Você está na página 1de 11

IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 62, NO.

2, FEBRUARY 2015 423

Modeling of Wide Bandgap Power


Semiconductor DevicesPart I
Homer Alan Mantooth, Fellow, IEEE, Kang Peng, Student Member, IEEE,
Enrico Santi, Senior Member, IEEE, and Jerry L. Hudgins, Fellow, IEEE
(Invited Paper)

Abstract Wide bandgap power devices have emerged as an frequency of operation. This has created a growing need for
often superior alternative power switch technology for many power device technologies that can deliver high temperature,
power electronic applications. These devices theoretically have high-power density, and high-frequency operation. For
excellent material properties enabling power device operation at
higher switching frequencies and higher temperatures compared example, a variety of applications in the aircraft, automotive,
with conventional silicon devices. However, material defects and energy exploration industries require power conversion
can dominate device behavior, particularly over time, and this systems to operate at an ambient temperature significantly
should be strongly considered when trying to model actual > 200 C, far beyond Si material limits. Consequently, a new
characteristics of currently available devices. Compact models generation of so-called wide bandgap semiconductor devices
of wide bandgap power devices are necessary to analyze and
evaluate their impact on circuit and system performance. has emerged as viable replacements for the current Si-based
Available compact models, i.e., models compatible with circuit- power devices.
level simulators, are reviewed. In particular, this paper presents A substantial amount of research and development activity
a review of compact models for silicon carbide power diodes and has occurred over the past 20 years in the silicon carbide (SiC)
MOSFETs. and gallium nitride (GaN) fields. These efforts have deliv-
Index Terms Gallium-nitride (GaN), modeling, power device ered several classes of power semiconductor devices, with
modeling, power semiconductor devices, silicon-carbide (SiC), voltage ratings from 30 V to 15 kV and growing, for a wide
wide bandgap. variety of applications. Circuit and system researchers have
collaborated in parallel to demonstrate the efficacy of these
I. I NTRODUCTION
device technologies in demonstration designs ranging from

P OWER electronics can be defined as the application of


solid-state electronics to condition, control, and convert
electric power. With increasing concern for energy delivery
laptop chargers (200 V) to plug-in hybrid electric vehicle
battery chargers (1200 V) to fault current limiters for the
electric power grid (10 kV). The ability of these devices to
and environmental protection, power electronics is playing an switch at higher frequencies than their Si counterparts has
increasingly important role in human society. Power semi- led to system-level benefits in volume or efficiency even at
conductor devices are the core solid-state components in operational temperatures where Si is suitable.
the overall power conversion system, and they most often At present, SiC and GaN are the most promising among
consume the largest portion of power losses in that system. all wide bandgap semiconductor materials [1]. Table I [2]
Therefore, the development of power semiconductor devices compares the material properties of Si, SiC, and GaN. The SiC
has been the driving force in the progress of power elec- and GaN have almost a three times larger bandgap (3 eV)
tronics systems. Recently, silicon (Si)-based devices have compared with Si (1 eV). The breakdown electric field of
dominated the power device market due to mature and well- SiC and GaN is one order of magnitude higher than that
established fabrication technology for Si. However, there of Si. The higher breakdown electric field enables the design
have been applications where Si power electronics have not of wide bandgap power devices with thinner and higher doped
made the inroads necessary for widespread deployment due voltage-blocking layers. For unipolar power devices, this can
to insufficient thermal capabilities, voltage breakdown, or yield a lower ON-state voltage drop and conduction loss. For
bipolar power devices, this can lead to a shorter switching
Manuscript received August 12, 2014; revised October 20, 2014; accepted
October 23, 2014. Date of publication November 26, 2014; date of current time and lower switching loss. The high thermal conductivity
version January 20, 2015. This work was supported by the Office of Naval of SiC, together with the large bandgap, allows SiC-based
Research, Arlington, VA, USA, under Grant N00014-08-1-0080. The review devices to operate at temperatures easily exceeding 200 C.
of this paper was arranged by Editor N. Ohtani.
H. A. Mantooth is with the University of Arkansas, Fayetteville, AR 72701 All of these properties make wide bandgap semiconductor
USA (e-mail: mantooth@uark.edu). devices a promising alternative to Si-based devices.
K. Peng and E. Santi are with the Department of Electrical Engineer- Substantial improvements in GaN and SiC material quality
ing, University of South Carolina, Columbia, SC 29208 USA (e-mail:
pengk@email.sc.edu; santi@engr.sc.edu). over the past several years have led to the availability of
J. L. Hudgins is with the Department of Electrical Engineering, University commercial SiC and GaN power devices. In particular, SiC
of NebraskaLincoln, Lincoln, NE 68588 USA (e-mail: jhudgins2@unl.edu). Schottky diodes, MOSFETs, and junction gate FETs (JFETs)
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org. have reached the market. Although GaN-based devices can
Digital Object Identifier 10.1109/TED.2014.2368274 theoretically offer better performance than SiC, the lack
0018-9383 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
424 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 62, NO. 2, FEBRUARY 2015

TABLE I turn-OFF thyristors, and IGBTs, were modeled in addi-


M ATERIAL P ROPERTIES tion to bipolar junction transistors (BJTs) and SiC junction
transistors [6].
This paper presents a survey of recent progress on
SiC- and GaN-based power semiconductor devices along with
their device models. In Section II, an overview of device mod-
eling requirements is given, and five different device model
levels are described. An SiC power devices are described
in Section III, while GaN power devices are presented
in Section IV. Throughout this paper, a description of reli-
ability issues associated with these wide bandgap devices is
provided. Some assertions regarding outstanding issues and
future trends of power device modeling are provided in the
of good-quality substrates hinders the development of high concluding section.
voltage vertical GaN power devices. However, the interest in
GaN power devices by industry is increasing.
II. C OMPACT D EVICE M ODEL R EQUIREMENTS
Nevertheless, material defects are a root cause of device
failures. These defects and corresponding secondary effects The basic objective in compact device modeling is to
are exacerbated by additional energy from high electric fields, achieve a predictive description of the current flow through
large current densities, increasing temperature, and interface the device as a function of the applied voltages and currents,
stress of material layers. For example, nonmicropipe defects in environmental conditions, such as temperature and radiation,
the bulk and epitaxial layers of SiC have been reported to limit and physical characteristics, such as geometry, doping levels,
the current and voltage ratings in devices and have contributed and so on. Understanding this objective and the mathematical
to degraded performance and failure [3]. The bottom line formulation that is consistent with the numerical algorithms
question is whether SiC and GaN can meet or exceed Si used by circuit simulators to analyze circuits and systems,
Failure In Time (FIT) (109 failures per device operation hour) is the first step in understanding how to create compact
levels to take advantage of a projected significant increase models that are predictive of device behavior and also run at
in power density in wide bandgap devices. This improved computationally attractive speeds. For device models, at least
power density typically means higher operating temperatures, a good approximation to the actual relationship of the elec-
greater thermal cycling, higher impressed electric fields - trical variables needs to be obtained. A compromise between
all in opposition to device reliability. Yet presently, some computational speed and model accuracy is usually made. The
second-generation SiC Schottky diodes are reported to have required accuracy and simulation time are crucial factors con-
a failure rate of < 0.15 ppm. In addition, some SiC junction- sidered by device model designers when making this tradeoff.
barrier Schottky (JBS) diodes have achieved a failure rate A very simple device model generally provides fast simulation
of <0.5 failures per billion of operational hours (0.5 FIT) [4]. speed but loses physics insight into power device behavior and
Meanwhile, looking at comparison with silicon devices, simulation accuracy. In contrast, a very complicated physics-
Si insulated gate bipolar transistor (IGBT) modules are around based model is usually preferred for accurate simulation of
tens to a few FIT. Gate drivers now have a higher FIT device behavior, but is very time-consuming and not very
(and, therefore, worse reliability) than Si IGBT devices, and suitable for circuit simulation.
Si Integrated Gate Commutated Thyristor (IGCT) operation in The predictive nature of a good compact model varies
rail inverter applications has indicated > 45 years mean time depending on the intended use of the model. In the case
between failures and an associated low FIT [5]. Therefore, of power semiconductor device models, the intended use is
SiC diodes have been shown to meet the reliability levels of typically as a power switch in power electronics applications.
Si; but because material defects can dominate device behavior, Therefore, it is not necessary to model first derivative
particularly over time, this should be strongly considered when characteristics, such as device transconductance and output
trying to model actual characteristics of wide bandgap devices, conductance, to the degree of accuracy expected in a Berkeley
particularly devices other than diodes. short-channel insulated gate FET model for analog integrated
The commercially available wide bandgap semiconductor circuit design. The primary considerations are predicting the
power devices indicate that a significant breakthrough in power dc I V characteristics of the device along with switching
electronics has been achieved over the past decades. As power characteristics resulting from attributes, such as device capaci-
devices have progressed, power device simulation is becoming tances, carrier lifetimes, and conductivity modulation of lightly
an increasingly significant method for studying device perfor- doped, voltage-blocking regions.
mance. Significant work has been devoted to the creation and A basic device model selection procedure is shown in Fig. 1.
validation of compact device models for the purpose of power As a first step, the device model requirements are determined
electronic circuit simulations. As many of the SiC devices based on the balance between model accuracy and simulation
were created, researchers developed SiC models, starting with time. According to the device modeling requirements, an
diodes and static induction transistors and then proceeded appropriate device model is chosen for characterization. Static
to MOSFETs and JFETs. Others, such as thyristors, gate I V and CV measurements are performed to extract model
MANTOOTH et al.: MODELING OF WIDE BANDGAP POWER SEMICONDUCTOR DEVICESPART I 425

1) Behavioral Models: These models simulate power


devices without considering their physical mechanism
of operation, and usually they are implemented using
mathematical fitting methods. Their relevant model para-
meters have no direct physical meaning. The advantage
of these behavioral models is that they are simple and
less time-consuming, although their accuracy is low,
Fig. 1. Device model selection procedure: measurements, parameter extrac- especially when the device operates under conditions
tion, and model validation. different from those used for parameter curve fitting.
2) Semiphysics Models: Such models are partly based on
parameters that determine the accuracy of the device model. device physics. For simplicity, in many cases, the stan-
Pulsed I V measurements are used to avoid self-heating dard low voltage device models, which are available in
issues. Based on these measurements, model parameters are circuit simulation tools, such as SPICE and SABER, are
extracted to achieve a best fit to the measured data. Model val- adapted to address high-voltage power device modeling.
idation is performed to check the effectiveness of the resulting Therefore, the physical meaning of some device model
device model. For power devices, a double-pulse test circuit is parameters and equations might be lost. Some behavioral
often used to validate the dynamic behavior of device models equations may also be included in these semiphysics
and is sometimes used for device characterization. Based on models.
the validation against experimental results, the selected model 3) Physics-Based Models: These models are based on
may be either accepted or rejected. If the device model is semiconductor physics. The description of electrical
rejected, a new device model needs to be chosen; and the and thermal behavior is obtained by solving physics
same process described above must be repeated. equations with some simplifications.
High computational speed and accuracy are required for 4) Seminumerical Models: These models are in between
compact models of power semiconductor devices. For the physics-based and numerical models. For example, for
development of power device models, it is critical that certain some power devices, especially bipolar power devices,
prominent device features be captured in the model if they carrier distribution in different regions of the devices is
dominate the static or switching device performance. For determined by the ambipolar diffusion equation (ADE)
example, MOS capacitances have significant impact on the under high-level injection. The solution to the ADE is
dynamic behavior of devices with isolated gates (MOSFET, obtained using a numerical method, such as Fourier
IGBT, and MOS-controlled thyristor). These inherent capaci- series, Laplace transformation, or other methods, so
tances vary strongly with device terminal voltages. In addition, that the spatial carrier distribution is calculated by the
for bipolar power devices (p-i-n diodes, IGBT, BJT, and model.
thyristor), conductivity modulation is a key effect that reduces 5) Numerical Models: Numerical simulation tools
the resistance of the thick, lightly doped semiconductor layer. (SILVACO, Sentaurus TCAD, MEDICI, etc) are
During its conduction period, the large injected carrier con- available for power devices and circuit simulation [10].
centration exceeds the background doping concentration; and The 2-D numerical simulation greatly contributes to
the resistance of the region is reduced dramatically. Therefore, the development of power devices. These numerical
the conductivity modulation effect should be considered for models can provide very accurate results, but they are
bipolar device modeling. Due to high power loss, power complicated, computationally intensive, and require
devices heat up during operation. The device characteristics are detailed information on material properties and device
typically a strong function of junction temperature, so that an geometry.
accurate model must capture this relationship. For this reason, The most commonly used levels are behavioral and
electrothermal device models are sometimes used, while static physics-based models. Ideal switch models are often used
thermal models (i.e., device characteristics are temperature to simulate overall circuit behavior when perhaps comparing
dependent, but temperature is constant during simulation) circuit topologies. Once the topology is selected and detailed
are sufficient for most instances. Moreover, breakdown in converter design begins, a physics-based model is soon desir-
power devices might occur, typically under extreme tran- able when component interactions, switching waveforms, and
sient conditions. The most common breakdown mechanism thermal management considerations need to be addressed. Any
is avalanche breakdown, due to impact ionization; but punch- model-based design flow will employ different device models
through breakdown and Zener breakdown are also possible in at different stages in the design process [11].
some cases [7]. The process of developing a physics-based model is shown
Depending on the modeling method used, available mod- in Fig. 2. The initial phase involves device characterization
els of power semiconductor devices can be categorized into through experimental evaluation of samples along with numer-
five different levels, starting with the simplest behavioral ical device simulations of a representative device structure.
models and then moving to more complex physics-based Through the analysis of these results, model equations are
models [8], [9]. Assuming such models exist, then the model formulated from physics principles of charge transport and
selection procedure of Fig. 1 can be exercised to obtain the storage, field effects, junctions, depleted regions, interface
appropriate model for the task. traps, and so on. These relationships are simplified to a
426 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 62, NO. 2, FEBRUARY 2015

imperfections in SiC material. Stacking faults originate from


the BPD and have been shown to cause an increase in
the forward voltage, under constant current, in bipolar SiC
devices and an increase in reverse-biased leakage current, and
a decrease in forward-biased conduction current in unipo-
lar devices. The main source of BPDs in the epitaxial
layer is from propagation of BPDs present in the substrate.
At present, the density is < 10 cm2 due to improvements
in bulk and epitaxial growth [19]. The TED are usually
1-D defects on the wafer surface that occur by insertion or
removal of an extra half plane of atoms between two lattice
atomic planes. Typically, the density of these defects is several
Fig. 2. Model development methodology for physics-based models (Level 3). thousands per cm2 . Diodes with higher TED densities have
higher leakage current and slightly lower breakdown voltage
compared with devices without dislocations [20].
lumped version or to a 1-D version for computational speed.
Overcurrent during the ON-state causes an increase in the
Sometimes 2-D effects are captured in a simplified manner,
recombination energy being absorbed that then creates and fur-
giving rise to so-called 1.5-D models. A large-signal model is
ther extends existing material defects. These increased defects
constructed and the equations parameterized based on observ-
lead to more charge carrier recombination, thus delivering
ability from experimental measurements. The model is then
more energy to the crystal and increasing the effective forward
implemented in the appropriate format for the target simulator.
voltage drop (increasing the power dissipated in the device).
Modeling tools such as ModLyng are very useful at this stage
In particular, BPDs expand into triangular-shaped stacking
because they allow for graphical construction of the model
faults [21]. The 48 off-orientation of the crystallographic
and subsequent code generation into a variety of formats for
c-axis for the epitaxial layer can cause BPD growth that is
various simulators, such as Saber, HSpice, and Spectre [12].
inclined to the surface, originating from the epilayer, and sub-
Then, the model is parameterized by fitting it to the measured
strate interface with a growing stacking fault. For a SiC BJT,
data through model parameter extraction and then simulating
interface states near the emitter act as recombination sites and
it under test configurations, such as clamped-inductive load,
affect the current gain. Stacking faults grow from the surface
to compare the measured with the simulated results. If the
and degrade the current gain by up to 50% after a few hours
requisite effects are captured, then the model is complete.
of operation. This can be mitigated by proper processing of
If they are not, then the model must be enhanced and part
surface passivation layers [21]. Failure of Schottky diodes has
of the process repeats.
been reported from dv/dt stress that causes breakdown due
to excess charge generation in the space charge region, under
III. SiC P OWER D EVICES AND T HEIR M ODELS reverse bias, in an area with dislocation defects [22]. The JBS
Due to the significant improvements in SiC material diode structures have been shown to handle surge current den-
growth technology, SiC-based power devices are available for sities at least twice as high as for Schottky diodes [23], [24].
high power, high temperature, and high frequency applica- However, operation of JBS diodes has shown an increased
tions. Micropipes in SiC have been nearly eliminated (now forward voltage over time due to stacking faults increasing
< 2.5 per cm2 ). Other defects are likely to be resolved within charge carrier recombination and impeding charge flow, just
several years. These include threading screw dislocation (TSD) as in other SiC bipolar devices [21].
(TSD density is currently 1000 cm2 ) [13], basal plane Problems exist with aluminum (Al) at operating tempera-
dislocation (BPD) (BPD density is currently a few to tens tures > 200 C [4]. There is a phenomenon termed interlayer
of cm2 ) [14], and threading edge dislocation (TED) (TED dielectric erosion (IDE) where Al electrodes erode and pene-
density is presently 1001000 s of cm2 ) [15]. Comparatively, trate the underlying interlayer dielectric and finally reach the
GaN can have bulk defect densities exceeding 106 cm2 . poly-Si gate contact, resulting in a gatesource short-circuit.
Currently, several manufacturers have produced SiC diodes The erosion progression depends on the temperature and the
that have a FIT of < 0.5 [16]. However, all non-Si devices are quality of the silicate glass. To prevent IDE, Si doped (1 wt%)
conduction-current limited in their ratings because of material Al can be used to effectively reduce dissolution and movement;
quality. or use of a barrier metal can prevent the movement of Al into
Substrate screw dislocations can propagate into an adjoining the underlying material. An example of this is to deposit 50 nm
epilayer depending on the epitaxial growth method. One dislo- of tantalum (Ta) and 150 nm of tantalum nitride as a barrier
cation has been shown to reduce the breakdown voltage rating dielectric over the silicate glass interlayer.
by 5%35% and soften the breakdown characteristic. Highly Another problem is Al spearing in poly-Si. The Al spears
localized current may concentrate in the dislocation resulting (or spikes) are observed in the gate located between the
in microplasmas. The defects may also cause a reduction Al-interconnect and the gate after storage at an elevated
in the charge carrier lifetime. The density of these types of temperature. This is caused by massive movement of Al of
dislocation defects is currently 10003000 cm2 [17], [18]. the gate interconnecting with the poly-Si gate. The Al spears
Basal plane defects (BPD) are one of the troublesome extend to the channel region along the poly-Si gate and
MANTOOTH et al.: MODELING OF WIDE BANDGAP POWER SEMICONDUCTOR DEVICESPART I 427

TABLE II
P UBLISHED SiC P OWER D IODE M ODELS

Fig. 3. Cross-sectional views of (a) SiC Schottky diode, (b) SiC p-i-n diode,
and (c) SiC hybrid diode.

compromise the gate oxide, causing a dielectric breakdown


failure and finally a gate-to-source short circuit.
Nickel silicide (Ni2 Si) contacts have low resistance and
are used as ohmic contacts on n-type 4H-SiC. A com-
mon technique for fabricating these contacts is to deposit
a thin nickel (Ni) film followed by rapid thermal annealing
at 950 C1050 C. Surface segregation of carbon due to
this annealing causes the upper metallization layer to sep-
arate from the semiconductor, thus delaminating the entire
contact. To avoid this reaction, the surface segregated carbon
is removed before the metallization process can be performed.
However, delamination can still occur probably due to move-
ment of carbon back to the surface over time. Ni2 Si contact
degradation also occurs when the NiSi-based ohmic contacts
interact with the Al interconnects due to Al reacting with
Ni2 Si, decomposing it and forming NiAl and Si.

A. SiC Power Diodes


The power diode is a key component in power electronics
applications. Basically, SiC power diodes can be classified into
three types: 1) SiC Schottky diode; 2) SiC p-i-n diode; and
3) SiC merged p-n-Schottky diode (MPS), also called JBS.
1) SiC Schottky Diodes: The Schottky diode is a very
attractive unipolar device formed by an electrically
nonlinear contact between a metal and a semiconductor
bulk region, as shown in Fig. 3(a). It has low ON-state
voltage drop, almost zero reverse recovery charge and
extremely high switching speed, but its reverse leakage
current is large, especially at high temperature, due to
its lower built-in potential barrier. In addition, the max-
imum breakdown voltage of practical Schottky diodes
is limited by the increase in drift region resistance with
breakdown voltage in the absence of conductivity mod-
ulation effects. The commercially available Si Schottky
diodes are usually rated at maximum breakdown
voltages < 100 V. Beyond this limit, Si Schottky diodes
cannot compete with Si p-i-n diodes due to the unac-
ceptably large ON-state resistance of Schottky diodes.
In comparison with their Si counterparts, novel SiC
Schottky diodes can block thousands of volts because
SiC material has a much larger breakdown electric field. carrier injection in the epitaxial drift region result-
2) SiC p-i-n Diodes: SiC p-i-n diodes consist of a highly ing in conductivity modulation. However, conductivity
doped n-type substrate, a lightly doped n-type epitax- modulation causes significant reverse recovery current
ial layer with specified thickness, and a highly doped during switching, which is undesirable because it causes
p-type region for the anode, which is shown in Fig. 3(b). additional turn-ON loss in the active switch.
The advantage of the SiC p-i-n diode is low ON-state 3) SiC JBS/MPS Diodes: Hybrid diodes (JBS or MPS)
voltage drop in high current conduction, due to minority combine the attractive benefits of the low ON-state
428 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 62, NO. 2, FEBRUARY 2015

TABLE II curve fitting equations as a function of temperature. However,


(CONTINUED.) P UBLISHED SiC P OWER D IODE M ODELS no dynamic characteristics were validated with this model.
Another electrothermal behavioral model for SiC MPS diodes
has been reported in [48]. Mathematical equations with no
direct physical meaning and several curve-fitting parameters
were used to describe the static behavior of the diode under
forward and reverse bias voltages, and thermal network mod-
eling is based on the analysis of the thermal constant spectrum
of the temperature response of the diode.

C. Semiphysics-Based Models
In [44], the conventional p-n junction Shockley equation
was used to model forward conduction and reverse leakage
currents for SiC JBS. The temperature dependence of series
resistance was modeled as a second-order relationship with the
difference between the device junction temperature and ambi-
ent temperature. However, the reverse recovery characteristic
was not included in this model. The charge-controlled model
was used in [35] to accurately model the dynamic effects of
parasitic device capacitance. In [47], the Shockley equation
was also applied to model a SiC Schottky diode; and the effects
of image force barrier lowering and tunneling on reverse
leakage current were considered. In [36], the forward current
voltage characteristic of a SiC Schottky diode was modeled
voltage drop of a Schottky contact and the high blocking using standard Schottky diode equations based on thermionic
voltage of p-i-n diodes and have become common- emission theory. Note that the parameters used, such as
place starting with the second generation of SiC diode Schottky barrier height and effective Richardson constant,
structures. Hybrid diodes have Schottky-like ON-state have a clear physical meaning. The reverse leakage current
and switching behavior and p-i-n-like OFF-state char- was modeled using a parallel conductance. Zhang et al. [41]
acteristics at the same time. An MPS or JBS presented a SiC Schottky diode model using the standard
diode consists of interleaved Schottky and p+ doped Schottky diode equation, including the temperature effect on
regions [Fig. 3(c)]. For very small ON-state voltage carrier mobility. The parameter extraction procedure of this
drop, only Schottky contact areas are active. On the approach to model a SiC Schottky diode was presented in [49].
other hand, when the diodes are reverse-biased, Schottky However, this model was not complete, because the tempera-
contact conduction areas are electrically shielded by ture and reverse bias voltage effects on reverse leakage current
p-i-n junction depletion regions. An additional advantage were not included. In [38][40], modified Schottky diode
of hybrid diodes is increased surge current capability, models were proposed, considering the dependence of the
due to the conduction of the p-i-n areas at higher forward reverse leakage current on the electric field and temperature.
voltages.
To evaluate the behavior of power electronics systems, D. Physics-Based Models
accurate power diode device models are required for power
converter designers. Consequently, accurate and reliable power The Mantooth unified diode model is capable of describing
diode models are necessary. Most of the SiC power diode SiC p-i-n, Schottky, and MPS diodes under different
models published have been reviewed, and they are listed simulation conditions, including forward bias, emitter
chronologically in Table II. recombination, conductivity modulation, breakdown behav-
ior, reverse recovery, and temperature effects [26][30]. The
temperature dependence of the forward series resistance,
B. Behavioral Models semiconductor material bandgap, and carrier mobility was
In [43], the linear model of a SiC Schottky diode, including considered in this model. In [31], the lumped charge method
a dc voltage drop and a series resistor, has been reported. was used to model SiC p-i-n diodes. The main idea of this
Linear equations have been used to approximate the tem- approach is that the n-drift region is divided into a number
perature effects on the dc voltage drop and on series resis- of small subregions, and charge control equations are applied
tance. In [34], the exact same method has been applied to to each subregion. For a high voltage JBS diode, a novel
estimate power loss characteristics of a SiC Schottky diode. physics-based model was developed in [42]. The forward
A behavioral model of SiC merged p-i-n Schottky diodes was ON -resistance was more accurately approximated, consider-
presented in [32] and [33]. The forward and reverse current ing the current spreading from anode to cathode due to
versus voltage relationships were described using exponential the presence of nonconducting p+ regions on the anode
equations, and model parameters were approximated using side [Fig. 3(c)]. The electric field at Schottky contact and
MANTOOTH et al.: MODELING OF WIDE BANDGAP POWER SEMICONDUCTOR DEVICESPART I 429

the FowlerNordheim tunneling currents were considered to This high blocking voltage was achieved using an 85-m thick
model the reverse leakage current. and 8 1014 cm3 doped n-type drift epilayer. A specific
ON -resistance of 123 mcm2 was obtained with a gate

E. Seminumerical-Based Models voltage bias of 18 V, and an effective channel carrier mobility


of 22 cm2 /Vs was measured. However, 10-kV SiC MOSFETs
The forward I V characteristics of 4H-SiC p-i-n diodes are not mature enough to be commercially available now.
were investigated using an analytic model describing the For oxides in SiC, the electric field stress is determined by
minority carrier distributions in different diode regions the oxide thickness and gate voltage. Choices are limited due
(high-doped regions and low-doped drift region) [46]. In this to targets for threshold voltage and transconductance. Due to
model, the solution to the ADE in a lightly doped drift permittivity differences between SiO2 and SiC, the electric
region was derived under high injection conditions. The field strength in the oxide is typically three times higher than
doping dependence of bandgap narrowing, incomplete dop- in the SiC. Control of the pitch in p-body modifies this field
ing ionization, carrier lifetime, and mobility have been distribution. Various intrinsic defects not related to dopants or
considered. The accuracy of this model has been verified impurities are observed at the SiC/SiO2 interface. An SiC has a
by comparisons with numerical simulations and experimen- higher surface density of atoms per unit area compared with Si,
tal results under different temperature and conduction cur- resulting in a higher density of dangling Si- and C-bonds, and
rent conditions. However, this model is inconvenient for carbon cluster at the interface. Defects located in the near-
implementation into circuit simulators, due to its com- interfacial oxide layer may appear in the energy gap of SiC
plicated carrier distribution solution. In [45] and [50], and act as traps for electrons. Low channel mobility in 4H-SiC
a Fourier-based solution to the ADE was used to develop is directly linked to very high interface state densities at the
p-i-n diode models. The second-order partial differential car- SiO2 /SiC boundary. Interface state density reduction plays a
rier diffusion equation was turned into an infinite number of critical role in increasing channel mobility and improving high
first-order linear differential equations with series coefficients temperature performance, as well as the reliability of power
implemented by equivalent RC components. The Fourier SiC MOSFETs and IGBTs.
series of first-order equations is truncated to implement this Reduction of the interface state density and improvement in
model. device reliability can be achieved by the use of nitric oxide
postoxidation annealing. This reduces carbon clusters at the
F. Numerical Models interface, thus increasing the channel mobility and the dielec-
The finite-element simulation tool SILVACO was used to tric quality. One promising solution is to create a carefully
simulate a 10 kV SiC p-i-n diode in [37], and simulation formed oxide-nitride-oxide dielectric on 4H-SiC [4], [21].
results have been compared with test results from the actual An accurate model of a SiC MOSFET is necessary for
device. A good agreement between the SILVACO model and device evaluation, system design, and power converter behav-
the experimental results was demonstrated. ior prediction. Therefore, much effort is needed to model
SiC MOSFETs. A few SiC MOSFET models have been
published and are listed chronologically in Table III.
G. SiC Power MOSFET
A SiC vertical power MOSFET is a next-generation H. Behavioral Models
switching device expected to replace conventional Si power
To model the turn-ON dynamic characteristics of a
switching devices in many applications because it can operate
SiC MOSFET, a behavioral MOSFET model has been
with lower power loss, at a higher switching frequency and at
proposed in [71]. The SiC MOSFET model consists of a
higher operating temperatures. In the past ten years, a lot of
drainsource resistance and three constant interelectrode
effort has been devoted to developing SiC power MOSFETs;
capacitances. The drainsource resistance can be switched
and great progress has been achieved. The first SiC power
from infinity to a finite small value and vice versa, according
MOSFET was demonstrated in 1994 in the form of a vertical
to the transistors gate voltage. The parasitic inductances from
trench gate structure (UMOSFET) [51]. The reported device
device packages have been considered to accurately simulate
had 150 V blocking voltage and 3.3 mcm2 specific
the transient ringing. A behavioral SiC MOSFET model is
ON -resistance. It suffered from a high electric field in the gate
built in [72] using polynomial functions to create a model
oxide at the trench corner and low inversion carrier mobility.
that estimates device loss.
To avoid these drawbacks in a UMOSFET, a SiC planar gate
MOSFET was fabricated using a double implantation MOS
process to form p-base (DMOSFET) [52]. A diagram of a typi- I. Semiphysics-Based Models
cal UMOSFET and DMOSFET are shown in Fig. 4. A 6H-SiC The conventional Shockley lateral MOSFET equations,
DMOSFET was based on a 10-m thick and 6.5 1015 cm3 which can describe the static I V characteristic curve, have
doped n-type drift layer, which can support a reverse blocking been used to model a SiC vertical power MOSFET. These
voltage of 760 V. Later on, DMOSFETs with higher reverse equations are implemented, for example, in the level 1
blocking voltages have been reported. A 10-kV 4H-SiC power MOSFET model in PSpice. Since the Shockley MOSFET
DMOSFET was demonstrated [53]. This was the highest model was originally developed for lateral structure devices in
reported blocking voltage for SiC power MOSFETs [53]. Si and germanium (Ge), many researchers have proposed some
430 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 62, NO. 2, FEBRUARY 2015

TABLE III TABLE III


P UBLISHED SiC P OWER MOSFET M ODELS (CONTINUED.) P UBLISHED SiC P OWER MOSFET M ODELS

threshold voltage and device transconductance. The reduced


carrier mobility in the drift layer and JFET region was
considered in modeling ON-state resistance. This proposed
model was validated from 25 C to 200 C. On the basis
of the same model, Sun et al. [69] and Lu et al. [70]
investigated a modified SiC MOSFET model that can simulate
device behavior at low temperature (25 C). The effect of
negative gate drive voltage on gatesource capacitance was
also included in this proposed model. In [67], to account
for the device self-heating effect, a Foster RC network was
used for thermal modeling and coupled with the electri-
cal model that used Shockley lateral MOSFET equations.
Moreover, another large-signal model for 6H-SiC MOSFET
with temperature compensation has been proposed [58]. The
temperature-dependent compensating currents were considered
to be in parallel with the MOS channel current between the
drain and source. These three temperature-dependent currents
were the body leakage current, channel current change due to
modification to these equations for their proposed SiC vertical the gate threshold voltage variation and the channel current
power MOSFET models. In [60], a simple SPICE model change due to drain, and source contact resistance variation.
for SiC power DMOSFETs was proposed with some specific In [74], a temperature-dependent voltage source is added in
modifications in the conventional Si lateral MOS channel mod- series with the gate of the standard MOSFET model, to sim-
eling approach. Temperature-dependent voltage and current ulate the gate threshold voltage variation with operating tem-
sources were added to describe the temperature-dependent gate perature. To include temperature-dependent carrier mobility
MANTOOTH et al.: MODELING OF WIDE BANDGAP POWER SEMICONDUCTOR DEVICESPART I 431

Fig. 4. (a) Cross-sectional views of SiC UMOSFET. (b) SiC DMOSFET [40].

behavior and avalanche effect, two parallel current sources


are added in parallel with the standard MOSFET model.
Fig. 5. Cross-sectional views of SiC DMOSFET for modeling.
Sakurai and Newton [73] introduced a simple nth power law
MOSFET model. This model had the excellent feature of
fast parameter extraction, and was originally proposed for
high power and temperature conditions. This poor inversion
short-channel CMOS. In [62] and [66], this nth power law
surface mobility is due to the large number of interface
MOSFET model was used to simulate high-voltage SiC power
defects during the MOS thermal oxidation process. A com-
MOSFETs, improving the accuracy of forward I V curves.
prehensive physical model was given, incorporating interface
trap densities, Coulombic interface trap scattering, surface
J. Physics-Based Models roughness scattering, phonon scattering, velocity saturation,
An analytical model for a DMOSFET was developed and so forth [61]. Coulomb scattering and surface roughness
in [56] and [57]. This model was proposed based on regional scattering reduce surface mobility in the subthreshold and
analysis of carrier transport in the channel and drift regions. linear region, whereas the saturation current is limited by
The drift region was divided into three parts, as shown carrier saturation velocity.
in Fig. 5: 1) an accumulation Region A; 2) a drift Region B
with a varying cross section area; and 3) a drift Region C L. Numerical Models
with a constant cross-region area. The voltage drops across
A 2-D device model was created using the SILVACO
these regions were calculated as a function of the electric
ATLAS simulator to investigate a 4H-SiC 1200 V power
field in these regions. Another physics-based model for a
MOSFET [68]. This model included recombination effects,
SiC DMOSFET was proposed in [63], where the nonuniform
bandgap narrowing, impact ionization, and lattice heating.
current distribution in the JFET region was modeled using
A comprehensive numerical model considering carrier mobil-
a nonlinear voltage source and a resistance network. One
ity and interface states was established in [59]. The channel
advantage of this novel DMOSFET model was that only a
current distribution versus bias voltage was solved numerically
single set of equations was used to describe device operation
in two dimensions.
in both the linear and saturation regions.
McNutt et al. [54], [55] presented a compact model for
SiC power MOSFET. The expression for the channel current IV. C ONCLUSION
in this model was the sum of two distinct components with This paper begins a comprehensive review of available
different gate threshold voltages, providing additional flexi- devices and models in the power semiconductor space. The
bility for device modeling. Due to the unique feature of the most prominent devices, the SiC diode and MOSFET, were
channel current description, this model could reproduce the included here. Part II of this paper will complete the lesser
gradual transition from linear to saturation region observed used SiC power devices and describe what is available in
in SiC power MOSFETs. The parameter extraction was done GaN devices and models as well.
using a software package called IMPACT. In [64] and [65],
a new parameter extraction strategy for the McNutt model was R EFERENCES
introduced, which relied on the device datasheet data only,
[1] J. L. Hudgins, G. S. Simin, E. Santi, and M. A. Khan, An assessment of
eliminating the need for experimental characterization. wide bandgap semiconductors for power devices, IEEE Trans. Power
Electron., vol. 18, no. 3, pp. 907914, May 2003.
[2] B. J. Baliga, Silicon Carbide Power Devices. Singapore:
K. Seminumerical-Based Models World Scientific, 2005.
[3] K. Shenai, P. G. Neudeck, M. Dudley, and R. F. Davis, High-power
Low channel mobility of electrons in the surface inversion switching in semiconductorsWhat is beyond silicon thyristor? in
layer is a severe limitation for a SiC MOSFET operating at Proc. IEEE Energytech, May 2011, pp. 16.
432 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 62, NO. 2, FEBRUARY 2015

[4] S. Tanimoto and H. Ohashi, Reliability issues of SiC power MOSFETs [29] T. R. McNutt, A. R. Hefner, H. A. Mantooth, J. L. Duliere,
toward high junction temperature operation, Phys. Status Solidi A, D. W. Berning, and R. Singh, Parameter extraction sequence for silicon
vol. 206, no. 10, pp. 24172430, 2009. carbide Schottky, merged PiN Schottky, and PiN power diode models,
[5] P. K. Steimer, H. E. Gruning, J. Werninger, E. Carroll, S. Klaka, and in Proc. 33rd IEEE Power Electron. Specialists Conf., Jun. 2002,
S. Linder, IGCTA new emerging technology for high power, low cost pp. 12691276.
inverters, IEEE Ind. Appl. Mag., vol. 5, no. 4, pp. 1218, Jul./Aug. 1999. [30] T. R. McNutt, A. R. Hefner, Jr., H. A. Mantooth, J. L. Duliere,
[6] J. Millan, P. Gordignon, X. Perpina, A. Perez-Tomas, and J. Rebollo, D. W. Berning, and R. Singh, Physics-based modeling and characteri-
A survey of wide bandgap power semiconductor devices, IEEE Trans. zation for silicon carbide power diodes, Solid-State Electron., vol. 50,
Power Electron., vol. 29, no. 5, pp. 21552163, May 2014. no. 3, pp. 388398, Mar. 2006.
[7] R. Kraus and H. J. Mattausch, Status and trends of power semiconduc- [31] R. Kolessar and H. P. Nee, A new physics-based circuit model for
tor device models for circuit simulation, IEEE Trans. Power Electron., 4H-SiC power diodes implemented in SABER, in Proc. 16th IEEE
vol. 13, no. 3, pp. 452465, May 1998. Appl. Power Electron. Conf. Expo., Mar. 2001, pp. 989994.
[8] E. Santi, J. L. Hudgins, and A. Mantooth, Variable model levels for [32] M. Zubert, M. Napierlska, G. Jablonski, L. Starzak, M. Janicki, and
power semiconductor devices, in Proc. Summer Comput. Simulation A. Napieraski, Static electro-thermal model of SiC merged p-i-n
Conf. (SCSC), Jul. 2007, pp. 276283. Schottky diodes, in Proc. 10th Int. Seminar Power Semiconductors,
[9] K. Sheng, B. W. Williams, and S. J. Finney, A review of IGBT models, Sep. 2002, pp. 227232.
IEEE Trans. Power Electron., vol. 15, no. 6, pp. 12501266, Nov. 2000. [33] M. Zubert, L. Starzak, G. Jablonski, M. Napierlska, M. Janicki, and
[10] H. Ohashi and I. Omura, Role of simulation technology for the progress A. Napieraski, Novel SPICE dynamic model of SiC merged PiN
in power devices and their applications, IEEE Trans. Electron Devices, Schottky diodes, in Proc. 18th Int. Conf. Mixed Design Integr. Circuits
vol. 60, no. 2, pp. 528534, Feb. 2013. Syst., Jun. 2011, pp. 541544.
[11] P. R. Wilson and H. A. Mantooth, Model Based Engineering of Complex [34] B. Opineci and L. M. Tolbert, Characterization of SiC Schottky diodes
Electronic Systems. London, U.K.: Elsevier, Mar. 2013, p. 511. at different temperatures, IEEE Power Electron Lett., vol. 1, no. 2,
[12] H. A. Mantooth, A. M. Francis, Y. Feng, and W. Zheng, Modeling tools pp. 5457, Jun. 2003.
built upon the HDL foundation, in Proc. IET Comput. Digital Techn., [35] M. Giesselmann, R. Edwards, S. Bayne, S. Kaplan, and E. Shaffer,
Sep. 2007, pp. 519527. Forward and reverse recovery spice model of a JBS silicon carbide
[13] X. Ma, Superscrew dislocations in silicon carbide: Dissocia- diode, in Proc. 26th Int. Power Modulator Symp. High-Voltage Work-
tion, aggregation, and formation, J. Appl. Phys., vol. 99, no. 6, shop, May 2004, pp. 364367.
pp. 063513-1063513-6, 2006. [36] Z. Ouennoughi, R. Weiss, S. Benlala, and H. Ryssel, Extracting of sili-
[14] N. Ohtani et al., Propagation behavior of threading dislocations during con carbide Schottky diode model parameters using lateral optimization
physical vapor transport growth of silicon carbide (SiC) single crystals, method including the parallel conductance, in Proc. 5th Int. Conf. Adv.
J. Crystal Growth, vol. 286, no. 1, pp. 5560, 2006. Semiconductor Devices Microsyst., Oct. 2004, pp. 271274.
[15] S. I. Maximenko et al., Effect of threading screw and edge dislocations [37] R. L. Thomas, M. Morgenstern, and S. B. Bayne, Silvaco modeling
on transport properties of 4HSiC homoepitaxial layers, J. Appl. Phys., of a 10 kV SiC p-i-n diode, in Proc. 26th Int. Conf. Power Modular
vol. 108, no. 1, p. 013708, 2010. Symp., May 2004, pp. 567570.
[16] M. K. Das, Commercially available cree silicon carbide power devices: [38] J. Zarebski and J. Dabrowski, SiC Schottky power diode modelling
Historical success of JBS diodes and future power switch prospects, in in SPICE, in Proc. 12th IEEE Int. Conf. Electron., Circuits Syst.,
Proc. CS MANTECH Conf., 2011, pp. 315317. Dec. 2005, pp. 14.
[17] G. Manolis, Optical and Electrical Properties of Highly Excited 3C- [39] J. Zarebski and J. Dabrowski, SPICE modelling of power Schottky
SiC Crystals and Heterostructures, Ph.D. dissertation, Vilnius Univ., diodes, Int. J. Numer. Model., Electron. Netw., Devices Fields, vol. 21,
Vilnius, Lithuania, Mar. 2013, pp. 2631. no. 6, pp. 551561, 2008.
[18] X. Ma, Superscrew dislocations in silicon carbide: Dissociation, aggre- [40] J. Zarebski and J. Dabrowski, Modelling power Schottky diodes, in
gation, and formation, J. Appl. Phys., vol. 99, no. 6, p. 063513, 2006. Proc. Int. Conf. Modern Problems Radio Eng., Telecommun. Comput.
[19] N. Ohtani et al., Propagation behavior of threading dislocations during Sci., Feb. 2006, pp. 9093.
physical vapor transport growth of silicon carbide (SiC) single crystals, [41] H. Zhang, L. M. Tolbert, and B. Ozpineci, System modeling and char-
J. Crystal Growth, vol. 286, no. 1, pp. 5560, 2006. acterization of SiC Schottky power diodes, in Proc. IEEE Workshops
[20] S. I. Maximenko et al., Effect of threading screw and edge dislocations Comput. Power Electron., Jul. 2006, pp. 199204.
on transport properties of 4HSiC homoepitaxial layers, J. Appl. Phys., [42] L. Zhu and T. P. Chow, Analytical modeling of high-voltage 4H-SiC
vol. 108, no. 1, p. 013708, 2010. junction barrier Schottky (JBS) rectifiers, IEEE Trans. Electron Devices,
[21] M. Treu, R. Rupp, and G. Solkner, Reliability of SiC power devices and vol. 55, no. 8, pp. 18571863, Aug. 2008.
its influence on their commercializationReview, status, and remain- [43] B. Ozpineci, M. S. Chinthavali, L. M. Tolbert, A. S. Kashyap, and
ing issues, in Proc. IEEE Int. Rel. Phys. Symp. (IRPS), May 2010, H. A. Mantooth, A 55-kW three-phase inverter with Si IGBTs and SiC
pp. 156161. Schottky diodes, IEEE Trans. Ind. Appl., vol. 45, no. 1, pp. 278285,
[22] K. Acharya and K. Shenai, On the rating of SiC Schottky power Jan./Feb. 2009.
rectifiers, in Proc. Power Electron. Technol. Conf., 2002, pp. 672677. [44] J. Wang, Y. Du, S. Bhattacharya, and A. Q. Huang, Characterization,
[23] M. Treu et al., A surge for current stable and avalanche rugged SiC modeling of 10-kV SiC JBS diodes and their application prospect in
merged pn Schottky diode blocking 600 V especially suited for PFC X-ray generators, in Proc. IEEE Energy Convers. Congr. Expo.,
applications, Mater. Sci. Forum, vols. 600602, pp. 935603, Oct. 2007. Sep. 2009, pp. 14881493.
[24] B. Heinze, J. Lutz, M. Nuemeister, R. Rupp, and M. Holz, Surge cur- [45] A. T. Bryant et al., Physical modelling of large area 4H-SiC PiN
rent ruggedness of silicon carbide Schottky- and merged-PiN-Schottky diodes, in Proc. IEEE Energy Convers. Congr. Expo., Sep. 2009,
diodes, in Proc. 20th Int. Symp. Power Semiconductor Devices ICs, pp. 986993.
May 2008, pp. 245248. [46] S. Bellone, F. G. Della Corte, L. F. Albanese, and F. Pezzimenti,
[25] M. Kneifel, D. Silber, and R. Held, Predictive modeling of SiC-device An analytical model of the forward IV characteristics of 4H-SiC
power Schottky diode for investigations in power electronics, in Proc. p-i-n diodes valid for a wide range of temperature and current,
8th IEEE Appl. Power Electron. Conf. Expo., Mar. 1996, pp. 239245. IEEE Trans. Power Electron., vol. 26, no. 10, pp. 28352843,
[26] H. A. Mantooth and J. L. Duliere, A unified diode model for circuit Oct. 2011.
simulation, IEEE Trans. Power Electron., vol. 12, no. 5, pp. 816823, [47] S. Ahmed, H. A. Mantooth, M. Mudholkar, and R. Singh, Character-
Sep. 1997. ization and modeling of SiC junction barrier Schottky diode for circuit
[27] T. R. McNutt, A. R. Hefner, Jr., H. A. Mantooth, J. Duliere, simulation, in Proc. IEEE 14th Workshop Control Modeling Power
D. W. Berning, and R. Singh, Silicon carbide PiN and merged PiN Electron. (COMPEL), Jun. 2013, pp. 15.
Schottky power diode models implemented in the Saber circuit sim- [48] L. Starzak et al., Behavioral approach to SiC MPS diode electrothermal
ulator, IEEE Trans. Power Electron., vol. 19, no. 3, pp. 573581, model generation, IEEE Trans. Electron Devices, vol. 60, no. 2,
May 2004. pp. 630638, Feb. 2013.
[28] T. R. McNutt, A. R. Hefner, H. A. Mantooth, J. Duliere, D. W. Berning, [49] R. Fu, A. Grekov, K. Peng, and E. Santi, Parameter extraction
and R. Singh, Silicon carbide PiN and merged PiN Schottky power procedure for a physics-based power SiC Schottky diode model, in
diode models implemented in the Saber circuit simulator, in Proc. 32nd Proc. 28th Annu. IEEE Appl. Power Electron. Conf. Expo., Mar. 2013,
IEEE Power Electron. Specialists Conf., Jun. 2001, pp. 21032108. pp. 545552.
MANTOOTH et al.: MODELING OF WIDE BANDGAP POWER SEMICONDUCTOR DEVICESPART I 433

[50] L. Lu, A. Bryant, E. Santi, J. L. Hudgins, and P. R. Palmer, Physical [71] P. Alexakis, O. Alatise, L. Ran, and P. Mawby, Modeling power
modeling and parameter extraction procedure for p-i-n diodes with converters using hard switched silicon carbide MOSFETs and Schottky
lifetime control, in Proc. 41st IAS IEEE Ind. Appl. Conf., Oct. 2006, barrier diodes, in Proc. 15th Eur. Conf. Power Electron. Appl. (EPE),
pp. 14501456. Sep. 2013, pp. 19.
[51] J. W. Palmour, J. A. Edmond, H. Kong, and C. Charter, Vertical power [72] A. Merkert, T. Krone, and A. Mertens, Characterization and scalable
devices in silicon carbide, in Proc. Silicon Carbide Rel. Mater., Inst. modeling of power semiconductors for optimized design of traction
Phys. Conf., Oct. 1994, pp. 499502. inverters with Si- and SiC-devices, IEEE Trans. Power Electron.,
[52] J. N. Shenoy, J. A. Cooper, and M. R. Melloch, High-voltage double- vol. 29, no. 5, pp. 22382245, Jan. 2014.
implanted power MOSFETs in 6H-SiC, IEEE Electron Device Lett., [73] T. Sakurai and A. R. Newton, A simple MOSFET model for circuit
vol. 18, no. 3, pp. 9395, Mar. 1997. analysis, IEEE Trans. Electron Devices, vol. 38, no. 4, pp. 887894,
[53] S. H. Ryu et al., 10-kV, 123-m cm2 4H-SiC power DMOSFETs, Apr. 1991.
IEEE Electron Device Lett., vol. 25, no. 8, pp. 556558, Aug. 2004. [74] V. dAlessandro et al., SPICE modeling and dynamic electrothermal
[54] T. McNutt, A. Hefner, A. Mantooth, D. Berning, and S. Ryu, Silicon simulation of SiC power MOSFETs, in Proc. IEEE 26th Int. Symp.
carbide power MOSFET model and parameter extraction sequence, in Power Semiconductor Devices ICs (ISPSD), Jun. 2014, pp. 285288.
Proc. IEEE 34th Annu. Power Electron. Specialist Conf., Jun. 2003,
pp. 217226.
[55] T. R. McNutt, A. R. Hefner, H. A. Mantooth, D. Berning, and
S.-H. Ryu, Silicon carbide power MOSFET model and parameter
extraction sequence, IEEE Trans. Power Electron., vol. 22, no. 2,
pp. 353363, Mar. 2007. H. Alan Mantooth (S83M90SM97F09)
[56] M. Hasanuzzaman, S. K. Islam, L. M. Tolbert, and B. Ozpineci, Model received the Ph.D. degree from the Georgia Institute
simulation and verification of a vertical double implanted (DIMOS) of Technology, Atlanta, GA, USA, in 1990.
transistor in 4H-SiC, in Proc. 7th Int. Multi-Conf. Power Energy Syst., He joined the faculty of the Department of Elec-
2003. trical Engineering, University of Arkansas, Fayet-
[57] M. Hasanuzzaman, S. K. Islam, L. M. Tolbert, and B. Ozpineci, Design, teville, AR, USA, in 1998, where he is currently a
modeling, testing and SPICE parameter extraction of DIMOS transistor Distinguished Professor.
in 4H-silicon carbide, Int. J. High Speed Electron. Syst., vol. 16, no. 2,
pp. 733746, 2006.
[58] M. Hasanuzzaman, S. K. Islam, and L. M. Tolbert, Effects of tem-
perature variation (300600 K) in MOSFET modeling in 6Hsilicon
carbide, Solid-State Electron., vol. 48, no. 1, pp. 125132, Jan. 2004.
[59] S. K. Powell, N. Goldsman, J. M. McGarrity, J. Bernstein, C. J. Scozzie,
and A. Lelis, Physics-based numerical modeling and characterization of
6H-silicon-carbide metaloxidesemiconductor field-effect transistors, Kang Peng received the B.S. degree in electri-
J. Appl. Phys., vol. 92, no. 7, pp. 40534061, Oct. 2007. cal engineering from Hunan University, Changsha,
[60] J. Wang et al., Characterization, modeling, and application of China, and the M.S. degree in electrical engineering
10-kV SiC MOSFET, IEEE Trans. Electron Device, vol. 55, no. 8, from the Huazhong University of Science and Tech-
pp. 17981806, Aug. 2008. nology, Wuhan, China, in 2008 and 2011, respec-
[61] S. Potbhare, N. Goldsman, A. Lelis, J. M. McGarrity, F. B. McLean, and tively. He is currently pursuing the Ph.D. degree
D. Habersat, A physical model of high temperature 4H-SiC MOSFETs, with the University of South Carolina, Columbia,
IEEE Trans. Electron Devices, vol. 55, no. 8, pp. 20292040, Aug. 2008. SC, USA.
[62] N. Phankong, T. Funaki, and T. Hikihara, A static and dynamic model His current research interests include power semi-
for a silicon carbide power MOSFET, in Proc. 13th Eur. Conf. Power conductor devices modeling and application.
Electron. Appl., Sep. 2009, pp. 110.
[63] R. Fu, A. Grekov, J. Hudgins, A. Mantooth, and E. Santi, Power SiC
DMOSFET model accounting for nonuniform current distribution in
JFET regions, IEEE Trans. Ind. Appl., vol. 48, no. 1, pp. 181190,
Jan./Feb. 2012.
[64] M. Mudholkar, M. Saadeh, and H. A. Mantooth, A datasheet Enrico Santi (S90M94SM02) received the
driven power MOSFET model and parameter extraction procedure Ph.D. degree from the California Institute of Tech-
for 1200 V, 20 A SiC MOSFETs, in Proc. 14th Eur. Conf. Power nology, Pasadena, CA, USA, in 1994.
Electron. Appl. (EPE), 2011, pp. 110. He has been with the University of South Carolina,
[65] M. Mudholkar, S. Ahmed, M. N. Ericson, S. S. Frank, C. L. Britton, Columbia, SC, USA, since 1998, where he is cur-
and H. A. Mantooth, Datasheet driven silicon carbide power MOSFET rently an Associate Professor. His current research
model, IEEE Trans. Power Electron., vol. 29, no. 5, pp. 22202228, interests include modeling and simulation of semi-
May 2014. conductor power devices, control of power elec-
[66] Y. Cui, M. Chinthavali, and L. M. Tolbert, Temperature dependent tronics systems, and advanced modeling of power
Pspice model of silicon carbide power MOSFET, in Proc. 27th systems.
Annu. IEEE Appl. Power Electron. Conf. Expo. (APEC), Feb. 2012,
pp. 16981704.
[67] S. Yin, T. Wang, K. J. Tseng, J. Zhao, and X. Hu, Electro-thermal
modeling of SiC power devices for circuit simulation, in Proc.
39th Annu. Conf. IEEE Ind. Electron. Soc. (IECON), Nov. 2013,
pp. 718723. Jerry L. Hudgins received the Ph.D. degree from
[68] B. N. Pushpakaran, S. B. Bayne, and A. A. Ogunniyi, Electro thermal Texas Tech University, Lubbock, TX, USA, in 1985.
transient simulation of silicon carbide power MOSFET, in Proc. 19th He is currently the Chair of the Department
IEEE Pulsed Power Conf. (PPC), Jun. 2013, p. 1. of Electrical Engineering with the University of
[69] K. Sun, H. Wu, J. Lu, Y. Xing, and L. Huang, Improved modeling of Nebraska, Lincoln, NE, USA. His current research
medium voltage SiC MOSFET within wide temperature range, IEEE interests include power electronic device characteri-
Trans. Power Electron., vol. 29, no. 5, pp. 22292237, May 2014. zation and modeling, power electronics design, and
[70] J. Lu, K. Sun, H. Wu, Y. Xing, and L. Huang, Modeling of SiC renewable energy systems.
MOSFET with temperature dependent parameters and its applications,
in Proc. 28th Annu. IEEE Appl. Power Electron. Conf. Expo. (APEC),
Mar. 2013, pp. 540544.

Você também pode gostar