Eoff
5
Err
1
10
100
Gate resistance Rg (Ω)
Switching loss (mJ)
Switching loss (mJ)
10
VCC =300 V, VGE = ± 15 V,
Rg = 1.6 Ω, Tj = 125 °C
10
Eon
Eoff
5
Err
0
0
50
100
150
Current IC, IF (A)
1,000
(a) Mode B
(a) Mode B
15
Eon
15
0
250
VCC =300 V, VGE =±15 V,
IC , IF =100 A, Tj =125 °C
20
200
VCC =300 V, VGE =±15 V,
IC , IF =100 A, Tj =125 °C
5
Eoff
Eon
Err
0
0.1
250
1
10
Gate resistance Rg (Ω)
(b) Mode C
(b) Mode C
Fig.4 Current dependence of switching loss
(c) Path in which the main IGBTs switch and the
RB-IGBTs operate in reverse recovery (Mode C)
For the 3-level inverter operation, basic operation
is in mode B and mode C. Figure 3 shows the turn-on,
turn-off and reverse recovery waveforms in mode B
for a module at VCC = 300 V, IC = 100 A, Rg = 24 Ω and
Tj =125 °C.
The switching loss is 3.0 mJ at turn-on, 4.1 mJ at
turn-off, and 1.67 mJ at reverse recovery, and no turnoff surge that exceeded the rated voltage was found.
Figure 4 shows the current dependence of the
switching loss, and Fig. 5 shows the gate resistance
dependence of the switching loss. As described above,
the reverse-recovery loss characteristics when the RBIGBT is in reverse-recovery mode C are no different
from when a conventional FWD is in reverse-recovery
mode B.
100
Fig.5 Gate resistance dependence of switching loss
U
V
W
P
M
N
Fig.6 Main terminal arrangement
2.3 Package
A conventional compact package (EconoPIM™*2 3 /
TM
*2: EconoPIM is a trademark or registered trademark of
Infineon Technologies AG.
52
Vol. 58 No. 2 FUJI ELECTRIC REVIEW
Fig.7 Terminal shape
PC-Pack3) was selected for the newly developed IGBT
module for A-NPC circuits. As a result, the package
has the following characteristics.
(a) Main terminals P, M, N
Layout allows for easy placement of snubber
capacitors (between P-M, between M-N) to reduce
surge voltage (see Fig. 6)
(b) Terminal shape
2 types of terminal shapes (solder pin, press-fit
pin) can be selected according to customer needs
(see Fig. 7)
(c) Environmentally friendly
Lead-free and compliant with RoHS directive*3
3. Power Dissipation
Figure 8 compares the power dissipation per phase
of a conventional 2-level inverter, an NPC 3-level inverter and an A-NPC 3-level inverter when operating
under the same conditions.
The power dissipation was computed with the V
Series ratings of 100 A / 1,200 V (EconoPIM™ 3) for the
conventional 2-level inverter and using the V Series
ratings of 100 A / 600 V (EconoPIM™ 3) for the NPC
3-level inverter. Operating conditions for the 20 kVA
inverter were fc = 7 kHz, DC voltage = 700 V, and output
current = 30 A (rms). As a result, the A-NPC 3-level inverter exhibited the least power dissipation, 51% less
than the conventional 2-level inverter and 33% less
than the NPC 3-level inverter. Viewed individually,
these inverters have the following characteristics.
The 2-level inverter has the smallest conduction
loss since it has only one device that conducts current.
However, because the DC voltage is twice that of a
3-level inverter, switching loss accounts for 76.6% of
the total power dissipation.
The NPC 3-level inverter has the largest conduction loss since there are two conducting devices in each
current flow path. With three levels, however, the DC
voltage is halved and the switching loss is less than
half that of a 2-level inverter. Consequently, the conduction loss accounts for 54.8% of the power dissipation per phase. As the carrier frequency decreases,
*3: RoHS directive: European Union (EU) directive on restriction of the use of certain hazardous substances in
electrical and electronic equipment
IGBT Module Series for Advanced-NPC Circuits
Switching loss
120
20 kVA inverter
fC =7 kHz, VDC = 700 V,
IO (RMS value) = 30 A
108.1
100
112.9
(76.6%)
80
48.9
(45.2%)
72.3
60
Conduction
loss
40
20
34.4
(23.4%)
0
2-level
inverter
34.9
(48.2%)
59.2
(54.8%)
NPC 3-level
inverter
issue: Power Semiconductor contributing in energy and environment region
(b) Press-fit pin
147.3
140
37.5
(51.8%)
A-NPC 3-level
inverter
Fig.8 Comparison of power dissipation for various inverters
Power dissipation per phase (W)
(a) Solder pin
Power dissipation per phase (W)
160
400
350
400
NPC 3-level
inverter
2-level inverter
250
200
150
100
A-NPC 3-level
inverter
50
0
0
5
10
15
20
Carrier frequency fC (kHz)
25
30
Fig.9 Carrier frequency dependence of power dissipation
conduction loss accounts for a higher percentage of total power dissipation.
With an A-NPC 3-level inverter, because an RBIGBT has larger conduction loss than an ordinary
IGBT, the conduction loss is 8.7% larger than that of
a 2-level inverter, but can be reduced to about 37%
less than an NPC 3-level inverter. On the other hand,
the switching loss is smaller as in the case of the NPC
3-level inverter. As a result, in contrast to the loss in
the NPC 3-level inverter, the percentages of conduction
loss and switching loss become equal, and even if the
carrier frequency changes, the power dissipation never
exceeds that of the NPC 3-level inverter.
Figure 9 shows the carrier frequency dependence of
power dissipation. In the region of carrier frequencies
of 5 kHz or higher, the power dissipation is less for a
3-level inverter than a 2-level inverter. In the region
of carrier frequencies lower than 5 kHz, the 2-level inverter appears to have less power dissipation, but the
noise filter attached to the inverter apparatus is larger
for the 2-level inverter than a 3-level inverter, and as a
result, the total loss (power dissipation including fixed
loss and filter loss) generated by the entire inverter apparatus, the 3-level inverter has less power dissipation.
53
4. Postscript
This paper has presented an overview and described characteristics of Fuji Electric’s IGBT module
series for advanced-NPC circuits. This product supports applications of several tens of kVA, and will surely satisfy customer requests for high efficiency, small
size and ease of use.
In the future, Fuji Electric will expand the product
lineup of this IGBT module series for advanced-NPC
circuits, and intends to develop modules in response to
requests for higher efficiency in UPSs and the like.
54
References
(1) Nabae, A. et al. “A New Neutral-Point-Clamped PWM
Inverter,” IEEE Trans. on I. A., 1981, vol.IA-17, no.5,
p.518-523.
(2) Takei, M. et al. “The Reverse Blocking IGBT for Matrix
Converter with Ultra-Thin Wafer Technology,” Proc. of
ISPSD ‘03, 2003, p.156-159.
(3) Komatsu, K. et al. “New IGBT Modules for Advanced
Neutral-Point-Clamped 3-Level Power Converters,”
Proc. of IPEC ‘10, 2010, p.523-527.
Vol. 58 No. 2 FUJI ELECTRIC REVIEW
Direct Liquid Cooling IGBT Module for
Automotive Applications
Takahisa Hitachi † Hiromichi Gohara † Fumio Nagaune †
A compact insulated gate bipolar transistor (IGBT) module with low thermal resistance and direct liquid cooling,
system has been developed to contribute to reducing the size of user systems. Thermal fluid simulations were used
to optimize the liquid cooling fin shape. Square pin fins were selected because of their overall outstanding performance, including heat dissipation performance, cooling liquid flow velocity, and pressure loss. Under optimized liquid
flow conditions, measured values of chip temperature were within 2% of simulation values, confirming the accuracy
of the simulations. From these results, this IGBT module for direct liquid cooling realizes a 30% reduction in thermal
resistance and allow a 40% reduction in size compared to the conventional configuration.
1. Introduction
Hybrid electric vehicles (HEVs), currently the most
prevalent type of fuel-efficient cars, have the dual power sources of a internal combustion engine and an electric motor. During acceleration, the motor assists the
engine, while during deceleration, regenerative braking acts to charge the battery to improve the fuel efficiency. Because a gasoline engine is used, HEVs can
be operated without reliance on quick charger or other
infrastructure, and their further popularization in the
future is anticipated.
This paper introduces an automotive insulated
gate bipolar transistor (IGBT) module suitable for
use in drive system inverters in HEVs and electric
vehicles (EVs). In order to meet many diverse needs,
Fuji Electric plans to create a series of products having
specific output power classes, suitable for motor output
ratings of up to about 100 kW, and to contribute to the
development of fuel-efficient systems for customers
such as automobile manufacturers and electric equipment manufacturers.
2. Background
Types of hybrid systems include a 2-motor type
equipped with both a drive motor and a generator motor, and a 1-motor type that assists the engine and
performs regenerative braking. The 2-motor type is
installed mainly in mid-size or large passenger cars
and results in a relatively large improvement in fuel
efficiency. The 1-motor type is installed mainly in
compact cars and although a dramatic improvement
in fuel efficiency is not expected, is a small lightweight
and relatively low-cost system. In a 2-motor type hy†
Fuji Electric Co., Ltd.
brid system, motors having a relatively large output of
50 kW or greater are used, while for the 1-motor type,
motors with output power of 20 kW or less are often
used.
Electric vehicles have a battery and a motor as
their power source, and do not generate harmful exhaust gas since they do not use fossil fuels when being
driven. Additionally, because they have a high wellto-wheel efficiency (efficiency from the primary energy
source to actual driving of the vehicle), electric vehicles
exhibit a large energy-saving effect and are thought to
be the ultimate fuel-efficient car. Motors installed in
electric vehicles range from 50 kW to 100 kW according
to the body size. A significant reduction in the cost of
their batteries, however, is needed in order for electric
vehicles to achieve wide-spread adoption. In the meantime, plug-in hybrid electric vehicles (PHEVs) that
have been developed to realize higher fuel efficiency
performance are being commercialized. Batteries can
be charged from household AC power sockets, and
through using electric vehicles for short-distance travel
and hybrid vehicles that combine an electric motor
with engine output for long-distance travel, a dramatic
reduction in CO2 emissions and an increase in cruising
distance are anticipated.
In this way, many types of systems are equipped
with a motor to save energy, and their output capacities vary from small to large. Accordingly, IGBT modules of various current capacities and voltage classes
are used in the inverters in these systems.
A hybrid system is configured from such components as a power control unit (PCU) that controls the
power converting function, a motor, a battery and so
on. Inside the PCU, IGBTs play a major role in acting
as the main switch of an inverter that outputs threephase alternating current.
Many varieties of IGBT modules have been devel-
55
issue: Power Semiconductor contributing in energy and environment region
ABSTRACT
oped for industrial and consumer applications, and
their product lineup is also diverse in terms of current
capacity. Because the specifications concerning durability of these IGBT modules were not entirely suitable
for automotive applications, however, they could not
be used as-is, and instead, custom products were often
developed and installed. In the future, as electric drive
technology becomes indispensible for reducing fuel consumption and various customer requirements for IGBT
modules are received, their use is expected to increase
further.
3. Product Concept and Specifications
3.1 Product concept
While meeting diverse customer requirements, We
have also considered products that incorporate ideas of
its own. With the goal of “contributing to miniaturization of the customer’s system,” Fuji Electric has significantly improved the heat dissipation performance
of IGBT modules while maintaining the required basic
performance in order to enhance the power density.
To reduce the size of the module, a direct liquidcooling structure was adopted and thermal fluid simulations were carried out in order to optimize the fin
shape and the liquid flow. A 30% reduction in thermal
resistance, compared to a module with a conventional
structure, was confirmed. Additionally, reducing the
active area of the power chip (die shrink) was also
studied. As a result, We determined that a module
size 40% smaller than that of a module with a conventional structure could be realized, and began product
development. Use of the latest version “V-series” chip,
which has already begun to be mass-produced for industrial applications, as the power chip provides an
even greater effect.(1)
Next, the design of a module that conforms to the
usage conditions (battery voltage: 300 V, max. current
(RMS value): 200 A, carrier frequency: 10 kHz) actually requested by customers was examined, and the
results of evaluation of prototypes will be described
below.
3.2 Product specification
Figure 1 shows the appearance of the IGBT module
and Fig. 2 shows its equivalent circuit diagram.
This product is characterized by low thermal resistance as a result of a direct liquid cooling structure,
and the high current density has enabled the package
size to be reduced significantly. The module has external dimensions of 105× 108 (mm), which is approximately 40% smaller than our previous comparable
product having a conventional structure. Figure 3 compares cross-sections of the conventional structure and
the liquid cooling structure, and Fig. 4 compares their
thermal resistances.
In the conventional structure, thermal grease is
used to reduce the contact thermal resistance between
a copper base and the cooling fin surface. Thermal
grease has a large thermal resistance even at thick-
Wire
Terminal
Sealing
Chip material
Wire
Sealing
Chip material
Case
Insulated
substrate
with metal
pattern
Al fin
Solder
Grease layer
(a) Conventional structure
Base
Base cover
Fin Solder
(b) Direct liquid cooling structure
Fig.3 Cross-sectional comparison of conventional structure
and direct liquid cooling structure
(a) Surface
(b) Underside
P
9C
3G
2E
5G
4E
U
7 T1
6E
V
W
8 T2
11 G
13 G
15 G
10 E
12 E
14 E
Fig.2 Equivalent circuit of the module
56
NTC
1G
Thermal resistance comparison (%)
120
Fig.1 Appearance of M651 (650 V/ 400 A) module
100
Solder
Approx. 30%
reduction
Insulated
substrate
with metal
pattern
60
40
Solder
Base
20
0
N
Chip
80
Grease
Cooling fin
Conventional
structure
Direct liquid
cooling structure
Fig.4 Thermal resistance comparison of conventional structure and direct liquid cooling structure
Vol. 58 No. 2 FUJI ELECTRIC REVIEW
4. Direct Liquid Cooling Technology
4.1 Investigation of heat dissipation performance
With the direct liquid cooling structure, the heat
dissipation performance changes according to such factors as the fin shape and the flow direction of the cooling liquid. Fluid simulations were performed for various fin shapes and flow paths in order to optimize the
cooling system, and details are presented below.(2)
The IGBT module in an inverter system fulfills the
function together with connected smoothing capacitor,
circuit board, harness, current sensor and the like. So
as to accommodate different system designs from many
Round pin fin
customers, the fin shape was selected to be a pin-type
fin through which cooling fluid could flow in either the
width or depth direction. (see Fig.6) Selection of the
pin fin cross-sectional shape necessitated a comprehensive examination of the heat dissipating performance,
cooling liquid flow velocity and pressure loss, and Fuji
Electric investigated the cross-sections of typical round
and square pin fins.
4.2 Optimization of flow path
The flow path was investigated using a thermal
fluid simulator (Icepak*1), and the junction temperature, flow velocity of the cooling liquid and pressure
loss were computed and compared. Class 2 antifreeze
coolant (LLC) 50% is used as the cooling liquid, and
the model of the IGBT module shown in Fig. 6 is used
as the analysis model. The fins are formed on the underside of the insulated substrate where the heat-generating chip is located. One IGBT chip and one FWD
chip are provided in each arm. A single output phase
(half bridge) is provided on each insulated substrate.
In consideration of the inverter system commonly used
and so that the cooling liquid will flow evenly to the fin
area, the depth direction shown in Fig. 6 was selected
as the direction of fluid flow. Figure 7 shows the structure of a cooling liquid jacket that compares the flow
velocities of the cooling liquid.
With the fluid flow direction shown in Fig. 6, the
flow of cooling liquid in the fin area becomes nearly
constant and uniform. A flow velocity distribution that
is uniform, from the liquid inlet to below the insulated
substrate, is obtained for each phase, and the cooling
performance in each phase is predicted to be equal.
Moreover, in order to reduce the pressure loss, because
Square pin fin
Liquid inlet
Liquid outlet
Fig.5 Cooling fin shape
Fluid flow
direction
IGBT
Liquid inlet
Flow
velocity
Fast
Output terminal
Slow
A
B
C
D
FWD
E
F
Pressure loss
: 6.0 kPa
Liquid outlet
Insulated substrate
Fig.7 Fluid flow velocity distribution in cooling liquid jacket
PN terminal side
Base
Fig.6 Thermal fluid simulator model
Direct Liquid Cooling IGBT Module for Automotive Applications
*1: Icepak is a trademark or registered trademark of USbased ANSYS, Inc. and its subsidiaries.
57
issue: Power Semiconductor contributing in energy and environment region
nesses of several tens of μm, and was unable to conduct waste heat efficiently from the module to the cooling fins. Additionally, an adequate cooling effect could
not be obtained since the fluid flow path of the cooling
fins did not match the distribution of heat generated
by the mounted device. With the newly developed direct liquid cooling structure, the cooling effect was
improved by integrating the copper base and the cooling fins so as to eliminate the thermal grease layer.
Moreover, by arranging the fins in a high density configuration directly beneath the power chip, which is a
heat-generating body, the capacity for heat dissipation
between the fins and the cooling liquid is increased.
The result is that the thermal resistance between the
power chip and the cooling liquid was reduced by approximately 30% compared to that of the conventional
structure. By improving the cooling efficiency, the device can be made with higher current density so that
more current can flow through a single chip (see Fig.5).
the cooling liquid traverses a short distance in passing
through the fin area, and the flow velocity of the cooling liquid must be constant in the fin area, the flow
path of Fig. 7 is found to enable efficient cooling.
4.3 Selection of the fin shape
Using the cooling liquid jacket of Fig. 7, thermal
fluid simulations were performed to select the fin
shape in consideration of its effect on junction temperature and pressure loss. The analysis conditions assumed loss at the time of inverter operation, heat generation of 258 W by the IGBT and 31 W by the FWD,
a cooling liquid of LLC 50%, a cooling liquid flow rate
of 10 L / min, and a cooling liquid temperature of 65 °C.
Figure 8 shows the simulation results of the rise in
junction temperature. The junction temperature was
verified and compared for each fin shape. The maximum junction temperature was 141.6 °C for round pin
fins and 136.0 °C for square pin fins, while the pres-
Round pin fin
Square pin fin
Liquid inlet
Liquid inlet
Liquid
outlet
Liquid outlet
Temperature
High
Temperature
High
Low
Low
sure loss was 4.8 kPa for round pin fins and 6.0 kPa for
square pin fins. Figure 9 compares the IGBT junction
temperature and pressure loss for round pin fins and
square pin fins. The round pin fins have a smaller fin
volume density, and therefore the pressure loss is less.
On the other hand, because the square pin fins have a
larger surface area, the junction temperature decreases, but the fin volume density increases and the pressure loss becomes greater. The difference in pressure
loss was determined to be about 1 kPa, which is not
considered to be a significant difference, and therefore
square pin fins were selected so as to fully utilize the
cooling performance of direct liquid cooling.
4.4 Actual measurement results of junction heat generation
To confirm the validity of the simulation of the
cooling performance, the rise in junction temperature
of an actual sample was verified. So that the measurement conditions matched those of the simulation, the
following conditions were used.
™Loss: IGBT 258 W, FWD 31 W
™Cooling liquid: LLC 50%
™Flow rate: 5 to 15 L / min
™Cooling liquid temperature: 65 °C
In consideration of the aforementioned results, the
water jacket used for the measurements was fabricated
based on the model of Fig. 7. (see Fig. 10)
A comparison of the square pin fin simulation
and measurement results in the case of a flow rate of
10 L / min is shown in Table 1. Columns A through
F in Table 1 show the junction temperatures of each
Fig.8 Simulation results of junction temperature rise
18
Round
pin
155
15
150 Square
12
145
9
140
6
135
3
130
0
pin
Pressure loss (kPa)
IGBT junction temperature (°C)
160
Fig.10 Prototype of cooling liquid jacket
Table 1
0
5
10
15
20
Flow rate (L/min)
Fig.9 Flow rate dependence of round pin fin and square pin fin
IGBT junction temperature and pressure loss (simulation)
58
Comparison of junction heat generation by simulation
and actual measurement (IGBT)
(Units: °C)
F
A
B
C
D
E
Actual
measurement
133.6
137.6
138.4
139.1
136.6
137.7
Simulation
136.7
137.4
137.1
137.6
136.9
136.9
Difference
2.3%
0.1%
0.9%
1.1%
0.2%
0.6%
Vol. 58 No. 2 FUJI ELECTRIC REVIEW
0.40
12
FWD
0.30
9
IGBT
0.20
6
0.10
3
0
0
5
10
15
20
0
Flow rate (L/ min)
Fig.11 Flow rate dependence of thermal resistance and pressure loss (actual measured values)
5. Postscript
A direct liquid cooling-type IGBT module for automotive applications has been introduced. The 400 A
rated product introduced in this paper will serve as a
stepping stone for Fuji Electric’s planned development
of high-current rated 600 A and 900 A products in the
future. We intends to develop a series of products that
are widely applicable to inverter systems for motor
outputs up to about 100 kW.
We will continue to develop highly reliable, highperformance modules that facilitate system design for
an increasing number of users, and to reduce the environmental impact of automobiles.
chip in Fig. 6. The difference for each phase is at most
about 2%, which confirms the equivalence of the simulation and the measurement results. From the measured results, the thermal resistance was calculated
to be 0.27 K / W (IGBT average value). While taking
the measurements, the flow rate was varied in order
to confirm the flow rate dependence of thermal resistance. Figure 11 shows the flow rate dependence (actual measured values) of thermal resistance and pressure loss for flow rates of 5 to 15 L / min.
Comparing the 5 L / min and 15 L / min flow rates,
it can be seen that the IGBT and the FWD both ex-
References
(1) Nakano, H. et al. 600 V trench-gate IGBT with Micro-P
structure (Proceedings of the 21th International
Symposium on Power Semiconductor Devices and ICs).
2009, p.132-135.
(2) Nagaune, F. et al. Small Size and High Thermal
Conductivity
IGBT
Module
for
Automotive
Applications. PCIM Europe 2011, p.785-790.
Direct Liquid Cooling IGBT Module for Automotive Applications
59
issue: Power Semiconductor contributing in energy and environment region
15
Pressure loss (kPa)
Thermal resistance (junction–liquid) (K/W)
0.50
hibited a decrease in thermal resistance by about 10%
at 15 L / min, and that by increasing the flow rate, the
heat dissipation performance was found to improve.
Increasing the flow rate, however, results in a greater
loss of pressure, and therefore optimization of the
pump performance during use and of the flow path design are needed.
Expanded Lineup of High-Power 6th Generation
IGBT Module Families
Takuya Yamamoto † Shinichi Yoshiwatari † Hiroaki Ichikawa †
ABSTRACT
To respond to growing demand in the renewable energy sector, including wind and solar power, Fuji Electric has
expanded the lineup of modules in its high-power insulated gate bipolar transistor (IGBT) module families. These
new high-power modules feature 6th-generation “V-Series” IGBTs. Operation is guaranteed at maximum junction temperatures up to 175 °C, and the modules deliver industry leading low on-voltage and low switching loss.
Reliability is higher than conventional products due to the application of the latest packaging technology, including ultrasonic welded terminals and highly reliable lead-free solder.
1. Introduction
2. Product Lineup
Insulated gate bipolar transistor (IGBT) modules
are used widely due to their advantages of low loss,
high breakdown resistance, ease of drive circuit design
and so on. In the field of high-voltage and high-power
device applications, the heretofore widely-used gate
turn-off (GTO) thyristors are being replaced with IGBT
modules, and IGBT modules are being applied widely
to high-power inverters and high voltage inverter
units.
In recent years, for the prevention of global warming, the market for renewable energy (wind power
generation, solar power generation) has been growing rapidly. In this field, power conversion equipment
has progressed to higher capacities, and in particular,
the need for high-power IGBT modules has increased
greatly. For applications in this field, Fuji Electric has
previously developed the high power module (HPM)
and PrimePACKTM *1 product series.(1)(2)
Recently, in response to diverse customer
needs, Fuji Electric has expanded the HPM and
PrimePACKTM product series. Equipped with Fuji
Electric’s 6th generation “V-Series” IGBTs(3), these
products achieve the industry’s leading level of low
on-voltage and, at the same time, low switching loss.
Additionally, the latest package technology is applied
to realize high power density and high reliability.
This paper presents an overview and describes
the characteristics of Fuji Electric’s “V-Series HPM
Family” of high-power 6th generation IGBT modules.
Figure 1 shows the appearance of the V-Series
HPM Family packages. The PrimePACKTM product
series consists of 2-in-1 and chopper module circuit
*1: PrimePACK is a trademark or registered trademark of
Infineon Technologies AG.
TM
†
60
Fuji Electric Co., Ltd.
M272
M271
(a) PrimePACKTM
M152/ M156
M151/ M155
M256/ M278
(b) HPM
Fig.1 Appearance of V-Series HPM Family packages
V-Series HPM Family product lineup
Product
lineup
E-type
PrimePACKTM
P-type
E-type
Rated
voltage
(V)
Product type
2MBI600VXA-120E-50
2MBI900VXA-120E-50
2MBI900VXA-120P-50
1,200
1,400
2MBI650VXA-170E-50
650
2MBI1000VXB-170E-50
1,700
1,000
1,600
1MBI2400VC-120*1
2,400
1,200
Industrial-use HPM
172 × 89 × 38
M272
250 × 89 × 38
M271
172 × 89 × 38
M272
250 × 89 × 38
M271
172 × 89 × 38
M272
250 × 89 × 38
M151
130× 140 × 38
M152
190× 140 × 38
M256
130× 140 × 38
Base material
Base thickness
Al2O3
Copper
3 mm
Si3N4
Copper
5 mm
AlN
Al SiC
1-in-1
2,400
3,600
600
2MBI600VG-120*1
E-type
Traction-use HPE
1,200
1MBI1600VC-120*1
1MBI3600VD-120*
M271
Insulating
substrate
1,400
1MBI1200VC-120*1
1
Package size
(mm)
Chopper
1MBI1000VXB-170EH-50
1MBI2400VD-120*1
2-in-1
650
1MBI650VXA-170EL-50
1MBI1000VXB-170EL-50
TBD*1
Package
type
1,000
2MBI1000VXB-170EA-50
1MBI650VXA-170EH-50
E-type
900
2MBI1400VXB-120P-50
2MBI1400 VXB-170P-50
Circuit
configuration
600
2MBI1400VXB-170E-50
P-type
Rated
current
(A)
issue: Power Semiconductor contributing in energy and environment region
Table 1
2MBI800VG-120*1
800
2MBI1200VG-120*1
1,200
1MBI1200VC-170E
1,200
1MBI1600VC-170E
1,600
1MBI2400VC-170E
2,400
1MBI2400VD-170E
2,400
1MBI3600VD-170E
3,600
2MBI600VG-170E
600
2MBI800VG-170E
800
2MBI1200VG-170E
1MBI1200VR-170E*2
1,700
130× 140 × 38
M152
190× 140 × 38
M256
130× 140 × 38
M155
130× 140 × 38
M156
190× 140 × 38
M278
130× 140 × 38
1-in-1
2-in-1
1,200
1,600
1MBI2400VR-170E*2
2,400
1MBI2400VS-170E*2
2,400
1MBI3600VS-170E*
3,600
2MBI600VT-170E*2
M151
1,200
1MBI1600VR-170E*2
2
2-in-1
1-in-1
600
2MBI800VT-170E*2
800
2MBI1200VT-170E*2
1,200
2-in-1
*1 : TBD:To Be Determined
*2 : underdevelopment
configurations, 1,200 V and 1,700 V class ratings, and
current capacities of 600 to 1,400 A. The HPM product series consists of 1-in-1 and 2-in-1 module circuit
configurations, 1,200 V and 1,700 V class ratings, and
current capacities of 600 to 3,600 A. Table 1 lists the
lineup of the V-Series HPM Family product series.
of Tjmax = 175 °C for momentary abnormal states, and
guarantees normal operation at an operating temBy improving reliability
perature of Tjop = 150 °C.
and breakdown resistance during high-temperature
operation, each of these temperatures was increased
by 25 °C compared to those of the 5th generation
“U-Series” IGBT modules.
3. Electrical Characteristics
3.1 IGBT chip characteristics
Incorporating a V-Series IGBT, the V-Series HPM
Family of products guarantees non-continuous operation up to a maximum chip junction temperature
Because a high-power IGBT module will instantaneously cut off a large current, the surge voltage
generated at turn-off is large. For the V-Series HPM
Expanded Lineup of High-Power 6th Generation IGBT Module Families
61
250
1,200
200
1,000
150
800
600
P-type
400
100
E-type
50
200
0
2.5
3.0
3.5
4.0
4.5
5.0
Switching loss: Eon, Eoff, Err (mJ/ pulse)
1,400
300
2,000
Switching loss: Eon, Eoff, Err (mJ/ pulse)
Tj = 25 °C
VCC = 900 V
IC = 115 A
Collector current (A)
Collector-emitter voltage (V)
1,600
2,000
Tj =150 °C
Tj =125 °C
1,000
Eoff
2MBI1400VXB-170E-50
0
0
1,000
2,000
Collector current IC (A)
(a) E-type
Time (μs)
Fig.2 Comparison of IGBT turn-off switching waveforms
Collector current IC (A)
3,000
Tj = 25 °C
125 °C
150 °C
2,000
Rated
current
1,000
Eon
Err
0
6.0
5.5
VCC =900 V,
Rg (on)= 0.47 Ÿ,
Rg (off)=0.68 Ÿ
Tj =150 °C
Tj =125 °C
3,000
VCC =900 V,
Rg (on)= 0.47 Ÿ,
Rg (off)=0.68 Ÿ
E off
E on
1,000
E rr
2MBI1400VXB-170P-50
0
0
1,000
2,000
Collector current IC (A)
(b) P-type
3,000
2MBI1400VXB-170E-50
0
0
1
2
3
4
Collector-emitter voltage VCE (V)
(a) E-type
5
voltage. Electrical characteristics are described below
for the example of a 1,700 V / 1,400 A module.
3,000
Collector current IC (A)
Tj = 25 °C
125 °C
150 °C
3.2 V-I characteristics
Figure 3 shows IC vs. VCE(sat) characteristics of the
module. Comparing the E-type and the P-type reveals
that at the rated current of IC = 1,400 A and Tj = 125 °C,
the characteristic of the P-type is about 0.4 V lower.
2,000
Rated
current
1,000
3.3 Switching characteristics
2MBI1400VXB-170P-50
0
0
1
2
3
4
Collector-emitter voltage VCE (V)
(b) P-type
5
Fig.3 VCE(sat)-Ic characteristics
Family, in addition to the previous (E-type) lineup of
V-Series IGBT chips, an IGBT chip product lineup
(P-type) having soft switching characteristics was
newly developed by adjusting the IGBT chip characteristics for applications in the high-power device field.
Figure 2 shows a comparison of the switching waveforms at turn-off for E-type and P-type 1,700 V-IGBT
chips. Compared to the E-type, the P-type has a slower di /dt at turn-off, and achieves a lower turn-off surge
62
Fig.4 Switching loss vs. current characteristic
Figure 4 shows the switching loss vs. current
characteristic. In terms of turn-on loss and reverse
recovery loss, the E-type and the P-type are the same,
but the turn-off loss is about 1.8 times larger for the
P-type.
As described above, the V-Series HPM Family contains two types of product lines with different IGBT
chip characteristics so that suitable products can be
provided for the drive conditions of our customers.
4. Package Structure
Power conversion equipment in the renewable energy field and elsewhere must have high reliability in
order to provide a stable supply of electric power.(4) The
V-Series HPM Family uses the latest package technology to ensure long-term reliability.
Vol. 58 No. 2 FUJI ELECTRIC REVIEW
4.1 Application of ultrasonic terminal welding technology
Figure 6 shows the external appearance and a
cross-sectional view of an ultrasonically welded terminal. This product uses ultrasonic terminal welding
to bond copper terminals directly to the copper circuit
Cu circuit
pattern
Wire bonding
Chip
Solder layer
Insulating
substrate
4.2 Improved power cycling capability
As shown in Fig. 5, thermal cycle stress occasionally causes cracks to form in the solder layer between
the copper base and the copper pattern under the substrate. With the PrimePACKTM series, tolerance to
high temperature cycling is achieved by using highly
Copper pattern
under substrate
Solder layer
Copper terminal tensile strength (%)
Copper terminal
pattern. In a conventional solder joint structure, the
greatest amount of stress is concentrated in the solder
layer due to difference in the coefficients of thermal expansion of the solder material and the copper material.
As a result, failure may result whereby cracks form in
the solder layer and the copper terminal is pulled out.
Figure 7 shows a comparison of the results of copper
terminal tensile strength tests before and after a thermal cycle test (test conditions: −40 to + 150 °C repeatedly). For the conventional solder joint, an approximate
50% decrease in tensile strength from the initial value
was confirmed after 300 cycles. On the other hand, almost no decrease in tensile strength was observed in
the case of ultrasonic welding. This is because the copper terminals and the copper circuit pattern are bonded together directly with ultrasonic terminal welding,
and there is no difference in the coefficients of thermal
expansion at the joint surface.
Base plate (copper)
Fig.5 Cross-sectional schematic view of an IGBT module
Table 2
Technologies and materials applied to the V-Series
HPM Family
HPM
PrimePACKTM
Industrial
use
Traction
use
Ultrasonic
welding
Solder
welding
Solder
welding
Insulating substrate
Al2O3
Si3N4
AlN
Solder material under
insulating substrate
Sn-Sb
Sn-Pb
Sn-Pb
Base material
Base thickness
Copper
3 mm
Copper
5 mm
AlSiC
5 mm
Terminal welding
method
140
120
100
50%
decrease
80
60
40
20
0
Initial
After 300
cycles
Ultrasonic welding
Initial
After 300
cycles
Solder joint
(conventional method)
Fig.7 Copper terminal tensile strength test results
107
Weld layer
Number of cycles
PrimePACKTM, industrial-use HPM
Copper terminal
Copper circuit pattern
(a) External appearance
Traction-use HPM
10
5
104
103
30
(b) Cross section
Fig.6 External appearance and cross-sectional view of
ultrasonically welded terminal
106
Fig.8
Expanded Lineup of High-Power 6th Generation IGBT Module Families
Previous product
Al2O3 +Cu
Si3N4 +Cu
AlN +AlSiC (ongoing test)
40
50
60
¨ TC (°C)
70
80
90
Tc power cycling capability
63
issue: Power Semiconductor contributing in energy and environment region
Figure 5 shows a cross-sectional schematic view
of an IGBT module. Connecting a conducting / blocking electrical load to an IGBT module causes thermal
stress is generated in the junction of the IGBT. The
use of materials having a low coefficient difference of
thermal expansion in the junction ensures high thermal cycling capability. Table 2 lists the technologies
and materials applied to the V-Series HPM Family.
The PrimePACKTM series uses ultrasonic welding technology and highly reliable lead-free solder material to
achieve higher reliability than in previous products.
The HPM product line uses a 5 mm-thick base, or an
AlSiC base for traction applications, to achieve even
longer term reliability.
crack-resistant Sn-Sb solder.
In traction-use HPMs, to ensure even higher reliability, an AlN substrate is used as the insulating substrate and an AlSiC base is used as the base material.
AlSiC is a composite of Al and SiC, and having a coefficient of thermal expansion close to that of the AlN substrate, achieves higher thermal cycling capability and
power cycling capability than in the case of a copper
base. In the simulated tests of actual operation shown
in Fig. 8, improved thermal cycling (ΔTc power cycling)
capability was realized. The V-Series HPM Family has
a power cycle capability of greater than 10,000 cycles
at ΔTc =80 °C, and realizes more than twice the ΔTc
power cycling capability as the previous product.
4.3 Improved environmental durability of molded case
When the surface of a molded case is placed under
a high electric field, dust and moisture adhering to
the molded case surface cause the surface to become
carbonized, and form a conductive path (track). This
degrades the insulating performance and may lead to
breakdown of the insulation. Wind and solar power
generating equipment are often installed in high humidity environments containing large amounts of dust
and salt. So that an IGBT module can be used in such
an environment while maintaining high reliability, the
development of a molded case on which a carbonized
conductive path is not easily formed is needed. This
product series uses a mold resin having a high comparative tracking index (CTI) of ≥ 600 to ensure high
anti-tracking performance.
4.4 Reduction of internal inductance
The V-Series HPM Family introduced in Section 3
achieves electrical characteristics suitable for application in the high capacity field. Most power conversion
equipment used in the high capacity field is required
64
to be able to block large currents instantaneously. For
this purpose, reducing the internal inductance Lm of
the product to reduce the surge voltage is very important. In this product series, the collector and emitter
terminals, which are main terminals, are located in
close proximity to one another so as to actively utilize
the mutual interactions of the magnetic field and reduce Lm.
5. Postscript
This paper has introduced the “V-Series HPM
Family” which incorporates “V-Series” IGBTs and realizes significantly improved reliability. Fuji Electric
is confident that these modules will be able to support
the diverse needs of the high-power device field, as well
as the needs of the renewable energy field for which a
rapidly growing market is being formed.
Fuji Electric will continue to strive to advance the
level of semiconductor technology and package technology so as to respond additional needs, and to develop
new products that will contribute to the progress of
power electronics.
References
(1) Yamamoto, T. et al. New High Power 2-in-1 IGBT
Module. FUJI ELECTRIC REVIEW. 2011, vol.57, no.3,
p.82-86.
(2) Nishimura, T. et al. High Power IGBT Modules. FUJI
ELECTRIC REVIEW. 2009, vol.55, no.2, p.51-55.
(3) Takahashi, K. et al. New Lineup of V-Series IGBT
Modules. FUJI ELECTRIC REVIEW. 2010, vol.56,
no.2, p.56-59.
(4) Morozumi, A. et al. “Reliability of Power Cycling for
IGBT Power Semiconductor Module.” Conf. Rec. IEEE
Ind. Appl. Conf. 36th. 2001. p.1912-1918.
Vol. 58 No. 2 FUJI ELECTRIC REVIEW
New Lineup of Large-Capacity “V-Series”
Intelligent Power Modules
Naoki Shimizu † Tatsuya Karasawa † Kazumi Takagiwa †
To meet the diversifying needs for power control, Fuji Electric has developed a family of large-capacity intelligent
power modules (IPMs). These products with high-performance, new-generation IGBT chips, new control ICs and
lower package inductance are able to reduce total power loss and radiated noise, and increase current capacity. A
new solder material and divided direct copper bonding (DCB) are employed to enable a Tc power cycling capability
significantly enhanced. Terminals and screw hole positions are compatible with existing products, allowing existing
products to be replaced with the new products without major design changes.
1. Introduction(1)(2)
2. “V Series IPM” Product Lineup(1)(2)
Recently, as essential items for conserving energy
and reducing CO2 emissions in the industrial field,
high efficiency power converting equipment is being
used more and more. Additionally, the requirements
for standard insulated gate bipolar transistor (IGBT)
modules that integrate an IGBT chip and a free wheeling diode (FWD) chip into a single package are becoming more diverse. An intelligent power module (IPM)
integrates a control IC for internal drive and protection functions into a standard IGBT module. With
an IPM, optimized drive control can be performed so
that the IGBT can be driven and provided with high
reliability protection with low dissipation loss and low
noise. IPMs are used in a wide range of applications,
such as in motor driven equipment (numerical control
(NC) machine tools, general-purpose inverters, servos,
elevators, etc.), uninterruptible power supplies (UPS),
power conditioning systems (PCS) for solar energy
generation, and the like where low dissipation loss and
low noise are strongly required.
Since beginning to commercialize IPMs in 1988,
Fuji Electric has responded to market requests for lower power dissipation, lower noise and smaller size with
each successive generation of devices. In recent years,
through developing a “V-Series” IPM using a nextgeneration trench gate structure field stop (FS) type
“V-Series” IGBT chip, even lower power dissipation
loss and smaller size have been realized. This paper
describes Fuji Electric’s new lineup of large capacity
“V-Series” IPMs and the large capacity series of IPMs
(P631 package).
At present, Fuji Electric’s V-Series of IPMs is
available in the four packages (small capacity: P629,
medium capacity small size: P626, medium capacity
low profile: P630, large capacity: P631) as shown in
Fig. 1. All of these packages comply with the RoHS
Fig.1 Appearance of “V-Series” IPM packages
†
*1: RoHS directive: European Union (EU) directive on the
restriction of the use of certain hazardous substances in
electrical and electronic equipment
Fuji Electric Co., Ltd.
P630
P631
P629
P626
(a) Package appearance
P629
package
1
4
7 10 15
P626
package
1
P
5
9
13 19
P630
package
N
1
5
9 13 19
U
V
P631
package
1 5 9 13 19
P1
P2
B
PU
V W N
L49.5×W70
×H12.5
(mm)
N U
V
W
L50×W87
×H12
(mm)
P B
W
L84×W128.5
×H14
(mm)
N1
U
V
W
N2
L110× W142
× H27
(mm)
(b) Type of package
65
issue: Power Semiconductor contributing in energy and environment region
ABSTRACT
directive*1 Fuji Electric plans to expand further the
V-Series IPM product lineup, increasing the capacity
beyond that of the previous “R-Series” IPMs, to rated
currents of 20-400 A for the 600 V rated voltage series,
and to 10-200 A for the 1,200 V rated voltage series.
In addition to the protective functions of overcurrent protection, short-circuit protection, control supply
protection in control circuit and IGBT chip over heat
protection, which are the same protective functions as
had been provided previously, a cause identification
function based on the width of the alarm output has
been newly added. In the P629 package, the previous
method of N-line current detection based on shunt resistance has been changed to a method of IGBT sense
current detection, enabling protection in the case of a
ground fault in which current flows only through the
upper arm element.
Table 1 lists the product lineup and functions of
the V-Series of IPMs.
Table 1
Rated
voltage
600 V
1,200 V
3. Large-capacity “V-Series” IPM Product
Overview
3.1 Development goals
Development goals for the large-capacity V-Series
of IPMs are as follows.
(1) Reduction of total dissipation loss
(2) Improvement of tradeoff relation between switching loss and radiation noise
(3) Expanded current rating (400 A / 600 V, 200 A/
1,200 V)
(4) Shorter deadtime
(5) Separate alarm output signal for each cause
(6) Upper arm alarm output (P631: alarm control terminal added for upper arm)
(7) Maintain package compatibility (tapped hole locations, guide pins)
(8) Compliance with RoHS directive
(9) Reduction of internal inductance
(10) Improvement of ΔTc power cycling capability
“V-Series” IPM product lineup and functions
Product type
Rated
current
Internal functions*
Upper
Both upper and lower arms
arm
Drive
UV
TjOH
OC
ALM
Lower
arm
ALM
6-in-1
7-in-1
20 A
6MBP20VAA060-50
–
○
○
○
○
–
○
30 A
6MBP30VAA060-50
–
○
○
○
○
–
○
50 A
6MBP50VAA060-50
–
○
○
○
○
–
○
50 A
6MBP50VBA060-50
–
○
○
○
○
○
○
75 A
6MBP75VBA060-50
–
○
○
○
○
○
○
50 A
6MBP50VDA060-50
7MBP50 VDA060-50
○
○
○
○
○
○
75 A
6MBP75VDA060-50
7MBP75 VDA060-50
○
○
○
○
○
○
100 A
6MBP100VDA060-50
7MBP100 VDA060-50
○
○
○
○
○
○
150 A
6MBP150VDA060-50
7MBP150 VDA060-50
○
○
○
○
○
○
200 A
6MBP200VDA060-50
7MBP200 VDA060-50
○
○
○
○
○
○
200 A
6MBP200VEA060-50
7MBP200VEA060-50
○
○
○
○
○
○
300 A
6MBP300VEA060-50
7MBP300VEA060-50
○
○
○
○
○
○
400 A
6MBP400VEA060-50
7MBP400VEA060-50
○
○
○
○
○
○
10 A
6MBP10VAA120-50
–
○
○
○
○
–
○
15 A
6MBP15VAA120-50
–
○
○
○
○
–
○
25 A
6MBP25VAA120-50
–
○
○
○
○
–
○
25 A
6MBP25VBA120-50
–
○
○
○
○
○
○
35 A
6MBP35VBA120-50
–
○
○
○
○
○
○
50 A
6MBP50VBA120-50
–
○
○
○
○
○
○
25 A
6MBP25VDA120-50
7MBP25VDA120-50
○
○
○
○
○
○
35 A
6MBP35VDA120-50
7MBP35VDA120-50
○
○
○
○
○
○
50 A
6MBP50VDA120-50
7MBP50VDA120-50
○
○
○
○
○
○
75 A
6MBP75VDA120-50
7MBP75 VDA120-50
○
○
○
○
○
○
100 A
6MBP100VDA120-50
7MBP100VDA120-50
○
○
○
○
○
○
100 A
6MBP105VEA120-50
7MBP100VEA120-50
○
○
○
○
○
○
150 A
6MBP150VEA120-50
7MBP150VEA120-50
○
○
○
○
○
○
200 A
6MBP200VEA120-50
7MBP200VEA120-50
○
○
○
○
○
○
Package
model
P629
P626
P630
P631
P629
P626
P630
P631
* : Drive: IGBT drive circuit, UV: Control supply undervoltage protection, TjOH: IGBT chip over heat protection, OC: Overcurrent protection, ALM: Alarm output
66
Vol. 58 No. 2 FUJI ELECTRIC REVIEW
(1) Reduction of total dissipation loss
To improve equipment controllability, as is requested by customers, the IPM dissipation loss must
be reduced in order to realize a higher carrier frequency and larger output current. Moreover, a reduction
in dissipation loss enables the cooling system for the
equipment to be simplified, such as with smaller-sized
air-cooling fins and fans, and also contributes to a lower overall low cost of the equipment.
Figure 2 compares the total dissipation loss during PWM inverter operation for a V-Series IPM and
the previous product (R-Series IPM), both of which
are 300 A / 600 V devices. The V-Series IPM realizes
at least 20% lower dissipation loss than the previous
product.
With the V-Series IPM, in the case of a
300 A / 600 V device, the static loss Psat and the turnoff loss Poff of the IGBT account for approximately 50%
of the total dissipation loss during inverter operation.
The on-voltage VCE(sat) and turn-off loss Eoff characteristics that determine these two types of loss have a
tradeoff relationship with the short-circuit withstand
capability of the IGBT. Improving the tradeoff is a
key point for reducing the dissipation loss. V-Series
IGBT chips for standard IGBT modules feature an
optimized surface structure to reduce the resistance
of the drift layer and make the chip thinner, thereby
lowering VCE(sat) and improving Eoff (3). On the other
hand, IGBT chips for V-Series IPMs feature an even
finer surface structure and an improved tradeoff relationship with VCE(sat) and Eoff (4). As a result of the finer
structure, however, because VCE(sat) decreases and the
current flows with greater ease, short-circuit current
will increase and the short-circuit withstand capability
(allowable time) will decrease. Therefore, a chip that
features an improved tradeoff relationship between
VCE(sat) and Eoff can be used to speed up the short-circuit protection function.
Ed = 300 V, VCC = 15 V, Tj =125 °C,
IC (rms)= 100 A, fC = 5/ 10 kHz, fo =50 Hz
138.7 W
− 23.2%
100
106.5 W
105 W
Prr
Pf
Pon
50
Poff
Psat
0
R-Series
IPM
V-Series
IPM
5 kHz
90
− 26.6%
77.1 W
R-Series
V-Series
IPM
IPM
10 kHz
Fig.2 Comparison of total dissipation loss in 300 A/ 600 V
products
New Lineup of Large-Capacity “V-Series” Intelligent Power Modules
Noise level (dB)
Total dissipation loss (W)
150
For the 600 V series, because the rated current is
larger and the difference between rated voltage and
working voltage is smaller than the 1,200 V series, the
design must take into consideration the surge voltage. In addition to reducing the internal impedance of
the package, as will be described later, by shifting the
VCE(sat) and Eoff tradeoff to the low VCE(sat) side where
the turn-off di / dt is smaller, the device was optimized
so that the surge voltage would be equivalent to that of
previous products.
(2) Improvement of tradeoff relation between switching loss and radiation noise
A tradeoff relationship exists between switching
loss and radiation noise. To improve this relationship,
the internal capacitance of the IGBT was reduced,
the temperature dependence of the control IC was improved, and the internal circuit wiring pattern of package was optimized(5). As a result, in a relative comparison of radiation noise using inverter test equipment,
the peak radiation noise in a 300 A / 600 V product was
reduced by approximately 3 dB compared to the previous product as is shown in Fig. 3.
(3) Expanded current rating (400 A / 600 V, 200 A/
1,200 V)
With the V-Series large-capacity IPM (P631), the
two power chips (IGBT and FWD) used in parallel
in the R-Series large-capacity IPM (P612) are integrated into a single chip, and the chip size is miniaturized and optimized to reduce the total area of the
300 A / 600 V power chip by 32%. By configuring the
control circuit as a two-level structure positioned above
the power unit, the product series could be expanded
to 400 A / 600 V and 200 A / 1,200 V in an package of
equivalent size and having the same mounting position
as previous products. This was a first in the industry.
(4) Shorter deadtime
For the purpose of preventing upper and lower arm
short circuits, the inverter control is provided with a
deadtime interval. Shortening the deadtime interval
is important for improving waveform distortion and rotational unevenness. With V-Series IPMs, the control
IC switching time has been optimized and temperature
dependence has been improved to reduce the minimum
deadtime interval to 1 μs from the previous value of 2.4
R-Series IPM (P612)
7MBP300RA060-50
MAX Peak =77.95 dB
60
30
0
30
V-Series IPM (P631)
7MBP300VEA060-50
MAX Peak =74.45 dB
50
70
90
Frequency (MHz)
110
130
Fig.3 Comparison of radiation noise
67
issue: Power Semiconductor contributing in energy and environment region
3.2 Characteristics(1)(2)
TjOH
UV
OC
Alarm output for
2 ms interval
0V
0V
Alarm output for
4 ms interval
0V
Alarm output for
8 ms interval
Overcurrent
VCC: 15 V/ div
IC
0A
0V
Overheat protection
occurred
Undervoltage
t: 2 ms / div
Fig.4 Alarm cause identification function
Unit: mm
4
7 10 16
P
N
U
V
9 13 19
P1
B
N
5
B
N1
W
Control terminal
(4) (Solder between terminal and PCB)
Printed circuit board (PCB)
(5) (Solder between PCB and
mounted components)
P2
95
P
1
95
1
Resin case
Copper circuit
N2
U
V
121
121
60.96
60.96
W
27.1
27.1
Ceramic
(a) P612 package
Main terminal
(3) (Solder between terminal
and insulating substrate)
(b) P631package
Insulating
substrate
Copper
circuit
IGBT chip
(2) (Solder between chip and
copper circuit)
Base plate (copper)
(1) (Solder between insulating
substrate and base plate)
Fig.5 Package dimensions
Fig.6 Internal structure of P631 package
μs.
(5) Separate alarm output signal for each cause
With V-Series IPMs, the widths of outputted alarm
pulses differ according to the alarm cause (see Fig. 4).
As a result, the cause of the protective alarm can be
identified easily, and after the equipment has been
stopped by the IPM alarm, cause analysis and restoration can be performed in a shorter amount of time.
terminals and the insulating substrate, (4) the junction between control terminals and the printed circuit
board, and (5) the junction between the printed circuit
board and mounted components. The P631 uses leadfree solder at all of these locations, and is compliant
with the RoHS directive.
4. Package
4.1 Package dimensions
Figure 5 shows the dimensions of the P631 and
P612 packages. The P631 is provided with the same
mounting holes, terminal locations and height dimensions as the existing P612, and maintains package compatibility when being installed in equipment.
Additionally, an alarm output terminal has been added
to each phase of the upper arm, and the terminal pitch
and total width of the control terminals was made common.
4.2 Compliance with RoHS directive
Figure 6 shows the internal structure of the P631
package. In a conventional IPM, lead is used primarily at soldering locations. Solder material is used at
five locations: (1) the junction between the insulating
substrate and base plate, (2) the junction between the
chip and copper circuit, (3) the junction between main
68
4.3 Reduction of inductance
Figure 7 shows a schematic drawing of the internal wiring, and Fig. 8 compares the results of measurement of the internal inductances of the P631 and
P612 packages. Aiming to reduce internal inductance
through the mutual inductance effect, a parallel plate
configuration employing overlapping P and N line terminal bars was used. As a result, the internal inductance is reduced by approximately 22% compared to
the P612 package, and an effect is obtained whereby
the radiation noise is lower and the turn-off surge does
not become excessive, as described in Section 3.
4.4 Improvement of
Tc power cycling capability
IGBT modules are typically formed by soldering together a base having a heat dissipating surface and an
insulated circuit board. Because these materials have
different coefficients of thermal expansion, stress is
repeatedly generated in the soldered junction between
the materials whenever the temperature changes. In
a ΔTc power cycling test in which the case temperature
was varied, it was confirmed that solder in the junction
Vol. 58 No. 2 FUJI ELECTRIC REVIEW
Overlapping portion
P
N
(a) P612 package
P
P
N
N
P
N
(b) P631 package
Fig.7 Schematic of internal wiring in package
300 A/ 600 V actual measurement values obtained by connecting to
one side of a snubber capacitor located between P and N terminal bars
100
Inductance L (nH)
78.8
28.2
Upper
arm
50
0
− 22.6%
61.0
24.0
50.6
Lower
arm
37.0
P612
package
P631
package
cycling capability at ΔTc = 80 K has been improved by
more than twice as that of the P612.
5. Postscript
This paper has described Fuji Electric’s new lineup of large-capacity “V-Series” IPMs and the largecapacity series of IPMs (P631 package). The V-Series
of IPMs realize low power dissipation, low noise and
RoHS compliance, as requested in the marketplace,
and also provide such value-added features as larger
capacity, shorter deadtime, an alarm cause identification function, and package compatibility.
In the future, Fuji Electric intends to continue to
improve the performance and expand the lineup of
available packages, and will focus on developing products capable of contributing to the conservation of energy and protection of the environment.
References
(1) Motohashi, S. et al. “The 6th Gen. Intelligent Power
Module,” in Proc. 2011 PCIM, p.161-166.
(2) Shimizu, N. et al. V Series’ Intelligent Power Modules.
FUJI ELECTRIC REVIEW. 2010, vol.56, no.2, p.60-64.
(3) Kobayashi, Y. et al. “The New concept IGBT-PIM with
the 6th generation V-IGBT chip technology,” in Proc.
2007 PCIM.
(4) Momose, M. et al. “A 600 V Super Low Loss IGBT with
Advanced Micro-P Structure for the next Generation
IPM,” in Proc. 2010 ISPSD.
(5) Onozawa, Y. et al. “Development of the next generation
1200 V trench-gate FS-IGBT featuring lower EMI noise
and lower switching loss,” in Proc. 2007 ISPSD, p.1316.
Fig.8 Comparison of results of internal inductance
measurement
New Lineup of Large-Capacity “V-Series” Intelligent Power Modules
69
issue: Power Semiconductor contributing in energy and environment region
between the insulating substrate and the base plate
will crack, expand and ultimately break.
With the P631 package, the insulating substrate
has been subdivided and miniaturized to ease the
stress on the solder. Additionally, the solder material
used in the junction between the insulating substrate
and the base plate has been changed to material having higher mechanical strength. As a result, ΔTc power
Hybrid Si-IGBT and SiC-SBD Modules
Masayoshi Nakazawa † Toshiyuki Miyanagi † Susumu Iwamoto †
ABSTRACT
Fuji Electric has developed hybrid modules that combine silicon-insulated gate bipolar transistor (Si-IGBT) and
silicon carbide-Schottky barrier diode (SiC-SBD) for high-efficiency inverter applications that contribute to energy
savings. The SiC-SBD chip was developed jointly with the National Institute of Advanced Industrial Science and
Technology, a public research institute, and the Si-IGBT chips are the latest 6th-generation “V-Series” IGBTs from
Fuji Electric. The product lineup are 600 V class rated at 50/ 75/ 100 A, and 1,200 V class rated at 35/ 50 A. Inverter
power loss in the 1,200 V 50 A class has been reduced by 23% compared to the V-series module.
1. Introduction
To prevent global warming, the further reduction
of greenhouse gases (such as CO2) is a pressing issue. The greatest benefit of using power electronics
technology to reduce greenhouse gases is that electrical power becomes more energy-efficient. In such an
undertaking, increasing the efficiency of inverters is an
important factor. To do so requires technical innovation of the power devices, circuits, controllers and other
components used in inverters. Since power devices are
the main elements used in inverters, there is increasing demand for low loss power devices that realize
higher efficiency. The insulated gate bipolar transistor
(IGBT) is such a power device, and the use of silicon
(Si) IGBT and free wheeling diode (FWD) chips is common. However, Si semiconductor device performance
is approaching its theoretical limits based on its material properties, and future breakthroughs that achieve
significantly lower loss cannot be expected. For this
reason, wide bandgap (WBG) semiconductors, which
exhibit material properties superior to those of Si semiconductors, are promising.
Fuji Electric is advancing the development of silicon carbide (SiC) semiconductor devices, which are a
type of WBG semiconductors. With an SiC device,
higher breakdown voltage and lower on-state resistance can be achieved than with a Si device. In theory,
the on-state resistance of a SiC can be made lower
than that of a Si device having the same breakdown
voltage. For this reason, SiC devices do not necessarily
need to be bipolar devices as was essential for reducing the on-state resistance in high-voltage Si devices.
Bipolar devices are associated with the injection of
minority carriers and generally exhibit greater switch†
70
Fuji Electric Co., Ltd.
ing energy than unipolar devices. Thus, unipolar devices are desirable for reducing the switching energy.
Accordingly, the use of a SiC unipolar device enables
low on-state resistance and low switching energy to be
achieved simultaneously.
2. Product Overview
Figure 1 shows an internal circuit diagram of a
power integrated module (PIM). Si-IGBT and silicon
carbide schottky barrier diode (SiC-SBD) hybrid modules, which enabled to reduce power loss than ever
before, have been developed using SiC-SBD unipolar
devices as FWDs. A chip developed in collaboration
with the National Institute of Advanced Industrial
Science and Technology is used as the SiC-SBD, and
Fuji Electric’s latest chip, a 6th generation “V-Series”
IGBT chip is used as the Si-IGBT.
Figure 2 shows the appearance, and Table 1 lists
the product lineup of the Si-IGBT and SiC-SBD hybrid
modules. For high efficiency inverter applications,
50 A, 75 A and 100 A rated products have been de-
Thermistor
Converter
Brake chopper
Inverter
SiC-SBD
Fig.1 PIM internal circuit diagram
at 50 A
2.0
V-Series PND
1.8
SiC-SBD
1.6
1.4
0
50
100
Temperature (°C)
150
issue: Power Semiconductor contributing in energy and environment region
Forward voltage (V)
2.2
200
Fig.4 Temperature characteristics of the forward voltage
L122 × W62 × H17 (mm)
Fig.2 Appearance of the hybrid Si-IGBT/ SiC-SBD module
10−1
10−2
Lineup of hybrid Si-IGBT / SiC-SBD modules
Rated voltage (V)
Rated current (A)
Hybrid module model
50
7MBR50VB060S-50
600
75
7MBR75VB060S-50
100
7MBR100VB060S-50
1,200
25
7MBR25VB120S-50*
35
7MBR35VB120S-50
50
7MBR50VB120S-50
* : To Be Determined
Leakage current (A)
Table 1
150 °C
V-Series module
10−3
Hybrid module
10−4
10−5
Hybrid module
Room
temperature V-Series module
10−6
10−7
10−8
0
200
400
600
800
Voltage (V)
1,000
1,200
Fig.5 Leakage current characteristics
100
150 °C
Forward current (A)
Room temperature
SiC-SBD
SiC-SBD
V-Series PND
V-Series PND
50
Current range of actual use
0
0
1
2
3
Forward voltage (V)
4
5
Fig.3 Forward characteristics
veloped for the 600 V series, and 35 A and 50 A rated
products have been developed for the 1,200 V series.
Product performance and characteristics are presented
below for the 1,200 V / 50 A product, as a representative example.
3. Static Characteristics
3.1 Forward characteristics
Figure 3 shows the forward characteristics of a SiCSBD and a V-Series PN junction diode (V-Series PND),
and Fig. 4 shows the temperature characteristics of the
Hybrid Si-IGBT and SiC-SBD Modules
forward voltage Vf at the rated current of 50 A. In the
current range that is actually used, the Vf of the SiCSBD is the same as that of the V-Series PND. From
Fig. 4, it can be seen that at temperatures higher than
100 °C, the V-Series PND exhibits a negative temperature coefficient with reducing Vf. A device with a negative temperature coefficient is prone to current imbalance when in a multi-parallel connection. On the other
hand, the SiC-SBD, which has a strong positive temperature characteristic, is unlikely to create a current
imbalance, even when in a multi-parallel connection.
3.2 Leakage current characteristics
Figure 5 compares the leakage current characteristics of the hybrid module and a V-Series module. The
leakage current of the hybrid module is about 1,000
times as large as that of the V-Series module when at
room temperature, but is only slightly less than that of
the V-Series module when at 150 °C. As shown in Fig.
6, the leakage current of the hybrid module is nearly
constant at temperatures of around 100 °C and below,
and increases similarly as the V-Series at higher temperatures. The SiC-SBD, which has a wide bandgap
and therefore very few thermally excited carriers, is
less affected by temperature rise. The increase in leakage current at temperatures above 100 °C is a result of
71
Leakage current (A)
10
Reverse recovery energy (mJ/ pulse)
10−1
at 1,200 V
−2
10−3
10−4
Hybrid module
10−5
10
−6
V-Series module
10−7
10−8
0
50
100
Temperature (°C)
150
200
Fig.6 Temperature characteristics of leakage current
20
VCC =600 V, VGE =± 15 V, Rg =15 1, 150 °C
10
V-Series module
Hybrid module
70% reduction
0
0
25
50
Current (A)
75
100
Fig.8 Current characteristic of reverse recovery loss
VCC = 600 V, VGE = ± 15 V, Rg = 15 1, 150 °C
VCC =600 V, VGE =± 15 V, Rg = 15 1, 150 °C
Ia: 25 A/ div
at 50 A
VGE: 10 V/div
0
0
at 50 A
IC: 25 A/ div
Vka: 200 V/ div
VCE: 200 V/div
0
t: 200 ns / div
t: 200 ns / div
(a) Hybrid module
(a) Hybrid module
0
0
0
(b) V-Series module
the leakage current of the IGBT becoming dominant.
Thus, even if the SiC-SBD exhibits a large leakage current at room temperature, because its leakage current
during high temperature operation is the same as that
of a V-Series device, the SiC-SBD is similarly able to
operate at junction temperatures of up to 175 °C.(1)
4. Switching Characteristics
4.1 Reverse recovery characteristics
Figure 7 compares reverse recovery waveforms of
the hybrid module and a V-Series module. The hybrid
module exhibits extremely lower reverse recovery peak
current. This behavior can be explained from little injection of minority carriers since the SIC-SBD is a unipolar device. As can be seen in Fig. 8, at the rated current of 50 A, reverse recovery energy can be reduced
significantly by 70% compared to the V-Series module.
4.2 Turn-on characteristics
The reverse recovery peak current of the FWD is
72
Fig.9 Turn-on waveforms
20
Turn-on energy (mJ/pulse)
Fig.7 Reverse recovery waveforms
(b) V-Series module
VCC =600 V, VGE = ± 15 V, Rg =15 1, 150 °C
Hybrid module
10
V-Series module
0
54% reduction
0
25
50
75
100
Current (A)
Fig.10 Current characteristics of turn-on energy
reflected in the turn-on peak current of the IGBT in
the opposing arm, and a reduction in turn-on energy
can be attained as a result of the reduction in reverse
recovery energy.
Figure 9 compares turn-on waveforms of the hybrid
Vol. 58 No. 2 FUJI ELECTRIC REVIEW
layer of the SiC-SBD has a much lower resistance
VCC =600 V, VGE =± 15 V, Rg =15 1, 150 °C
at 50 A
Ia: 20 A/ div
4.3 Turn-off characteristics
Figure 11 shows a comparison of the turn-off waveforms of the hybrid module and V-Series module. At
the rated current of 50 A, the turn-off surge voltage
of the hybrid module is 47 V lower than that of the
V-Series module. In general, the surge peak voltage
can be defined by Equation (1), and if the IGBT element characteristics and the main circuit inductance
are equivalent, the difference in the surge voltages is
originated from the difference of the transient on-state
voltages of the diodes. Figure 12 compares transient
on-state waveforms of the diodes. Because the drift
0
0
90 V
(b) V-Series module
at 50 A
0
t: 200 ns / div
(a) Hybrid module
VCC = 600 V, VGE = ± 15 V, Rg = 15 1, 150 °C
VGE: 10 V/ div
Vka: 200 V/ div
45 V
issue: Power Semiconductor contributing in energy and environment region
module and V-Series module. As in the reverse recovery waveform, there is extremely lower current peak.
At the rated current of 50 A, as is shown in Fig. 10, the
turn-on energy can be reduced significantly by 54%
compared to the V-Series module.
Fig.12 Diode transient on-state recovery waveforms
675 V
IC : 10 A/ div
20
VCC =600 V, VGE =± 15 V, Rg =15 1, 150 °C
Turn-off energy (mJ/pulse)
0
t: 200 ns / div
(a) Hybrid module
0
722 V
Hybrid module
V-Series module
800
10
700
47 V reduction
600
0
0
900
0
25
50
75
Surge peak voltage (V)
VCE : 200 V/ div
500
100
Current (A)
(b) V-Series module
Fig.13 Turn-off energy and surge peak voltage/ current
characteristics
Fig.11 Turn-off waveforms
Inverter power loss (W)
100
Bus voltage (Vbus)= 600 V, Output current (rms value) (Io)= 35 A, Output frequency (fo)=50 Hz
Power factor (cosφ)= 0.9, Control factor (λ)= 1.0, 3-arm modulation
27%
reduction
23%
reduction
50
Prr
Pf
Pon
Poff
16% reduction
Psat
0
V-Series
module
Hybrid
module
Carrier frequency 4 kHz
V-Series
module
Hybrid
module
Carrier frequency 8 kHz
V-Series
module
Hybrid
module
Carrier frequency 12 kHz
Fig.14 Inverter power loss
Hybrid Si-IGBT and SiC-SBD Modules
73
than that of the V Series PND, the transient on-state
voltage of the SiC-SBD is reduced from 90 V to 45 V.
Therefore, as shown in Fig. 13, the surge voltage at
turn-off can be kept low and the turn-off energy can be
reduced.
VSp = VCC +L s $
dIC
+VTR ………………………………(1)
dt
VSp : Surge peak voltage
VCC : Applied voltage
Ls : Inductance of main circuit
IC : Collector current
VTR : Transient on-state voltage
5. Inverter Power Loss
Figure 14 shows the calculated results of inverter
power loss in the newly developed hybrid module and
the V-Series module. When the carrier frequency is
8 kHz, the total power loss of the hybrid module can be
reduced significantly by 23% for that of the V-Series
module. Moreover, because the rate of loss reduction
increases with hisher carrier frequency, the hybrid
module is more effective for applications involving high
frequency operation.
hybrid module that combines a SiC-SBD, developed in
collaboration with the National Institute of Advanced
Industrial Science and Technology, and a 6th generation “V-Series” IGBT chip, which is Fuji Electric’s latest chip. By greatly reducing the power loss of the device itself, this product is expected to contribute significantly to the achievement of higher efficiency inverters. In the future, Fuji Electric intends to expand the
lineup of products that use SiC chips, and to contribute
to efforts for preventing global warming.
The authors wish to thank all parties concerned
at the Advanced Power Electronics Research Center of
the National Institute of Advanced Industrial Science
and Technology (AIST) for their cooperation in the
development of the SiC-SBD chip. A portion of this
development work was carried out under the AIST
Industrial Transformation Research Initiative “SiC
Device Practical Application Verification Based on
Mass Production Prototype” in fiscal years 2008, 2009
and 2010.
References
(1) Takahashi, K. et al. New Lineup of V-Series IGBT
Modules. FUJI ELECTRIC REVIEW. 2010, vol.56,
no.2, p.56-59.
6. Postscript
This paper introduced the Si-IGBT and SiC-SBD
74
Vol. 58 No. 2 FUJI ELECTRIC REVIEW
Packaging Technologies for SiC Power Modules
Masafumi Horio † Yuji Iizuka † Yoshinari Ikeda †
Wide bandgap materials such as silicon carbide (SiC) and gallium nitride (GaN) are attracting attention as materials for next-generation power semiconductor devices. Fuji Electric is currently developing new packaging technologies to take full advantage of SiC devices. Compact and highly reliable power modules with low thermal resistance
and high-temperature operating capability can be realized by replacing aluminum wire bonding, solder joints and silicone gel encapsulating structures with copper pin connections, silver sintering joints and epoxy resin molding structures. Improved performances of prototype all-SiC modules and SiC diode modules with new packaging technologies
have been evaluated.
1. Introduction
Recently, power modules have been widely used in
industrial, consumer products and hybrid and electric
vehicle applications. As the performance of Si devices
used in the power modules is approaching its theoretical limits, wide bandgap (WBG) devices such as silicon
carbide (SiC) and gallium nitride (GaN) are attracting
attention. WBG devices have some advantages such as
higher dielectric breakdown voltage, lower loss, higher
switching and higher temperature operation capability
compared to Si devices.
This paper introduces new packaging technologies
that take full advantage of the features of WBG devices, especially SiC devices.
paths are formed from a DCB substrate, a power board
and Cu pins instead of bonding wires. Additionally,
this new structure uses epoxy resins as the encapsulating material instead of silicone gel. The features and
technologies of new structure that uses these new materials will be discussed below.
2.1 Downsizing
Power modules are being made smaller and smaller with the rise in power density of power chips and
the reduction of thermal resistance resulting from
improved packaging technology. In the conventional
structure, aluminum wire requires a certain amount
of area for bonding. Large current capacity power
Silicone Gel
2. Features and technology of SiC Power Module
Package
Aluminum Wire
Power Chip
DCB Substrate
Terminal
Resin Case
Fuji Electric is presently evaluating SiC metaloxide-semiconductor field-effect transistors MOSFETs
and SiC Schottky barrier diodes (SBDs) to be implemented in SiC modules. New packaging technology introduced in this paper can also be applied to other SiC
devices such as SiC insulated gate bipolar transistors
(IGBTs).
Figure 1 shows the aluminum wire bonding structure, which is currently the mainstream structure
for power modules, and the newly developed package
structure.
In the conventional structure, the main current
paths are formed from aluminum bonding wires and
the direct copper bonding (DCB) substrate. On the
other hand, in the newly developed structure, circuit
†
Fuji Electric Co., Ltd.
Cu Base
Solder
Ceramic Substrate
(a) Alminum wire bonding strucrture
Epoxy Resin
Cu Pin
Power Board
Terminal
Thick Cu
Block
DCB Substrate
SiN Ceramic Substrate
(b) Newly developed strucrture
Fig.1 Comparison of power module structure
75
issue: Power Semiconductor contributing in energy and environment region
ABSTRACT
modules require larger number of wires and result in
many aluminum wires on the DCB substrate. This is
a barrier to realize the high-density packaging i.e. the
downsizing of power modules.
In the newly developed structure, Cu pins are used
to connect power board to power chips, instead of using
aluminum wires. The power board has a printed circuit board structure and current flows through its Cu
pattern and Cu pins.
Because this structure allows current to flow in a
vertical direction with respect to a power chip, power
chips can be located close to each other. Furthermore,
two layers of current paths, on both the DCB substrate
and the power board, contribute to downsizing of the
power module.
2.2 Low thermal resistance
In order to downsize power modules having high
power density and to prevent temperature rise of the
power chips, the thermal resistance of the power module package must be reduced. An alumina ceramic
substrate is generally used in conventional structure
shown in Fig. 1. This alumina ceramic acts as a large
thermal barrier in the conventional structure due to
its low thermal conductivity of about 20 W / (m•K). In
order to reduce this large thermal resistance of the alumina ceramic, various developments have been undertaken.(1),(2)
In the development of this new structure, a much
thicker Cu block is bonded to the silicon nitride (SiN)
ceramic substrate in order to realize a further reduction of thermal resistance. The reason why a SiN ceramic substrate is used is because SiN has a larger
thermal conductivity than alumina ceramic as well as
large strength to endure the stress that occurs with
Thermal Resistance (K/W)
0.5
0.4
0.3
Other
Components
0.2
0.1
0
2.3 High temperature operation
One of the advantages of SiC devices is the capability to operate at high temperature. When a power chip
can operate at high temperature, the cooling cost can
be reduced and the overall system size can be reduced
as a result of a downsized cooling system. In order to
realize high temperature operation, it is necessary to
improve the high temperature withstanding capability
of the packaging components, especially the bonding
material and the encapsulating material.
Currently, a tin (Sn)-silver (Ag) compound of
lead (Pb)-free solder is generally used in power modules. The solidus line temperature*1 of that solder is
below 250 °C. Because higher temperatures generally cause solder to deteriorate more quickly, a higher
solidus temperature and melting point are preferred.
However, a high temperature process would cause
large stress and strain in the components. In order to
reconcile these contradicting issues, Ag-sinter material
is now being developed to apply as a bonding material.
Figure 3 shows a cross-sectional view of the bonding
area with Ag-sinter material.
Ag-sinter material exhibits superior characteristics
and is capable of bonding under relatively low temperatures (ca. 300 °C) and has a high melting point of ca.
962 °C that is the same as Ag bulk material after sintering. Also, its thermal conductivity is larger by one
digit than that of Sn-Ag compound solder.
Regarding encapsulating material, conventional
silicone gel can withstand the duration of the lifetime
(generally 10 years) of a power module at temperatures
below 150 °C, but has difficulty in enduring if the operating temperature increases. Therefore, the newly
developed structure uses a new epoxy resin that has
glass transition temperature of over 200 °C.
Ceramic
Substrate
(Alumina)
(窒化けい素)
(SiN)
Conventional
Structure
New Structure
Fig.2 Comparison of power module thermal resistance
*1: The solidus line temperature is the temperature at
which alloyed metal starts to melt, and the liquidus line
temperature is the temperature at which alloyed metal
melts completely. For pure metal, these temperatures
are the same, and are called the melting point.
76
thick Cu block during the thermal cycle. Figure 2 compares the power module thermal resistance of the conventional alumina ceramic structure with that of the
newly developed SiN ceramic structure, when implemented with the same size power chips. The thermal
resistance of the ceramic substrate in the new structure is one-fourth that of the conventional structure.
The new structure enables a 50% reduction in the
thermal resistance of the overall structure.
Power Chip
Ag-sinter material
DCB Substrate
Fig.3 Cross-sectional view of Ag-sinter bonding area
Vol. 58 No. 2 FUJI ELECTRIC REVIEW
Improved reliability at high temperature is needed in order to realize high temperature operation of the
power modules using SiC devices. The primary factors affecting the reliability of a power module having
a conventional structure are the lifetime of the bonding area between aluminum wire and power chip electrodes, and the lifetime of the solder layer.(3) With the
new structure, because aluminum wires are replaced
by Cu pins, the factors affecting reliability are different
from those of the conventional structure. A power cycling test, an important reliability test for power modules, was carried out with Si devices to compare reliability with that of the conventional structure. Figure
4 compares the power cycling capability of the new
structure and the conventional structure.
This test results shows that the new structure
has more than 30 times the power cycling capability
as compared to the conventional structure under the
conditions of chip junction temperature swings (ΔTj) of
125 K and 150 K. One of the reasons for the improve-
ment in power cycling capability is the epoxy resin
molding structure. Figure 5 compares the stress occurring at the bonding material underneath the power
chip in the cases of silicone gel and epoxy resin (result
of FEM analysis). The vertical axis in Fig. 5 shows the
values of stress occurring at the bonding material underneath power chip with silicone gel structure as 100.
This results shows that the epoxy molding structure reduces the stress at the bonding material by
about half. It is speculated that the epoxy resin molding structure suppresses deformation of the power
chips and the DCB substrate, for which expansion and
constriction are repeated freely in silicone gel structure. A thermal cycling test was also carried out under
the temperature range between −40 °C and 150 °C, and
the results show that the new structure has a greater
than 3,000 cycle capability.
3. Prototype Modules with New Structure
The following two types of prototype modules were
designed with the new structure that realizes downsizing, low thermal resistance, high temperature operation and high reliability.(4)~(6)
107
106
100
times
105
104
Conventional Structure
New Structure
30 times
(a) All-SiC Module
2-in-1, 100A, 1,200 V
102
10
50
100 125150 200
Swing of chip junction temperature 6Tj (K)
1,000
24.7 mm
103
34 mm
Number of cycles (cycle)
F(t) = 1%line
62.6 mm
92 mm
Si Module
All-SiC Module
with conventional structure
with new structure
(b) Comparison of footprint size
Fig.4 Comparison of power cycling capability
Fig.6 External view of All-SiC module and its footprint size
100
Total Switching Loss (a.u.)
Stress (a.u.)
80
60
40
20
0
Silicone Gel
Structure
Epoxy Resin Molding
Structure
Fig.5 Comparison of stress occuring at bonding material
underneath power chips
Packaging Technologies for SiC Power Modules
1.0
Conventional Structure
with Si devices
0.8
40 to 70% reduction
(due to SiC devices)
0.6
0.4
Conventional Structure
with SiC devices
17% reduction
(due to new structure)
New Structure with SiC devices
0.2
0
0
20
40
60
Gate Resistance Rg (Ω)
80
Fig.7 Loss evaluation by switching tests
77
issue: Power Semiconductor contributing in energy and environment region
2.4 High reliability
3.2 SiC diode module
The SiC diode module shown in Fig. 8(a) is configured as a SiC SBD diode module with 2 arms in series
and is rated at 400 A and 1200 V. The dimensions are
93.5 mm (length) × 30.6 mm (width) × 17.0 mm (height).
Compared to a 100 A Si diode module with a conventional structure, this SiC diode module has four times
the current density with about the same footprint size
(see Fig. 8(b)).
(a) SiC diode module
26 mm
30.6 mm
4. Postscript
92 mm
93.5 mm
SiC diode module
Si diode module
with new structure
with conventional structure
(400 A/ 1,200 V)
(100 A/ 1,200 V)
(b) Comparison of footprint size
Fig.8 External view of SiC diode module and its footprint size
3.1 All-SiC module
The all-SiC module shown in Fig. 6(a) is 2-in-1
module rated at 100 A and 1200 V and equipped with
SiC MOSFETs and SiC SBDs. The module has dimensions of 62.6 mm (length) × 24.7 mm (width) × 19.0 mm
(height). Compared to a Si module having conventional structure and the same rating, the all-SiC module
has a footprint of about half the size (see Fig. 6(b)).
The reasons for this small footprint size the all-SiC
module are the new structure with Cu pin connection
technology for the high-density mounting of power
chips, the low thermal resistance structure for high
thermal dissipation, and the low loss SiC devices and
high temperature withstanding components having
high reliability at high temperature operation.
Figure 7 shows the switching test results with different gate resistances of Si and SiC devices and different module structures in order to clarify the improvement in characteristics of the new structure and SiC
devices.
The loss reduction with SiC devices is 40 to 70%,
and a further 17% loss reduction is attributed to the
new structure. The small parasitic inductance of the
new structure as a result of the downsized current
path is thought to limit the surge voltage and thereby
reduce the switching loss.
78
A new power module structure with Cu pin connections, Ag-sinter bonding and epoxy resin molding has
been developed. This packaging technologies are able
to take full advantage of the superior characteristics of
SiC devices. Moreover, a prototype all-SiC module and
SiC diode module were designed and their improved
characteristics were evaluated.
In the future, Fuji Electric will verify the superior
characteristics of this new power module through field
tests and will continue to advance technology for the
conservation of energy.
References
(1) Nishimura, Y. et al. Development of a Next-generation
IGBT Module using a New Insulating Substrate. Fuji
Electric Review. 2005, vol.51, no.2, p.52-56.
(2) Kobayashi, Y. et al. New Concept IGBT-PIM Using
Advanced Technologies. Fuji Electric Review. 2007,
vol.53, no.3, p.69-72.
(3) Morozumi, A. et al. Reliability Design Technology for
Power Semiconductor Modules. Fuji Electric Review.
2001, vol.47, no.2, p.54-58.
(4) Horio, M. et al. “New Power Module Structure with
Low Thermal Resistance and High Reliability for SiC
Devices.” PCIM Europe. 2011, Proceeding CD, p.229234.
(5) Ikeda, Y. et al. “Investigation on Wirebond-less Power
Module Structure with High-Density Packaging and
High Reliability, International Symposium on Power
Semiconductor Devices and IC’s.” 2011, Proceeding CD,
p.272-275.
(6) Horio, M. et al. “Ultra Compact, Low Thermal
Resistance and High Reliability Module Structure
with SiC Schottky Barrier Diodes.” IEEE Applied
Power Electronics Conference and Exposition. 2011,
Proceeding CD, p.1298-1300.
Vol. 58 No. 2 FUJI ELECTRIC REVIEW
“Super J-MOS” Low Power Loss
Superjunction MOSFETs
Takahiro Tamura † Mutsumi Sawada † Takayuki Shimato †
Fuji Electric has developed superjunction MOSFETs with an optimized surface design that delivers lower switching loss. In these “Super J-MOS”chips, gate length and channel density were adjusted to optimize the gate-to-drain
capacitance and threshold voltage, thus achieving lower turn-off loss. For devices rated at 600 V/ 20 A/ 0.19 , an
extremely low turn-off loss of 160 μJ at the turn-off dV/ dt of 10 kV/μs was realized. Power efficiency is over 94.0%,
enabling compliance with the 80 PLUS certification.
1. Introduction
Recently, in response to heightened interest in
global environmental protection through conserving
energy and reducing CO2 emissions, lower levels of
power loss are sought in so-called IT equipment, such
as PCs and servers. In order to reduce the power loss
in IT equipment, the power conversion devices used
in IT equipment must be made more efficient, and the
technology that enables this higher efficiency is power
semiconductors.
The power semiconductors installed in power conversion equipment operate as switching devices, and
their power loss consists of conduction loss while the
device is in its on-state, and switching loss when the
device changes from its on- to off-state or from its offto on-state. To achieve higher efficiency and lower loss
in power conversion equipment, both types of loss must
be reduced.
This paper reports on Fuji Electric’s successful development of a “Super J-MOS” (Superjunction
MOSFET) that achieves low switching loss as a result
of optimization of the SJ-MOSFET surface structures
from a theoretical perspective and an improved tradeoff relationship between turn-off loss Eoff, generated
when the device changes from its on- to off-state, and
the value of turn-off dV / dt, which indicates the timechange of drain-to-source voltage at the time of turnoff.
2. Characteristics of the “Super J-MOS”
The use of an SJ-MOSFET is one way to reduce
both conduction loss and switching loss. Compared to
†
Fuji Electric Co., Ltd.
a conventional power MOSFET, the SJ-MOSFET features a significantly improved trade-off relationship
between the device breakdown voltage BVDSS and the
specific on-resistance Ron•A, and because the conduction loss can be reduced dramatically, SJ-MOSFETs
are being used more and more in power conversion
equipment.
Figure 1 shows schematic cross-sectional views
of a SJ-MOSFET and a conventional MOSFET. The
SJ-MOSFET has a structure in which p-pillars and npillars are arranged alternately in the drift region. By
narrowing the width of each pillar, the impurity concentration in the drift region can be increased without
decreasing the breakdown voltage, and therefore the
on-resistance can be reduced.(1)~(4)
Furthermore, because an SJ-MOSFET has signifi-
Source
n+
p-base
Gate
p+ n+
n
n+
Source
n+
p-base
Gate
Lg
p+ n+
n p
n
p
n
n+
Drain
Drain
(a) Conventional MOSFET
(b) SJ-MOSFET
Fig.1 Schematic cross-sectional views of power MOSFETs
*1: Miller period: See supplemental explanation 2 on page
88.
79
issue: Power Semiconductor contributing in energy and environment region
ABSTRACT
3. Optimization of Surface Structures
3.1 Design concept
In order to improve the tradeoff relationship between Eoff and turn-off dV / dt in a SJ-MOSFET, it is
necessary to reduce turn-off dV / dt under conditions
of constant Rg. Focusing on the reduction of turn-off
dV /dt, characteristics of the tradeoff between Eoff and
turn-off dV / dt were improved in accordance with the
following equation.
Assuming that the gate-source capacitance during
turn-off is constant within the Miller period, turn-off
dV /dt can be expressed as in Eq.(1).
…………………………………(1)
ID : Drain current
gfs : Transconductance
VDS : Drain-to-source voltage
From Eq.(1), it can be understood that when Rg,
ID and VDS are constant, increasing CGD and decreasing the threshold voltage Vth are effective for reducing
turn-off dV /dt. CGD is determined by the distance between the p-bases, i.e., the gate length Lg, and therefore Lg should be lengthened in order to increase CGD.
100
Rg = 91 1
100
10
0.5
Turn-off
dV/ dt =10 kV/ μs
1.0
1.5
Gate length Lg (a.u.)
Fig.2 Lg dependence of Eoff and turn-off dV/ dt
80
10
1
2.0
Simulations based on the design concept were performed to estimate the Lg dependence of Eoff. Figure
2 shows the Lg dependence of Eoff and turn-off dV /dt.
Eoff values are shown for the case that turn-off dV /dt is
10 kV / μs, and turn-off dV / dt values are shown for the
case that Rg is 91 Ω. Additionally, the values of Lg are
relative to the value of Lg prior to optimization of the
structure.
As shown in Fig. 2, turn-off dV / dt can be reduced
by lengthening Lg, and the resulting decrease in the
value of Eoff was confirmed. When Lg increases above
1.4, the Eoff value shows nearly no improvement and
remains nearly unchanged. This is thought to be
caused by the lengthening of the Miller period and increased loss that occurs when CGD is increased, causing the turn-off time to become longer and the feedback
capacitance to increase.
3.3 Threshold voltage dependence of turn-off loss
Next, the Vth dependence of Eoff and turn-off dV /dt
was calculated. The results are shown in Fig. 3. As in
the calculation of the Lg dependence, the values of Vth
are relative to the value of Vth prior to optimization of
the structure. Additionally, an estimation of the Vth
dependence was calculated based on the simulation described in Section 3.2 and using an optimal Lg design
value of 1.4.
As shown in Fig. 3, turn-off dV / dt decreases as Vth
becomes smaller, and accordingly, a decrease in the
value of Eoff was confirmed. If Vth becomes too small,
however, a problem may occur in which the device
turns-on unintentionally due to noise. In optimizing
the design for Vth, it is necessary to be careful so as not
to reduce Vth too much in order to prevent malfunction
of the device.
Based on the above results and using Lg = 1.4 and
Vth = 0.75 as optimal structure values, the tradeoff rela-
100
1,000
Turn-off dV / dt (kV/μs)
Turn-off loss Eoff (μJ)
1,000
3.2 Gate length dependence of turn-off loss
Turn-off loss Eoff (μJ)
ID
+Vth
gfs
dV
=
dt
CGD $VDS $Rg
Moreover, because Vth is determined according to the
impurity concentration of the p-base region, Vth can be
decreased by reducing the impurity concentration of
the p-bases.
Turn-off
dV/ dt=10 kV/ μs
100
10
Rg = 91 1
10
0.5
1.0
Threshold voltage Vth (a.u.)
Turn-off dV / dt (kV/μs)
cantly smaller Ron•A than a conventional MOSFET,
its gate-to-drain capacitance CGD is also significantly
smaller. As a result, there is a problem of the CGD becoming too small, causing the gate controllability to decrease and the turn-off dV / dt to increase. Additionally,
if the gate resistance Rg is increased in order to reduce
the turn-off dV / dt, the Miller period*1 will lengthen
and the loss will increase. As a result, the tradeoff relationship between Eoff and turn-off dV / dt deteriorates.
Accordingly, if the tradeoff relationship between Eoff
and turn-off dV / dt can be improved, and the turn-off
loss decreased, a low power loss device that realizes
both low conduction loss and low switching loss can be
realized. In fact, the Super J-MOS is a realization of
such a device.
1
1.5
Fig.3 Vth dependence of Eoff and turn-off dV / dt
Vol. 58 No. 2 FUJI ELECTRIC REVIEW
2.5
Turn-off dV/ dt =10 kV/ μs
2.0
1.0
0.5
1.5
1.0
issue: Power Semiconductor contributing in energy and environment region
Turn-off loss Eoff (a.u.)
Specific on-resistance Ron‡A (a.u.)
1.5
0.5
0
Super J-MOS
Competitor A’s
product
Competitor B’s
product
Fig.4 Comparison and evaluation of Ron•A
0
Super J-MOS
Competitor A’s
product
Competitor B’s
product
Fig.5 Comparison and evaluation of Eoff
tionship between Eoff and turn-off dV / dt was improved.
PFC circuit
Forward converter
4. Super J-MOS Performance
4.1 Evaluation of on-resistance
For SJ-MOSFETs rated at 600 V / 20 A / 0.19 Ω, the
specific on-resistances Ron•A at rated voltages of the
Super J-MOS and competitors’ products were compared and evaluated. Figure 4 shows the evaluation
results. The values of Ron•A are relative to the value
of the Super J-MOS. With the Super J-MOS, Ron•A
values equal to or better than those of competitors’ SJMOSFETs were confirmed.
4.2 Evaluation of switching loss
Next, Eoff was evaluated. With the Super J-MOS,
the Eoff value is 160 μJ when turn-off dV / dt is 10
kV/ μs, and this extremely small Eoff was realized
through structural optimization. Eoff values when
turn-off dV / dt is 10 kV / μs were compared and evaluated for the Super J-MOS and competitors’ products,
and the results are shown in Fig. 5. As in the case of
Fig. 4, Eoff values are relative to the value of the Super
J-MOS.
As shown in Fig. 5, the Super J-MOS is affected by
the structural optimization and the results showed an
Eoff value significantly lower than those of competitors’
products.
5. Investigation in Electrical Equipment
As described above, by optimizing the surface
structures, the Super J-MOS was confirmed to exhibit
excellent levels of Ron•A and Eoff. Next, to verify the
power efficiency when using a Super J-MOS, a Super
J-MOS was installed in the power factor correction
(PFC) circuit of a 400 W-ATX power supply as shown
in Fig. 6, and the power efficiency was evaluated. The
same evaluation was also performed for company A’s
SJ-MOSFET, which exhibited lowest turn-off loss
“Super J-MOS” Low Power Loss Superjunction MOSFETs
Super J-MOS
Fig.6 Configuration of PFC circuit
among the competitors’ products. The values of power
supply loss and power supply efficiency that were obtained were compared and evaluated (see Fig. 7). All
the devices that were evaluated were rated at 600 V/
0.19 Ω.
As shown in Fig. 7(a), in comparison to company
A’s product, the Super J-MOS exhibits lower loss especially during turn-off, and this contributes greatly to a
reduction in total power supply loss.
Moreover, as shown in Fig. 7(b), highly efficient
power supply operation is realized with the Super
J-MOS, and when the power supply has a 50% load
factor, the power efficiency is at the high level of 96%.
Furthermore, in the load factor range from 20% to
100%, the power efficiency was at least 94% or higher.
This result conforms to the “80 PLUS(6) ”*2 standard,
and indicates that the Super J-MOS possess characteristics that can contribute to improvement of the power
*2: “80 PLUS” : A standard promoting higher efficiency in
power supplies and is defined by an independent private organization (http: / / www.80plus.org). In the power
supplies used in PC and servers, 80 PLUS certification
indicates that the power conversion efficiency is 80% or
greater at load factors of 20%, 50% and 100%.
80 PLUS is a trademark or registered trademark of USbased Ecos Consulting Inc.
81
Power loss (a.u.)
1.5
Input voltage: AC115 V
Output voltage: 390 V
1.0
Turn-off
loss
0.5
Turn-on
loss
Conduction
loss
0
Super J-MOS
Competitor A’s product
(a) Comparison of power loss in power supply
(at output power of 400 W)
Power efficiency (%)
100
98
Super J-MOS
96
94
92
Competitor A’s product
0
20
40
60
Load factor (%)
80
100
(b) Comparison of power efficiency
Fig.7 Results of investigation in electric equipment
efficiency of a power converter.
6. Postscript
By optimizing the surface structures of the SJMOSFET, the “Super J-MOS,” which features an improved tradeoff relationship between turn-off loss and
turn-off dV / dt and low switching loss, was developed.
82
Increasing the gate-to-drain capacitance and lowering the threshold voltage was confirmed to result in
a lower turn-off dV / dt and an improved tradeoff relationship between turn-off loss and turn-off dV / dt. By
optimizing the surface structures of the device, the
turn-off loss was found to be 160 μJ when turn-off
dV / dt is 10 kV / μs, and this is an excellent level for a
SJ-MOSFET. Additionally, as a result of installing a
Super J-MOS in the PFC circuit of a 400 W-ATX power
supply and then conducting an evaluation, power supply operation exhibiting much higher efficiency than
that of competitors’ SJ-MOSFETs was found to be possible.
Targeting applications in the communication and
PC server power supply market, Fuji Electric is currently moving forward with efforts to reduce loss
and increase the efficiency of the 600 V-rated Super
J-MOS. Fuji Electric intends to continue to improve
device performance in the future through device miniaturization and the like.
References
(1) Fujihira, T. “Theory of Semiconductor Superjunction
Devices,” Jpn. J. Appl. Phys., 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) Onishi, Y. et al. “24 mΩcm2 680 V Silicon
Superjunction MOSFET,” Proc. ISPSD’ 02, 2002, p.241244.
(4) Saito, W. et al. “A 15.5 mΩcm2-680 V Superjunction
MOSFET Reduced On-Resistance by Lateral Pitch
Narrowing,” Proc. ISPSD’ 06, 2006, p.293-296.
(5) Baliga, B. J. Modern Power Devices, John Wiley &
Sons, Inc., 1987, p.305-314.
(6) ECOS Consulting. http: / / www.80plus.org. (Refer to
Jul. 29, 2011)
Vol. 58 No. 2 FUJI ELECTRIC REVIEW
P-Channel Power MOSFETs for Space Applications
Masanori Inoue † Akio Kitamura † Shuhei Tatemichi †
Fuji Electric has added a family of p-channel power MOSFETs to its lineup of power MOSFETs for space applications. Depending on the application, designers can now choose between the existing n-channel power MOSFETs
and the new chips, allowing them to reduce the system part count and achieve higher system reliability. Like the nchannel power MOSFETs, the new chips use quasi-plane junction technology to lower the resistance of the drift layer, thus lowering the on-resistance. A low-temperature process is used to form gate oxide films in all diffusion layers,
achieving high total ionizing dose (TID) tolerance.
1. Introduction
technology of these products are introduced below.
That the benefits of utilizing outer space for such
applications as communication satellites, meteorological satellites, GPS and earth observation have permeated our daily lives is well-known. The electronic
devices and switching power supplies installed in artificial satellites are required to be highly efficient so as
to efficiently utilize limited power in outer space and to
have a reduced number of components so as to ensure
system reliability. Moreover, the power metal-oxidesemiconductor field-effect-transistor (MOSFET), a key
device for power conversion, is required to be a low
loss device as well as to have high reliability against
ionizing radiation(1) and to be resistant to high-energy
charged particles (heavy particles)(2) and the like in a
space environment.
Fuji Electric has previously developed and commercialized
n-channel
high-reliability
power
MOSFETs for space applications(3). Recently, p-channel power MOSFETs for space applications have
been newly added to Fuji Electric’s product lineup.
Compared to the n-channel power MOSFET, the pchannel power MOSFET has on-resistance that is 2 to
3 times larger in principle, and this is a disadvantage,
but because its polarity is reversed, high-side switches
and other circuits can be configured more simply. For
this reason, the p-channel power MOSFET has the
advantage of allowing the number of components to
be decreased, thereby enhancing reliability and enabling a reduction in the size and weight of the overall system. Based on the technology cultivated with
n-channel power MOSFET for space applications, Fuji
Electric has developed and commercialized p-channel
MOSFETs for space applications. The features and
2. Product Features
†
Fuji Electric Co., Ltd.
Table 1 lists Fuji Electric’s product lineup of pchannel power MOSFETs, and Table 2 shows the differences in requirements for power MOSFETs for
consumer-use and power MOSFETs for space applications. The MOSFETs for space applications maintain
the equivalent low on-resistance of consumer power
MOSFETs while providing the capability to tolerate a
space environment.
2.1 Tolerance to ionizing radiation (TID tolerance)
Ionizing radiation is present in the environment
along the orbit of an artificial satellite or the like.
Generally, if a power MOSFET normally used at
ground level is used in an environment of ionizing
radiation, the breakdown voltage will decrease and
the gate threshold voltage Vth that controls the on-off
control of the power MOSFET will shift. The marketplace requirement for ionizing radiation tolerance is
1,000 Gy, which is the equivalent exposure as 10 years
of geostationary orbit, and the ability to tolerate this
level of exposure is ensured with the newly developed
p-channel power MOSFET. Ionizing radiation tolerance is assessed by evaluating the change in characteristics that occurs when a product is actually irradiated
with ionizing radiation. Figure 1 shows the evaluation
results of breakdown voltage BVDSS and Vth, which are
particularly susceptible to ionizing radiation. BVDSS
does not change at all. Also, the shift in Vth is limited
to within the range of the specification.
2.2 Heavy particle tolerance (SEE tolerance)
In space, heavy particles emitted from solar winds,
supernova explosions and the like fly back and forth.
83
issue: Power Semiconductor contributing in energy and environment region
ABSTRACT
Table 1
Product list
Product type
VDSS
(V)
lD
(A)
lD (Pulse)
(A)
RDS(on)
max. *1
(Ω)
PD *2
(W)
VGS
(V)
VGS(th)
(V)
Qg
max.
(nC)
Radiation
level
(krad)
SEE *3
LET MeV /
(mg / cm2)
Package
type
Mass
(g)
JAXA R 2SJ1 A0l
− 100
− 42
− 168
0.045
250
±20
−2.5
to
−4.5
230
100
37
TO-254
9.3
JAXA R 2SJ1 A02
− 100
− 25
− 100
0.097
125
±20
− 2.5
to
− 4.5
95
100
37
TO-254
9.3
JAXA R 2SJ1 A03
− 100
− 11
− 44
0.226
62.5
±20
− 2.5
to
− 4.5
40
100
37
TO-254
9.3
JAXA R 2SJ1 A04
− 100
− 42
− 168
0.038
250
±20
− 2.5
to
− 4.5
230
100
37
SMD-2
3.3
JAXA R 2SJ1 A05
− 100
− 29
− 116
0.09
150
±20
− 2.5
to
− 4.5
95
100
37
SMD-1
2.6
JAXA R 2SJ1 A06
− 100
− 13
− 52
0.219
70
±20
− 2.5
to
− 4.5
40
100
37
SMD-0.5
1.0
JAXA R 2SJ1 A07
− 200
− 35
− 140
0.091
250
±20
− 2.5
to
− 4.5
230
100
37
TO-254
9.3
JAXA R 2SJ1 A08
− 200
− 16
− 64
0.21
125
±20
− 2.5
to
− 4.5
95
100
37
TO-254
9.3
JAXA R 2SJ1 A09
− 200
− 7.5
− 30
0.487
62.5
±20
− 2.5
to
− 4.5
40
100
37
TO-254
9.3
JAXA R 2SJ1 A10
− 200
− 37
− 148
0.084
250
±20
− 2.5
to
− 4.5
230
100
37
SMD-2
3.3
JAXA R 2SJ1 A11
− 200
− 18
− 72
0.203
150
±20
− 2.5
to
− 4.5
95
100
37
SMD-1
2.6
JAXA R 2SJ1 A12
− 200
− 8.5
− 34
0.48
70
±20
− 2.5
to
− 4.5
40
100
37
SMD-0.5
1.0
*1 : RDS(On): VGS = − 12 V,
C
*2 : PD: TC = 250 °
*3 : SEE: Kr, Energy: 520 MeV, Range: 63 μm, VDS = rated VDS, VGS =+5 V
Table 2
Requirements of power MOSFET for space
applications
Consumer-use
MOSFET
Space-use
MOSFET
Ionizing radiation tolerance
(TID tolerance)
×
◎
Heavy particle tolerance
(SEE tolerance)
×
◎
Long-term reliability
Requirements
Electrical characteristics
○
◎
Breakdown
voltage
200 V
200 V
Onresistance
◎
◎
TID:Total ionizing dose
SEE:Single event effect
◎: Fully meets requirements
○: Meets requirements
× : Does not meet requirements
The phenomenon in which a single incident heavy particle can cause degraded characteristics or permanent
damage in a biased MOSFET is generically referred to
as a “single event effect” (SEE). The probability of impact in space by a heavy particle with a larger mass is
84
lower, but such a particle would have a large amount
of energy and would have a significant effect on a
MOSFET. The magnitude of the energy that a heavy
particle would transfer to a MOSFET is described
by the linear energy transfer (LET) of the particle.
Additionally, the SEE tolerance also depends on the biased state of the power MOSFET at the time when the
heavy particle comes in, and the higher the VDS and
VGS, the greater the susceptibility to damage.
Therefore, the SEE tolerance is generally expressed with the magnitude of the LET and the useable areas of VDS and VGS. Figure 2 is shown for the
power MOSFET at a LET value of 37 MeV / (mg/cm2).
Because a reverse bias of + 5 V or greater is not used
with VGS, in the actual useable area, the tolerance extends to the rated value of VDS. The probability that
a heavy particle having this LET value would collide
with the MOSFET corresponds to approximately once
every 200 years (estimated assuming an orbit altitude
of 550 km and an orbital inclination angle of 31°), and
Vol. 58 No. 2 FUJI ELECTRIC REVIEW
Radiation source: 60CO a-rays,
Radiation dose: 1,000 Gy (360 Gy/ h)
Bias conditions (during and after irradiation)
™VDS =0 V, VGS =< 20 V
<250
™VDS = < 160 V, VGS = 0 V
<200
VDS V
< 250
<150
<100
< 200
Usable area
Lower limit value: <200 V
<50
Bias
< 150
0
VGS = < 20 V, VDS =0 V
VGS = 0 V, VDS = < 160 V
500
Total radiation dose (Gy)
(a) Breakdown voltage BVDSS
0
0
+5
+10
+ 15
+20
+ 25
VGS V
1,000
Fig.2 Tolerance to heavy particles (SEE tolerance)
<6
Vth (V)
ID= < 1 mA
<5
Upper limit: <4.5 V
<4
SMD-2
<3
SMD-1
Lower limit value: <2.5 V
<2
Bias
<1
0
TO-254
0
SMD-0.5
VGS = < 20 V, VDS =0 V
VGS = 0 V, VDS = < 160 V
500
Total radiation dose (Gy)
(b) Threshold voltage Vth
1,000
Fig.1 Tolerance to ionizing radiation (TID tolerance)
extremely high reliability against heavy particles is
provided.
2.3 Long-term reliability
A metallic hermetically sealed package is used to
ensure long-term reliability of the package. In order
to improve thermal cycle tolerance, the package frame
(part into which the MOSFET chip is mounted) is
made from CuW (copper tungsten), which has a coefficient of thermal expansion extremely close to that of
silicon (the material from which the MOSFET chip is
made). Additionally, the hermetically sealed package
is made hollow (Fig. 3) and filled with dry nitrogen to
protect the power MOSFET chip from exogenous degradation modes.
3. Technology Applied to the MOSFETs for
Space Applications
The technology developed for the previously commercialized 2nd generation n-channel power
MOSFETs for space applications was also applied to
these p-channel MOSFETs. This section introduces
the technology for ensuring TID tolerance and low onresistance.
P-Channel Power MOSFETs for Space Applications
Fig.3 Internal structure of power MOSFETs for space
applications
3.1 Application of low-temperature process
The phenomenon of a Vth shift is caused by electrical charge becoming trapped in the oxide film as a result of ionizing radiation.
When a MOSFET is exposed to ionizing radiation,
electrons and holes are induced in its oxide film. The
holes, in particular, become trapped in the oxide film
and form a fixed positive electrical charge that appears
in the electrical characteristics as a shift in Vth.
From prior research, it is known that the greater
the thermal history of exposure to high-temperatures,
the greater amount of charge that will be trapped in
an oxide film. In a consumer-use MOSFET, an efficient
manufacturing process and stable electrical characteristics can be provided through the use of a gate electrode (on which a gate oxide film has also been formed)
as a self-aligned mask. With this process, however,
the gate oxide film acquires a thermal history of hightemperature exposure during the diffusion layer fabrication process, and the oxide layer ends up trapping a
large amount of charge.
Therefore, a process flow in which the gate oxide
film is formed after fabrication of the entire diffusion
layer was applied to MOSFETs for space applications
85
issue: Power Semiconductor contributing in energy and environment region
BVDSS (V)
ID= < 1 mA
LE: 37 MeV/ (mg/ cm2)
Ion: Kr, Energy: 520 MeV
Range: 63.1 +m,
TA=25+/ <5 °C
Fluence: 3E5+/<5% ions /cm2
Initial oxidation
40
Fabrication of gate electrode
30
Fabrication of diffusion layer
Fabrication of gate oxide film
Fabrication of gate electrode
Fabrication of inter-layer insulation film
Fabrication of inter-layer insulation film
Fabrication of surface electrode
Fabrication of surface electrode
(a) Consumer-use MOSFET
(b) MOSFET for space applications
Fig.4 Process flow of consumer-use MOSFET and MOSFET
for space applications
Source
p+ n+ p+
n base
Gate
p+ n+ p+
n base
p+ n+ p+
n base
p<
p+
Drain
Fig.5 Cross-sectional view of active part of p-channel power
MOSFET for space applications
(see Fig. 4). By applying this process, the gate oxide
film acquired a lower temperature thermal history and
a TID tolerance of 1,000 Gy was ensured.
3.2 Low on-resistance technology
The quasi-planer-junction (QPJ)(4) technology used
in Fuji Electric’s consumer-use MOSFETs was also applied to this MOSFET.
With QPJ technology, n-bases are arranged in a
dense configuration and the interval between the nbases is minimized to flatten the electric field and obtain a breakdown voltage close to that of a planar pn
junction (see Fig. 5). As a result, this technology lowers
the resistivity of the drift layer and reduces the on-
Ron‡A (m1cm2)
Fabrication of gate oxide film
Fabrication of diffusion layer
86
50
Initial oxidation
Competitor’s product
20
<25%
200 V class
p-channel power
MOSFET
100 V class
10
70 80 90 100
200
300
400
BVDSS (V)
Fig.6 Ron•A versus BVDSS tradeoff
resistance.
The application of this QPJ technology resulted in
a 25% reduction in on-resistance per unit area compared to that of a competitor’s product (see Fig. 6).
4. Postscript
This paper has introduced Fuji Electric’s p-channel power MOSFET for space applications. These
MOSFETs are tolerant of the space environment, i.e.,
they exhibit good tolerance of ionizing radiation, heavy
particles and so on, and also realize the low on-resistance of consumer-use power MOSFETs.
Fuji Electric will continue to endeavor to reduce
on-resistance further, improve SEE tolerance, and to
contribute to the promotion of space development in
Japan and overseas.
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.
(3) Inoue, M. et al. High Reliability Power MOSFETs for
Space Applications. FUJI ELECTRIC REVIEW. 2010,
vol.56, no.2, p.69-73.
(4) Yamada, T. et al. Power MOSFET ‘Super FAP-G
Series’ for Low-Loss, High-Speed Switching. FUJI
ELECTRIC REVIEW. 2001, vol.47, no2, p.41-44.
Vol. 58 No. 2 FUJI ELECTRIC REVIEW
Supplemental Explanation
3-level inverter technology
Figure in which an inverter is wired to the intermediate potential (N) of the DC power source, is known as
the neutral-point-clamped (NPC) method. The naming
of this method originates from the fact that the voltage
applied to the switch element is always clamped to half
the DC voltage Ed.
Compared to the NPC method, the advanced-NPC
(A-NPC) method enables a simpler circuit configuration because the series-connected insulated gate bipolar transistors IGBTs have twice the rated voltage as
the IGBTs used with the NPC method, and an reverse
blocking IGBT is connected between the intermediate potential point (N) of the DC power source and an
intermediate point (U) of the series-connected IGBTs.
Having fewer current-conducting elements, the A-NPC
method has the advantages of realizing lower loss and,
when configuring the inverter, of requiring fewer number of power sources for the gate driving circuit.
A multilevel-type inverter, as typified by a 3-level
inverter, has many advantages over a typical 2-level
inverter. As shown in the Figure, the voltage waveform
at the conversion output part of the 2-level inverter
is ± Ed pulse width modulated (PWM) pulses centered
about the zero point, but in the case of a 3-level inverter, is combined PWM pulses of ± Ed / 2 and ± Ed centered about the zero point. Because the output waveform of the 3-level inverter more closely resembles
a sine wave, the size of the LC filter used to convert
the output waveform into a sine wave can be reduced.
Furthermore, the switching loss occurring in a switch
element is roughly halved and the noise generated by
the equipment is also reduced, because the width of
voltage fluctuation per one-time switch operation is
half that of the 2-level inverter. The use of a 3-level
inverter having these characteristics is effective for realizing smaller size and higher efficiency in a system.
For 3-level inverters, the method shown in the
3-level inverter
2-level inverter
NPC method
A-NPC method
Converter
Converter
Converter
LC filter
L
Ed
Ed
C
LC filter
LC filter
L
L
Ed
N
N
U
C
C
RB-IGBT
Output line-to-line voltage waveform
Output line-to-line voltage waveform
Filter output
Ed
Converter output
Ed / 2
Filter output
Ed / 2
Converter output
Figure Comparison of 2-level inverter and 3-level inverter circuits and voltage waveforms
Supplemental Explanation
87
Supplemental Explanation
Supplemental explanation 1
88
Vth
Time
VDS
In the switching of a semiconductor device, the
gate capacitance is charged and discharged to turn-on
and turn-off the device. At such times, changes in the
drain-source voltage VDS cause the gate-drain capacitance CGD to change, and an interval occurs in which
the gate-source voltage VGS for charging and discharging CGD becomes flat. This is known as the Miller period. The Figure shows a schematic of the L-load turnoff switching waveform of a power metal-oxide-semiconductor field-effect-transistor. As shown in Figure,
when VGS decreases, at the time t0 when VGS = VDS,
a depletion layer begins to expand on the gate to the
drain and VDS begins to increase. As a result, CGD
decreases and the Miller period is appeared. At the
time t1 when VDS reaches the power source voltage, the
depletion layer has stopped to expand, CGD stops decreasing and the Miller period ends. At the end of the
Miller period, VGS and the drain current ID begin to
decrease. The duration of the Miller period depends on
the product CGD and the gate resistance Rg, and therefore it is important that the device be designed such
that loss does not increase with an extended Miller
period.
VGS
Miller period
Time
ID
Supplemental explanation 2
t0
Miller period
t1
Time
Figure Schematic diagram of L-load turn-off switching
waveform
Vol. 58 No. 2 FUJI ELECTRIC REVIEW
Overseas Subsidiaries
North America
Fuji Electric Corp. of America
M
arketing, installation and repair of electrical machinery, control
systems and electronic components
Tel +1-732-560-9410
URL http://www.fujielectric.com/fecoa/
EU
Fuji Electric Motor (Dalian) Co., Ltd.
Manufacture of motors
Hoei Hong Kong Co., Ltd.
M
arketing, installation and repair of electrical machinery, control
systems and electronic components
Tel +852-2369-8186
URL http://www.hoei.com.hk/eng/index.php Fuji Electric (Changshu) Co., Ltd.
Fuji Electric Europe GmbH
M
anufacture and sales of electromagnetic contactors and thermal
relays
M
arketing, installation and repair of electrical machinery, control
systems and electronic components
Tel +86-512-5284-5642
URL http://www.csfe.com.cn/
Tel +49-69-6690290
URL http://www.fujielectric.de/ Fuji Electric France S.A.S
Manufacture and sale of measuring and control devices
Tel +33-4-73-98-26-98
URL http://www.fujielectric.fr
East Asia
Fuji Electric FA (Asia) Co., Ltd.
Fuji Electric FA Korea Co., Ltd.
S
ales of power receiving and distribution/control equipment and drive
control equipment
Tel +82-2-780-5011
URL http://www.fujielectric.co.kr/
Shanghai Fuji Electric Transformer Co., Ltd.
Manufacture and sales of molded case transformers
Tel +86-21-5718-7705
URL http://www.csfuji.com/
Marketing of power distributors and control equipment
Tel +852-2311-8282
URL http://www.fea.hk/
Fuji Electric Dalian Co., Ltd.
Manufacture of low-voltage circuit breakers
Tel +86-411-8762-2000
Fuji Electric Taiwan Co., Ltd.
S
emiconductor devices, power distribution and control equipment,
sales of control, drive and rotating equipment
Southeast and South Asia
Fuji Electric Asia Pacific Pte. Ltd.
M
arketing, installation and repair of electrical machinery, control
systems and electronic components
Tel +65-6533-0010
URL http://www.fujielectric.com.sg/ Fuji Electric Philippines, Inc.
Manufacture of semiconductor devices
Tel +886-2-2515-1850
Tel +63-2-844-6183
Fuji Electric (China) Co., Ltd.
Fuji Electric (Malaysia) Sdn. Bhd.
M
arketing of locally manufactured or imported products in China, and
export of locally manufactured products
Tel +86-21-5496-1177
URL http://www.fujielectric.com.cn/
Fuji Electric (Shenzhen) Co., Ltd.
Manufacture of magnetic disks
Tel +60-4-403-1111
URL http://www.fujielectric.com.my/ Fuji Electric Semiconductor (Malaysia) Sdn.
Bhd.
Manufacture and marketing of photoconductive drums
Manufacture of semiconductor devices
Tel +86-755-2734-2910
URL http://www.szfujielectric.com.cn/FUJIWebSite/index.
html
Tel +60-4-494-5800
Wuxi Fuji Electric FA Co., Ltd.
URL http://www.fujielectric.com.my/ Fuji Electric Power Supply (Thailand) Co., Ltd.
Manufacture and marketing of inverters in China
M
anufacture and sale of small to mid-size UPS and internal electrical
parts
Tel +86-510-8815-2088 Tel +66-0-2909-5998
Fuji Electric Hong Kong Co., Limited
Fuji Electric India Pvt.Ltd.
Sales of semiconductor devices and photoconductors
Tel +852-2664-8699
URL http://www.szfujielectric.com.cn/hkeng/company/index.
htm Shanghai Fuji Electric Switchgear Co., Ltd.
M
anufacture and sales of switching equipment, monitoring and
control appliances and related facilities
Tel +86-21-5718-1234
URL http://www.csfuji.com/
Sales of drive control equipments and semiconductor devices
Tel +91-22-4010 4870
URL http://www.fujielectric.co.in
Printed on recycled paper
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement