Professional Documents
Culture Documents
Body Diode
AbstractIn this paper, the static and switching cost of external anti-parallel diodes. However, a significant
characterizations of a SiC MOSFETs body diode are presented. issue of body diode utilization is its reverse recovery. The
The static characterization of SiC MOSFETs body diode is reverse recovery is due to high concentrations of injected
carried out using a curve tracer and a double pulse test bench is carriers stored in the drift region during conduction. During
built to characterize the inductive switching behavior of SiC the switch turn-off transition, some carriers are swept away
MOSFETs body diode. The reverse recovery of SiC MOSFETs from the drift region, resulting in reverse recovery current. The
body diode is shown at different forward conduction currents, reverse recovery current leads to additional switching loss in
junction temperatures and current commutation slopes. In order the complementary power switch [9].
to evaluate the performance of SiC MOSFETs body diode in
different applications, an accurate physics-based diode model is
introduced to perform simulations of SiC MOSFETs body
diode. The parameter extraction procedure for this body diode
model is given. The validation of the body diode model shows
good agreement between simulation and experimental results,
which proves the accuracy of the model.
I. INTRODUCTION
Fig. 1. Cross-sectional structure schematic of SiC power DMOSFET
SiC power MOSFET is a very good candidate for high-
switching-frequency and low-loss power conversion In this paper, a complete performance characterization of
applications. The lower on-resistance makes SiC power SiC MOSFETs body diode is carried out. The study is
MOSFETs an ideal choice in high power applications, offering conducted for a 1200V/36A SiC MOSFET from Cree Inc. The
similar conduction loss as Si IGBTs while operating at a much static characterization is done using a curve tracer. For
higher switching frequency [1]-[3]. The switching loss of a SiC dynamic switching characterization, a double pulse tester
power MOSFET is much lower than that of a Si IGBT or Si (DPT) printed circuit board (PCB) with an inductive load is
GTO for the same voltage and current ratings, due to its lower built. The reverse recovery of SiC MOSFETs body diode is
device capacitance [4] [5]. In inductive hard switching, SiC shown at different current commutation slopes, forward
MOSFETs body diode might be used if no external anti- conduction currents, and junction temperatures. In addition, a
parallel diode is connected [6] [7]. For example, in a Fourier-based-solution physics-based model for SiC
synchronous buck converter the inductor current flows through MOSFETs body diode is proposed. The parameter extraction
the lower MOSFETs body diode during the dead time periods procedure for SiC MOSFETs body diode model is presented.
[8]. In order to utilize the body diode of SiC MOSFET, a Finally, the model is validated by comparing simulated results
complete characterization (static and dynamic) of SiC with experimental results under inductive switching condition.
MOSFETs body diode is required. In addition, a circuit-
oriented device model is needed to evaluate the performance of
II. CHARACTERIZATION OF SIC MOSFET BODY DIODE
SiC MOSFETs body diode in power converter design.
The body diode in a SiC power MOSFET is a p-i-n diode, A. Static characterization
as shown in Fig.1. The low-doped drift region is sandwiched In this section, static characterization of SiC MOSFETs
between drain and source, creating a vertical diode structure. body diode is described. Static characteristics (I-V) are
This p-i-n diode can be utilized to conduct current through the measured with a Tektronix 371A curve tracer. The device
SiC power MOSFET in third quadrant operation. It is under test (DUT) is a SiC MOSFET C2M0080120D from Cree
desirable to utilize MOSFET body diodes to avoid additional Inc. rated at 1200V/36A.
This work was supported by the Office of Naval Research under grant
N00014-14-1-0165
\
Fig. 4. Static characteristics of SiC MOSFET body diode with positive gate-
source voltages at 25oC and 125oC
Fig. 2. Static characteristics of SiC MOSFET body diode with positive gate-
source voltages
Fig. 3. Static characteristics of SiC MOSFET body diode with negative gate-
source voltages
2128
Fig.6 shows the on-state resistance curve as a function of IXYS Corporation is used as the SiC MOSFET gate driver
junction temperature, when Vgs=20V, and Ids=18A. The on- with 9A maximum source/sink drive current. A Pearson coil
state resistance is 79.6 m at 25 oC junction temperature, while (model 2878) is used to measure the body diode current Id of
on-state resistance increases to 134.8 m at 150 oC junction high-side SiC MOSFET. A 250 H ferrite-core inductor is
temperature. used as the load inductor for inductive switching experiments.
Fig.7 shows body diode built-in potential curve as a
function of junction temperature, when Vgs= -5V. The body
diode series resistance as a function of junction temperature is
shown in Fig.8, when Vgs= -5V.
2129
The body diode current waveforms in Fig.14 are measured
with different forward conduction currents at 20 gate
resistance and room temperature. The reverse peak current and
reverse charge increase with forward conduction current. A
higher forward conduction current requires more free charges
in the drift region, which results in a larger reverse recovery.
Fig. 11. Body diode current waveforms at different low-side MOSFET gate
resistances (block voltage: 500V, forward conduction current: 30A, room
temperature)
Fig. 15. Body diode reverse peak current and reverse recovery charge at
different forward conduction currents (block voltage: 500V, low-side gate
resistance: 20, room temperature)
Fig. 13. Body diode reverse recovery times and di/dt at different low-side
MOSFET gate resistances (block voltage: 500V, forward conduction current:
30A, room temperature)
2130
Fig. 19. Body diode reverse recovery times at different junction temperatures
Fig. 16. Body diode reverse recovery times at different forward conduction
currents (block voltage: 500V, low-side gate resistance: 20, room (block voltage: 500V, forward conduction current: 30A, gate resistance: 20)
temperature)
III. DEVELOPMENT OF BODY DIODE MODEL
3) Varied junction temperatures
The DUT is heated by attaching it to a heat spreader, whose The body diode model uses a Fourier series solution of the
temperature is controlled by a thermal controller Eurotherm 94. ambipolar diffusion equation (ADE) in the drift layer to find
At each temperature operating point, the device is heated for a the carrier distribution in that region. This carrier distribution is
long enough time to ensure that MOSFETs junction used to find the conductive voltage drop in the drift region,
temperature is the same as the case temperature. Fig.17 shows accounting for conductivity modulation.
the experimental waveforms at varied junction temperatures. Fig.20 shows the general arrangement of the carrier
Both the reverse peak current and reverse recovery charge distribution in n- drift region, including an un-depleted carrier
increase with junction temperature. storage layer and two depletion layers [10]. The carrier storage
layer is sandwiched between two depletion layers. When the
body diode is on, the two depletion layers shrink, and the
carrier storage layer occupy the whole drift region. When the
body diode is off, the two depletion layers expand from the two
ends of the drift region, and some free carriers are swept from
the carrier storage region. The depletion layers start to support
a voltage and the body diode becomes reverse-biased.
2131
coefficients p0pk. The Fourier-series based solution is given V0 V2 V2n
by: R0 R2 R2n
Reven
k ( x x1 ) C0 C2 C2n
p ( x , t ) = p0 (t ) + pk ( t ) cos (2)
k =1 ( x2 x 1 ) I0 I2 I2n
C1 C3 C2n+1
x2
2 k ( x x1 )
pk ( t ) =
x2 x1 p ( x, t ) cos
x1
( x2 x 1 )
dx (4) I1
p ( x , t )
I3
p ( x , t )
I2n+1
I odd = D x2 + x1
x x
By substituting equations (3) and (4) into equation (1), the
Fourier-series coefficients pk are determined in an infinite set
of 1st order linear differential equations. The boundary Fig. 21. Equivalent circuits used to calculate the carrier density representing
conditions at the boundaries of the undepleted region (x1 and x2) the coefficients of the Fourier series solution to ADE
are required, which give the gradients of the carrier densities.
The boundary conditions are given by:
A. Voltage components
p 1 In I p The voltage drop across the body diode Vak is comprised of
|x1 = |x (5) several components, including the voltages Vj1 and Vj2 across
x 2qA Dn Dp 1
junctions J1 and J2, the voltages Vd1 and Vd2 across two
depletion layers, and the voltage Vn- across the n- drift region.
p 1 In I p
|x2 = |x (6)
x 2qA Dn D p 2
2132
accuracy and simulation speed is made by selecting the number where hn is the recombination parameter, and Px2 is the carrier
of segments. density at the boundary x2.
The displacement current Idisp2 in n+ end region is
calculated by:
1 dVd 2
I disp 2 = A (17)
Wd 2 dt
where Wd2 is the depletion width at junction J2.
The drift region voltage Vn- in carrier storage region is TABLE I. BODY DIODE MODEL PARAMETER LIST
calculated by:
Symbol Description
Id x2 x1 M 1 P p P
Vn = [
qA( n + p ) M k = 0 PT ( k ) PT ( k 1)
ln( T ( k ) )] VT ( n
PT ( k 1) n + p
) ln( x 2 )
Px1
A (cm )2
Active chip area
(s) High-level minority carrier lifetime
WN- (m) Drift region width
(11) NN- (cm-3) Doping concentration in drift region
hp (cm4/s) Recombination parameter in P region
where Id is the body diode conduction current, A is the active hn (cm4/s) Recombination parameter in n+ region
chip area, and M is the number of segments in carrier storage
region. n and p are the electron mobility and hole mobility,
respectively. PT(k) is the carrier density at segment boundary The parameter extraction procedure includes an initial
points. parameter estimation from the manufacturers datasheets, and a
parameter refinement based on the measured waveforms [11].
The depletion layer voltages Vd1 and Vd2 are derived using
feedback from the carrier densities Px1 and Px2 at boundaries: A. Initial parameter estimation
0 if Px1 > 0 1) Active chip area A:
Vd 1 = { (12) The active chip area can be obtained from datasheets or be
KF Px1 otherwise
roughly estimated from the maximum current density J (about
0 if Px 2 > 0 300A/cm2) and current rating Id. The active chip area A is
Vd 2 = { (13) calculated by:
KF Px 2 otherwise
where KF is the feedback constant. Id
A= (18)
J
B. Current components
2) High-level minority carrier lifetime :
The electron current In1 in the p end region is given by: The high-level minority carrier lifetime is estimated by:
I n1 = qAhp Px21 (14) Q rr
= (19)
IF
where hp is the recombination parameter, and Px1 is the carrier
density at the boundary x1. where Qrr is the reverse recovery charge in device datasheet,
The displacement current Idisp1 in the p end region is and IF is the forward conduction current in the datasheet.
calculated by: 3) Low-doped drift region width WN-:
1 dVd 1 The ionization coefficients for electrons and holes, which
I disp1 = A (15) are electric field dependent, are given by:
Wd 1 dt
n , p = a exp( b / E ) (20)
where Wd1 is the depletion width at junction J1.
The hole current Ip2 in n+ end region is given by: where a and b are the constants, and E is the electric field.
Assuming avalanche breakdown in an abrupt junction, the
I p 2 = qAhn Px22 (16) equation for the breakdown voltage VBD as a function of the
2133
constants a and b, and n- drift region width WN-, can be given The electron mobility n with temperature dependence is
by: given by:
bWN 300 2.7
VBD = (21) n = n 0 ( ) (23)
ln( aWN ) T
The drift region width WN- can be derived from equation where n0 is electron mobility at room temperature (300 K).
(21), using a=3.15 106 cm-1, and b=1.04 107 V/cm. The The hole mobility p with temperature dependence is given
breakdown voltage VBD is the voltage rating in the by:
manufacturers datasheet plus some safety margin.
300 2.7
4) Doping concentration NN- in n- drift region: p = p0 ( ) (24)
From the empirical effective impurity doping concentration T
in n- drift region, the doping concentration NN- is assumed to where p0 is hole mobility at room temperature (300 K).
be 61015 cm-3.
The intrinsic concentration ni with temperature dependence
5) Recombination parameters hn and hp: is given by:
The recombination parameters hn and hp control the carrier
charge in the carrier stored region, and the amount of carrier 20800
ni = 1.70 1016 T 1.5 / exp( ) (25)
charge in the drift region is reduced with higher recombination T
parameters. An initial estimate of 10-14cm4/s is made for both
recombination parameter hn and hp. The junction temperature during operation is determined
using thermal RC equivalent circuits. The junction temperature
calculated from the thermal equivalent circuit is used to update
B. Refinement of parameter values
temperature-dependent parameters in the model. The new
With the values of Qrr and IF obtained from switching tests, values of temperature-dependent parameters are used to
the high-level minority carrier lifetime is refined. The high- calculated body diode current and voltage.
level minority carrier lifetime is the critical parameter
affecting the reverse recovery current waveforms. A better
V. MODEL VALIDATION
match can be achieved by refinement of high-level minority
carrier lifetime . After this, the low-doped drift region width The extracted gate-to-source switching loop and drain-to-
WN- and doping concentration NN- can be altered to improve source switching loop parasitic inductance of the PCB layout
the matching of diode voltage waveforms. are used in Pspice simulation together with SiC MOSFET
model and SiC MOSFET body diode model to validate the
device model for SiC MOSFET body diode. The SiC
MOSFET model used in simulation is from the manufacturer
Cree Inc., and the parasitic inductances of the PCB layout are
extracted using FastHenry, which is a finite difference software
tool for PCB parasitic element extraction [12]. Fig.25 shows
the simulation circuit, including the extracted parasitic
elements (red).
C. Temperature dependence
The temperature dependent equation for high level minority
carrier lifetime is given by:
T 1.5
=0( ) (22)
300
where 0 is high level minority carrier lifetime at room
temperature (300 K). Fig. 25. Equivalent circuit used for inductive switching
2134
Fig.26 shows the comparison of body diode turn-off I-V curves of SiC MOSFETs body diode at varied junction
voltage and current waveforms between experiment (dashed) temperatures are given. The dynamic characteristics of SiC
and simulation (solid) at room temperature 25oC. The DC MOSFETs body diode are tested based on a double pulse test
supply voltage is 500V, and forward conduction current of bench. The switching behavior of SiC MOSFETs body diode
body diode is 30A. The results illustrates a very good matching at different current commutation slopes, forward conduction
between simulation and experiment. In the experimental diode currents and junction temperatures is demonstrated. The device
voltage waveform, the diode voltage starts increasing and model of body diode is described in detail. The parameter
reaches a small value (about 40V) before the diode current extraction procedure for this model is introduced, which only
reaches the reverse peak current. This voltage drop is caused requires data from the manufacturers datasheets and one
by parasitic inductances from PCB layout and device packages. simple switching waveform measurement. Finally, the
The diode begins to block reverse voltage when diode current comparison between simulation and experiment proves the
reaches the reverse peak current. accuracy of the body diode model and the parameter extraction
method over a wide junction temperature.
REFERENCES
[1] H.A.Mantooth, K.Peng, E.Santi, and J.L.Hudgins. Modeling of wide
bandgap power semiconductor devices-part I, IEEE Transactions on
Electron Devices,vol.62, no.2,pp.423-433, Feb.2015.
[2] E.Santi, K.Peng, A.Mantooth and J.L.Hudgins. Modeling of wide
band-gap power semiconductor devices-part II, IEEE Transactions on
Electron Devices,vol.62, no.2,pp.434-442, Feb.2015.
[3] R.Fu, A.E.Grekov, K. Peng and E.Santi, Parameter extraction
procedure for a physics-based power SiC Schottky diode model, IEEE
Transactions on Industry Applications,vol.50, no.5,pp.3358-3568,Sep-
Oct.2014.
[4] K.Peng, and E.Santi, Performance projection and scalable loss model
Fig. 26. Comparison of body diode turn-off voltage and current waveforms of SiC MOSFETs and SiC Schottky diodes, IEEE Electric Ship
between experiment (solid) and simulation (dashed) at 25oC Technologies Symposium, ESTS 2015, pp.281-286, Jun.2015.
[5] K.Peng, S.Eskandari, and E.Santi, Analytical loss model for power
Fig.27 illustrates the comparison of body diode turn-off converters with SiC MOSFET and SiC Schottky diode pair, IEEE
voltage and current waveforms between experiment (solid) and Energy Conversion Congress and Exposition, ECCE 2015, pp.6153-
simulation (dashed) at 150oC. The DC supply voltage is 500V, 6160, Sep.2015.
and forward conduction current of body diode is 30A. A good [6] J. Jordan, V.Esteve, E.Sanchis-Kilders, E.J.Dede, E.Maset, J.B.Ejea, and
matching between experiment and simulation proves the A. Ferreres, A comparative performance study of a 1200V Si and SiC
MOSFET intrinsic diode on an induction heating inverter, IEEE
accuracy of the model over a wide junction temperature range. Transactions on Power Electronics, vol.29, no.5, pp.2550-2562, May.
2014.
[7] A. Bolotnikov, J.Glaser, J.Nasadoski, P.Losee, S.Klopman, A.Permuy
and L.Stevanovic, Utilization of SiC MOSFET body diode in hard
switching applications, Materials Science Forum, Vol. 778-780, pp.
947-950, Feb.2014.
[8] T.Funaki, M.Matsushita, M.Sasagawa, T.Kimoto, and T. Hikihara, A
study on SiC devices in synchronous rectification of DC-DC converter,
Applied Power Electronics Conference and Exposition, APEC 2007,
pp.339-344, Mar.2007.
[9] Z.Wang, J.Ouyang, J.Zhang, X.Wu and K.Sheng, Analysis on reverse
recovery characteristics of SiC MOSFET intrinsic diode, IEEE Energy
Conversion Congress and Exposition, ECCE 2014, pp. 2832-2837,
Sep.2014.
[10] T. Gachovska, J.L.Hudgins, A.Bryant, E.Santi, H.A.Mantooth and A.
K.Agarwal, Modeling, simulation, and validation of a Power SiC BJT,
IEEE Transactions on Power Electronics,vol.27,no.10,pp.4338-4346,
Fig. 27. Comparison of body diode turn-off voltage and current waveforms Apr.2012.
between experiment (solid) and simulation (dashed) at 150oC [11] X.Kang, A.Caiafa, E.Santi, J.L.Hudgins, and P.R.Palmer, Parameter
extraction for a power diode circuit simulator model including
temperature dependent effects, Applied Power Electronics Conference
VI. CONCLUSIONS and Exposition, APEC 2002, pp.452-458, Mar.2002.
The static and dynamic characterizations of SiC [12] R.Fu, A.E.Grekov, K. Peng and E.Santi, Parasitic modeling for
accurate inductive switching simulation of converters using SiC
MOSFETs body diode are provided. To our knowledge, this is devices, IEEE Energy Conversion Congress and Exposition, ECCE
the first complete characterization of SiC MOSFETs body 2013,pp.1259-1265, Sep. 2013
diode in the literature. The static characterization of SiC
MOSFETs body diode is carried out using a curve tracer. The
2135