0% found this document useful (0 votes)
74 views9 pages

Hong2020 Article PressurelessSilverSinteringOfS

Download as pdf or txt
Download as pdf or txt
Download as pdf or txt
You are on page 1/ 9

JOM, Vol. 72, No.

2, 2020
https://doi.org/10.1007/s11837-019-03815-y
Ó 2019 The Minerals, Metals & Materials Society

PROGRESS WITH LEAD-FREE SOLDERS

Pressureless Silver Sintering of Silicon-Carbide Power Modules


for Electric Vehicles

WON SIK HONG ,1,4 MI SONG KIM,1 CHULMIN OH,1


YONGJIN JOO,2 YOUNGSEOK KIM,3 and KYOUNG-KUK HONG3

1.—Electronic Convergence Materials and Device Research Center, Korea Electronics Technology
Institute (KETI), Seongnam-si, Gyeonggi-do 13509, Republic of Korea. 2.—Reliability Test
Division, iA Powertron, Inc., Incheon 21314, Republic of Korea. 3.—Environment & Energy
Research Team, Research & Development Division, Hyundai Motor Company, Hwaseong-si,
Gyeonggi-do 16082, Republic of Korea. 4.—e-mail: wshong@keti.re.kr

Pressureless silver (Ag) sintering was optimized at 250°C in vacuum and


nitrogen gas atmosphere with silicon carbide (SiC) chips, and silicon nitride
active metal-brazed substrates (AMB). A 1200-V/200-A power module was
developed using a pressureless Ag-sintered live SiC metal–oxide–semicon-
ductor field-effect transistor (MOSFET) device and a Si3N4 AMB substrate
module, and diverse reliability tests were performed. The void content and the
bonding layer thickness of the Ag sinter joints were 1.2–3.4% and 68.5 lm.
The bonding strength of the Ag sinter joints after thermal cycling testing
(TCT) were slightly decreased due to the crack generation. In contrast, high-
temperature storage testing (HTST) was carried out for a long time at high
temperature, the sintering process occurred continuously, and the shear
strength increased by 31% after HTST. The drain–source on-resistance
(RDS(ON)) of a SiC MOSFET power module before and after the TCT and power
cycle test was similar to the as-sintered state without change.

conditions.3,9 The pressurized sintering is a method


INTRODUCTION
of directly pressing the top surface of the device
Various power modules have been developed to using a pressure jig during the sintering process in
improve the power conversion efficiency of high- order to increase the densification of the sintered
power devices. In particular, high heat resistance joint and lower the void content. However, the
bonding technology has been actively studied along direct pressurization of the chip has the disadvan-
with a technology to replace conventional silicon tage of damaging the pattern on the top of the
chips with silicon-carbide (SiC) chips, which have device. In addition, because several to several dozen
wide band gaps.1–4 SiC devices cannot use lead-free power devices with several different thicknesses are
solder as a bonding material because of the temper- sintered in one power module package, it is very
ature rise of 250°C due to the heat generated during difficult to pressurize all the chips at the same time
operation; instead, gold (Au)-tin (Sn) or Au-based while maintaining a constant thickness at a uniform
alloys must be used. However, because Au alloys are pressure. Therefore, in consideration of mass pro-
very expensive, it is inconvenient to commercialize duction, a pressureless sintering process that is not
these devices.3,4 Therefore, silver (Ag) sintering and directly pressurized during sintering is considered
transient liquid phase (TLP) bonding technology are to be more advantageous.
being studied as alternative bonding techniques.1–8 The TLP bonding process has the advantage of
Ag has a very high melting point of 961°C, enabling utilizing existing lead-free soldering processes and
the use of heat-resistant solid-state sinter joints at equipment. TLP bonding is an interconnecting
very high temperatures. In addition, the Ag sinter method in which a joint formed by using lead-free
bonding process can be carried out a temperature solder at a low temperature is transferred to an
range of 200–250°C, which is lower than the melting intermetallic compound (IMC) of Sn-Cu, Ag-Sn, or
point of Ag, under pressurized or pressureless Ni-Sn systems, and can be safely used at high

(Published online October 10, 2019) 889


890 W. S. Hong, M. S. Kim, Oh, Joo, Y. Kim, and K.-K. Hong

temperatures without melting.5–9 However, it is results confirmed that the commercialization of the
difficult to confirm the change of all joints into IMCs sintered SiC chip/Si3N4 AMB substrate modules is
by nondestructive analysis methods, which is nec- feasible.
essary for mass production. Therefore, considering
current commercialization, the most suitable bond- EXPERIMENTAL PROCEDURES
ing method for SiC devices is the sinter joints
Ag Paste, SiC Device, and Si3N4 AMB Sub-
process.4–12
strate
When sintering is performed using an Ag paste
with a SiC element, the material of the substrate is A commercial Ag paste (CT2700R7S; Kyocera
also very important. Because thermo-mechanical Chemical, Japan) containing 90 wt.% Ag powder
fatigue cracks are generated in the joint due to the was used. Figure 1a and b shows photographs of the
warpage of the substrate, it is very important to use Ag-paste printed on a Si3N4 AMB substrate and the
a substrate with small warpage, a coefficient of SiC chip mounted on the Ag-paste printed substrate
thermal expansion (CTE) similar to that of SiC, and before sintering. The SiC chips were used with a
an excellent heat dissipation property to improve 1200-V/200-A and 25-mmX live MOSFET device
the reliability of the module. The CTE of SiC is (CPM2-1200-0040B; Cree, USA) and a SiC dummy
2.7 9 10 6/K, and the contents of alumina (Al2O3), die. The chip was 180 lm thick, 4.04 mm wide, and
AlN, and silicon nitride (Si3N4) are 6.7 9 10 6/K, 6.44 mm long. The top and back were sputtered with
4.5 9 10 6/K, and 2.6 9 10 6/K, respectively. Si3N4 Al (4 lm) and Ni (0.8 lm)/Ag (0.6 lm), respectively.
is the most reliable substrate material because it A Si3N4 AMB substrate with Ag finish was used.
has a CTE most similar to that of SiC and has a The substrate was 12.0 mm wide, 24.0 mm long,
bending strength of approximately 800 MPa.13 and 0.92 mm thick [0.3 (Cu)/0.32 (ceramic)/0.3 (Cu)
In this study, a 1200-V/200-A metal-oxide semi- mm]. The ZTA AMB substrate was composed of
conductor field-effect transistor (MOSFET) power Al2O3 + ZrO2 mixed ceramics (Amosense, Korea),
module package for automobiles was developed and was 12.0 mm wide, 24.0 m long, and 0.72 mm
using pressureless Ag sinter interconnection tech- thick [0.2 (Cu)/0.32 (ceramic)/0.2 (Cu) mm]. The
nology. The MOSFET module was assembled using base-plate material was C1220 Cu (90.0 mm 9 45.0
a Ag-sintered SiC chip/Si3N4 active metal-brazed mm, 3.0 mm thick), and had a thickness of 150 lm;
(AMB) substrate module, aluminum (Al) wire, cop- a metal mask (stainless steel, SUS304) with a 70%
per (Cu) terminal, Cu base plate, and plastic case. printing area window was used.
Moreover, the pressureless Ag sinter bonding pro-
cess was optimized at 250°C in a vacuum and Pressureless Ag-Sintering Condition
nitrogen gas atmosphere using an Ag paste, SiC
Figure 1c and d shows the vacuum bonding
chips, and Si3N4 AMB substrates. The void maxi-
furnace (Ajeon-010; Ajeon Heating Industrial,
mum content of the sinter joint was 3.4%, and the
Korea) used for the pressureless Ag sintering pro-
initial shear strength, bonding layer thickness
cess. Figure 1e shows the temperature and pressure
(BLT), and densification were 32.7 MPa, 68.5 lm,
profiles of the pressureless sintering process. The
and 91.9%, respectively. To compare the reliability
sintering was carried out at 250°C without pressure
of the Ag sinter joints, thermal cycling tests (TCT),
in vacuum and nitrogen atmosphere for 90 min.
high-temperature storage tests (HTST), and power
cycling tests (PCT) were conducted for 500 cycles,
Reliability Test Conditions
700 h, and 1000 cycles, respectively.14–16 After the
TCT and HTST, the bond strengths were 25.7 MPa To compare the reliability of the sinter joints
and 42.9 MPa, respectively and the densification before and after Ag sintering, TCTs, HTSTs, and
was 90.3% and 95.4%, respectively. Furthermore, power cycling tests were conducted. The 1200-V/
the bond strength after the HTST was higher. These 200-A SiC MOSFET power modules were utilized
results indicate that the sintering process is contin- for the reliability tests, with samples assembled
uously generated during the long test at high using an Ag-sintered SiC chip/Si3N4 AMB sub-
temperature, and the densification of the joint is strate module, aluminum wire, Cu terminal, and Cu
higher, which leads to an increase in the joint base plate with a plastic case, as shown in Fig. 2.
strength. The drain–source on-resistance (RDS(ON)) The TCT temperature ranged between 50°C
was measured after the reliability tests. The RDS(ON) and 150°C (DT = 250°C) with a soak time of
after the TCT and PCT tests was 6.63–6.97 mmX, 10 min for 500 cycles using a thermal cycling
which was similar to the initial value of 6.60–6.99 chamber (TSA-11A; Espec, Japan). The HTST was
mmX. The microstructure and interface of the SiC carried out at 200 ± 3°C for 700 h using a high-
chip/AMB substrate Ag-sinter joints were observed temperature storage oven (OF-22GW; Jeio Tech,
through scanning electron microscopy (SEM) and Japan). Power cycling was carried out at 50–100°C
energy dispersive x-ray spectroscopy (EDS). The (DTc = 50°C) and 1–6 min dwell time at each
Pressureless Silver Sintering of Silicon-Carbide Power Modules for Electric Vehicles 891

Fig. 1. Photographs of pressureless sintering process: (a) Ag-paste printing on a Si3N4 AMB substrate, (b) SiC chip mounting on an Ag-paste printed
substrate, (c, d) pressureless sintering machine set-up and samples, respectively, and (e) temperature and pressure profiles for Ag sintering.

Fig. 2. Schematic diagrams of (a) components, (b) cross-sectional structure, (c) before case molding assembly, and photographs of (d) before
and (e) after plastic case molding assembly.

temperature for 1000 cycles, as shown in Fig. 3, reached 150°C, the power supply was stopped and
using a power cycling test machine (Ontest, allowed to cool until the temperature of the case
Korea). The test method of power cycling for the was 50°C. This procedure was repeated for 1000
1200-V/200-A SiC MOSFET modules was carried cycles.16 After finishing each reliability test, the
out according to the JEDEC standards: 20 V RDS(ON) was measured with a power device ana-
applied at the gate and then 160 A applied at lyzer for circuit design (B1506A; Keysight Tech-
the low side, and once the device’s temperature nologies, USA).
892 W. S. Hong, M. S. Kim, Oh, Joo, Y. Kim, and K.-K. Hong

Fig. 3. Photographs of the power cycling test: (a) pressureless Ag-sintered 1200-V/200-A SiC MOSFET power module, (b) schematic of the test
process, (c) power cycling test machine, and (d) power cycling test.

Densification Measurement of Ag-Sinter sinter joints; finally, the bonding strength was
Joints decreased after the TCT. Therefore, the TCT is
more effective than the HTST for evaluating the
After the pressurized sintering, the cross-sec-
sinter joints.
tional area of the Ag-sinter joints was examined.
The cross-sectional morphology was obtained and
Bonding Layer Thickness, Densification,
the porosity of the Ag sinter joints was measured
and Microstructure of Pressureless Ag-Sinter
using the i-solution software (Image & Microscope
Joints
Technology, USA). The densification of the Ag-
sinter joints was calculated as the ratio of the Figure 5a–c shows the cross-sectional SEM micro-
porosity to the sinter joint area. graphs used to measure the BLT and densification
of the Ag sinter joints after the TCT and HTST
RESULTS AND DISCUSSION compared with the as-sintered state, and Fig. 5d
shows the measurement results of the BLT and the
Void Content and Shear Strength of Pressure-
densification. As shown in Fig. 5d, the initial BLT of
less Ag-Sintered SiC Chip/Si3N4 AMB Sub-
the Ag-sinter joints was 68.5 lm; after the TCT and
strate Joints
HTST, the BLT was 68.4 lm and 75.1 lm,
Figure 4a shows a photograph of the pressureless respectively.
Ag-sintered SiC Chip/Si3N4 AMB substrate sample The initial densification of the Ag-sinter joint
and Fig. 4b and c are non-destructive x-ray analysis measured from Fig. 5a–c was 91.9%, and, after the
images. The void content was 1.2–3.4% of the sinter TCT and HTST, it was 90.3 and 95.4%, respectively.
joint layer, which was the average of the three The densification remained similar after the TCT,
samples. whereas it increased after the HTST, similar to the
Figure 4d compares the void content and the void content. Because the TCT can generate thermo-
shear strength before and after the TCT and HTST. mechanical fatigue cracks at the Ag-sinter joints
The bond strength of the SiC chip/Si3N4 substrate of due to the CTE, the densification after the TCT was
the as-sintered sample was 32.7 MPa and the void similar or lower than the initial one. Similar to the
content was 4.1%. After the TCT, the bond strength principle of void reduction after the TCT, the
decreased to 25.7 MPa and, after the HTST, it densification also increased after the test. The
increased to 42.9 MPa, compared with the as-sin- densification increased because the HTST was
tered value. The void contents were 5.0% and 3.3% applied at a high temperature for a long time, so
after the TCT and HTST, respectively. It should be that sintering based on the solid-state diffusion
noted that a void content of 10% or less is accept- bonding occurred continuously during the test.
able for mass production. Because a high tempera- Figure 6 shows SEM images of the (a) as-sintered
ture was continuously maintained during the SiC chip/ZTA AMB substrate joint and (b–d) mag-
HTST, the Ag sintering process occurred, which nified images. It was confirmed that the sinter joint
led to the increase in the shear strength of the Ag was densely bonded. In addition, it can be seen that
sinter joints. However, the TCT caused a mismatch sintering was successfully performed at the inter-
of the coefficient of thermal expansion (CTE) in the face between the back side of the SiC chip with the
Pressureless Silver Sintering of Silicon-Carbide Power Modules for Electric Vehicles 893

Fig. 4. (a) Photograph of pressureless Ag-sintered SiC chip/AMB substrate module, (b, c) non-destructive x-ray analysis images, and (d) a graph
of shear strength and void content comparison of Ag sinter joints before and after the TCT and the HTST.

Ag-coated surface and the Ag paste. In particular, during the TCT induced by the difference in CTE
an Ag-Cu alloy layer was formed at the interface between the SiC, sintered-Ag, and substrate. The
between the Ag sinter matrix and the Ag-finish shape of the crack indicates a typical fatigue
surface of the substrate.17,18 This Al-Cu alloy layer crack.23 The pores were observed in a grain, and
is considered to be formed by inter-diffusion some voids existed at the triple point of particles.
between the Cu layer and the Ag coating layer of Figure 7d shows the microcrack evolution at the
the substrate.17,18 In the Ag-sinter joint, it was triple contact points of the Ag-sinter joint due to the
observed that the network structure formed thermal fatigue during the TCT. The microcracks
between the powders was sufficient to reach a were mainly observed at the triple contact point of
densification of 91.9%. During the sintering process the particles. The triple contact points around the
of the solid state, the network formation process of pores or voids were in the continuous sintering
the Ag powders is carried out through the original process, those areas acting as a fracture site where
point contact, neck growth, pore-channel closure, cracks could occur. As the operation progressed,
pore rounding, densification and pore shrinkage, these microcracks propagated inside the sinter joint
and pore coarsening.3,17,19–22 When these sintering and the bond strength decreased.
processes are successfully performed, the densifica-
tion can be 90% or more. RDS(ON) of SiC MOSFET Power Module After
Figure 7a shows SEM images of the Ag sinter the Power Cycling
joint of the SiC chip/AMB substrate after the TCT. Figure 8 shows SEM images of the 1200-V/200-A
Figure 7b–d shows an enlarged view of the left side SiC MOSFET power module Ag-sinter joints after
and the AMB substrate side of the SiC chip, and the the PCT for 1000 cycles and after the TCT for 500
sintered-Ag matrix, respectively. Figure 7b and c cycles were compared to those in the as-sintered
shows that cracks are generated in the sinter joint state. The RDS(ON) of SiC MOSFET power module
due to the repeated thermo-mechanical stress was measured before and after the reliability tests.
894 W. S. Hong, M. S. Kim, Oh, Joo, Y. Kim, and K.-K. Hong

Fig. 5. Cross-sectional SEM micrographs for measuring the bonding layer thickness and densification of the pressureless Ag-sinter joints: (a) as-
sintered, (b) after 500 thermal cycles, and (c) after 700 h high-temperature storage; and (d) BLT and densification measurement results.

Fig. 6. SEM images of (a) as-sintered SiC chip/ZTA AMB substrate joint and a magnified view of (b) SiC die side, (c) Ag sinter joint, and (d)
substrate side.
Pressureless Silver Sintering of Silicon-Carbide Power Modules for Electric Vehicles 895

Fig. 7. SEM images of (a) Ag sinter joint of SiC chip/AMB substrate after the TCT. Enlarged views of (b) the left side, (c) the AMB substrate side
of the SiC chip, and (d) the sintered Ag, respectively.

The RDS(ON) represents the total resistance in the to the chip joint, so that a crack occurs more rapidly.
path between the drain and source in a MOSFET For these reasons, the PCT test is suitable for
module and consists of a number of resistances. evaluating the reliability of the die attachment
Because the sinter joint is bonded to the SiC chip, joint, and the TCT test is suitable for verifying the
its resistance is included in RDS(ON). Therefore, if a reliability of a large-area joint. Comparing the
defect occurs in the Ag-sinter joints, the on-resis- microstructures of the SiC chip/Ag powder sinter
tance will increase. The RDS(ON) is the basis for the joints in Fig. 8, the porosity or micro-pores of the
maximum current rating of the MOSFET and is also sinter joints after the PCT test can be seen to be
associated with current losses. Thus, the power larger and higher in the PCT test samples than the
conversion efficiency of the power module is TCT samples. This is because the PCT test induces
increased when the RDS(ON) is reduced. Finally, it more severe conditions for the sinter joints, as the
is important to observe the change in the RDS(ON) to device is directly driven to generate heat.
observe the bonding state of the module.
The RDS(ON) specification of the 1200-V/200-A SiC CONCLUSION
MOSFET device was 8 mmX and the initial RDS(ON)
was 6.60–6.99 mmX. After the TCT and PCT, the Pressureless Ag sintering was optimized using a
RDS(ON) was 6.77–6.92 X and 6.63–6.97 X, respec- SiC chip on a Si3N4 AMB substrate at 250°C for
tively. These results indicated that, after the relia- 90 min without pressure in a vacuum and nitrogen
bility test, the RDS(ON) remained constant during atmosphere. The void content and shear strength of
module operation, which in turn indicates that the the SiC chip/AMB substrate sinter joints were 3.3–
sinter joint was well maintained. 5.0% and 32.7 MPa, respectively. The BLT of the
The magnified SEM images of the SiC chip side pressureless Ag sinter joint was maintained at a
and the Si3N4 AMB substrate side indicated that constant thickness of 68.4–75.1 lm. To evaluate the
the sinter joint was well formed in the as-sintered sinter joint reliability, the shear strength, BLT, and
joints and that the network microstructure was densification were measured after the TCT and
maintained after the PCT and the TCT. HTST. The shear strengths were 25.7 MPa and
The densification of the sinter joints after PCT 42.9 MPa after the TCT and HTST, respectively.
and TCT was similar, and cracks were not observed. The shear strength after the HTST was increased
The PCT test drives the device directly to maintain compared to the initial strength. Because sintering
high and low temperatures, thus applying the worst progresses continuously during the HTST, the
drive conditions to the sinter joints of the chip. On densification of the sinter joints increased to
the other hand, because the TCT test applies heat to 95.4%, which caused an increase in the shear
the entire module, more stress is applied to a larger strength. The microcracks were generated at the
area, such as a substrate or a base plate, in addition Ag-sinter joint due to the thermal fatigue during the
896 W. S. Hong, M. S. Kim, Oh, Joo, Y. Kim, and K.-K. Hong

Fig. 8. SEM images of the Ag sinter joints of the 1200-V/200-A SiC MOSFET power module before and after the reliability tests: (a–c) as-
sintered state, (d–f) after power cycling test for 1000 cycles, and (g–i) after the TCT for 500 cycles. Magnified SEM images of (b, e, h) the SiC
chip side and (c, f, i) the Si3N4 AMB substrate side.

TCT. The microcracks were observed to occur at the that, after the reliability test, the RDS(ON) was
triple bonding point of the grains. After a long time, maintained constant during operation because the
these microcracks propagated inside the sinter sinter joint was well maintained. The densifications
joint, leading to a decrease in the bond strength. after PCT and TCT were similar, and no cracks
The initial RDS(ON) of the 1200-V/200-A SiC were observed. The PCT test conditions are more
MOSFET module was 6.60–6.99 mmX. After the unfavorable for sinter joints than those of the TCT
TCT and PCT, the RDS(ON) were 6.77–6.92 X and because the device is directly driven to apply heat.
6.63–6.97 X, respectively. These results indicated Therefore, the porosity and microstructures of the
Pressureless Silver Sintering of Silicon-Carbide Power Modules for Electric Vehicles 897

sinter joints after the PCT test were slightly larger 9. C. Chen, D. Kim, Z. Wang, Z. Zhang, Y. Gao, and K. Suga-
than before the test; however, the sinter joints were numa, Ceram. Int. 45, 9573 (2019).
10. H. Zhang, W. Li, Y. Gao, H. Zang, J. Jiu, and K. Suganuma,
maintained. Therefore, the commercialization of J. Electron. Mater. 46, 5201 (2017).
SiC MOSFET modules without pressure sintering 11. H. Chin, K. Cheong, and A. Ismail, Metall. Mater. Trans. B
is considered feasible. 41, 824 (2010).
12. M.S. Kim, C. Oh, and W.S. Hong, J. Weld. Join. 37, 15
ACKNOWLEDGEMENTS (2019).
13. Maruwa Co., LTD. www.maruwa-g.com/e/products/cerami
This research was supported by the Korea Eval- c/ceramic-substrate-4.html. Accessed on 20th Jan 2019.
uation Institute of Industrial Technology (KEIT) 14. S. Noh, H. Zhang, and K. Suganuma, Materials 11, 2531
and the Ministry of Trade, Industry & Energy (2018).
(MOTIE) of the Republic of Korea (No. 10063263). 15. K. Suganuma, S. Sakamoto, N. Kagami, D. Wakuda,
K.S. Kim, and M. Nogi, Micronelectron. Reliab. 52, 375
We thank Professor Yongil Kim of Sungkyunkwan (2012).
University for his help with the power terminal 16. JEDEC Standard, JESD22-A122A, Power Cycling (JEDEC
bonding of the power module assembly. Solid State Technology Association, Arlington, VA, 2016).
17. W.S. Hong and S.S. Cha, J. Microelectron. Packag. Soc. 19,
REFERENCES 67 (2012).
18. S. Divinski, M. Lohmann, and C. Herzig, Acta Mater. 49,
1. T. Kunimune, M. Kuramoto, S. Ogawa, T. Sugahara, S.
249 (2001).
Nagao, and K. Suganuma, Acta Mater. 89, 133 (2015).
19. J.S. Hirschhorn, Introduction to Powder Metallurgy (New
2. R. Khazaka, L. Mendizabal, D. Henry, and R. Hanna, IEEE
York: The Colonial Press, 1969), pp. 155–273.
Trans. Power Electron. 30, 2456 (2015).
20. S.-K. Lin, S. Nagao, E. Yoko, C. Oh, H. Zhang, Y. Liu, S. Lin,
3. W.S. Hong, M.S. Kim, D. Kim, and C. Oh, J. Electron. Mater.
and K. Suganuma, Sci. Rep. 6, 34769 (2016).
48, 122 (2019).
21. C. Oh, S. Nagao, T. Kunimune, and K. Suganuma, Appl.
4. J. Biela, M. Schweizer, S. Waffler, and J. Kolar, IEEE Trans.
Phys. Lett. 104, 161603 (2014).
Ind. Electron. 58, 2872 (2011).
22. C. Oh, S. Nagao, and K. Suganuma, J. Mater. Sci.: Mater.
5. Y. Gao, A. Huang, S. Krishnaswami, J. Richmond, and A.
Electron. 26, 2525 (2015).
Agarwal, IEEE Trans. Ind. Appl. 44, 887 (2008).
23. R. Shioda, Y. Kariya, N. Mizumura, and K. Sasaki, J.
6. C. Chen, Z. Zang, C. Choe, D. Kim, S. Noh, T. Sugahara, and
Electron. Mater. 46, 1155 (2017).
K. Suganuma, J. Electron. Mater. 48, 1106 (2019).
7. H. Zhang, C. Chen, S. Nagao, and K. Suganuma, J. Electron.
Mater. 46, 1055 (2017). Publisher’s Note Springer Nature remains neutral with re-
8. S.W. Yoon, M.D. Glover, and K. Shiozaki, IEEE Trans. gard to jurisdictional claims in published maps and institutional
Power Electron. 28, 2448 (2013). affiliations.

You might also like