Technology for Planar Power Semiconductor Devices … for Planar Power Semiconductor Devices Package...
-
Upload
nguyenduong -
Category
Documents
-
view
220 -
download
0
Transcript of Technology for Planar Power Semiconductor Devices … for Planar Power Semiconductor Devices Package...
Technology for Planar Power Semiconductor Devices Package with
Improved Voltage Rating
by
Jing Xu
Dissertation submitted to the Faculty of the Virginia Polytechnic Institute and State
University in partial fulfillment of the requirements for the degree of
DOCTOR OF PHILOSOPHY
in
ELECTRICAL ENGINEERING
Khai. D. T. Ngo
Daan van Wyk
Guoquan Lu
Hardus Odendaal
Yilu Liu
Douglas Nelson
Dec. 5, 2008
Blacksburg, Virginia
Keywords: High voltage, Planar package, Embedded Power, SiC diode, MEDICI
Copyright 2008, Jing Xu
Technology for Planar Power Semiconductor Devices Package with
Improved Voltage Rating
Jing Xu
ABSTRACT
The high-voltage SiC power semiconductor devices have been developed in recent
years. They cause an urgent in the need for the power semiconductor packaging to have
not only low interconnect resistance, less noise, less parasitic oscillations, improved
reliability, and better thermal management, but also High-Voltage (HV) blocking
capability.
The existing power semiconductor packaging technologies includes wire-bonding
interconnect, press pack, flip-chip technology, metal posts interconnected parallel plates
structure (MIPPS), dimple array interconnection (DAI), power overlay (POL)
technology, and embedded power (EP) technology. None of them meets the
requirements of low profile and high voltage rating.
The objective of the work in this dissertation is to propose and design a high-voltage
power semiconductor device packaging method with low electric field stress and low
profile to meet the requirements of high-voltage blocking capability. The main
contributions of the work presented in this dissertation are:
1. Understanding the electric field distribution in the package.
The power semiconductor packaging is simulated by using Finite Element Analysis
(FEA) software. The electric field distribution is known and the locations of high electric
field concentration are identified.
2. Development of planar high-voltage power semiconductor device packaging method
With the proposed structure in the dissertation, the electric field distribution of a planar
device package is improved and the high electric field intensity is relieved.
3. Development of design guidelines for the proposed planar high-voltage device
packaging method.
The influence of the structure dimensions and the material properties is studied. An
optimal design is identified. The design guideline is given.
iii
4. Fabrication and experimental verification of the proposed high-voltage device
packaging method
A detailed fabrication procedure which follows the design guideline is presented. The
fabricated modules are tested by using a high power curve tracer. Test results verify the
proposed method.
5. Simplification of the structure model of the proposed device package
The package structure model is simplified through the elimination of power
semiconductor device internal structure model. The simplified model can be simulated
by a non-power device simulator. The simulation results of the simplified model match
the simulation results of the complete model very well.
iv
Acknowledgements
First and foremost, I would like to express my deepest gratitude to my academic
advisors, Dr. J. D. van Wyk and Dr. Khai D. T. Ngo. Dr. van Wyk introduced me into
this wonderful packaging world and guided me to the research of the topic. In the first
three years of my study, with his greatest patience, he leads me to this very interesting
area, about which I have great passion finally, although knew nothing from the
beginning. Even after his retirement, Dr. van Wyk gave me many support and mentoring
on the research. To me, he is like both an advisor and a grandfather. In the last two years
of my study, Dr. Ngo gave me tremendous supervising on the research. His brilliant
ideas, his illuminating comments and suggestions, and his strict training lead the
completion of this dissertation. I am impressed by his broad thoughts and diligent
attitude. It is such a great experience to be both professors’ student for these years. Both
of them showed me their great intuition, broad knowledge and accurate judgment.
Without their unconditional support in guiding me to conduct and complete this research,
this dissertation would never have come into existence. The most precious thing I learned
from them is the attitude toward research, which can be applied to every other aspect of
life, too.
I also would like to thank Dr. Zhenxian Liang and Mr. Dan Huff. Dr. Liang taught me
everything about embedded power, which is the basis of my dissertation. Dan gave me
tremendously unconditional support from every aspect of the packaging lab, and this
work wouldn’t be done without his help.
I am grateful to have Dr. GQ Lu, Dr. William Hardus Odendaal, Dr. Yilu Liu, and Dr.
Doug Nelson as my doctoral committee members. I owe them a special note of
appreciation for their encouragement and expertise throughout my graduate studies.
I am indebted to the National Science Foundation and Center for Power Electronics
Systems (CPES) at Virginia Tech for funding this research (under award number EEC-
9731677) and providing research facilities. I also appreciate the support from all the staff
in CPES, especially Dan Huff. I wish to thank all the colleagues and friends at CPES for
their insightful discussions, guidance, and helps during my stay at CPES. I will always
cherish our friendships.
v
Table of Contents
ABSTRACT................................................................................................................................... ii
Acknowledgements ...................................................................................................................... iv
Table of Contents .......................................................................................................................... v
List of Figures.............................................................................................................................. vii
List of Tables ...............................................................................................................................xii
Chapter 1: Introduction.......................................................................................................- 1 - 1.1. Requirements for the power semiconductor packaging in power electronics
applications .............................................................................................................................- 1 -
1.2. Review of existing packaging technologies of the power semiconductor devices and
modules ...................................................................................................................................- 2 -
1.2.1. Wire bonding ...................................................................................................... - 3 -
1.2.2. Press Pack ........................................................................................................... - 4 -
1.2.3. Flip-Chip Technology......................................................................................... - 4 -
1.2.4. Metal-Posts Interconnected Parallel Plates (MPIPPs) Technology.................... - 6 -
1.2.5. Dimple Array Interconnect (DAI) Technology .................................................. - 7 -
1.2.6. Power Overlay (POL) Technology ..................................................................... - 8 -
1.2.7. Embedded Power Technology ............................................................................ - 8 -
1.2.8. Summary of existing power semiconductor packaging of the power
semiconductor devices and modules................................................................................. - 10 -
1.3. Research objective and approach..............................................................................- 11 -
1.4. Dissertation outline ...................................................................................................- 12 -
Chapter 2: Survey of Strategies for Reduction of Electric Field in Semiconductor Die
and Packages. ..........................................................................................................................- 13 - 2.1. Introduction...............................................................................................................- 13 -
2.2. Overview of classical techniques for high electric field intensity reduction in power
semiconductors [34][35] .......................................................................................................- 13 -
2.3. Overview of high electric field reduction techniques in wire bonding packages [37][38]
...................................................................................................................................- 16 -
2.4. Summary...................................................................................................................- 19 -
Chapter 3: Development of High-Voltage Embedded Power (HVEP) Technology via
MEDICI......... ..........................................................................................................................- 21 - 3.1. Introduction...............................................................................................................- 21 -
3.2. Evaluation of two LVEP structures ..........................................................................- 23 -
3.3. Parametric study of LVEP #2 ...................................................................................- 26 -
3.4. Medium-Voltage Embedded Power (MVEP) structure............................................- 31 -
3.5. Development of High-Voltage Embedded Power (HVEP) structure .......................- 38 -
3.6. Geometry impact on HVEP structure .......................................................................- 41 -
3.7. Summary...................................................................................................................- 47 -
Chapter 4: Design Guidelines for the HVEP package and Fabrication Process of a
HVEP package ........................................................................................................................- 49 -
vi
4.1. Design guidelines for the HVEP package.................................................................- 49 -
4.1.1. Introduction to engineering design ................................................................... - 49 -
4.1.2. Design practice in the HVEP package .............................................................. - 50 -
4.1.3. Design guidelines for the HVEP package......................................................... - 51 -
4.1.4. A design example.............................................................................................. - 53 -
4.2. Fabrication process of a HVEP package...................................................................- 55 -
4.2.1. Devices and materials ....................................................................................... - 55 -
4.2.2. Design of the fabrication procedure.................................................................. - 56 -
4.2.3. Fabrication process and experimental results ................................................... - 59 -
4.3. Summary...................................................................................................................- 63 -
Chapter 5: Simplified Modeling of the HVEP Package..................................................- 65 - 5.1. Model simplification.................................................................................................- 65 -
5.2. Simulation in MAXWELL 2D..................................................................................- 67 -
5.3. Summary...................................................................................................................- 70 -
Chapter 6: Summary..........................................................................................................- 72 - 6.1. Contributions.............................................................................................................- 72 -
6.2. Recommendations.....................................................................................................- 76 -
Reference....... ..........................................................................................................................- 77 -
Appendix Mathematical Modeling of the Simplified Model of HVEP........................- 81 - A1 Conformal Mapping and Schwarz-Christoffel Transformation......................................- 81 -
A2 Model of HVEP package ................................................................................................- 84 -
A3 The problem ....................................................................................................................- 86 -
vii
List of Figures
Fig. 1.1 Power semiconductor package using wire bonding technology
[14]............................................................................................................ - 1 -
Fig. 1.2 Pictures of the IGBT devices with press pack interconnect [18] © 2004
IEEE. (a) Asymmetric 4500 V/1000 A Press Pack IGBT along with a
submodule comprising 12 IGBT chips. (b) Direct pressure contact on the
gate............................................................................................................ -4-
Fig. 1.3 Examples of BGA package. (a) MOSFET BGA [21]. (b) BGA for
lateral MOSFET [22] © 2004 IEEE. (c) FlipFET [23]. (d) BGA for
Volterra integration module [24]..................................................................... -5-
Fig. 1.4 Flip-chip on flex technology [25] © 2001 IEEE. (a) Picture of a
module. (b) Cross-section schematic............................................................... -6-
Fig. 1.5 Metal-posts interconnected parallel plates technology [29] © 1999
IEEE. (a) Picture of MPIPPs module. (b) Cross-section schematic............... -7-
Fig. 1.6 Dimple array interconnect technology [31]. (a) Picture of DAI. (b)
Schematic cross-section............................................................................ -7-
Fig. 1.7 Cross-sectional schematic of a GE-POL structure.................................... -8-
Fig. 1.8 Embedded power technology. (a) Picture of an IPEM using embedded
power technology. (b) Cross-sectional schematic of an IPEM using
embedded power technology.................................................................... -9-
Fig. 2.1 Electric field crowding for a planar diffused junction.............................. -13-
Fig. 2.2 Cross-section of edge termination with a single floating field ring.......... -14-
Fig. 2.3 Cross-section of edge termination with multiple field rings..................... -14-
Fig. 2.4 Planar junction with field plate................................................................. -15-
Fig. 2.5 Cross section of field plate with step oxide.............................................. -15-
Fig. 2.6 Cross section of field plate with step oxide.............................................. -16-
Fig. 2.7 Cross section of resistive field plate......................................................... -16-
Fig. 2.8 Cross Edge termination design combining floating field rings with field
plate........................................................................................................... -16-
Fig. 2.9 Cross-sectional diagram of an IGBT module package using wire
bonding..................................................................................................... -17-
Fig. 2.10 Cross Electric field concentration due to proximity between track and
bonding..................................................................................................... -17-
Fig. 2.11 Influence of a resistive layer..................................................................... -18-
Fig. 2.12 Influence of modification of metal edge................................................... -19-
Fig. 2.13 Module reinforcement at the copper edges............................................... -19-
Fig. 3.1 Cross sectional view of an IPEM.……………………………………… -22-
Fig. 3.2 Enlarged cross-sectional schematic of the package of a semiconductor
device using LVEP.…………………………………………………….. -22-
viii
Fig. 3.3 Cross-sectional schematic of a 10 kV PiN SiC diode [42]....................... -23-
Fig. 3.4 Electric field and potential distribution for an abrupt parallel-plane
P+/N junction [34].................................................................................... -23-
Fig. 3.5 Simulation model and result of LVEP structure in MEDICI. (a)
Dimensions and materials of the structure. (b) Distribution of
equipotential lines..................................................................................... -25-
Fig. 3.6 Electric field intensity in the insulation materials of LVEP structure
when the applied voltage on the cathode is 13.5 kV. (a) Along the x
direction (lateral direction) line AB in the polyimide. (b) Along the y
direction (vertical direction) line CD in the embedded insulation
material.Electric field intensity in the insulation materials of LVEP
structure when the applied voltage on the cathode is 13.5 kV. (a)
Along the x direction (lateral direction) line AB in the polyimide. (b)
Along the y direction (vertical direction) line CD in the embedded
insulation material. ................................................................................... -26-
Fig. 3.7 Simulation model and result of LVEP #2 structure in MEDICI. (a)
Dimensions and materials of the structure. (b) Distribution of
equipotential lines..................................................................................... -27-
Fig. 3.8 Electric field intensity in the insulation materials of LVEP #2 structure
when the applied voltage on the cathode is 13.5 kV. (a) Along the x
direction (lateral direction) line AB in the polyimide. (b) Along the y
direction (vertical direction) line CD in the embedded insulation
material..................................................................................................... -28-
Fig. 3.9 Comparison of the maximum electric field intensity between two
LVEP structures........................................................................................ -28-
Fig. 3.10 Parameters in the LVEP#2 structure study............................................... -29-
Fig. 3.11 Variation of maximum electric field intensity with the thickness of
polyimide. (a) Inside the polyimide at the upper corner. (b) Inside the
embedded insulation material at the upper and lower corner................... -29-
Fig. 3.12 Comparison of the equipotential lines distribution between two
different thicknesses of polyimide. (a) D = 100µm. (b) D = 500 µm..... -30-
Fig. 3.13 Variation of maximum electric field intensity with the dielectric
constant of polyimide. (a) Inside the polyimide. (b) Inside the
embedded insulation material................................................................... -31-
Fig. 3.14 Comparison of the equipotential lines distribution between two
different dielectric constant of polyimide. (a) εr = 3. (b) εr = 1............. -31-
Fig. 3.15 Variation of maximum electric field intensity with the dielectric
constant of embedded insulation material. (a) Inside the polyimide. (b)
Inside the embedded insulation material................................................... -32-
Fig. 3.16 Comparison of the equipotential lines distribution between two
different dielectric constant of embedded insulation material. (a) εr’ =
4. (b) εr’ = 15........................................................................................... -32-
ix
Fig. 3.17 Simulation model and result MVEP structure in MEDICI. (a)
Dimensions and materials of the structure. (b) Distribution of
equipotential lines..................................................................................... -34-
Fig. 3.18 Electric field intensity in the insulation materials of MVEP structure.
(a) In the polyimide. (b) In the embedded insulation material................ -35-
Fig. 3.19 Comparison of the maximum electric field intensity between two
LVEP structures and the MVEP structure................................................ -36-
Fig. 3.20 Parameters in the MVEP structure study.................................................. -36-
Fig. 3.21 Variation of maximum electric field intensity with the thickness of
polyimide. (a) Inside the polyimide at the upper corner. (b) Inside the
embedded insulation material................................................................... -36-
Fig. 3.22 Comparison of the equipotential lines distribution between two
different thicknesses of the polyimide. (a) D = 100 µm. (b) D = 500
µm............................................................................................................. -37-
Fig. 3.23 Variation of maximum electric field intensity with the dielectric
constant of polyimide. (a) Inside the polyimide at the upper corner. (b)
Inside the embedded insulation material................................................... -37-
Fig. 3.24 Comparison of the equipotential lines distribution between two
different dielectric constant of polyimide. (a) εr = 4. (b) εr = 8............. -37-
Fig. 3.25 Equivalent circuit model of LVEP#2. (a) Structure of LVEP#2. (b)
Series capacitor model of a small area in the LVEP#2 structure.............. -38-
Fig. 3.26 Simulation model and result of HVEP structure in MEDICI. (a)
Dimensions and materials of the structure. (b) Distribution of
equipotential lines..................................................................................... -39-
Fig. 3.27 Electric field intensity in the insulation materials of HVEP structure.
(a) In the polyimide. (b) In the embedded insulation material................ -40-
Fig. 3.28 Comparison of the maximum electric field intensity of different
structures................................................................................................... -40-
Fig. 3.29 Parameters in the HVEP structure study................................................... -41-
Fig. 3.30 Variation of maximum electric field intensity at the corner of the
device with the thickness of the polyimide............................................... -42-
Fig. 3.31 Variation of maximum electric field intensity with the position of the
inserted metal layer. (a) Maximum electric field intensity in two
insulation materials. (b) Distribution of Equipotential lines at t =
200µm. ..................................................................................................... -42-
Fig. 3.32 Distribution of equipotential lines with different combination of the
position and the voltage of the inserted metal layer. (a) t = 0µm and V
= 0 V. (b) t = 200 µm and V = 5 kV. ................................................. -43-
Fig. 3.33 Variation of maximum electric field intensity with different
combination the pos of the inserted metal layer and its voltage............... -44-
x
Fig. 3.34 Variation of maximum electric field intensity with the width of the
inserted metal layer. (a) Maximum electric field intensity in the
polyimide. (b) Maximum electric field intensity in the embedded
insulation material..................................................................................... -45-
Fig. 3.35 Distribution of equipotential lines with different width of the inserted
metal layer. (a) w = 500µm. (b) w = -400 µm....................................... -45-
Fig. 3.36 Variation of the maximum electric field intensity along the horizontal
line at y = -100 µm with the angle of the polyimide .............................. -46-
Fig. 3.37 Distribution of equipotential lines with different angle of the
polyimide. (a) θ = 30°. (b) θ = 90°........................................................ -46-
Fig. 3.38 Comparison of the maximum electric field intensity of different
structures................................................................................................... -47-
Fig. 4.1 Engineering design process of HVEP package......................................... -51-
Fig. 4.2 A design example. (a) Physical layout design. (b) Inserted metal layer
design. (c) Polyimide layer design. (d) Top metallization design.......... -54-
Fig. 4.3 Picture and dimensions of a 2 kV SiC diode............................................ -55-
Fig. 4.4 Major fabrication steps of the LVEP package.......................................... -58-
Fig. 4.5 Major fabrication steps of the HVEP package.......................................... -59-
Fig. 4.6 Fabrication process of the HVEP package: step 1. Ceramic cutting........ -60-
Fig. 4.7 Fabrication process of the HVEP package: step 2. Chip embedding....... -60-
Fig. 4.8 Fabrication process of the HVEP package: step 3. Shim soldering.......... -60-
Fig. 4.9 Fabrication process of the HVEP package: step 4. Masking.................... -61-
Fig. 4.10 Fabrication process of the HVEP package: step 5. Thin layer copper
deposition.................................................................................................. -61-
Fig. 4.11 Fabrication process of the HVEP package: step 6. Interlayer patterning. -61-
Fig. 4.12 Fabrication process of the HVEP package: step 7. Top metallization
deposition.................................................................................................. -61-
Fig. 4.13 Fabrication process of the HVEP package: step 8. Electroplating and
etching....................................................................................................... -62-
Fig. 4.14 Test result of the fabricated HVEP package............................................. -62-
Fig. 4.15 Cross-sectional picture of one fabricated HVEP package........................ -63-
Fig. 4.16 ESEM picture of one fabricated HVEP package. (a) Copper shim edge.
(b) Package edge....................................................................................... -63-
Fig. 5.1 Model of the power semiconductor device. (a) Model in MEDICI. (b)
Simplified model....................................................................................... -66-
Fig. 5.2 Simplified model of the HVEP package................................................... -66-
Fig. 5.3 Voltage distribution on JTE...................................................................... -67-
Fig. 5.4 Simplified model for mathematical calculation and simulation............... -67-
Fig. 5.5 Comparison of the voltage distribution on JTE in MEDICI and
MAXWELL. (a) Voltage distribution in MEDICI. (b) Voltage
distribution in MAXWELL....................................................................... -68-
xi
Fig. 5.6 Comparison of the electric field intensity at different y locations.
Upper: MEDIC; Lower: MAXWELL 2D (a) y = 0. (b) y = 10 µm. (c)
y = 50 µm. (d) y = 100 µm. (e) y = 200 µm. (f) y = 300 µm................. -69-
Fig. 5.7 Explanation of two values in Fig. 5.6....................................................... -69-
Fig. 5.8 Errors of the simulation results in MAXWELL compared to the results
in MEDICI................................................................................................ -70-
Fig. A.1 Schwarz-Christoffel transformation. (a) Polygon (original plane) (b)
Upper-half plane (destination plane)........................................................ -82-
Fig.A.2 Elliptic function of first kind. (a) Rectangle (original plane) (b)
Upper-half plane (destination plane)........................................................ -82-
Fig.A.3 Transformation of the simplified structure. (a) Simplified structure
(rotate 90°). (b) Transformed structure.................................................... -84-
Fig.A.4 Equipotential lines approximation............................................................ -85-
Fig.A.5 Transformations. (a) Original rectangle (u-v). (b) Upper-half plane
after Elliptic Jacobian transformation (s-t). (c) Upper-half plane after
linear-fractional transformation (m-n). (d) Destination rectangle after
elliptic integral of the first kind (ξ-ζ)....................................................... -86-
Fig.A.6 Variation of aspect ratio with the parameter k.......................................... -87-
Fig. A.7 Error generation in the forward mapping. (a) upper-half plane in s-t
plane. (b) p-q plane after exponential transformation. (c) m-n plane
after linear-fractional transformation. (d) Destination rectangle in ξ-ζ
plane.......................................................................................................... -88-
xii
List of Tables
Table I Properties of silicon and 4H-SiC........................................................ -2-
Table II Comparison of different power semiconductor device packaging
technologies ....................................................................................... -11-
Table III Dimensions of the simulation model and material properties ........... -24-
Table IV Simulation setting in MEDICI .......................................................... -24-
Table V Parameters of the LVEP#2 structure study......................................... -27-
Table VI Parameters of the MVEP structure study............................................ -33-
Table VII Parameters of the HVEP structure study............................................ -41-
Table VIII Comparison of designed values and fabricated values of HVEP
package............................................................................................... -62-
- 1 -
Chapter 1: Introduction
1.1. Requirements for the power semiconductor packaging in power
electronics applications
Power electronics devices and modules are the key components in power systems for
maneuvering the storage, conversion, and conditioning of massive electromagnetic
energy. Marketing demands on higher power density, higher frequency, higher current,
higher voltage, and higher temperature are the driving forces on pushing the packaging
technologies to a higher level.
Electronic packages provide protection (mechanical, chemical and electromagnetic),
heat dissipation, and power and signal distribution for electrical components and their
interconnections [1]. Hierarchically, the packaging begins at the interface of the
semiconductor chip itself, which is considered the power semiconductor packaging. The
power semiconductor packaging deals with the attachment of one or more bare chips to a
substrate, the interconnection from these chips to package leads, and the encapsulation.
The power semiconductor interconnection plays an important role because it directly
interfaces with the chips that contain millions of transistor circuits electrically, thermally
and mechanically.
In the specific application of power semiconductor devices and modules, the power
semiconductor packaging must fulfill very different requirements compared to those for
microelectronic Integrated Circuit (IC) chips. First, the current and power handling
capability per bond or joint increases by several orders of magnitude. This requires a
larger cross-section area for these interconnections. Second, the voltage rating varies
with the power semiconductor and the applications and can be as high as several
kilovolts. This leads to high breakdown voltage demands on the insulation and the
encapsulation. Third, since these devices operate at high frequencies, in order to maintain
high efficiency, parasitic noise must be reduced. Therefore, the conduction path must be
minimized. Furthermore, the increased power density drives the power semiconductor
packaging to improve its role in heat dissipation [2]. Last, reliability of the power
- 2 -
semiconductor interconnections is vital in ensuring that the electronic assemblies have an
extended lifetime.
Silicon Carbide (SiC) has gained tremendous interest as a promising wide-bandgap
material for high power and high temperature applications. Table I shows the
comparison of the properties between Silicon (Si) and 4H-SiC [3]. The high critical
breakdown electric field lends the SiC material the development of High-Voltage (HV)
devices. Substantial progress has been made in the recent years in SiC power device
development, especially in high-voltage area. The SiC PiN diodes of 10 kV or even
higher blocking voltage have been reported [4]-[8], and the SiC MOSFETs of 10 kV are
also demonstrated and evaluated in recent years [9]-[12].
With the increasing of the voltage rating of the SiC semiconductor devices, the power
semiconductor packaging is required to have not only low interconnect resistance, less
noise, less parasitic oscillations, improved reliability, and better thermal management, but
also HV blocking capability, which increases the electric field stress in the insulation and
encapsulation in the package.
1.2. Review of existing packaging technologies of the power
semiconductor devices and modules
The existing power semiconductor packaging of the power semiconductor devices and
modules include wire-bonding, press pack, flip-chip technology, metal posts
TABLE I
PROPERTIES OF SILICON AND 4H-SIC
Property Si 4H-SiC
Bandgap, Eg (eV) 1.12 3.26
Electron mobility, µn (cm2/Vs) 1400 800
Hole mobility, µp (cm2/Vs) 450 140
Intrinsic carrier concentration, ni (cm-3
) at 300 K 1.5×1010
5×10-9
Electron saturated velocity, vnsat (x 107 cm/s) 1.0 2.0
Critical breakdown electric field, Ecrit (kV/mm) 25 220
Thermal conductivity, Θ (W/cm⋅K) 1.5 3.0-3.8
Dielectric constant 11.8 9.8
- 3 -
interconnected parallel plates structure (MIPPS), dimple array interconnection (DAI),
power overlay (POL) technology, and embedded power (EP) technology.
1.2.1. Wire bonding
Wire bonding is the most common chip-level interconnect technology in power
electronics nowadays. Fine wires are attached between the I/O pads on a chip and its
associated package pins. This technology originated from AT&T’s bean lead bonding in
the 1950s. It accounted for over 90% of all of chip-to-package interconnections formed
in 1999 [13].
In a power semiconductor package, the power semiconductor dice are solder-attached
to a copper heat spreader or substrate, and the source and gate terminals are connected
from the Aluminum (Al) chip bonding pads to the nickel-plated copper leads by
Aluminum wires. The top of the dice and the bonding wires are encapsulated with
molding compound for insulation, mechanical support and protection. Fig. 1.1 shows a
picture of wire bonded power semiconductor without the encapsulation [14].
The advantages of the wire bonding technology include the highly flexibility chip-to-
package interconnection process, the high yield interconnection processing, the easily
programmed bonding cycles, the very large industry infrastructure supporting the
technology, and the rapid advances in equipment, tools, and material technology. The
disadvantages include the slower interconnection rates due to the point-to-point
processing of each wire bond, the long chip-to-package interconnection length, the
degraded electrical performance, the larger footprint required for chip-to-package
Fig. 1.1. Power semiconductor package using wire bonding technology [14].
- 4 -
interconnection, the potential for wire sweep during encapsulation over molding, and the
limited heat dissipation capability.
1.2.2. Press Pack
Press pack is one of the technologies without using the bonding wires. It was originally
developed by Fuji, Toshiba and ABB [15]-[18], and applied to the high-power devices,
such as diodes, thyristors, GTOs, etc., in the area of HVDC, SVC, transportation system,
and motor drives. The press pack structure has evolved to a sandwiched structure and
involves the Molybdenum (Mo) strain buffers, gate connection and insulation, Copper
(Cu) pole-piece and a ceramic ring. Fig. 1.2 shows the pictures of the IGBT devices with
press pack interconnection [18].
The press pack technology has a proven reliability record for the applications that
require extended thermal and power cycling reliability. Since there is no soldering
process is involved, this method eliminates the residual stress on each material. It also
enables the redundant converter design and double-sided cooling. The concerns of press
pack technology includes its higher manufacturing cost, the more complicated mounting
arrangement to maintain the pressure, and the damage caused by the pressure force.
1.2.3. Flip-Chip Technology
(a) (b)
Fig. 1.2. Pictures of the IGBT devices with press pack interconnect [18] © 2004 IEEE. (a) Asymmetric
4500 V/1000 A Press Pack IGBT along with a submodule comprising 12 IGBT chips. (b) Direct
pressure contact on the gate.
- 5 -
Flip-chip technology is another available technology for the power semiconductor
packaging. It firstly emerged in IC as low-cost, high density and reliable
interconnections [19][20]. The key material in the structure is the solder balls. The Ball
Grid Array (BGA) is one of the applications of Flip-chip technology. Some examples are
shown in Fig. 1.3 [21]-[24]. The advantages of BGA include higher interconnect density,
better electrical performance than wire-bond due to short interconnection path, better
thermal performance by using thermal vias, and the reduced placement and handling
issues. One big drawback of BGA is the difficulty to inspect the formed solder joints
after assembling, which is very time-consuming.
The flip-chip on flex is the extension of the flip-chip technology and the flex circuitry
technology to the power electronics packaging and takes the advantages of these two
technologies [25]-[28]. As shown in Fig. 1.4, the module using flip-chip on flex
technology has a multilayer structure. The bottom layer is a direct-bond copper (DBC)
substrate which provides both the thermal and electrical path. The top layer of the
structure is a double-sided flexible copper-clad laminate which is an adhesiveless
composite of polyimide file bonded to copper foil. A novel triple-stacked solder
(a) (b)
(c) (d)
Fig. 1.3. Examples of BGA package. (a) MOSFET BGA [21]. (b) BGA for lateral
MOSFET [22] © 2004 IEEE. (c) FlipFET [23]. (d) BGA for Volterra integration module
[24].
- 6 -
bumping technique is used to connect the power devices to the flexible substrate with a
circuit pattern for gate drive components. The module is encapsulated in underfill
polymer materials to reduce the thermo-mechanical stresses imposed on the solder joints.
Finally, the gate drive and control circuits are built on the top of the flex circuitry using
either flip-chip or surface-mount technology. This power semiconductor packaging
reduces the parasitcis, minimizes the thermal rise, and realizes a three-dimensional (3D)
structure. However, this technology relies on the solder balls for the interconnection,
thus it has the same problem as the BGA technology.
1.2.4. Metal-Posts Interconnected Parallel Plates (MPIPPs) Technology
The metal-posts interconnected parallel plates (MPIPPs) technology is another power
semiconductor packaging [29][30]. It uses the directly bonded metal posts to replace the
bonding wires, as shown in Fig. 1.5. An Aluminum Nitride (AlN) DBC is employed as
the base substrate. The power devices are attached to the substrate directly by soldering.
The copper (Cu) posts of different sizes and heights are soldered on the top of the devices
and the copper metallization on the AlN substrate. The top DBC plate is etched to a
desired pattern and attached onto the copper posts.
This technology has lower DC resistance than the conventional wire bonding, and it is
also conclude that the parasitic noise is much lower than the wire bonding technology.
However, this technology requires the solderable devices through the metallization of the
Aluminum (Al) contact.
Integrated Communications, Gate Drives and Protection
DeviceDevice
Double-sided FlexDevice Device
Solder bump Encapsulant
DeviceDevice-
Double-sided FlexDevice Device
Heat Spreader
Adhesive
DeviceDevice
Double-sided FlexDevice Device
Underfill
DeviceDevice-
Device Device
Heat Spreader
DBC
Integrated Communications, Gate Drives and Protection
DeviceDevice
Double-sided FlexDevice Device
Solder bump Encapsulant
DeviceDevice-
Double-
Device Device
Heat Spreader
Adhesive
DeviceDevice
Double-sided FlexDevice Device
Underfill
DeviceDevice-
Device Device
Heat Spreader
DBC
Integrated Communications, Gate Drives and Protection
DeviceDevice
Double-sided FlexDevice Device
Solder bump Encapsulant
DeviceDevice-
Double-sided FlexDevice Device
Heat Spreader
Adhesive
DeviceDevice
Double-sided FlexDevice Device
Underfill
DeviceDevice-
Device Device
Heat Spreader
DBC
Integrated Communications, Gate Drives and Protection
DeviceDevice
Double-sided FlexDevice Device
Solder bump Encapsulant
DeviceDevice-
Double-
Device Device
Heat Spreader
Adhesive
DeviceDevice
Double-sided FlexDevice Device
Underfill
DeviceDevice-
Device Device
Heat Spreader
DBC
(a) (b)
Fig. 1.4. Flip-chip on flex technology [25] © 2001 IEEE. (a) Picture of a module. (b) Cross-section
schematic.
- 7 -
1.2.5. Dimple Array Interconnect (DAI) Technology
The dimple array interconnect (DAI) is a 3D power semiconductor packaging
technique. The electrical interconnection is established by the formation of solder bumps
between device electrodes and the preformed array of dimples on a metal sheet, as shown
in Fig. 1.6 [31]. The result is a low-profile planar interconnection that is suitable for
multilayer integration with other components. With the exception of fabrication the
dimpled copper sheet, the DAI process follows the typical solder joint fabrication
process.
This technique reduces the interconnect parasitics and increases thermal capability of
the power semiconductor packaging. However, in a joint formed by controlled collapse
bonding (CCB), the mismatch of the coefficient of thermal expansion (CTE) between the
Al Pad
Passivation
UBM
Solder
Bump
Solder Dimple on Copper Flex
Underfill
Silicon ChipAl Pad
Passivation
UBM
Solder
Bump
Solder Dimple on Copper Flex
Underfill
Silicon Chip
(a) (b)
Fig. 1.6. Dimple array interconnect technology [31]. (a) Picture of DAI. (b) Schematic cross-section.
Metal PostsEncapsulant
AlN
CuHybrid Driver Circuit
Heat Sink (MMC, Al, etc)
Power device
Device DeviceDevice Device
or AluminaHeat Sink (MMC, Al, etc)
Device DeviceDevice Device
Metal PostsEncapsulant
AlN
CuHybrid Driver Circuit
Heat Sink (MMC, Al, etc)
Power device
Device DeviceDevice Device
or AluminaHeat Sink (MMC, Al, etc)
Device DeviceDevice Device
(a) (b)
Fig. 1.5. Metal-posts interconnected parallel plates technology [29] © 1999 IEEE. (a) Picture of
MPIPPs module. (b) Cross-section schematic.
- 8 -
dimpled copper sheet and the silicon causes thermal-mechanical problems which are the
stress concentrated along the high-angle joint edge on the silicon.
1.2.6. Power Overlay (POL) Technology
The Power Overlay (POL) Technology developed by GE is another approach which
eliminates the wire bonding through the utilization of the metalized Cu vias/polyimide to
achieve power and control interconnection [32][33]. As shown in Fig. 1.7, power
semiconductor devices are soldered to a Direct Bonded Copper (DBC) substrate from the
backside. The difference in device thickness is compensated by copper shims. A thin
layer of polyimide sheet is laminated over the die after vias are laser machined or
mechanically punched through the film. The whole top surface is metalized
(electroplated) with copper. Circuit patterns are achieved by the application of photo-
resist and chemical etching processes.
The POL technology eliminates the wire bonds with metallurgical interconnections. It
has better electrical performance through reduced interconnections parasitics. The
thermal performance is improved through the reduced number of thermal interfaces and
double-sided heat removal. The reduced profile and more flexible packaging provide the
options for stacking additional circuits. However, this POL structure has not been
successfully tested for high-power applications.
1.2.7. Embedded Power Technology
Fig. 1.7. Cross-sectional schematic of a GE-POL structure.
- 9 -
Another multi-layer integration technology which can be applied to the power
semiconductor packaging is the Embedded Power Technology. A multi-layer structure
that uses ceramic substrates, screen-printable dielectric materials and solder is employed
to enclose the power semiconductor devices, as shown in Fig. 1.8 (b). The power
semiconductor devices are mounted in the openings on the ceramic frame with adhesive
polymer that surrounds the device edges. The interlayer dielectric is a kind polymer that
is screen printed with designed pattern, for example, the vias corresponding to the pads of
the devices. The deposited Cu pattern is on the top as the interconnect layer. In order to
handle high current, a copper electroplating process is employed to thicken the deposited
Cu seed layer. This multi-layer planar stacking construction not only provides the power
semiconductor packaging through metallurgical contact to aluminum pads on the power
semiconductor devices, but also carries the wiring for the top electric circuitry. Fig. 1.8
(a) is the picture of an Integrated Power Electronics Module (IPEM) that uses the
Embedded Power Technology.
(a)
Heat Spreader
C-ChipS-ChipS-Chip
Componen
t
I/O PinsComponentComponent
Base Substrate
Ceramic Solder
Via
Heat Spreader
C-ChipS-ChipS-Chip
Componen
t
Componen
t
I/O PinsComponentComponentComponent
Base Substrate
Ceramic SolderSolder
Via
(b)
Fig. 1.8. Embedded power technology. (a) Picture of an IPEM using embedded power technology. (b)
Cross-sectional schematic of an IPEM using embedded power technology.
- 10 -
This technology features the building-up of interconnections on planar power
semiconductor devices which are embedded in a ceramic frame. The integrated power
stage similar to the Integrated Circuits (ICs) is fabricated through a thin- and thick-file
approach, which is an extension of wafer level process technology. The compact
interconnection also eliminates the bond wires and one-side contact interface. Both
electrical and thermal performance is improved by the low profile structure. Beside all
the merits of the POL technology, the passive components can be easily embedded in the
ceramic frame as the power semiconductor devices in the Embedded Power Technology,
which is not realized in the POL technology.
1.2.8. Summary of existing power semiconductor packaging of the power semiconductor devices and modules
Wire bonding is the most mature and low-cost technology. However, its large
footprint, high parasitics, and other disadvantages make it less attractive. The press pack
is widely used as well, but the concerns of this technology come from the pressure and
the structure itself. The flip-chip on flex technology, the metal-posts interconnected
parallel plates technology, and the dimple array interconnect technology are 3D
integration technology. They are good for power semiconductor packaging as well as the
multi-chip system packaging. The elimination of the wire bonds makes the electrical
parasitics much less than the wire bonding package. However, the thermo-mechanical
stress and the process make them less favorable. The power overlay technology and the
embedded power technology utilizes the metallurgical contact to form a multi-layer
stackable structure. The low profile, low parasitics, high thermal performance, and the
flexibility in packaging are the common advantages of these two technologies.
Compared to the POL technology, the embedded power technology uses the thin- and
thick-film approach in the IC fabrication process, and enables both power semiconductor
device and passive components to be integrated in one package. With the consideration
of the integration of more components in one package, the embedded power technology
is a good candidate for both power semiconductor packaging and system packaging.
The voltage rating, current rating, and availability of these technologies are compared
in Table II.
- 11 -
1.3. Research objective and approach
The high-voltage packaging involves both structure optimization and material
improvement in the package. The structure optimization is to minimize the electric field
intensity hotspots and make the electric field distribution in the structure as even as
possible. The material study focuses on the high dielectric strength material selection and
application. In addition, the proper selection of the power semiconductor packaging for
the high-voltage system packaging is also required to be considered.
The embedded power technology is an attractive candidate for both power
semiconductor and system packaging, thus it is selected as a candidate for the high-
voltage packaging. This research proposes the high-voltage packaging that is suitable for
both power semiconductor packaging and the system packaging derives from the
embedded power technology, so called high-voltage embedded power (HVEP)
technology.
In designing for the high-voltage package, the objective is to build even electric field
distribution and keep the profile as low as possible. With the in-depth study on the
electric field distribution in the package using the embedded power technology, hotspots
are identified and analyzed. Furthermore, after a parametric study on the dimensions and
the material properties in the structure, a design guideline is suggested. This guideline
can facilitate the future design of the package, thus reduce the design cycle time.
TABLE II
COMPARISON OF DIFFERENT POWER SEMICONDUCTOR DEVICE PACKAGING TECHNOLOGIES
Voltage Rating Current Rating Commercialization
Wire Bonding > 6 kV > 4 kA Yes
Press Pack 6.5 kV > 1 kA Yes
Flip Chip on Flex 42 V 6.5 A No
Metal Post interconnected parallel
plate
400 V 10 A No
Dimple array interconnect 600 V 50 A No
Power Overlay 2.4 kV 25 A No
Embedded Power 400 V 10 A No
- 12 -
1.4. Dissertation outline
This dissertation consists of seven chapters including necessary background, motivation
and the objectives of the research effort in Chapter 1. Chapter 1 also gives the review of
existing power semiconductor packaging technologies for the power semiconductor
devices and modules. Chapter 2 surveys the existing methods to relieve the high electric
field intensity in high-voltage power semiconductor devices and the wire bonding
packages.
In Chapter 3, the low-voltage package using embedded power technology (LVEP) is
studied in details. The materials are evaluated and the electric field distribution is
simulated in MAWELL. Based on the simulation results, simple solutions are discussed
as well. Chapter 4 presents the simulation in MEDICI of LVEP simulated in
MAXWELL in Chapter 3, and compares the results in two different simulators firstly.
Then the medium-voltage embedded power (MVEP) technology and the high-voltage
embedded power (HVEP) technology are proposed. Next, the dimensions and material
properties in the HVEP structure are studied. In Chapter 5, the design guideline of HVEP
structure is suggested and the fabrication process of the HVEP module is proposed. The
HVEP structure and the design guidelines are verified by a fabricated 2 kV SiC diode
package in house. In Chapter 6, a simplified model of the power semiconductor device is
proposed. The feasibility of the application of the conformal mapping to the simplified
structure to calculate the electric field distribution mathematically is studied. Simulations
in MAWELL with the simplified model is done and compared to the simulation results in
MEDICI for the HVEP structure. Finally, Chapter 7 contains a summary of the research
and the contribution of this work towards new packaging technology for high-voltage
devices and modules.
- 13 -
Chapter 2: Survey of Strategies for Reduction of Electric
Field in Semiconductor Die and Packages
2.1. Introduction
Inside the power semiconductor, the PN junction bares the high voltage applied to the
device. Many measures have taken to reduce the electric field crowding at the PN
junction. These measures are reviewed firstly in this Chapter.
There are ways that have been reported to increase the voltage capability of a package
using wire bonding technology. The high electric field concentration is studied through
simulation and several solutions are developed to limit the peak electric field intensity in
the literature. These methods are discussed secondly.
2.2. Overview of classical techniques for high electric field intensity
reduction in power semiconductors [34][35]
From the theoretical analysis, there is an upper bound for the breakdown voltage of
parallel-plane (semi-infinite) junction in power devices. However, in practice this value
can be reduced by the presence of high electric field concentration at the edges of the
device. As shown in Fig. 2.1, which is a planar diffused junction, there is electric field
crowding at the edges.
An elegant approach to reduce the electric field at these edges is by the placement of
floating field rings, which is shown in Fig. 2.2. The objective of this method is to extend
the depletion boundary along the surface and reduce the electric field at the main
Fig. 2.1. Electric field crowding for a planar diffused junction
- 14 -
junction. A single field ring reduces depletion layer curvature, reduces the electric field
crowding, and enhances the breakdown voltage. An even greater improvement in
breakdown voltage can be obtained by using multiple float field rings working in
conjunction with each other. The depletion layer extends gradually away from the main
junction, as shown in Fig. 2.3.
Another method for reducing the depletion layer curvature involves changing the
surface potential. One way for achieving this is to place a metal field plate at the edge of
the planar junction, as illustrated in Fig. 2.4. Because the shape of the depletion region
depends on the surface potential, the depletion layer spreading at the surface is sensitive
to the potential applied to the filed plate. At the edge of the field plate, small oxide
thickness creates high peak electric field, but in the vicinity of the junction, thin oxide is
Fig. 2.3. Cross-section of edge termination with multiple field rings.
Fig. 2.2. Cross-section of edge termination with a single floating field ring.
- 15 -
helpful to decrease the peak electric field created by the surface charges and the radius
effect. Thus a multiple step oxide method is proposed as shown in Fig. 2.5.
Semi-insulating POlycrystalline Silicon (SIPOS) are thin SiOx films (O<x<2),
deposited by means of low pressure chemical vapor deposition on suitable substrates
(silicon or sapphire). Its conductivity varies in a wide range and depends on the oxygen
content as illustrated in Fig. 2.6 [36]. Due to the potential difference between its two
terminals, which are the metal on the top of N+ doping and P+ doping, a resistive field
plate with this material, as illustrated in Fig. 2.7, induces the leakage current to flow from
one terminate to the other. If the resistance is uniform in the SIPOS, there is a uniform
electric field along the surface direction in the SIPOS film. This uniform electric field
will modulate the surface electric field and extend the depletion boundary.
A common practice is to combine the field plate with the floating field ring, as shown
in Fig. 2.8. Each floating field ring is connected to a field plate. The filed plate extends
beyond the edge of floating field ring. The field plate reduces the electric field crowding
at each of the floating field rings.
Fig. 2.5. Cross section of field plate with step oxide.
Fig. 2.4. Planar junction with field plate.
- 16 -
2.3. Overview of high electric field reduction techniques in wire
bonding packages [37][38]
In [37] and [38], the study shows the high electric field concentration location in a
wire-bonding IGBT module using finite element simulations in association with partial
Fig. 2.8. Cross Edge termination design combining floating field rings with field plate.
Fig. 2.7. Cross section of resistive field plate.
Fig. 2.6. Cross section of field plate with step oxide.
- 17 -
discharge measurements and proposes some solutions to lower the high electric field
concentration.
The high voltage IGBT module is shown in Fig. 2.9. One side of two IGBT or diode
chips is connected to the copper layer; the other side is connected by the bonding wires to
the copper traces on the top of the substrate. The whole module is encapsulated by a
silicon gel. In the simulation, the IGBTs were modeled with the dielectric constant and
resistivity of silicon.
The simulation results show that the high electric field concentrates at four locations:
(1) the metallization edge, (2) the silicon die, (3) from trace to trace and (4) from trace to
bonding wire. This is illustrated in Fig. 2.10. The original value of the maximum electric
field intensity at the metallization edge is about 40kV/mm when the applied voltage is 2
kV. There are several solutions proposed to relieve the high electric field concentration at
the metallization edge.
0.500E+07
0.400E+07
0.300E+07
0.100E+07
0.200E+07
0.500E+07
0.400E+07
0.300E+07
0.100E+07
0.200E+07
Fig. 2.10. Cross Electric field concentration due to proximity between track and bonding. © 2003 IEEE
Fig. 2.9. Cross-sectional diagram of an IGBT module package using wire bonding. © 2003 IEEE
- 18 -
One of these solutions is the deposition of a resistive layer on the top of the substrate.
This is shown in Fig. 2.11. The function of the deposited resistive layer is explained in
the papers with the following reasoning. The deposited resistive layer has lower
resistivity in comparison to the insulation material and the substrate. Thus leakage
current flows through this layer, and the maximum electric field intensity is limited. The
resistivity depends on the structure and other material properties. An optimal value of
about 107 Ωm is given in the papers for their structures, and the maximum electric field
intensity is reduced to 9 kV/mm. Furthermore, the resulting leakage current is low
because the resistive layer doesn’t connect the metallization on both sides of the
substrate. This provides more freedom to choose the value of resistivity.
The second solution is to modify the shape of the metallization edge as shown in Fig.
2.12. The high electric field at the metallization edge can be reduced by modifying its
curvature. Furthermore, the metallization can be embedded in the substrate. This
arrangement has much better electrical characteristics and the maximum electric field
intensity is reduced to 9.2 kV/mm.
The third solution is an alternative solution to the previous one. Instead of embedding
the metallization in the substrate, it can be reinforced by a material with high field
breakdown value as shown in Fig. 2.13. This solution doesn’t reduce the maximum
electric field strength.
Fig. 2.11. Influence of a resistive layer. © 2003 IEEE
- 19 -
2.4. Summary
In this Chapter, the classical high electric field reduction techniques in power
semiconductor are reviewed firstly. These techniques include the floating field ring, three
methods of field plate, and the combination of these two techniques. Then the measures
in a wire bonding package to reduce the high field concentration is introduced. Several
high field locations were found, and three solutions were presented in the paper. The
results showed that the maximum electric field can be reduced to less than 1/4 of the
original value.
From the measures utilized in the power semiconductor and the methods proposed for
the IGBT module package, it can be seen great effort has been taken to reduce the
curvature of the equipotential lines in the package besides using better material to replace
Fig. 2.13. Module reinforcement at the copper edges. © 2003 IEEE
Fig. 2.12. Influence of modification of metal edge. © 2003 IEEE
- 20 -
the existing material. Thus, it is important to know the equal-potential line distribution in
the structure as well as the electric field distribution.
In the simulation of the IGBT module package, the devices are modeled as a lump
material; the internal information of the IGBT device structure is not included. The study
on the low-voltage embedded power (LVEP) structure in the next chapter is simulated by
following the same method as in this chapter. The simulation provides a quick and
simple concept of the internal of the package, which is useful for the structure
improvement in the future.
- 21 -
Chapter 3: Development of High-Voltage Embedded Power
(HVEP) Technology via MEDICI
In the previous Chapter, an IGBT wire bonding module is evaluated in [37] and [38].
The IGBTs are modeled as a piece of lump Si material, which doesn’t reflect the real
situation inside the semiconductor devices. From the electromagnetic field theory, the
external electric field in the packaging material and internal electric field in the device are
not independent to each other. In the device packaging design, it is important to know
how the electric field inside the device influences the external electric field in the
packaging materials. Software which models the semiconductor device structure is
necessary to simulate the internal electric field of the device.
MEDICI is a powerful device simulation program that can be used to simulate the
behavior of semiconductor devices. It models the two-dimensional (2D) distributions of
potential in a device and can be used to predict electrical characteristics for arbitrary bias
conditions [39]. Thus, extensive 2D simulations using MEDICI are performed to study
the high-voltage planar package of a 10 kV SiC diode [43] based on the LVEP
technology.
In this Chapter, the Low-Voltage Embedded Power (LVEP) module and the SiC diode
used in MEDICI for the device package design are introduced first. Then, the LVEP
structure and an improved structure, which is called LVEP #2, are evaluated in MEDICI.
Next, the geometry parameters and material properties that could influence the electric
field are studied. Based on the simulation results, a Medium-Voltage Embedded Power
(MVEP) package is proposed, and the influence of the geometry parameters and material
properties to this package is simulated as well. A High-Voltage Embedded Power
(HVEP) is proposed and simulated. The influence of geometry parameters of this
structure is studied intensively.
3.1. Introduction
The existing embedded power technology is a planar integration technology which is
originally developed for the packaging of the Integrated Power Electronics Modules
(IPEMs). The IPEM integrates the switching devices, conductors, controls and sensors,
- 22 -
etc. into a standardized manufacturable compact subassembly with improved electrical,
thermal, and mechanical performance. In this approach, the power chips are embedded in
a special encapsulating structure, as well as that all interconnections are deposited planar
metallization as shown in Fig. 3.1. The key elements in this structure is the embedded
power stage which is comprised of ceramic frame, power chips, insulation material, and
the metallization circuit that supports the high density interconnections of all IPEM
components [40]-[42].
Since it is difficult to measure the electric field inside the IPEM shown in Fig. 3.1,
finite element analysis is employed to explore the fields inside the modules. The
simulation only focuses on the region in dashed-line rectangle shown above, which
represents the package of the semiconductor device. Fig. 3.2 is the enlarged picture of
this region.
A 10 kV SiC PiN diode is designed in [43], and its cross-sectional schematic is shown
in Fig. 3.3. The device has a 110 µm thick lightly doped N-type (N-) region with the
doping concentration of 6×1014
cm-3
. There is a novel multi-zone, single-implanted JTE
in the device to achieve the high blocking capability.
Ceramic carrierEmbedded insulation
Embedded chip Metallization Top insulation
Thick metallization
Thin metallization
Fig. 3.2. Enlarged cross-sectional schematic of the package of a semiconductor device using LVEP.
Heat Spreader
C-ChipS-ChipS-Chip
Component
I/O PinsComponentComponent
Base Substrate
Ceramic Solder
Via
Heat Spreader
C-ChipS-ChipS-Chip
Component
Component
I/O PinsComponentComponentComponent
Base Substrate
Ceramic SolderSolder
Via
Fig. 3.1. Cross sectional view of an IPEM.
- 23 -
When the power semiconductor is in the state of blocking voltage, the voltage is
dropped on the depletion region formed at the PN junction. The potential outside the
depletion region is identical in the P-type region and N-type region respectively, as
shown in Fig. 3.4. Varies Junction Termination Technologies (JTEs) are employed to
extend the boundary of depletion region laterally and reshape the distribution of the
electric field. Laterally, the voltage is dropped on the JTE region. The potential at the
top of the device is equal to the potential on the bottom of the device. This situation is
the same to all the semiconductors that use JTE to improve their blocking capability.
Thus, MEDICI is employed to simulate both semiconductor and its package to model this
lateral voltage drop.
3.2. Evaluation of two LVEP structures
The LVEP structure and its improved structure are revisited in MEDICI with the
mentioned 10 kV SiC PiN diode. The simulation model of half of the LVEP structure is
Fig. 3.4. Electric field and potential distribution for an abrupt parallel-plane P+/N junction [34].
Fig. 3.3. Cross-sectional schematic of a 10 kV PiN SiC diode [42].
- 24 -
shown in Fig. 3.5 (a). The dimension of the simulation model and the properties of the
materials are shown in Table III. The simulation settings are shown in Table IV.
Fig. 3.5 (b) shows the equipotential lines in the whole package. The total number of the
equipotential lines is 50, and the voltage difference between two lines is around 275 V.
Since the voltage is applied to the metallization on the top of the polyimide and the
cathode of the diode, and the thickness of the polyimide is much less than the thickness
of the ceramic, most of the equipotential lines bend down at the upper corner of the
device, and some of them also bend at the lower corner of the device. Thus two electric
field “hotspot” are located at the upper corner and the lower corner of the devices inside
the embedded insulation material and the polyimide respectively, as shown in the two
circles in Fig. 3.5 (b).
Fig. 3.6 (a) is the maximum electric field inside polyimide, which is measured along the
x direction from the left of the polyimide layer to the right, as the dash-dot line AB
shown in Fig. 3.5 (a). Fig. 3.6 (b) is the maximum electric field intensity inside the
embedded insulation material, which is measured along the y direction from the top of
this material to the bottom, as the dashed line CD shown in Fig. 3.5 (a). The highest
electric field intensity points in both polyimide and the embedded insulation material are
located at the upper corner of the device; in the embedded insulation material, the second
highest electric field intensity is at the lower corner of the device. The remaining electric
TABLE IV
SIMULATION SETTINGS IN MEDICI
Anode Cathode
Applied voltage (kV) 0 13.5
Meshing method Manual meshing
Simulation stop criteria Leakage current > 1 µA
Simulation method Variable step simulation
TABLE III
DIMENSIONS OF THE SIMULATION MODEL AND MATERIAL PROPERTIES
SiC diode Polyimide Embedded insulation material Ceramic
Thickness (µm) 115.5 100 508 508
Width (µm) 500 1100 500 100
Dielectric constant 9.72 3 4 3.8
- 25 -
field peaks in the polyimide are nearby the locations of the main junction and JTEs. The
maximum electric field intensity in the polyimide is 5 times of its dielectric strength at
13.5 kV, and the maximum electric field intensity in the embedded insulation material is
as high as 24 times of its dielectric strength.
From the simulation result shown in Fig. 3.5, it is the bending of the equipotential lines at
upper and lower corner which makes high electric field intensity at these two locations.
An improved LVEP is proposed as shown in Fig. 3.7 (a), which is called LVEP #2 here.
(a)
(b)
Fig. 3.5. Simulation model and result of LVEP structure in MEDICI. (a) Dimensions and materials of
the structure. (b) Distribution of equipotential lines.
Cathod
Anode
y
x
SiC
Ceramic Embedded
Insulation
Material
100 µm
115.5 µm
508 µm
10 µm 500 µm
1000 µm
1100 µm
Polyimide
0
200
400
0.00 0.20 0.40 0.60 0.80 1.00
Distance
(µm)
Distance (mm)
Upper corner
Lower corner
A B
D
C
- 26 -
The bottom of the package is covered by metallization compared to LVEP to prevent the
bending of the equipotential lines at the lower corner. This structure is simulated in
MEDICI verify the idea. As shown in Fig. 3.7 (b), although the equipotential lines bend
down at the upper corner of the device with the same mechanism as it in the LVEP
structure, they bend to the right at the bottom of the LVEP #2 structure due to the
presence of the metallization on the bottom of the package. This reduces the high electric
field at the lower corner of the device significantly, as shown in Fig. 3.8. A comparison
of the simulation results between two LVEP structures is shown in Fig. 3.9.
3.3. Parametric study of LVEP #2
Although the electric field intensity at the lower corner of the device is reduced greatly,
its value and the field intensity at the other hotspot are still much higher than the
dielectric strength of two insulation materials. Both LVEP structures cannot survive at a
voltage of 13.5 kV. The LVEP structure involves several materials and has many
dimension variables. A parametric study is done to show the influence of the material
properties and the structure dimensions to LVEP #2 structure before any improvement is
made on it. Table V gives the parameters shown in Fig. 3.10 and the values.
E = 250 kV/mm
JTE1
JTE2
JTE3
Distance (mm)
0.00 0.20 0.40 0.60 0.80 1.00
E (
kV
/mm
)
050
100
150
200
25
0
E = 250 kV/mm
JTE1
JTE2
JTE3
Distance (mm)
0.00 0.20 0.40 0.60 0.80 1.00
E (
kV
/mm
)
050
100
150
200
25
0 E = 600 kV/mm
E = 250 kV/mm
Distance (um)
0.00 100 200 300 400 500
E (
kV
/mm
)
0.0
0200
400
600 E = 600 kV/mm
E = 250 kV/mm
Distance (um)
0.00 100 200 300 400 500
E (
kV
/mm
)
0.0
0200
400
600
(a) (b)
Fig. 3.6. Electric field intensity in the insulation materials of LVEP structure when the applied voltage
on the cathode is 13.5 kV. (a) Along the x direction (lateral direction) line AB in the polyimide. (b)
Along the y direction (vertical direction) line CD in the embedded insulation material.
A B C D
- 27 -
TABLE V
PARAMETERS OF THE LVEP#2 STRUCTURE STUDY
Nominal
value
Swept values Experimental
value
Polyimide thickness D (µm) 100 100-500 100
Dielectric constant of Polyimide εr 3 1-4 3
Dielectric constant of embedded insulation material εr’ 4 1-4 4
(a)
(b)
Fig. 3.7. Simulation model and result of LVEP #2 structure in MEDICI. (a) Dimensions and materials
of the structure. (b) Distribution of equipotential lines.
Anode
Cathode
10 µm
y
x
500 µm
1000 µm
1100 µm
Polyimide
SiC
Embedded
insulation
material Ceramic
100 µm
115 µm
508 µm
0
200
400
0.00 0.20 0.40 0.60 0.80 1.00
Distance
(µm)
Distance (mm)
Upper corner
Lower corner
A B
C
D
- 28 -
The first parameter studied is the thickness of the polyimide layer, D, as labeled in Fig.
3.10. Since the electric field is related to the thickness of the structure, increasing the
thickness will help to reduce the electric field intensity. The polyimide layer is thinner
than the ceramic, which leads most of the equipotential lines to bend down at the upper
corner of the device. The highest electric field in the polyimide is compared with
different thickness, as shown in Fig. 3.11. With the increasing of the thickness, more
equipotential lines distribute in the polyimide, which leads the peak electric field
intensity to reduce. This can be seen from the comparison shown in Fig. 3.12. When the
thickness of the polyimide is comparable to the thickness of the ceramic, further
0
100
200
300
400
500
600
SiC Polyimide (upper
corner)
Embedded
insulation material
(upper corner)
Embedded
insulation material
(lower corner)
Ceramic
E (k
V/m
m)
Dielectric strength LVEP LVEP #2
10x
4x
5x
24x
0
100
200
300
400
500
600
SiC Polyimide (upper
corner)
Embedded
insulation material
(upper corner)
Embedded
insulation material
(lower corner)
Ceramic
E (k
V/m
m)
Dielectric strength LVEP LVEP #2
10x
4x
5x
24x
Fig. 3.9. Comparison of the maximum electric field intensity between two LVEP structures.
E (
kV
/mm
)
050
10
015
020
02
50
0.20 0.40 0.60 0.80 1.00
Distance (mm)
E = 250 kV/mm
JTE3
JTE2
JTE1E (
kV
/mm
)
050
10
015
020
02
50
0.20 0.40 0.60 0.80 1.00
Distance (mm)
E = 250 kV/mm
JTE3
JTE2
JTE1 E (
kV
/mm
)
050
20
040
060
00
100 200 300 400 500
Distance (um)
E = 600 kV/mm
E = 100 kV/mm
(a) (b)
Fig. 3.8. Electric field intensity in the insulation materials of LVEP #2 structure when the applied
voltage on the cathode is 13.5 kV. (a) Along the x direction (lateral direction) line AB in the polyimide.
(b) Along the y direction (vertical direction) line CD in the embedded insulation material.
A B C D
- 29 -
increasing cannot help to reduce the field intensity very much. Thus, increasing the
thickness of the polyimide doesn’t convert the LVEP package to a HVEP package.
The second parameter studied is the dielectric constant of the polyimide. The influence
of this parameter is shown in Fig. 3.13. From the schematic of the structure, it can be that
the polyimide layer is in series with the embedded insulation material. Thus, when the
dielectric constant of the polyimide increases, more voltage drops on the embedded
insulation material, which means more equipotential lines bend down and cross the area
of the material and leads to higher electric field intensity at the upper corner of the
0
50
100
150
200
250
300
350
0 200 400 600D (um)
E (
kV
/mm
)
0
100
200
300
400
500
600
0 200 400 600D (um)
E (
kV
/mm
)
Upper corner Lower corner
(a) (b)
Fig. 3.11. Variation of maximum electric field intensity with the thickness of polyimide. (a) Inside the
polyimide at the upper corner. (b) Inside the embedded insulation material at the upper and lower
corner.
508 µm
L
D
10 µm
1000 µm
1100 µm
115.5 µm
x
y
Polyimide
Embedded insulation material
SiC
Ceramic
εεεεr
Anode
Cathode
D
εεεεr’
Fig. 3.10. Parameters in the LVEP#2 structure study.
- 30 -
device. This can be observed from Fig. 3.14. As the previous study showed, the
maximum electric field intensity inside the polyimide is at the upper corner of the device.
Thus, the maximum electric field intensity in the polyimide increases with the dielectric
constant of the polyimide.
When the dielectric constant of the embedded insulation material increases, the
variation of the maximum electric field intensity is the opposite because of the same
mechanism. This can be seen from the simulation results in Fig. 3.15 and Fig. 3.16.
From these two simulations of material property influence, it is observed that the
dielectric constant of these two insulation material should match with their thickness. In
Fig. 3.14 (a), the dielectric constant of the polyimide and the embedded insulation
material is 3 and 4 respectively, there are around 14 equipotential lines distributed in the
polyimide area which means the voltage drop is around 3.8 kV. In Fig. 3.14 (b), the
dielectric constant of the polyimide and the embedded insulation material is 1 and 4
respectively, there are around 26 equipotential lines distributed in the polyimide area
which means the voltage drop is around 7 kV. In this way, the thin polyimide layer with
low dielectric constant could have the same voltage distribution as the thick embedded
insulation material with high dielectric constant. Thus the voltage can distributes in the
whole insulation area evenly, which leads to a lower electric field intensity. However,
this material selection is so critically dependent on the development of the material.
The parametric study shows that thickening the polyimide layer cannot help to convert
the LVEP structure to a HVEP structure. Change of material according to the physical
Distance (mm)
0.00 0.20 0.40 0.60 0.80 1.00
Dis
tance (
µm
)
0
200
400
Distance (mm)
0.00 0.20 0.40 0.60 0.80 1.00
Dis
tance (
µm
)
0
200
400
Distance (mm)
0.00 0.20 0.40 0.60 0.80 1.00
Dis
tance (
µm
)
0
200
400
Distance (mm)
0.00 0.20 0.40 0.60 0.80 1.00
Dis
tan
ce (
µm
)
0
40
0
20
0
-20
0
-40
0
Distance (mm)
0.00 0.20 0.40 0.60 0.80 1.00
Dis
tan
ce (
µm
)
0
40
0
20
0
-20
0
-40
0
Distance (mm)
0.00 0.20 0.40 0.60 0.80 1.00
Dis
tan
ce (
µm
)
0
40
0
20
0
-20
0
-40
0
(a) (b)
Fig. 3.12. Comparison of the equipotential lines distribution between two different thicknesses of
polyimide. (a) D = 100µm. (b) D = 500 µm.
- 31 -
dimension helps to reduce the high electric filed intensity at corners of the device.
However, this is determined by the availability of different materials. Thus, new
structures have to be developed to block this high voltage.
3.4. Medium-Voltage Embedded Power (MVEP) structure
From the observation of the equipotential line distribution in two LVEP structures, it
can be seen that the presence of the metallization on the bottom of LVEP #2 structure
changes the bending direction of the equipotential lines and thus avoids the high electric
field intensity at the lower corner. The bending down of the potential lines makes the
Distance (mm)
0.00 0.20 0.40 0.60 0.80 1.00
Dis
tan
ce
(µ
m)
0
20
0
4
00
Distance (mm)
0.00 0.20 0.40 0.60 0.80 1.00
Dis
tan
ce
(µ
m)
0
20
0
4
00
Distance (mm)
0.00 0.20 0.40 0.60 0.80 1.00
Dis
tan
ce (
µm
)
0
20
0
400
Distance (mm)
0.00 0.20 0.40 0.60 0.80 1.00
Dis
tan
ce (
µm
)
0
20
0
400
(a) (b)
Fig. 3.14. Comparison of the equipotential lines distribution between two different dielectric constant of
polyimide. (a) εr = 3. (b) εr = 1.
0
50
100
150
200
250
300
350
0 2 4 6
Dielectric constant of polyimide
E (kV
/mm
)
0
100
200
300
400
500
600
700
800
900
0 2 4 6Dielectric constant of polyimide
E (
kV
/mm
)
Upper corner) Lower corner
(a) (b)
Fig. 3.13. Variation of maximum electric field intensity with the dielectric constant of polyimide. (a)
Inside the polyimide. (b) Inside the embedded insulation material.
- 32 -
high electric field intensity at the upper corner of the device, thus a structure that has less
potential lines bending down could help to relieve the high electric field intensity.
A Medium-Voltage Embedded Power (MVEP) structure is proposed and shown in Fig.
3.17 (a). The thickness of the polyimide on the top is increased to 400 µm based on the
parametric study of LVEP #2. In this structure, the bottom of the package is metalized
and connected to the cathode of the device. This metallization is also extended to the top
of the package through a via in the ceramic carrier. Thus, the metallization connected to
the anode and cathode is both on the top of the structure. The extended metallization
connected to the cathode alleviates the bending of equipotential lines at the upper corner
Distance (mm)
0.00 0.20 0.40 0.60 0.80 1.00
Dis
tan
ce
(µ
m)
0
20
0
40
0
Distance (mm)
0.00 0.20 0.40 0.60 0.80 1.00
Dis
tan
ce
(µ
m)
0
20
0
40
0
Distance (mm)
0.00 0.20 0.40 0.60 0.80 1.00
Dis
tan
ce
(µ
m)
0
20
0
40
0
Distance (mm)
0.00 0.20 0.40 0.60 0.80 1.00
Dis
tance (
µm
)
0
200
40
0
Distance (mm)
0.00 0.20 0.40 0.60 0.80 1.00
Dis
tance (
µm
)
0
200
40
0
Distance (mm)
0.00 0.20 0.40 0.60 0.80 1.00
Dis
tance (
µm
)
0
200
40
0
(a) (b)
Fig. 3.16. Comparison of the equipotential lines distribution between two different dielectric constant of
embedded insulation material. (a) εr’ = 4. (b) εr’ = 15.
0
50
100
150
200
250
300
0 5 10 15 20Dielectric constant of embedded
insulation material
E (
kV
/mm
)
0
100
200
300
400
500
600
700
0 5 10 15 20Dielectric constant of embedded
insulation material
E (
kV
/mm
)
Upper corner) Lower corner
(a) (b)
Fig. 3.15. Variation of maximum electric field intensity with the dielectric constant of embedded insulation
material. (a) Inside the polyimide. (b) Inside the embedded insulation material.
- 33 -
of the device, as given in Fig. 3.17 (b), which means the crowding of the electric field is
reduced too. This can be seen from Fig. 3.18, which is the electric field intensity drawing
in the polyimide and the embedded insulation material respectively. The maximum
electric field intensity in these two insulation materials at the upper corner of the device
is reduced to 1/4 – 1/3 of the value of the LVEP structure, and the maximum electric field
intensity at the lower corner of the device in epoxy is more than 10 times less than in the
original structure. Fig. 3.19 shows the comparison.
In the MVEP structure, the voltage between two pieces of top metalization is the
applied voltage. This requires that the insulation material between these two pieces of
metalization can stand the applied voltage. Thus, the minimum distance in the design in
between of two pieces metalization has to be greater than the applied voltage divided by
the dielectric strength of the insulation material, which is around 300 µm in the
simulation model.
A parametric study is performed as well to learn the influence of the material properties
and the dimensions. Table VI gives the parameters labeled in Fig. 3.20 and the values.
The first parameter is the thickness of the polyimide, as shown Fig. 3.21. From the
simulation results, it can be seen that with the increasing of the thickness, the highest
electric field decreases, which is the same as the LVEP #2. This mechanism can be seen
from Fig. 3.22. However, there is a phenomenon of saturation when the thickness
reaches certain value, which means further increasing of the thickness cannot help to the
reduction of the field crowding. The other parameter studied is the dielectric constant of
the polyimide, as shown in Fig. 3.23. The higher the dielectric constant is, the higher is
the maximum electric field intensity. This trend is the same as it in the LVEP #2
structure. This is also indicated by that more potential lines bend down to the area of the
embedded insulation material as the dielectric constant increases, as shown in Fig. 3.24.
It can be also found that the variation of the dielectric constant in this MVEP structure
TABLE VI
PARAMETERS IN THE MVEP STRUCTURE STUDY
Nominal value Swept values
Polyimide thickness D (µm) 400 100-500
Dielectric constant of Polyimide εr 3 4-8
- 34 -
doesn’t influence the electric field intensity at the lower corner of the device inside the
embedded insulation material with 500 µm thick polyimide.
The influence of these two parameters to the LVEP #2 and MVEP can be explained
using a simplified circuit. Since the polyimide and the embedded insulation material are
in series with each other, as the area in the dashed-line rectangle area in Fig. 3.25 (a),
these two can be simplified to two parallel plate capacitors Cp and Ce in series, as shown
in Fig. 3.25 (b). The capacitance is calculated from the dielectric constant, area and the
thickness in (3.1). The voltage drop on two series connected capacitors is shown in (3.2)
(a)
(b)
Fig. 3.17. Simulation model and result MVEP structure in MEDICI. (a) Dimensions and materials of
the structure. (b) Distribution of equipotential lines.
Anode
Polyimide
y
x
Embedded
insulation
material
Ceramic
400 µm
115 µm
508 µm
500 µm 1000 µm
1100 µm
10 µm
0
-200
-400
200
400
0.00 0.20 0.40 0.60 0.80 1.00
Distance (mm)
Distance
(µm)
SiC
A B
D
C
- 35 -
and (3.3). It is desired that the voltage drop on the capacitor made of embedded
insulation material is as low as possible, which means less equipotential lines bend down
into this material. From (3.3), this can be realized by decreasing the capacitance of the
capacitor made of polyimide Cp, which can be implemented by either decreasing the
dielectric constant or increasing the thickness. On the other hand, increasing the
dielectric constant of the embedded insulation material can do the same job. Thus, when
the thickness of the materials is fixed, it is desired to decrease the ratio between the
dielectric constant of the polyimide and the embedded insulation material.
D
AC rε= (3.1)
ep
ep
CC
CV
+= (3.2)
ep
p
eCC
CV
+= (3.3)
E = 90 kV/mmJTE1
JTE2
JTE3
Distance (mm) 0.00 0.20 0.40 0.60 0.80 1.00
E (
kV
/mm
)
025
50
75
10
01
25
E = 90 kV/mmJTE1
JTE2
JTE3
Distance (mm) 0.00 0.20 0.40 0.60 0.80 1.00
E (
kV
/mm
)
025
50
75
10
01
25
E = 150 kV/mm
E = 20 kV/mm
Distance (mm)0.00 100 200 300 400 500
E (
kV
/mm
)
05
010
01
50
E = 150 kV/mm
E = 20 kV/mm
Distance (mm)0.00 100 200 300 400 500
E (
kV
/mm
)
05
010
01
50
(a) (b)
Fig. 3.18. Electric field intensity in the insulation materials of MVEP structure. (a) In the polyimide. (b)
In the embedded insulation material.
Cathode
- 36 -
Anode
Cathodey
500 µm1000 µm
1100 µm
D
115.5 µm
x
10 µm
SiC
CeramicEmbedded insulation material
Polyimide εεεεr
Fig. 3.20. Parameters in the MVEP structure study.
0
50
100
150
200
250
300
350
0 200 400 600D (um)
E (k
V/m
m)
0
100
200
300
400
500
600
0 100 200 300 400 500 600
D (um)
E (
kV
/mm
)
Upper corner Lower corner
(a) (b)
Fig. 3.21. Variation of maximum electric field intensity with the thickness of polyimide. (a) Inside the
polyimide at the upper corner. (b) Inside the embedded insulation material.
0
100
200
300
400
500
600
SiC Polyimide (upper
corner)
Embedded
insulation
material (upper
corner)
Embedded
insulation
material (lower
corner)
Ceramic
E (
kV
/mm
)
Dielectric strength LVEP LVEP #2 MVEP
5x
2x6x
10x
4x
0.8x
24x
Fig. 3.19. Comparison of the maximum electric field intensity between two LVEP structures and the
MVEP structure.
- 37 -
Distance (mm) 0.00 0.20 0.40 0.60 0.80 1.00
Dis
tan
ce (
µm
)
0
400
20
0
-200
-40
0
Distance (mm) 0.00 0.20 0.40 0.60 0.80 1.00
Dis
tan
ce (
µm
)
0
400
20
0
-200
-40
0
02
00
40
0
0.00 0.20 0.40 0.60 0.80 1.00
Dis
tan
ce
(u
m)
Distance (mm)
20
04
00
(a) (b)
Fig. 3.24. Comparison of the equipotential lines distribution between two different dielectric constant of
polyimide. (a) εr = 4. (b) εr = 8.
020
04
00
0.00 0.20 0.40 0.60 0.80 1.00
Dis
tance (
um
)
Distance (mm) Distance (mm) 0.00 0.20 0.40 0.60 0.80 1.00
Dis
tance (
µm
)
0
400
200
-20
0
-400
Distance (mm) 0.00 0.20 0.40 0.60 0.80 1.00
Dis
tance (
µm
)
0
400
200
-20
0
-400
(a) (b)
Fig. 3.22. Comparison of the equipotential lines distribution between two different thicknesses of the
polyimide. (a) D = 100 µm. (b) D = 500 µm.
80
90
100
110
120
130
140
150
160
170
3 5 7 9
Dielectric constand of polyimide
E (
kV
/mm
)
0
50
100
150
200
250
300
350
400
450
3 5 7 9
Dielectric constant of polyimide
E (
kV
/mm
)
Upper corner Lower corner
(a) (b)
Fig. 3.23. Variation of maximum electric field intensity with the dielectric constant of polyimide. (a)
Inside the polyimide at the upper corner. (b) Inside the embedded insulation material.
- 38 -
3.5. Development of High-Voltage Embedded Power (HVEP)
structure
From the simulations in the previous section, it is known that the equipotential lines
bend down at the upper corner of the device because of the internal structure of the
device. This leads to the high electric field intensity at the corners of the device.
Although the MVEP structure has greatly improved the blocking capability of the
package, it is still cannot stand the voltage as high as 10 kV because the hotspot at the
upper corner of the device still exists.
To eliminate this high electric field crowding at the upper corner, one good method is
to prevent the equipotential lines from entering the area of the embedded insulation
material. A piece of metalization can be inserted to the device package to reshape the
electric field distribution. This is the idea of high-voltage embedded power (HVEP)
structure, as shown in Fig. 3.26 (a). Compared to the MVEP structure, the bottom
cathode metallization of the package is extended to the surface of the ceramic carrier and
the embedded insulation material, instead of to the top of the whole package. It has the
same potential as the cathode of the device, which is also the potential of the body of the
device in blocking mode. This inserted metal layer prevents the equipotential lines from
entering the embedded insulation material, and lets the entire voltage drop on the
0 V
Anode
Cathode y
500 µm1000 µm
1100 µm
D
115.5 µm
x
10 µm
SiC
14 kV CeramicEmbedded insulation
material
Polyimide εεεεr
εεεεr’
D’
Anode
Cathode y
500 µm1000 µm
1100 µm
D
115.5 µm
x
10 µm
SiC
14 kV CeramicEmbedded insulation
material
Polyimide εεεεr
εεεεr’
D’
Anode
Cathode y
500 µm1000 µm
1100 µm
D
115.5 µm
x
10 µm
SiC
14 kV CeramicEmbedded insulation
material
Polyimideεεεεr
εεεεr’
D’
Cp
Ce
(a) (b)
Fig. 3.25. Equivalent circuit model of LVEP#2. (a) Structure of LVEP#2. (b) Series capacitor model of
a small area in the LVEP#2 structure.
- 39 -
polyimide layer, which can be seen in Fig. 3.26 (b). This inserted metal layer is called
fileld-shaping metal layer.
(a)
(b)
Fig. 3.26. Simulation model and result of HVEP structure in MEDICI. (a) Dimensions and materials of
the structure. (b) Distribution of equipotential lines.
Anode
Cathode
y
x
10 µm 500 µm 1000 µm
1100 µm
508 µm
115 µm
400 µm Polyimide
Embedded
insulation
material
Ceramic
0
200
400
-200
-400
0.00 0.20 0.40 0.60 0.80 1.00
Distance
(µm)
Distance (mm)
5 µm
- 40 -
It can be observed that the electric intensity at the area adjacent to the JTEs in the
device is high. This high electric field intensity cannot be reduced by improving the
structure geometry because it depends on the internal structure of the device, which is
determined by the device manufacturers, and it is taken care by the passivation on the top
of the device from the manufacturers. Furthermore, since the field-shaping metal layer
prevents the equipotential lines from entering the embedded insulation material, even
with the presence of the voids in the embedded insulation material, it won’t degrade the
voltage rating of the package.
JTE3
JTE2
JTE1
E = 125 kV/mm
Distance (mm) 0.00 0.20 0.40 0.60 0.80 1.00
E (
kV
/mm
)
02
55
07
510
01
25
JTE3
JTE2
JTE1
E = 125 kV/mm
Distance (mm) 0.00 0.20 0.40 0.60 0.80 1.00
E (
kV
/mm
)
02
55
07
510
01
25
E (
kV
/mm
)0
12
30
100 200 300 400 500Distance (um)
E = 3.2 kV/mm
E = 0.2 kV/mm
(a) (b)
Fig. 3.27. Electric field intensity in the insulation materials of HVEP structure. (a) In the polyimide. (b)
In the embedded insulation material.
0
100
200
300
400
500
600
SiC Polyimide (upper corner)
Embedded insulation
material (upper corner)
Ceramic
E (
kV
/mm
)
Dielectric strength LVEP LVEP #2 MVEP HVEP
5x
24x
2x
6x
10x
4x
0.8x
0.9x 0.1x
Embedded insulation material
(lower corner)
Fig. 3.28. Comparison of the maximum electric field intensity of different structures.
- 41 -
3.6. Geometry impact on HVEP structure
There are several geometric parameters need to be determined in the design of a HVEP
structure. These parameters include the thickness of the polyimide – D, the position of
the inserted metal layer – t, the width of the field-shaping metal layer – w, and the angle
of the edge of the polyimide - θ, as shown in Fig. 3.29. Table VII gives the parameters
shown in Fig. 3.29 and the values
From the simulation results in Fig. 3.30, with the increasing of the thickness of the
polyimide, the maximum electric field at the upper corner of the device in the polyimide
layer decreases. It influences the electric field intensity in the area in which the
equipotential lines are uniform as well, as shown in Fig. 3.26.
TABLE VII
PARAMETERS IN THE HVEP STRUCTURE STUDY
Nominal value Swept values
Polyimide thickness D (µm) 400 400 - 1000
Field-shaping metal layer position t (µm) 0 0 - 200
Field-shaping layer width (µm) 0 -400 - 500
Polyimide edge angle θ (°) 90 30 - 90
Anode
Cathodey
500 µm1000 µm
1100 µm
D
115.5 µm
x
10 µm
SiC
CeramicEmbedded insulation material
Polyimide
t
w
θθθθ
flat area
Fig. 3.29. Parameters in the HVEP structure study.
- 42 -
When the field-shaping metal layer is extended from the cathode metallization on the
bottom of the package, its voltage is tied to 0 V. The impact of the variation of the
position of this layer on the maximum electric field intensity at the upper corner of the
device in the polyimide layer is shown in Fig. 3.31. From the simulation result shown in
Fig. 3.31 (b), the highest electric field intensity happens at the edge of the inserted metal
layer instead of the corner of the device. Furthermore, besides this value exceeds the
dielectric strength of the polyimide with the movement of the metal layer, the maximum
electric field intensity at the device corner in the embedded insulation material is
increases 3 times.
0
10
20
30
40
50
60
70
80
0 50 100 150 200t (um)
E (
kV
/mm
)
Embedded insulation material Polyimide Plat area
Distance (mm) 0.00 0.20 0.40 0.60 0.80 1.00
Dis
tan
ce (
µm
)
0
40
0
20
0
-200
-400
(a) (b)
Fig. 3.31. Variation of maximum electric field intensity with the position of the inserted metal layer. (a)
Maximum electric field intensity in two insulation materials. (b) Distribution of Equipotential lines at t =
200µm.
0
10
20
30
40
50
400 600 800 1000D (um)
E (
kV/m
m)
Polyimide Flat area
Fig. 3.30. Variation of maximum electric field intensity at the corner of the device with the thickness of the
polyimide.
- 43 -
The increased electric field intensity at the edge of the field-shaping metal layer can be
alleviated by thicken the polyimide layer. When t is 80 µm and D is 1000 µm, the
maximum electric field intensity at the edge of the inserted metal drops to 40 kV/mm.
However, this measure cannot help to reduce the maximum electric field intensity in the
embedded insulation material which is still 2 times higher than the maximum electric
field intensity when t is equal to 0.
Since the inserted metal layer is connected to the cathode metallization, its voltage is
13.5 kV. Actually the voltage applied to this metal layer can be different values. Fig.
3.32 shows the equipotential line distribution of two cases. As shown in Fig. 3.32 (b), the
equipotential lines bend down into the embedded insulation material, and the high
electric field intensity occurs at the upper corner. Fig. 3.33 shows the comparison of the
simulation results of different combination of the applied voltage and the position of the
field-shaping metal layer. When the field-shaping metal layer is tied to higher voltages,
the equipotential lines bend down into the embedded insulation material, which results
high electric intensity at the corner of the device inside both the polyimide and the
embedded insulation material. Thus, the best position of this metal layer is at the same
surface as the device and its voltage is tied to the cathode voltage.
Distance (mm) 0.00 0.20 0.40 0.60 0.80 1.00
Dis
tan
ce (
µm
)
0
400
200
-20
0 -4
00
Distance (mm) 0.00 0.20 0.40 0.60 0.80 1.00
Dis
tan
ce
(µ
m)
0
40
0
20
0
-20
0
-40
0
0 V
13.5 kV
5 kV
(a) (b)
Fig. 3.32. Distribution of equipotential lines with different combination of the position and the voltage of
the inserted metal layer. (a) t = 0µm and V = 0 V. (b) t = 200 µm and V = 5 kV.
- 44 -
The width of the -metal layer has great impact on the electric field distribution as well.
The definition of this parameter is the distance between the edges of the field-shaping
metal layer and the device. The origin is at the edge of the device. As shown in Fig.
3.34, there is an optimal point where the field-shaping metal is long enough to reach the
edge of the device which means w equals to 0. When the metal layer is too short which
means w is greater than 0, as can be seen from Fig. 3.35 (a), equipotential lines bend
down and the metal layer lose its shield effect. When the metal layer is too long which
means w is less than 0, as shown in Fig. 3.35 (b), it blocks the equipotential lines from
entering the embedded insulation material extremely well. However, the near-evenly
distributed equipotential lines on the surface of the device have to bend to reach the edge
of the metal layer, which creates high electric field concentration at the edge of the metal
layer and the surface of the device. Thus, the optimal width of the field-shaping metal
layer is realized when the edge of the field-shaping metal layer touches the edge of the
device, which means w equals to 0.
Furthermore, when the field-shaping metal layer slightly covers the edge of the device,
the maximum electric field intensity in the polyimide doesn’t increase dramatically, as
shown in Fig. 3.35 (a). Thus, it is preferred that there is little overlap between the field-
shaping metal layer and the device edge to that the field-shaping metal layer is apart from
the device edge.
0
200
400
600
800
1000
1200
1400
1600
1800
2000
Polyimide (upper corner) Embedded insulation material
(upper corner)
Embedded insulation material
(lower corner)
E (
kV
/mm
)
V = 13.5 kV, t = 0 um V = 13.5 kV, t = 100 um V = 5 kV, t = 0 um
V = 5 kV, t = 200 um V = 10kV, t = 100 um V = 10 kV, t = 200 um
Fig. 3.33. Variation of maximum electric field intensity with different combination the position of the
inserted metal layer and its voltage
- 45 -
In the previous simulations, the angle of the polyimide edge θ shown in Fig. 3.29 is
90°. Practically, this angle is usually less than 90° because the overflow of the
polyimide. Thus, the influence of different angles is simulated. In Fig. 3.36, the
maximum electric field intensity along a horizontal line at y = -100 µm is compared
among different angles. As it can be seen from Fig. 3.37, the equipotential lines are
compressed at the edge of the polyimide because of the small angle there. Thus, the
electric field at the same position in the area near the edge of the polyimide is higher
when the angle is small.
Distance (mm) 0.00 0.20 0.40 0.60 0.80 1.00
Dis
tance (
µm
)
0
400
200
-200
-400
Distance (mm) 0.00 0.20 0.40 0.60 0.80 1.00
Dis
tance (
µm
)
0
400
20
0
-200
-400
(a) (b)
Fig. 3.35. Distribution of equipotential lines with different width of the inserted metal layer. (a) w =
500µm. (b) w = -400 µm.
0
200
400
600
800
1000
-400 -200 0 200 400
w (um)
E (
kV
/mm
)
0
50
100
150
200
250
-400 -200 0 200 400
w (um)
E (
kV
/mm
)
Upper corner Lower corner
(a) (b)
Fig. 3.34. Variation of maximum electric field intensity with the width of the inserted metal layer. (a)
Maximum electric field intensity in the polyimide. (b) Maximum electric field intensity in the embedded
insulation material.
- 46 -
Although this angle is inevitable in the process, it can be moved to avoid the high field
crowding, as shown in Fig. 3.38 (a). Assume the angle of the polyimide edge is 30°, by
increasing its height, h, the maximum electric field of the horizontal line at y = -100 µm
is reduced. Thus, in reality, a “dam” can be built to lift the sharp angle of the polyimide
edge to avoid increasing the electric field intensity in that area.
Since the field-shaping metal layer and the top metallization are in parallel, an extra
capacitor in the device package is introduced. The capacitance can be estimated by using
(3.1).
AAD
AC r 66
10400
1085.836
12
=×
××==
−
−
ε nF
Distance (mm) 0.00 0.20 0.40 0.60 0.80 1.00
Dis
tance (
µm
)
0
400
20
0
-200
-400
Distance (mm) 0.00 0.20 0.40 0.60 0.80 1.00
Dis
tan
ce (
µm
)
0
40
0
20
0
-20
0
-40
0
(a) (b)
Fig. 3.37. Distribution of equipotential lines with different angle of the polyimide. (a) θ = 30°. (b) θ =
90°.
30
35
40
45
50
55
60
65
70
20 40 60 80 100
Theta (deg)
E (
kV
/mm
)
Fig. 3.36. Variation of the maximum electric field intensity along the horizontal line at y = -100 µm with
the angle of the polyimide .
- 47 -
Assume that the area is usually less than 100 mm2, the capacitance is less than 10 fF,
which is negligible.
3.7. Summary
In this Chapter, two LVEP structures are revisited in MEDICI with a 10 kV SiC diode
embedded in the structures. The MEDICI simulations show that the double-sided
metallization can help to reduce the high electric field intensity at the lower corner of the
device inside the embedded insulation material. Some parameters in the LVEP #2
structure are studied to obtain their influences on the maximum electric field intensity.
The higher thickness of the polyimide layer can decrease the maximum electric field at
the corners of the device in two insulation material. However, when the thickness of the
polyimide is comparable to the thickness of the ceramic, further increasing cannot help to
reduce the field intensity very much. From the material property point of view, low
dielectric constant of the polyimide is desired.
Based on the simulations of the LVEP structures, a MVEP structure is proposed. The
cathode metallization is extended to the top of the structure. The maximum electric field
intensity at the corners of the device inside two insulation material is reduced greatly.
However, at a voltage of 13.5 kV, these values are still higher than their dielectric
strength. Some parameters in the structure are investigated as well. It is shown that the
Distance (mm) 0.00 0.20 0.40 0.60 0.80 1.00
Dis
tan
ce (
µm
)
0
400
200
-200
-400
h
30°°°°
30
35
40
45
50
55
60
0 50 100 150 200
h (um)
E (
kV
/mm
)
(a) (b)
Fig. 3.38. Influence of the height of the angle of the polyimide when θ = 30°. (a) Structure. (b) Electric
field intensity variation with different values of the height.
- 48 -
influence of the thickness and the dielectric constant of the polyimide is the same as that
in the LVEP #2 structure. This can be explained by simplifying the structure to a circuit
that consists of two capacitors in series.
A HVEP structure is proposed finally. A thin layer of metal is field-shaping to the
middle of the structure and it is tied to the cathode metallization. This field-shaping
metal layer prevents the equipotential lines from entering the embedded insulation
material. Thus, the maximum electric field at the corners inside the two insulation
materials is reduced to the value below their dielectric strength. Furthermore, since there
is no electric field in the embedded insulation material, even with the presence of the
voids in the embedded insulation material, the voltage rating of the package won’t be
degraded.
The HVEP structure design involves several geometry parameters. They are the
thickness of the polyimide – D, the position of the field-shaping metal layer – t, the
extension length of the field-shaping metal layer – w, and the angle of the edge of the
polyimide - θ. The thickness of the polyimide is proportional to its dielectric strength
and the applied voltage in the flat area. The best position for the field-shaping metal
layer is on the same surface as the device, and the optimal voltage applied this metal
layer is the same as the cathode. Simulation also indicates that, the optimal width of the
field-shaping metal layer is that the edge of the field-shaping metal layer touches the edge
of the device. It can be found that the angle of polyimide edge is less than 90°
practically, and this smaller angle could influence the package voltage rating greatly.
Although this angle is unavoidable, it can be solved by the application of a thick “dam”.
It is preferred that the angle of the polyimide would be 90° or even larger with the aid of
this “dam”.
The meshing method in MEDICI is manual meshing. The number of meshes
influences the accuracy of simulation results. However, since the critical areas are
known in the structures, the meshing in those areas is much denser than other non-critical
areas. Thus, the accuracy is insured by this way.
- 49 -
Chapter 4: Design Guidelines for the HVEP package and
Fabrication Process of a HVEP package
The main task of engineers is to apply their scientific and engineering knowledge to the
solution of technical problems, and then to optimize those solutions within the
requirements and constraints set by material, technological, economic, legal,
environmental and human-related considerations. An engineering design process is a
process used by engineers to help develop products. The engineering design includes
many aspects and wide-range activities. Both engineer’s scientific knowledge and
experience play important roles in the engineering design. Guidelines are useful in the
old product re-design and the new product design. The idea behind the design guidelines
is that there is a fundamental set of the principles that determines good design practice.
These design guidelines can be derived from the knowledge when a set of judgments
based on the design rules can yield correct solutions to all problems.
In this Chapter, the general engineering design practice is reviewed first. Then the
practice of the engineering design of the HVEP package is introduced. Next, the design
guidelines for the HVEP package are derived. Based on the design guidelines, a HVEP
package for a 2 kV SiC diode is developed. After the package is designed, it fabrication
process is designed based on the device dimensions, materials, and the optimal rules in
the design guidelines, and the detailed the fabrication process is shown. Finally, the
package is tested at the rated voltage and the experimental results are presented.
4.1. Design guidelines for the HVEP package
4.1.1. Introduction to engineering design
The engineering design is defined as the process of devising a system, component or
process to meet desired needs. It is a decision-making process (often iterative), in which
the basic sciences, mathematics, and engineering sciences are applied to convert
resources optimally to meet a stated objective. Among the fundamental elements of the
- 50 -
design process are the establishment of objectives and criteria, synthesis, analysis,
construction, testing, and evaluation. The engineering design process involves ten steps.
They are: (1) identifying a need; (2) defining the problem; (3) conducting research; (4)
narrowing the research; (5) analyzing set criteria; (6) finding alternative solutions; (7)
analyzing possible solutions; (8) making a decision; (9) presenting the product; and (10)
communicating and selling the product [44].
Modeling is a technique used by the engineer to define a real situation for generating
quantitative solutions. Relevant approximations in the modeling are critical for obtaining
the best solution to the engineering problem. Thus, the engineer must have a clear
understanding of the physical basis of the real situation. Usually, more than one solution
exists, the best one is generally found based on the engineer’s scientific knowledge and
experience. An evaluation of the solutions to find the optimal one is necessary.
4.1.2. Design practice in the HVEP package
In [45], the design practice in a thermal design is introduced. Similarly, in the electrical
design of the high-voltage packaging, the first task is to identify the objectives of the
design, which is to keep the maximum electric field intensity of the insulation material at
or below its dielectric strength. Three general ways of analyzing the design situation
include analytical analysis, mathematical modeling, and experiments. Some common
solutions to achieve this objective include justified selection of high dielectric strength
material and the manipulation of the structure. Optimization of the design generally
concentrates on selectively choosing the best values of design parameters that optimize
performance reliability at lowest cost. Finally, the best design parameter quantity will be
chosen as the output of the design for prototype preparation.
The design process of a HVEP package can be illustrated in Fig. 4.1. The input to the
design includes the device dimensions, ratings, the electrical schematics, the packaging
material properties and dimensions, and the requirements from process. The electric field
modeling can be performed with analytical analysis or mathematical modeling. Since the
structure of the device and the package is complex, the mathematical modeling has to
simplify the structure and approximate the real case. Therefore, experimental validation
is always required to validate the accuracy of the model if mathematical modeling is used
- 51 -
as a tool for the electric field analysis. The numerical modeling is another method to
describe the behavior of the package. Compared to the mathematical method, it can
analyze much more complicated situation. Thus the numerical method is often used to
mathematically describe and obtain information on the electric field distribution. After
the parametric study, an optimized design is obtained. With the consideration of the
concurrent engineering design process in which the electrical, thermal, and process
constraints are performed in parallel, a good product design is achieved.
4.1.3. Design guidelines for the HVEP package
A design guideline is defined by a standard or principle by which to make a judgement
or determine a policy or course of action. Design guidelines can be from theory,
empirics, or practices. The role of the design guidelines is to provide some insight to the
design parameters during the design process. Although the engineer may be faced with
the available choices for the design, the key point in using the design guidelines is to
recognize the available information as well as the known constraints. By understanding
the behaviour of the electric field, the engineer can choose appropriate materials and
geometries to optimize the design.
The purpose of the design guidelines is to provide more suggestive and general rules in
determining and predicting the electric field distribution in the HVEP package. From the
Device information
Electrical layout
information
HVEP package design input
HVEP package modeling
Parametric studies
Design Optimization
Design Output
Check with Process
Constraints
Check with Electrical
constraints
Experimental Validation
Not Satisfied Satisfied
Satisfied
Not Satisfied
Fig. 4.1. Engineering design process of HVEP package.
- 52 -
simulations in Chapter 4, the guidelines for the design of HVEP package can be
summarized as the followings.
(1) Position of the field-shaping metal layer.
The position of field-shaping metal layer influences the electric field intensity at the
corners of the device in both insulation materials. When the field-shaping metal layer
moves up, it loses its effectiveness of shielding the embedded insulation material. Thus,
the electric field intensity increases in both insulation materials at the corners of the
device.
When the voltage applied to this layer is 0 V, the position of the field-shaping metal
layer should be as close as possible to the surface of the device. The best position is on
the same surface as the device.
(2) Width of the field-shaping metal layer.
When the field-shaping metal layer is nonadjacent to the edge of the device, the electric
field intensity at the device corners increases with the increasing of the distance between
the field-shaping metal layer and the edge of the device. In this situation, the metal layer
cannot prevent the equipotential lines from bending down into the embedded insulation
material. When the field-shaping metal layer overlaps the device, it blocks the
equipotential lines from bending down into embedded insulation material completely.
However, high electric field concentrates at the edge of this layer due to the bending of
equipotential lines, and the maximum electric field intensity increases dramatically with
the increasing of the overlapping length.
When the field-shaping metal layer touches the edge of the device, it gives the best
electric field distribution. It is more favourable that the metal layer overlaps little of the
device than that it is nonadjacent to the edge of the device.
(3) Voltage applied to the field-shaping metal layer
Different combination of applied voltage and position of the field-shaping metal layer
gives different electric field intensity at the corners of the device in the two insulation
materials. The higher applied voltage is, the less shielding effect is the field-shaping
metal layer.
The best applied voltage is the same as the cathode voltage.
(4) Thickness of the polyimide layer
- 53 -
The maximum electric field intensity at the upper corner in the polyimide decreases
with the increasing of the thickness. The electric field between the field-shaping metal
layer and the top metallization is nearly uniform. The minimum thickness of the
polyimide can be estimated by ratio between the rated voltage and the thickness.
Simulation is necessary for more accurate results.
(5) Angle of the polyimide edge
The small angle of the polyimide edge introduces higher electric field intensity in the
material. It is desired to keep that angle being 90°. This can be controlled by adding a
“dam” in the structure. When the “dam” is high enough, small angle doesn’t influence
the electric field too much. Thus, a “dam is needed in the package and its thickness is the
same as the polyimide layer.
The best position and the best width of the field-shaping metal layer are called optimal
rule.
4.1.4. A design example
The designed guideline in the previous section can be applied to a single device
package or a multi-device package. A design example is given in Fig. 4.2, which is the
package of an IGBT and its anti-parallel diode.
(1) From the input of the design, for example, the dimensions of the devices and
materials, the electrical rating and properties, and the electrical schematics, the physical
layout is designed, as shown in Fig. 4.2 (a).
(2) The layout of the field-shaping metallization is designed. From the design
guidelines above, the field-shaping metallization should be on the same surface as the
device, touches the edges of both devices and be tied to the bottom metallization, as
shown in Fig. 4.2 (b). Here the connection of the field-shaping metallization and the
bottom metallization is designed to be realized through a via.
(3) The polyimide layer is designed. From the design guideline, the minimum
thickness of the polyimide can be estimated. Simulation is necessary for more accurate
results. Dams for the device contacts are added to form 90° polyimide angles, as shown
in Fig. 4.2 (c).
- 54 -
(4) The layout of the top metallization is designed. With the consideration of the
electrical schematics, the top metallization and the arrangement of the components are
designed, as shown in Fig. 4.2 (d).
(5) After all the requirements and constraints are checked, some of the designed layouts
may be revised.
Ceramic
Diode IGBT
Embedded insulation material
AnodeCollector Gate
ViaJTEs
(a)
Ceramic
Diode IGBT
Metallization
Field-shaping Metal layer
(b)
Ceramic
Diode IGBT
Polyimide
Dam for anode Dam for collector Dam for gate
(c)
Ceramic
Diode IGBT
Polyimide
Lead/Pin Lead/PinComponents
(d)
Fig. 4.2. A design example. (a) Physical layout design. (b) Inserted metal layer design. (c) Polyimide
layer design. (d) Top metallization design.
- 55 -
4.2. Fabrication process of a HVEP package
From the design guidelines for the HVEP package, a package for a 2 kV SiC diode is
developed firstly. Then the designed package is fabricated and tested to validate the
design.
4.2.1. Devices and materials
The 2 kV SiC diode is from Cree Inc. Its picture and dimensions are shown in Fig. 4.3.
The thickness of this device is 381 µm (15 mil). The anode and the cathode metallization
is gold which is solderable. The JTEs are around the anode contact and covered by
passivation from the manufacturer.
The polyimide is Epo-Tek 600 from Epoxy Technology Inc. It is a one-component
polyimide system. The dielectric constant is 2.4, and its dielectric strength is 17 kV/mm.
This material needs to be cured at 150°C for 1 hour and then at 275°C for 1 hour [46].
The thickness of the polyimide is determined by the simulation. The device in the
simulations is the same device as previous section. The results show that the maximum
electric field intensity at the upper corner is less than 17 kV/mm when the thickness is
150 µm. Thus, the thickness of the polyimide layer is selected as 150 µm.
EP3HT from Master Bond Inc. is selected as the embedded insulation material. It is a
one-component, heat-curing epoxy adhesive/sealant. The ceramic carrier provides the
mechanical support to the device, and its thickness is 508 µm (20 mil). The metallization
is copper in the package.
3.5 mm
5 mm
1.5 mm
2.7 mm
Fig. 4.3. Picture and dimensions of a 2 kV SiC diode.
Anode
JTEs with
passivation on
the top
- 56 -
4.2.2. Design of the fabrication procedure
Although both the LVEP package and the HVEP package are based on the embedded
power technology, the HVEP package has more complicated structure than the LVEP
package. Therefore, the fabrication of the HVEP package needs to be re-designed.
The fabrication of the LVEP package is shown in Fig. 4.4. It includes 6 major steps.
First, the opening for the device is cut on the ceramic carrier by using a laser machine.
Second, the device is embedded in the ceramic and epoxy is dispensed around the device
to seal it in the ceramic. The surface of the device is aligned with the ceramic and the
epoxy on the top side. Third, the polyimide is patterned on the flat surface through
screen-printing. Fourth, thin layer of copper is deposited on the top surface by sputtering.
Fifth, the top metallization is thickened by electroplating for the electrical conduction.
Last, the unnecessary thin layer copper is etched away.
Besides all these steps, the fabrication of the HVEP package needs the deposition of the
double-sided metallization and the field-shaping metal layer, and the formation of the
“dam”. The designed fabrication procedure of the HVEP package is shown in Fig. 4.5.
The different steps between the LVEP package and the HVEP package are highlighted by
solid-line rectangle. The detailed fabrication steps are as followings.
(1) Layout design. The dimensions of the opening for the device are designed based on
the dimensions of the diode. The position of the via is determined by the mechanical
strength of the ceramic empirically, which requires at least 3 mm distance between
openings. Then the dimensions of the ceramic carrier is designed based on the opening
and the via. All the design drawings are done in CAD software.
(2) Ceramic cutting. A piece of ceramic is cut in this step by using the laser machine.
First, the design drawings are imported to the program that controls the laser machine.
Then parameters that control the power, repetitive rate, pulse width, and speed of the
laser are setup for cutting the ceramic. Finally, the laser is run by the program to
execute the command that cuts the opening and via in the design.
(3) Chip embedding. First, a piece of the Kapton tape is attached to one face of the
ceramic carrier right on the opening as the holder of the device. Second, the diode is
fitted into the opening with the anode adhered to the Kapton tape. Third, the EP3HT
- 57 -
material is dispensed to the space between the device and the ceramic by using a
dispenser, and it is cured at 160 °C for 10 minutes in an oven. Since the presence of the
field-shaping metal layer in the HVEP structure, the process control effort is greatly
reduced since the voids introduced in the EP3HT in this step are no longer a threat to the
package breakdown voltage.
(4) Shim soldering. First, the size of the copper shim is determined by the contact
dimensions of the device. Then a mask is designed to cover the size of the shim. Third, a
piece of copper is coated with photoresist. After 20 minutes soft bake at 80 °C, this piece
of copper is covered by the designed mask and exposed under the UV light for 20
minutes. This is called photolithography process. Forth, the exposed copper piece is
washed in the diluted D4000, and the ratio between D4000 and water is 1 to 20. After
this step, the shim area is protected by the photoresist on the copper piece. Fifth, the
copper piece is put into an etching tank and etched. The duration of etching is
determined by the thickness of the copper. Sixth, the photoresist on the copper shim is
rinsed off by Aceton after the etching is done. The copper shim from the etching
machine is soldered to the anode of the diode by using 180 °C Pb/Sn solder paste and
hotplate of 210 °C. Since the shape of the cross-section the shim is trapezoidal due to the
etching step, the larger area side is preferred to be soldered with the contact to form an
obtuse-angle dam for the polyimide layer.
(5) Field-shaping metal layer deposition. First, a piece of Kapton ring mask which fits
the dimensions of the JTEs of the device is cut by using the laser machine. Second, this
Kapton mask is adhered to the device and protects the JTE area. Third, the device
package is put in the chimer of a sputtering machine to deposit both titanium and copper.
The deposition duration of the titanium and copper is 45 minutes respectively, and the
thickness of these two metals is around 500- 600 µm and 1000 µm, respectively. Forth,
the Kapton ring mask is peeled off. In this step, from the simulation results in the
previous section, the Kapton mask is preferred to be smaller than the actual JTE
dimensions since it is difficult to precisely control the mask to be the same size as the
JTEs. The via is deposited as well.
(6) Interlayer patterning. First, a piece of Kaption mask is necessary to protect the top
of the copper shim. Second, a designed screen is loaded to a screen printer and the
- 58 -
device package is fixed to the platform of the screen printer with the anode side faced up.
Third, the polyimide is printed onto the anode side using the screen printer. Forth, the
Kapton mask is peeled off and the device packaging is put into an oven and cured
following the curing profile given in the datasheet. If thick polyimide layer is necessary,
this step can be performed repetitively.
(7) Sputtering. The device package is put into the chamber of the sputtering machine
and thin layers of metals are deposited on both sides of the package by two rounds of this
step.
(8) Electroplaing. First, a mask is designed for pattern of the anode-side metallization.
Second, the photolithography process is preceded on the device package as described in
step (4). Third, the device package is put into the copper-plating solution with the
desired pattern open. The current of the electroplating is set to 0.1 A since the area is
small.
(9) Micro--etching. The excessive sputtered copper and titanium are removed by the
micro-etchant respectively.
(10) Final cutting and assembling. The excessive ceramic carrier is cut by using the
laser machine. The external components or I/O pin can be soldered in this step.
1. Ceramic Cutting
6. Etching
Ceramic
2. Chip EmbeddingDeviceEpoxy
3. Interlayer Patterning
Polyimide
4. Sputtering
Thin-film copper
5. Electroplating
Thick-film copper
Fig. 4.4. Major fabrication steps of the LVEP package.
- 59 -
4.2.3. Fabrication process and experimental results
The fabrication process is shown in Fig. 4.6 through Fig. 4.13. The final package is
less than 1 mm thick. The Five HVEP package are fabricated and tested by using a high-
power curve tracer. The test results show that fabricated package can block the rated
voltage which is 2 kV, as it can be seen in Fig. 4.14, which is one of the results. In the
mean time, two LVEP package are fabricated and tested as well. Both of them failed at
1.1 kV-1.2 kV.
The cross-sectional view of one of the fabricated package is shown in Fig. 4.15. The
dimensions of the fabricated HVEP package are measured using an Environmental SEM
Device
Epoxy
Device
Epoxy
Via
Copper
shim
Thin-film
copper
Thicker
polyimide
Thicker
polyimide
Thin-film copperThin-film copper
Thick-film copperThick-film copper
1. Ceramic Cutting
6. Etching
2. Chip Embedding
3. Interlayer Patterning
4. Sputtering
5. Electroplating
3. Shim Soldering
4.Field-shapping Metal Layer Deposition
Kaptonring mask4. Masking of JTEs
Device
Epoxy
Device
Epoxy
Via
Copper
shim
Thin-film
copper
Thicker
polyimide
Thicker
polyimide
Thin-film copperThin-film copper
Thick-film copperThick-film copper
1. Ceramic Cutting
6. Etching
2. Chip Embedding
3. Interlayer Patterning
4. Sputtering
5. Electroplating
3. Shim Soldering
4.Field-shapping Metal Layer Deposition
Kaptonring mask4. Masking of JTEs
Fig. 4.5. Major fabrication steps of the HVEP package.
- 60 -
(ESEM). One of the measurement results is shown in Fig. 4.16. From Fig. 4.16 (a), it
can be clearly seen that the angle formed by the copper shim is greater than 90°. The
total thickness of the shim and the solder layer is around 155 µm. The thickness of the
polyimide at the shim edge is much thicker than the shim because the screen-printing
process cannot be well-controlled at the location. Fig. 4.16 (b) is the ESEM picture at the
package edge. At this location, the screen-printing process can be well-controlled.
Therefore the thickness of the polyimide is around 140 µm to 150 µm. A comparison of
the designed values and the fabricated values are shown in Table VIII.
3. Shim soldering
Top view
Copper shim (thickness = 0.254 mm)
3. Shim soldering
Top view
Copper shim (thickness = 0.254 mm)
Fig. 4.8. Fabrication process of the HVEP package: step 3. Shim soldering.
2. Chip embedding
Top view Bottom view
Anode (gold) Cathode (gold)
Epoxy Epoxy
2. Chip embedding
Top view Bottom view
Anode (gold) Cathode (gold)
Epoxy Epoxy
Fig. 4.7. Fabrication process of the HVEP package: step 2. Chip embedding.
1. Ceramic cutting
3 mm
5 mm
Ceramic
( thickness = 0.508 mm)1.8 kV SiC diode
(thickness = 0.381 mm)
Opening for the diode
Via for interconnection between inserted metal and cathode
1. Ceramic cutting1. Ceramic cutting
3 mm
5 mm
Ceramic
( thickness = 0.508 mm)1.8 kV SiC diode
(thickness = 0.381 mm)
Opening for the diode
Via for interconnection between inserted metal and cathode
Fig. 4.6. Fabrication process of the HVEP package: step 1. Ceramic cutting.
- 61 -
7. Top metallization sputtering
Anode + Copper shim
Top metallization
7. Top metallization sputtering
Anode + Copper shim
Top metallization
Fig. 4.12. Fabrication process of the HVEP package: step 7. Top metallization deposition.
6. Interlayer patterning
Anode + Copper shim
Polyimide
(thickness = 0.254 mm)
6. Interlayer patterning
Anode + Copper shim
Polyimide
(thickness = 0.254 mm)
Fig. 4.11. Fabrication process of the HVEP package: step 6. Interlayer patterning.
5. Inserted and bottom metallization sputtering
Top view
Anode + copper shim
Inserted metallization
Guard rings (un-
sputtered)
5. Inserted and bottom metallization sputtering
Top view
Anode + copper shim
Inserted metallization
Guard rings (un-
sputtered)
Fig. 4.10. Fabrication process of the HVEP package: step 5. Thin layer copper deposition.
4. Masking
Top view
Kapton ring (as a mask to prevent the sputtering on the guard rings)
4. Masking
Top view
Kapton ring (as a mask to prevent the sputtering on the guard rings)
Fig. 4.9. Fabrication process of the HVEP package: step 4. Masking.
- 62 -
TABLE VIII
COMPARISON OF DESIGNED VALUES AND FABRICATED VALUES OF HVEP PACKAGE
Designed value Fabricated value
Polyimide thickness D (µm) 150 230 @ shim edge
140 @ package edge
Shim thickness (µm) 150 155
Polyimide edge angle θ (°) 90 125 - 135
8. Final package
Top view Bottom view
Anode + Copper shim
Thick copper
8. Final package8. Final package
Top view Bottom view
Anode + Copper shim
Thick copper
Fig. 4.13. Fabrication process of the HVEP package: step 8. Electroplating and etching.
10
Leakage current (uA)
20
30
40
50
60
70
80
90
Blocking voltage (kV)0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
2.012 kV, 5 µA
Diode breakdown
Fig. 4.14. Test result of the fabricated HVEP package.
- 63 -
4.3. Summary
In this Chapter, the general engineering design practice is reviewed first. The design
procedure of the HVEP package is presented. From the parameters study in previous
Shim
Solder
SiC Diode
Polyimide
Ceramic
Polyimide
(a) (b)
Fig. 4.16. ESEM picture of one fabricated HVEP package. (a) Copper shim edge. (b) Package edge.
SiC Diode Ceramic Via
Shim Polyimide
Epoxy
Silicone gel
Solder mask
Metal clip
Fig. 4.15. Cross-sectional picture of one fabricated HVEP package.
- 64 -
Chapter, a general design guideline is proposed. A 2 kV SiC diode package is designed
by following the design guidelines.
The structure of the HVEP package is more complicated than that of the LVEP
package. Thus the fabrication procedure of the LVEP package needs to be re-designed to
satisfy the requirement of the HVEP package. The fabrication of the LVEP package is
revisited first. Then the new fabrication process for the HVEP package is proposed. In
the fabrication, the thickness of the polyimide which is through screen-printing, and the
accuracy of the alignment of the Kapton ring mask for the JTE area cannot not be easily
controlled.
Five HVEP packages are fabricated in house. The test results show that they can block
the rated voltage which is 2 kV. An environmental SEM measurement is done to
measure the dimensions of the fabricated package. The experimental results validate the
simulations and the design guidelines.
- 65 -
Chapter 5: Simplified Modeling of the HVEP Package
Simulation in MEDICI contains the internal information of the power semiconductor
devices, and its results are closer to the reality. However, the simulation is very time-
consuming. Usually it takes hours to run one simulation on a Linux system in the
previous Chapter. On the other hand, in the early stage of design, it is not necessary to do
this elaborated simulation. Simple and quick methods are needed to give an overall idea
of the design.
The complexity of the HVEP package partly comes from the internal structure of the
device. It is not easy to model this in a mathematical way or in an electromagnetic field
simulation software like MAXWELL 2D. Thus, the model of the power device needs to
be simplified first. Then the HVEP package can be either mathematically calculated or
simulated in a quick and simple way.
5.1. Model simplification
Although the internal structures are different from one power semiconductor device to
the other, the mechanism is the same when they are blocking voltages. The JTE relieves
the high electric field crowding at the corner of the main PN junction, and establishes a
horizontal voltage drop on the surface of the device instead of a vertical one.
When the device is blocking voltages, its surface has a voltage drop in certain scheme
and the rest of the device body has the same potential. From this mechanism, a
simplified model can be built as shown in Fig. 5.1. The JTE is modelled as a conductor
and its voltage is a function of the x-axis as in Fig. 5.1 (b). The bottom, the left and the
right of the device are at same potential. This model can be further simplified as a piece
of conductor that has different voltage added to different segments. Then the HVEP
structure can be simplified as well, as shown in Fig. 5.2. Since the area of the embedded
insulation material and the ceramic is fully shielded, this part can be eliminated and the
only the polyimide material and different voltage boundaries are left in the model.
Since the purpose of the JTE is to avoid electric field crowding in the structure, it
makes the voltage drop on it as linear as possible. Fig. 5.3 shows the voltage distribution
- 66 -
on the JTE of the device in the simulation. Therefore, the voltage distribution can be
simplified as a linear function of the distance in x direction. The final simplified model
can be used in the mathematical calculation and the simulation is shown in Fig. 5.4. Only
the polyimide area is of interest.
V = 0 V V = f(x)
x
V = 10 kV
Embedded
insulation
material
Ceramic
Polyimide
y
Fig. 5.2. Simplified model of the HVEP package.
P+P-JTE
N-
N+
V = 10 kV
V = 0 V
SiC
V = 0 V V = f(x)
x
V = 10 kV
Main
junction
JTE N body
(a) (b)
Fig. 5.1. Model of the power semiconductor device. (a) Model in MEDICI. (b) Simplified model.
- 67 -
5.2. Simulation in MAXWELL 2D
MAWELL 2D is a powerful electro-magnetic simulation software. It is user-interface
friendly, simple and fast when it runs static electric field simulations. However, it cannot
model the internal structure of the power semiconductor devices in the software. With
the simplified structure in Fig. 5.4, the internal information of the device is unnecessary,
then it is possible to perform the initial simulations in MAXWELL 2D.
The voltage distribution on the JTE is modelled in MAWELL 2D, as compared to
MEDIC results in Fig. 5.5. The origin is at point C, and the curve is reconstructed by 6
segments, as shown in (5.1).
V = 0 V V = f(x)
x
V = 10 kV
Polyimide
yA
B
C D E F
Fig. 5.4. Simplified model for mathematical calculation and simulation.
Distance (mm)
0.00 0.20 0.40 0.60 0.80 1.00
Voltage (
kV
)
5.0
0
2.5
7.5
1
2.5
10
Fig. 5.3. Voltage distribution on JTE.
- 68 -
( )
( )
( )
( )
=
+−=
+−=
+−=
+−=
=
13500
1300037.00599.0
500
1170032.00499.0
1300
780027.00499.0
3900
20017.00999.0
4600
0
6
5
4
3
2
1
V
xV
xV
xV
xV
V
>
<<
<<
<<
<<
<<
mmx
mmxmm
mmxmm
mmxmm
mmxmm
mmx
5.0
43.037.0
37.032.0
32.027.0
27.017.0
17.00
(5.1)
The electric field distribution along the x direction at different y positions in Fig. 5.4 is
shown in Fig. 5.6. The upper dashed- line gives the electric field intensity at the position
of the corner of the device, and the lower dashed- line gives the electric field intensity in
the uniform distribution area, as explained in Fig. 5.7. The error of the MAXWELL
simulation results is define in (5.2) and it is shown in Fig. 5.8.
%100×−
=MEDICI
MEDICIMAXWELLerror (5.2)
Distance (mm)
0.00 0.20 0.40 0.60 0.80 1.00
Voltage (
kV
)
5.0
0
2.5
7.5
1
2.5
10
12
.51
07
.55
.02
.50
Vo
lta
ge
(kV
)
0.11 0.33 0.55 0.990.77 1.1
Distance (mm)
Linear model
Segmental model
(a) (b)
Fig. 5.5. Comparison of the voltage distribution on JTE in MEDICI and MAXWELL. (a) Voltage
distribution in MEDICI. (b) Voltage distribution in MAXWELL.
- 69 -
V = 0 V V = f(x)
x
V = 10 kV
Polyimide
yA
B
C D E F
Fig. 5.7. Explanation of two values in Fig. 5.6.
200
E (
kV
/mm
)
0.2
Distance (mm)
150
50
10
0100
80
60
40
20
0.4 0.6 0.8 1.0
E (
kV
/mm
)
40 kV/mm
34 kV/mm
42 kV/mm
35 kV/mm
8
0E
(kV
/mm
)0.2
Distance (mm)6
02
04
08
06
04
02
00.4 0.6 0.8 1.0
E (
kV
/mm
)
40 kV/mm
34 kV/mm
40 kV/mm
35 kV/mm
50
E (
kV
/mm
)
0.2
Distance (mm)
30
10
20
50
40
10
0.4 0.6 0.8 1.0
E (
kV
/mm
)
40 kV/mm
34 kV/mm
40 kV/mm
35 kV/mm
40
60
20
30
(a) (b) (c)
E (
kV
/mm
)
0.2
Distance (mm)
30
10
20
40
0.4 0.6 0.8 1.0
E (
kV
/mm
)
37 kV/mm
34 kV/mm
37 kV/mm
34.5 kV/mm
40
20
30
E (
kV
/mm
)
0.2
Distance (mm)
30
10
20
10
0.4 0.6 0.8 1.0
E (
kV
/mm
)
35 kV/mm
34 kV/mm
33 kV/mm 34 kV/mm
20
30
35
25
15
E (
kV
/mm
)
0.2
Distance (mm)
30
10
20
10
0.4 0.6 0.8 1.0
E (
kV
/mm
)
33 kV/mm 34 kV/mm
30 kV/mm 34 kV/mm35
20
30
25
15
5
(d) (e) (f)
Fig. 5.6. Comparison of the electric field intensity at different y locations. Upper: MEDIC; Lower:
MAXWELL 2D (a) y = 0. (b) y = 10 µm. (c) y = 50 µm. (d) y = 100 µm. (e) y = 200 µm. (f) y = 300
µm.
Uniform distribution area x position of pink dash line value
- 70 -
From the comparison, it can be seen that the MAWELL simulation results give less
than 10% error compared to the MEDICI simulation. With the simplified model, the
simulation in MAXWELL can be used as initial simulation of the structure and it can
give an overall idea of the structure. More elaborated and accurate simulations need to be
performed in MEDICI.
5.3. Summary
Although the simulations in MEDICI include the internal information of the device,
and it gives an accurate and comprehensive electric field distribution, the simulation is
very time-consuming. In the early stage of design, it is not necessary to spend so long
time to do the simulation. On the contrary, a simplified structure is preferred to give a
simple, quick and clear overall understanding of the package first.
In this Chapter, the proposed HVEP structure is simplified. The device is replaces by a
conductor with 0 V and a conductor with a piece-wise linear voltage distribution in
series. In this simplified structure, only the polyimide area is of interest.
The simplified structure can be applied to perform the simulation in MAXWELL 2D.
After the voltage distribution on the JTEs is modelled, the simulation results in both
MEDICI and MAXWELL are presented. The MAWELL simulation results give less
than 10% error compared to the MEDICI simulation. With the simplified model, the
-10
-8
-6
-4
-2
0
2
4
6
0 10 50 100 200 300
y (um)
Err
or
(%)
Device corner Uniform area
Fig. 5.8. Errors of the simulation results in MAXWELL compared to the results in MEDICI.
- 71 -
simulation in MAXWELL can be used as initial simulation of the structure and it can
give an overall idea of the structure. More elaborated and accurate simulations need to be
done in MEDICI.
- 72 -
Chapter 6: Summary
The high-voltage SiC power semiconductor devices have been developed in recent
years. They cause an urge in the need for the power semiconductor packaging to have
not only low interconnect resistance, less noise, less parasitic oscillations, improved
reliability, and better thermal management, but also High-Voltage (HV) blocking
capability. With the ever-rising electric field intensity in the package caused by the high
voltage, the understanding the distribution of the electric field is becoming crucial to the
design, providing proper structure for high-voltage.
The primary objective of the high-voltage power semiconductor package is to limit the
maximum electric field intensity of the insulation materials to be lower than their
dielectric strength. A good understanding of electric field distribution can help in both
identifying potential dielectric failure and designing appropriate structure improvement
strategies for the package. Simulation tools and experimentation are commonly used to
perform the analysis. Advanced material that has high dielectric strength will also aid in
blocking high voltage of the package.
In this research, a thorough static electric field analysis is performed and an in-depth
understanding of the electric field distribution of the LVEP, MVEP and HVEP. The
objectives discussed in Chapter 1 are accomplished. The summary and contributions of
this research are discussed in the following.
6.1. Contributions
I. Understanding of the electric distribution in the Low-Voltage Embedded Power
(LVEP) package in MEDICI
a. The studies show that two electric field hotspots are at the upper and lower corner
of the device in the LVEP package.
The LVEP structure is revisited with a 10 kV SiC PiN diode. With the internal
structure of the device, the simulation shows two hotspots in both insulation materials are
at the upper and lower corner of the device. The internal structure of the device makes
the bending of the equipotential lines at both corners in the insulation materials.
- 73 -
b. Studies show the electric field distribution of the double-sided metallization LVEP
structure and the impact of the parameters in this structure on the electric field
distribution.
The double-sided metallization is proved to be an effective way to reduce the high
electric field concentration at the lower corner of LVEP structure. The higher thickness
of the polyimide layer can decrease the maximum electric field at the corners of the
device in two insulation material. However, when the thickness is comparable to the
thickness of the ceramic, it is less effective. From the material property point of view,
low dielectric constant of the polyimide is desired.
II. Development of a Medium-Voltage Embedded Power (MVEP) structure in
MEDICI
a. Development of a MVEP structure in MEDICI.
Based on the understanding of the LVEP structure, a MVEP structure is proposed. The
cathode metallization of the MVEP structure is extended to the top of the structure. The
maximum electric field intensity at the corners of the device inside two insulation
material is reduced greatly.
b. Studied show the impact of the parameters on the electric field distribution in the
MVEP structure.
The higher thickness of the polyimide layer can decrease the maximum electric field at
the corners of the device in two insulation material. However, when the thickness
reaches certain value, it is less effective. From the material property point of view, low
dielectric constant of the polyimide is desired. The influence of the thickness of the
polyimide and the properties of the materials is explained by simplifying the structure to
a circuit that consists of two capacitors in series.
III. Development of a High-Voltage Embedded Power (HVEP) structure in MEDICI
a. Development of a HVEP structure in MEDICI.
A thin layer of metal is inserted to the middle of the structure and it is tied to the
cathode metallization. This field-shaping metal layer is a shield to the embedded
insulation material. Thus, the maximum electric field at the corners inside the two
- 74 -
insulation materials is reduced to the value below their dielectric strength. Furthermore,
since the embedded insulation material is shielded by the field-shaping metal layer, even
with the presence of the voids in the embedded insulation material, it won’t degrade the
voltage rating of the package.
b. Studies show the impact of parameters on the electric field distribution in the
HVEP.
The thickness of the polyimide is proportional to its dielectric strength and the applied
voltage. The best position for the field-shaping metal layer is on the same surface as the
device, and the optimal voltage applied this metal layer is 0 V. Simulation also indicates
that the length of the field-shaping metal layer should neither exceed the edge of the
device nor be apart from the edge too much. The best result is obtained when the edge of
the field-shaping metal is right at the edge of the device. It can be found that the angle of
polyimide edge is less tan 90°, and this smaller angle could influence the package voltage
rating greatly. Although this angle is unavoidable, it can be solved by the application if a
thick enough “dam”.
IV. Development of design guidelines for the HVEP package and verification
a. Development of design guidelines for the HVEP structure.
The design procedure of the HVEP package is presented. Based on the simulations and
parameter study, the design guidelines are given. (1) The best position of the field-
shaping metal layer is on the same surface as the device. (2) The best width of the field-
shaping metal layer is when the field-shaping metal layer touches the edge of the device.
(3) The best applied voltage is the same as the cathode voltage. (4) The thickness of the
polyimide layer is determined by the rated voltage and the dielectric strength of the
material from calculation or simulation. (5) A “dam” that has the same thickness as the
polyimide is needed in the package to prevent the natural angle of polyimide caused high
electric field concentration.
b. Development of fabrication process of a HVEP package.
Due to the difference of the HVEP structure and LVEP structure, several steps are
added to the standard LVEP fabrication process and the dimensions are changed
accordingly. The shim soldering, field-shaping metallization and double-sided
- 75 -
metallization deposition, and thicker polyimide layer are the core difference between two
processes. A HVEP package for a 2 kV SiC diode is designed following the given design
guidelines and fabricated following the proposed procedure.
c. Experimental verification of the proposed HVEP structure and the design
guidelines.
Test results show that the fabricated 2 kV SiC diode package can block at least 2 kV.
The Environmental SEM image shows that the fabricated package follows the design
guidelines.
V. Development of a simplified model of HVEP package
a. Simplification of the model of the power semiconductor devices.
The model of power semiconductor devices is simplified to a conductor with difference
voltage on different segments. No more internal information of the device is necessary.
This simplified model can be applied to both mathematical analysis and non-
semiconductor simulator software, such as MAWELL to do quick and simple calculation
or simulation compared to MEDIC which is very time-consuming.
b. Investigation on the feasibility of the application of conformal mapping to the
electric field calculation in the HVEP package.
Conformal mapping is a very useful mathematical theory to calculation field problems.
Theoretically, after a series of the conformal mapping the simplified model of the HVEP
structure can be transformed to a rectangle in which the potential distribution is known.
Then this potential distribution can be inverse transformed back to the original simplified
model and the electric field intensity can be calculated accordingly. However, due to the
limitation of the mathematical software, severe error occurs in both forward and inverse
mapping procedure, which finally fails the effectiveness of the mapping.
c. Comparison of the simulation in MAWELL with the simplified model and that in
MEDICI.
From the comparison, it can be seen that the MAWELL simulation results give less
than 10% error compared to the MEDICI simulation. With the simplified model, the
simulation in MAXWELL can be used as initial simulation of the structure and it can
- 76 -
give an overall idea of the structure. More elaborated and accurate simulations need to be
done in MEDICI.
6.2. Recommendations
New materials. The selection of the insulation materials is important to achieve high-
voltage blocking capacity and better reliability. Although the current design of the HVEP
package is based on the current available insulation material with low dielectric strength,
the selection of the insulation materials should continue to search for emerging insulation
materials that have high dielectric strength.
Electrical performance. Although the package using embedded power technology has
a promising electrical performance of switching and conduction, it is necessary to test the
forward characteristics of the HVEP package since new layer is added to the LVEP
structure. These tests include the in-circuit switching speed, switching and conduction
loss, EMI, etc.
Reliability and thermo-mechanical performance. In the structure, the CTE mismatch of
the copper and the semiconductor will cause thermo-mechanical issues in the package.
The thermo-mechanical stress investigation, thermal performance, and cooling
technology are needed to be studied.
Expandability to HV module package. Embedded power technology can be used to
power semiconductor device package as well as the system package. Theoretically, the
HVEP device package can easily expended to a HV module package. However, issues
include the layout design, thermal problems, EMI, etc.
- 77 -
Reference
[1] R. R. Tummala, E. J. Rymaszewski and A. G. Klopfenstein, Microelectronics
Packaging Handbook, Part II, Semiconductor Packaging, Chapman & Hall, 1996,
pp. 12.
[2] S. Wen and G. Q. Lu, “Finite-element modeling of thermal and thermomechanical
behavior for three-dimensional packaging of power electronics modules,”
Proceeding of the 7th
Intersociety Conference on Thermal and Thermomechanical
Phenomena in Electornic Systems, Vol. II, Las Vegas, Nevada, May. 2000, pp. 303-
309.
[3] A. K. Agarwal, R. Singh, S. H. Ryu, J. T. Richmond, D. C. Capell, S. Schwas, B.
Moore, and J. W. Palmour, “600 V, 1-40 A, schottky diodes in SiC and their
applications,” available:
http://www.cree.com/products/pdf/PWRTechnicalPaper1.pdf
[4] R. Singh, K. G. Irvin, J. T. Richmond, and J. W. Palmour, “High-temperature
performance of 10 kilovolts, 200 amperes (pulsed) 4H-SiC pin rectifiers,” Proc. of
International Conference on Silicon Carbide and Related Materials (ICSCRM) 2001,
Tsukuba, Japan, 2001, pp. 1265-1268.
[5] Y. Sugawara, D. Takayama, K. Asano, R. Singh, J. W. Palmour, and T. Hayashi,
“12-19 kV 4H-SiC pin diodes with low power loss,” Proc. of the 13th
International
Symposium on Power Semiconductor Devices and ICs (ISPSD), Osaka, Japan, Aug.
2001, pp. 27-30.
[6] R. Singh, K. G. Irvine, D. C. Capell, J. T. Richmond, D. Berning, A. R. Hefner, and
J. W. Palmour, “Large area, ultra-high voltage 4H-SiC p-i-n rectifiers,” IEEE
Transactions on Electron Devices, vol. 49, no. 12, pp. 2308-2316, Dec. 2002.
[7] M. K. Das, J. J. Sumakeris, A. Krishnaswami, M. J. Paisley, A. K. Agarwal, and A.
Powell, “Latest advances in high voltage, drift free, 4H-SiC PiN diodes,” Proc. of
the 15th
International Symposium on Power Semiconductor Devices and ICs
(ISPSD), Cambridge, England, Apr. 2003, pp. 364-365.
[8] P. A. Losee, C. Li, Y. Wang, S. K. Sharma, T. P. Chow, I. B. Bhat, R. E. Stahlbush,
and R. J. Gutmann, “10 kV 4H-SiC PiN diodes designed for improved switching
performance using emitter injection control,” Proc. of the 19th
International
Symposium on Power Semiconductor Devices and ICs (ISPSD), Jeju, Korea, May.
2007, pp. 97-100.
[9] S. H. Ryu, S. Krishnaswami, M. O’Loughlin, J. T. Richmond, A. K. Agarwal, J. W.
Palmour, and A. R. Hefner, “10-kV, 123-mΩ⋅cm2 4H-SiC power DMOSFETs,”
IEEE Electron Device Lett., vol. 25, no. 8, pp. 556-558, Aug. 2004.
[10] S. H. Ryu, S. Krishnaswami, B. Hull, J. T. Richmond, A. K. Agarwal, and A. R.
Hefner, “10 kV, 5A 4H-SiC power DMOSFET,” Proc. of the 18th
International
Symposium on Power Semiconductor Devices and ICs (ISPSD), Naples, Italy, Jun.
2006, pp. 1-4.
- 78 -
[11] T. H. Duong, A. Rivera-Lopez, A. R. Hefner, and J. M. Ortiz-Rodriguez, “Circuit
simulation model for a 100 A, 10 kV half-bridge SiC MOSFET/JBS power module,”
Proc. of the 23th Applied Power Electronics Conference and Expostion (APEC),
Austin, Taxes, Feb. 2008, pp. 913-917.
[12] J. Wang, T. Zhao, J. Li, A. Q. Huang, R. Callanan, F. Husna, and A. K. Agarwal,
“Characterization, modeling, and application of 10-kV SiC MOSFET,” IEEE
Transactions on Electron Devices, vol. 55, no. 8, pp. 1798-1806, Aug. 2008.
[13] R. R. Tummala, Fundamentals of Microsystems Packaging, McGRAW Hill, 2001,
pp.346-347.
[14] New packaging concepts for low voltage power MOSFETs lead to performnace
improvemnet in advanced DC-DC converters, ST Microelectronics Application Note
AN2066, Nov. 2004.
[15] H. Kon, K. Nakayama, S. Yanagisawa, J. Miwa, and Y. Uetake, “The 4500 V-750 A
planar gate press pack IEGT,” Proc. of the 10th
International Symposium on Power
Semiconductor Devices and ICs (ISPSD), Kyoto, Japan, Jun. 1998, pp. 81-84.
[16] Y. Uchida, Y. Seki, Y. Takahashi, and M. Ichijoh, “Development of high power
press-pack IGBT and its application,” Proc. of the 22nd
International Conference on
Microelectronics (ICM), Nis, Yugoslavia, May, 2000, pp. 125-129.
[17] S. kaufmann, and F. Zwick, “10 kV IGBT press pack modules with series connected
chips,” Proc. of the 14th
International Symposium on Power Semiconductor Devices
and ICs (ISPSD), Santa Fe, New Mexico, Jun. 2002, pp. 89-92.
[18] S. Eicher, M. Rahimo, E. Tsyplakov, D. Schneider, A. Kopta, U. Schlapbach, and E.
Carroll, “4.5 kV press pack IGBT designed for ruggedness and reliabilty,” Proc. of
the 39th
Industy Applications Conference (IAS), Seattle, Washington, Oct. 2004, pp.
1534-1539.
[19] J. H. Hua, Flip Chip Technologies. New York: McGraw Hill, 1995.
[20] R. R. Tummala, E. J. Rymaszewski, and A. G. Klopfenstein, Microelectronics,
Packaging Handbook. New York: Chapman & Hall, 1999.
[21] A. Bindra, “BGA MOSFETs keep their cool at high power levels,” Electron Design.,
vol. 47, no. 19, pp. 43-45, Sep. 1999.
[22] Z. J. Shen, D. Okada, F. Lin, A. Tintikakis, and S. Anderson, “Lateral discrete power
MOSFET: enabling technology for next-generation, MHz-frequency, high-density
DC/DC converters,” Proc. of the 19th Applied Power Electronics Conference and
Expostion (APEC), Anaheim, California, Feb. 2004, pp. 225-229.
[23] H. Schofield and M. Standing, “Assembly of FlipFETTM
devices,” International
Rectifier (IR) Application Note AN-1011, Dec. 2000.
[24] A. Burstein, “Volterra’s 3rd generation integrated switching regulators for Intel
microprocessor power,” Intel Technology Symposium (ITS), Power Track
Presentation 8, Sep. 2002.
[25] X. Liu, S. Haque, and G. Q. Lu, “Three-dimensional flip-chip on flex packaging for
power electronics applications,” IEEE Transactions on Advanced Packaging, vol.
24, no. 1, pp. 1-9, Feb. 2001.
- 79 -
[26] J. Bai, G. Q. Lu, and X. Liu, “Flip-chip on flex integrated power electronics modules
for high-density power integration,” IEEE Transactions on Advanced Packaging,
vol. 26, no. 1, pp. 54-59, Feb. 2003.
[27] P. Clot, J. F. Zeberli, J. M. Chenuz, F. Ferrando, and D. Styblo, “Flip-chip on flex
for 3D packaging,” Proc. of the 24th
Electronics Manufacturing Technology
Symposium, Austin, TX, Oct. 1999, pp. 36-41.
[28] H. N. Shah, Y. Xiao, T. P. Chow, R. J. Gutmann, and E. R. Olson, “Power
electronics modules for inverter applications using flip-chip on flex-circuit
technology,” Proc. of the 39th
Industry Application Conference (IAS), Chattanooga,
TN, Oct. 2004, pp. 1526-1533.
[29] S. Haque, K. Xing, R. Lin, C. T. A Suchicital, G. Q. Lu, D. J. Nelson, D. Boroyevic,
and F. C. Lee, “An innovative technique for packaging power electronics building
blocks using metal posts interconnected parallel plates structure, ” IEEE
Transactions on Advanced Packaging, vol. 22, no. 2, pp. 136-144, May, 1999.
[30] S. haque, K. Xing, G. Q. Lu, D. J. nelson, D. Boroyevic, and F. C. Lee, “Packaging
for thermal management of power electronics building blocks using metal posts
interconnected parallel plate structure,” Proc. of the 6th
Intersociety Conference on
Termal and Thermomechanical Phenomena in Electronic Systems, Seattle, DC, May.
1998, pp. 392-398.
[31] S. Wen, D. Huff, and G. Q. Lu, “Dimple-array interconnect technique for packaging
power semiconducotr devices and modules,” Proc. of the 13th
International
Symposium on Power Semiconductor Devices and ICs (ISPSD), Osaka, Japan, Jun.
2001, pp. 69-74.
[32] R. Fisher, R. Filion, J. Burgess, and W. Hennessy, “High frequency, low cost, power
packaging using thin film power overlay technology,” Proc. of the 10th
Applied
Power Electronics Conference and Exposition (APEC), Dallas, TX, Mar. 1995, pp.
12-17.
[33] B. Ozmat, C. S. Korman, P. McConnelee, M. Kheraluwala, E. Delgado, and R.
Filion, “A new power module packaging technology for enhanced thermal
performance,” Proc. of the 7th Thermal and Thermomechanical Phenomena in
Electronic System (ITHERM), Las Vegas, NV, May, 2000, pp. 287-296.
[34] B. Jayant Baliga, Power Semiconductor Devices, Boston, MA: PWA publishing
company, 1996.
[35] P.A. Tove, “Methods of avoiding edge effects on semiconductor diodes,” J. Phys. D:
Appl. Phys., 15, pp. 517-536, 1982.
[36] P. Brüesch, Th. Stockmeier, and F. Stucki, P. A. Buffat, “Physical properties of
semi-insulating polycrystalline silicon: I. Structure, electronic properties, and
electrical conductivity,” Journal of Applied Physics, Vol. 73, no. 11, pp. 7677-7689,
Jun. 1993.
[37] D. Frey, J. L. Schanen, J. L. Auge, O. Lesaint, “Electric field investigation in high
voltage power modules using finite element simulations and partial discharge
measurements,” Proc. of the 38th
Industry Applications Conference (IAS), Salt Lake
City, UT, Oct. 2003, pp. 1000-1005.
- 80 -
[38] D. Frey, J. L. Schanen, J. L. Auge, O. Lesaint, “Electric field investigation in IGBT
power modules,” International Conference on Solid Dielectrics (ICSD), Toulouse,
France, Jul. 2004, pp. 864-867.
[39] http://www.synopsys.com/
[40] Z. Liang, F. C. Lee, G. Q. Lu, D. Borojevic, “Embedded power – a multilayer
integration technology for packaging of IPEMs and PEBBs,” International workshop
on Integrated Power Packaging (IWIPP), Waltham, MA, Jul. 2000, pp. 41-45.
[41] Z. Liang, J. D. van Wyk, “Functional integration technology,” Proc. of the 20th
Applied Power Electronics Conference and Exposition (APEC), Austin, TX, Mar.
2005, pp. 375-381.
[42] Z. Liang, J. D. van Wyk, “Planar metallization interconnected 3-D multi-chip
module,” Proc. of the 53rd
Electronic Components and Technology Conference
(ECTC), New Orleans, LA, May. 2003, pp. 1090-1094.
[43] P.A. Losee, C. Li, Y. Wang, S.K. Sharma, T.P. Chow, I.B. Bhat, R.E. Stahlbush, R.J.
Gutmann, “l0kV 4H-SiC PiN diodes designed for improved switching performance
using emitter injection control,” Proc. of the 19th
International Symposium on Power
Semiconductor Devices and ICs (ISPSD), Jeju, Korea, May. 2007, pp. 97-100.
[44] http://en.wikipedia.org/wiki/Engineering_design_process.
[45] Y. Pang, “Assessment of thermal behavior and development of thermal design
guideliens for integrated power electronics modules,” Ph.D. dissertation, Dept.
Mechanical Engineering, Virginia Polytechnic Institute and State University, VA,
2005.
[46] Datasheet, available:
http://www.epotek.com/SSCDocs/datasheets/600%20CFDS.PDF.
[47] P. K. Kythe, Computational Conformal Mapping, Boston: Birkhauser Boston, 1998.
[48] V.I. Ivanov, M.K. Trubetskov, Handbook of conformal mapping with computer-
aided visualization, Boca Raton : CRC Press, 1994.
[49] R. Schinzinger and P. A. A. Laura, Conformal mapping : methods and applications,
New York: Elsevier Science Pub. Co., 1991.
[50] M. R. Spiegel, Schaum's outline of theory and problems of complex variables, with
an introduction to conformal mapping and its applications, New York, McGraw-Hill
Book Co. 1964.
[51] R. Courant, Dirichlet's principle, conformal mapping, and minimal surfaces. With an
appendix by M. Schiffer, New York, Interscience Publishers, 1950.
[52] Nehari, Zeev, Conformal mapping, New York, McGraw-Hill, 1952.
[53] L. Bieberbach, Conformal mapping; translated by F. Steinhardt, New York, Chelsea
Pub. Co., 1953.
[54] T. A. Driscoll, and L. N. Trefethen, Schwarz-Christoffel mapping, New York :
Cambridge University Press, 2002.
[55] M. Walker, Conjugate functions for engineers; a simple exposition of the Schwarz-
Christoffel transformation applied to the solution of problems involving two-
dimensional fields of force and flux, London, Oxford University Press, 1933.
- 81 -
Appendix Mathematical Modeling of the Simplified Model of
HVEP
The electric field calculation is quite simple and straight-forward when the structure
and the material are simple enough. However, the voltage distribution of the simplified
model shown in Fig. 5.4 is irregular. Hence the electric field cannot be calculated
directly. Some further mathematical manipulation is needed.
A1 Conformal Mapping and Schwarz-Christoffel Transformation
A conformal mapping is a transform of geometric figures in which infinitely small
pieces are transformed into similar ones [47]-[55]. A map is called conformal at z0 if it
preserves oriented angles between curves through z0, as well as their orientation. A
Schwarz-Christoffel mapping is a transformation of the complex plane that maps the
upper half-plane conformally to a polygon. A polygon in w plane shown in Fig. A.1 (a)
has vertices at w1, w2,…, wn with corresponding interior angles a1, a2, …, an, respectively.
The points w1, w2, …, wn are mapped into points x1, x2, …, xn on the real axis of the z
plane shown in Fig. A.1 (b), respectively. A transformation which maps the interior R of
the polygon of the w plane on to the upper half plane R’ of the z plane and the boundary
of the polygon on to the real axis is given by (1) and (2),
( ) ( ) ( ) 11
2
1
1121 −−−
−−−=πππ na
n
aaxzxzxzA
dz
dwL (1)
( ) ( ) ( )( ) BdzxzxzxzAw na
n
aa+−−−= ∫
−−− 11
2
1
1121 πππ
L (2)
where A and B are complex constants.
- 82 -
Consider a rectangle in the W plane, which is symmetrical with respect to the imaginary
axis, as shown in Fig. A.2 (a), it can be conformally mapped onto the upper-half plane.
The boundary points B, 0, C = -b, 0, b are mapped into the boundary points -1, 0, 1 in the
R plane respectively. Let the image of the point E be a point of the real axis 1/k; due to
the principle of symmetry the image of the point A is the point -1/k. From the Schwarz-
Christoffel transformation, the function realizing the inverse mapping of the upper-half
plane onto the rectangle is:
( )( )∫−−
=r
k
dCw
0 2221
11 ξξ
ξ (3)
iH
b-b 0
C
EA
B
u
v
W Plane
C EA B
-1 1 1/k-1/k
t
s
R Plane
(a) (b)
Fig.A.2. Elliptic function of first kind. (a) Rectangle (original plane) (b) Upper-half plane (destination
plane)
u
v
w1
w2
w3
w4
a1a2
a3
a4
R
x
y
x1 x2 x3 x4
R’
(a) (b)
Fig. A.1. Schwarz-Christoffel transformation. (a) Polygon (original plane) (b) Upper-half plane
(destination plane)
- 83 -
The function
( )( )( )∫
−−=
r
m
dCmrF
0 221
11,
ξξ
ξ (4)
is called the elliptic integral of the first kind. The number m is called the parameter and
the number mk = is called the modulus of the elliptic integral. Provisionally suppose
that 0 < m < 1.
The value
( )( )( ) ∫∫
−=
−−=
2/
0 2
1
0 22 sin111,1
s
m
d
mss
dsmF
ϕ
ϕ (5)
is called the complete elliptic integral of the first kind and is designated as K(m) or
simply K. Thus the segment (-1, 1) is mapped into the segment (-K, K) by this function.
The point mkE /1/1 == in the R plane is mapped into the point
( )( )( ) ( )( )∫∫
−−+
−−==
k
mss
dsi
mss
dsmkFw
/1
1 22
1
0 22 1111,/1 (6)
The first term in this formula is the complete elliptic integral and the second term is
transformed by the change of the variable
1' 22 =+ ymms (7)
where mm −= 1' is the so-called complementary parameter of the elliptic integral, and
( )( ) ( )( )∫∫−−
=−−
1
0 22
/1
1 22 '1111 ymt
dt
mss
dsk
(8)
For brevity the value of the complete elliptic integral of the first kind of the
complementary parameter K(m’) is designated as K’. Therefore, the point
mkx /1/12 == is mapped into the point K+iK’, and the whole upper-half plane is
mapped onto the rectangle with the base 2K and the height K’.
The parameter m is unknown when searching for the function (3). It can be found from
the equation
( )( ) b
H
mK
mK=
−1 (9)
where H is the height of the rectangle in the w plane, and b is the half width of the
rectangle, as shown in Fig. A.2 (a).
- 84 -
Then the constant C1 is determined as
( )mK
bC =1 (10)
where C1 is the constant in the integration in (3).
A2 Model of HVEP package
Neither the equipotential lines nor the electric field can be mathematically plotted in the
simplified model in Fig. 5.4. It is desired to have a structure that either the electric field
distribution or the equipotential lines can be easily expressed or calculated. For example,
Assume that the aspect ration (AC/CD) is very small, then the equipotential lines can be
approximately straight, as shown in Fig. A.4.
Thus, the strategy in this section is to convert the structure shown in Fig. A.3 (a) to a
structure similar to Fig. A.3 (b) through a series of conformal mapping. Then, the easily
calculated potential distribution information can be transformed back to the original
V = 0 V
V = f(y)
x
V = 10 kV
Polyimide
y
A
B C
D
E
V = g(ξ)
ζ
V = 10 kVPolyimide
ξ
V = 0 V
A
C D
E
(a) (b)
Fig.A.3. Transformation of the simplified structure. (a) Simplified structure (rotate 90°). (b) Transformed
structure.
- 85 -
structure through the reverse transformation. Finally, both the potential and electric field
distribution in the original structure are known.
Through the elliptic integral of the first kind and its inverse mapping, the Elliptic
Jacobian functions, a rectangle and the upper-half plane can be transformed between each
other. Thus, this transformation is done firstly, as shown in Fig. A.5 (a) and (b). The
parameter 2km = is defined approximately as the function of the ratio
( ) ( ) bHmKmK //' = by mean of interpolation based on the dimensions of the original
structure.
The destination rectangle in ξ-ζ plane is shown in Fig. A.5 (d) and its corresponding
upper-half plane after the transformation are in Fig. A.5 (c). Thus there is one more
transformation is necessary to map the upper-half plane Fig. A.5 (b) to the upper-half
plane in Fig. A.5 (c). A linear-fractional transformation can perform this.
The transformation
δγ
βα
+
+=
z
zf , 0≠− βγαδ (11)
is called linear-fractional transformation. It can be considered as combinations of the
transformation of translation, rotation, stretching and inversion. This transformation
maps circle in z plane into circles in the w plane.
If z1, z2, z3 and z4 are four distinct points in the z plane, then the quantity
( )( )
( )( )13
23
24
14
zz
zz
zz
zzL
−
−
−
−= (12)
V = g(ξ)
ζ
V = 10 kV
ξ
V = 0 V
A
C D
EEquipotential lines
Fig.A.4. Equipotential lines approximation
- 86 -
is called the cross ratio of z1, z2, z3 and z4. This ratio is invariant under the
transformation, and this property can be used in obtaining specific transformation
mapping three points into three other points.
Thus, the destination rectangle can be mapped from the original rectangle through the
Elliptic Jacobian transformation, linear-fractional transformation, and elliptic integral of
the first kind.
A3 The problem
u
v
A
B C
D
E
-b b
H
d0
t
s
-1/k -1 +1 d1 1/k
A B C D E
(a) (b)
n
m
-1/p -1 +1b2 1/p
A B C D E
ξ
ζ
A
B
C D
E
-l l
G
b3
(c) (d)
Fig.A.5. Transformations. (a) Original rectangle (u-v). (b) Upper-half plane after Elliptic Jacobian
transformation (s-t). (c) Upper-half plane after linear-fractional transformation (m-n). (d) Destination
rectangle after elliptic integral of the first kind (ξ-ζ).
- 87 -
As mentioned in the previous section, the parameter m comes from the interpolation,
and there is an assumption that the aspect ratio of the destination rectangle is small.
However, it is found that the aspect ratio of the destination rectangle is about 1.5 when
the dimensions of the structure in Chapter 5 are applied.
Fig. A.6 shows the variation of the aspect ratio of both original rectangle and the
destination rectangle. It can be seen that in the whole range of k, the aspect ratio of the
destination rectangle is always greater than 1. It is also observed that when k is very
close to 1, the aspect ratio can be close to 0. Thus, in the last transformation, which is the
mapping from the upper-half plane to the destination rectangle, the p parameter has to be
very close to 1 to make the aspect ratio of the destination rectangle to be small.
Therefore, one more transformation has to be added to this mapping.
An exponential transformation can be applied. The exponential function of the
complex variables is defined as
yieyeeeew xxiyxz sincos +=== (13)
The aspect ratio of the destination rectangle is reduced to 0.2 after this transformation is
added. However, from the nature of the exponential function, it amplifies more on the
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10
0.5
1
1.5
2
2.5
3
k
Asp
ect ra
tio
G/2l
H/2b
Fig.A.6. Variation of aspect ratio with the parameter k.
- 88 -
large numbers than the small numbers. Due to the limitation of the mathematical
software, MATLAB, it cannot identify these amplified large numbers individually. This
causes big errors in the both forward mapping and inverse mapping. As shown in Fig.
A.7, point s1 is a small number, and points s2, s3, and s4 are large numbers. After the
exponential transformation, point s1 is mapped into point p1 in the p-q plane. However,
because the mapped points of s2, s3 and s4 are too large, the software recognizes them as
one point in the p-q plane, infinity. After the linear-fractional transformation and the
elliptic integral of the first kind are done, point s1 is mapped into a point somewhere on
the boundary of the destination rectangle, points s2, s3 and s4 are mapped to the same
point, iG. This wrong forward mapping gives the wrong information to the inverse
mapping as well. Points s2, s3, and s4 are merged to one point back to the original
rectangle after the inverse mapping. This mapping error is caused by the software and it
is inevitable.
Theoretically, the electric field distribution of the simplified structure shown in Fig. 5.4
can be calculated through a series of conformal mapping mathematically, however, the
limitation of the mathematical software causes inevitable error during the calculation,
which fails the mapping.
s1 s2 s3 s4
s
t
p1
p2
p3
p4
p
q
inf
(a) (b)
m1
m2
m3
m4
m
n
inf
ξ1+iζ1
ξ
ζ
iζ2 iζ3 iζ4iG
(c) (d)
Fig. A.7. Error generation in the forward mapping. (a) upper-half plane in s-t plane. (b) p-q plane after
exponential transformation. (c) m-n plane after linear-fractional transformation. (d) Destination rectangle
in ξ-ζ plane.