Japanese Journal of Applied Physics You may also like REGULAR PAPERS Effect of surface roughness of trench sidewalls on electrical properties in 4H-SiC trench MOSFETs - X-ray imaging and spectroscopy of nitrogen in the SiO2/SiC interface of the 4H–SiC MOSFET trench sidewalls Noritake Isomura, Kousuke Kitazumi, Keita Kataoka et al. To cite this article: Katsuhiro Kutsuki et al 2018 Jpn. J. Appl. Phys. 57 04FR02 - A review of manufacturing technologies for silicon carbide superjunction devices Run Tian, Chao Ma, Jingmin Wu et al. View the article online for updates and enhancements. - Nucleation Behavior of III/V Crystal Selectively Grown Inside Nano-Scale Trenches: The Influence of Trench Width S. Jiang, C. Merckling, A. Moussa et al. This content was downloaded from IP address 130.184.252.14 on 28/04/2025 at 17:24 REGULAR PAPER Japanese Journal of Applied Physics 57, 04FR02 (2018) https://doi.org/10.7567/JJAP.57.04FR02 Effect of surface roughness of trench sidewalls on electrical properties in 4H-SiC trench MOSFETs Katsuhiro Kutsuki1,2*, Yuki Murakami2, Yukihiko Watanabe1, Toru Onishi2, Kensaku Yamamoto3, Hirokazu Fujiwara2, and Takahiro Ito2 TOYOTA CENTRAL R&D LABS., Inc., Nagakute, Aichi 480-1192, Japan TOYOTA MOTOR Corporation, Toyota, Aichi 470-0309, Japan 3 Research Division 3, DENSO Corporation, Nisshin, Aichi 470-0111, Japan 1 2 *E-mail: kutsuki-k@mosk.tytlabs.co.jp Received October 2, 2017; revised November 16, 2017; accepted November 17, 2017; published online February 19, 2018 The effects of the surface roughness of trench sidewalls on electrical properties have been investigated in 4H-SiC trench MOSFETs. The surface roughness of trench sidewalls was well controlled and evaluated by atomic force microscopy. The effective channel mobility at each measurement temperature was analyzed on the basis of the mobility model including optical phonon scattering. The results revealed that surface roughness scattering had a small contribution to channel mobility, and at the arithmetic average roughness in the range of 0.4–1.4 nm, there was no correlation between the experimental surface roughness and the surface roughness scattering mobility. On the other hand, the characteristics of the gate leakage current and constant current stress time-dependent dielectric breakdown tests demonstrated that surface morphology had great impact on the long-term reliability of gate oxides. © 2018 The Japan Society of Applied Physics 1. Introduction Silicon carbide (SiC), a compound semiconductor of silicon and carbon, has superior physical properties to Si in achieving high blocking voltage, high-temperature operation, and low loss at high switching speed. SiC devices enable efficient use of electric power, allowing the reduction in the size of power supplies and the simplification of cooling systems.1) In addition, SiC is the only wide-gap semiconductor on which a SiO2 layer can be formed by conventional thermal oxidation as well as silicon. However, there are severe stopping issues regarding the MOS channel resistivity and gate oxide reliability toward the practical use of SiC-based MOS devices. To reduce the channel resistivity, we focused on the equation of drain current (ID) in MOSFETs. The drain current in the linear region can be described by 2) ID ¼ W eff COX ðVG Vth ÞVD ; L ð1Þ where W is the channel width, L is the channel length, μeff is the effective channel mobility, COX is the oxide capacitance, VG is the gate voltage, Vth is the threshold voltage, and VD is the drain voltage. A decrease in L causes Vth to decrease owing to the short-channel effect and punch-through in the off state depending on the impurity density in the channel region. Although an increase in COX by dielectric layers with high permittivity (high-k), e.g., HfAlON, was reported with appropriate oxide thickness in the application of power devices,3) the gate leakage current through a high-k layer due to a narrow band gap4) is a severe problem. Moreover, a decrease in Vth makes it difficult to design normally-off SiC MOSFETs in high-voltage power devices. For the above reasons, increasing W and μeff is concluded to be effective in increasing ID without disadvantageous effects. One of the best ways to increase W and μeff simultaneously is the use of trench MOSFETs (UMOSFETs). Trench MOSFETs enable us to shrink cell pitch, which results in the increase in channel width. In addition, when trench MOSFETs are fabricated on a Si-face, trench sidewalls face correspond to non-polar planes, e.g., ð1100Þ or ð1120Þ (m-face or a-face), which are reported to reveal high electron channel mobility compared with a (0001) Si face.5) In fact, many reports have demonstrated several types of trench MOSFETs with relatively small specific-on-resistance (RonA) values.3,6–17) However, the use of trench MOSFETs still has challenges. One of the challenges is a channel mobility lower than that expected from lateral MOSFETs. The reported field effect channel mobility is in the range of only 5–50 cm2 V−1 s−1. The limiting factors of channel mobility are also unclear. Another challenge is ensuring the reliability in gate dielectrics. How to control the shape and morphology of trenches is believed to be essential for the above-mentioned challenges. Since SiC is physically hard and chemically stable compared with Si, its etching rate is low even when using etching equipment employing high-density plasma, and the etched shape is difficult to control. Therefore, the transformation of SiC trenches has been reported at high temperatures.18–23) For instance, Kawada et al. proposed the two-step process of annealing at 1700 °C in SiH4=Ar atmosphere followed by annealing at 1500 °C in H2 atmosphere, which enabled simultaneous improvements in the shape of trenches and in the smoothness of substrate surfaces without significant etching.22) However, they did not pay attention to the surface morphology of trench sidewalls, which correspond to the channel region in MOSFETs. Moreover, the impact of the surface roughness of trench sidewalls on electrical properties has not been discussed yet. In this study, we investigated the surface roughness of trench sidewalls and its effects on electrical properties, especially channel mobility and the long-term reliability of gate oxides. To discuss the oxide reliability in detail, the results of current constant stress time-dependent dielectric breakdown (CCS-TDDB) tests were added to the original extended abstract.24) 2. Experimental methods The starting material was an n-type 4H-SiC epitaxial layer grown on a heavily doped n+-SiC (0001) substrate. The 04FR02-1 © 2018 The Japan Society of Applied Physics Jpn. J. Appl. Phys. 57, 04FR02 (2018) K. Kutsuki et al. AFM images Condition A Condition B Condition C (a) (b) (c) [0001] [1120] [ ] [ Fig. 1. (Color online) Schematic image of AFM analysis at trench sidewalls. 100nm ] 3.5 Height (nm) Scan area (400nm x 400nm) eff ¼ L @ID =@VD : W COX ðVG Vth Þ ð2Þ For further investigation, μeff was analyzed on the basis of the proposed mobility model including optical phonon scattering mobility (μOP).27) The optical phonon scattering has been taken into account for the bulk mobility of SiC, and reported to have great impact on the total mobility.28) Therefore, instead of acoustic phonon scattering mobility in conventional mobility models,29–33) μOP was applied to the improved mobility model with Coulomb scattering mobility (μC)32) and surface roughness scattering mobility (μSR)32) as follows: 1 1 1 1 ¼ þ þ ; ð3Þ eff C SR OP C T NS C 1þ ; ð4Þ C ¼ NT Nscr -10.0nm Line profiles (parallel) (d) (e) (f) Ra =0.40nm Ra =0.66nm Ra =1.4nm 0 -3.5 0 500 500 500 Distance (nm) Height (nm) isolated one-cell trenched MOSFETs with n-channel were fabricated in order to extract the channel resistance component and determine μeff. The surface roughness of trench sidewalls was controlled by high-temperature annealing after trench SiC etching (conditions A, B, and C). The high-temperature annealing was performed above 1000 °C after activation annealing, which was followed by a gate fabrication process. The deposition of a 75-nm-thick SiO2 layer as gate oxide was followed by nitridation. The nitridation process was highly effective, but the Dit values at SiO2=SiC interfaces were still above 2 × 1012 cm−2 eV−1 at EC − E = 0.2 eV. To evaluate the surface morphology of trench sidewalls, each sample was cleaved along the trench direction. Then, the gate electrode and gate oxide were removed. The scanning area of atomic force microscopy (AFM) was 400 nm square as shown in Fig. 1. The impurity concentration in the channel region (NA) was determined by secondary ion mass spectrometry (SIMS), and the thickness of the gate oxide and the taper angle of trenches are evaluated by cross-sectional transmission electron microscopy (TEM) (data not shown). The evaluation of electrical properties of isolated one-cell trench MOSFETs consists of the drain current–gate voltage (ID–VG) properties, gate leakage current–gate voltage (IG–VG) properties, and CCS-TDDB test results. The ID–VG characteristics were measured at temperatures in the range of −40 to 250 °C. The interface state density (Dit), Vth and μeff were calculated directly from the measured ID–VG charactesistics.25,26) Dit was calculated from the subthreshold swing of the measured ID–VG curves. The calculated Dit locates at the energy level in the range of 0.1 to 0.2 eV from the conduction band edge of SiC. According to Eq. (1), μeff was defined by 10.0nm 10 Line profiles (perpendicular) (h) (g) 5 (i) 0 -5 Ra =2.6nm Ra =0.95nm -10 0 500 Ra =2.4nm 500 500 Distance (nm) Fig. 2. (Color online) Panels (a)–(c) show AFM images of samples fabricated under conditions A, B, and C. The scanning area is 400 nm square. The crystal orientation and current direction are also described. Panels (d)–(f) are line profiles parallel to the current direction, which correspond to the AFM images (a)–(c). In addition, panels (g)–(i) are line profiles perpendicular to the current direction, which correspond to the AFM images (a)–(c). Ra values of line profiles are described in the figures. ; E2eff C 1 C ħ!OP ¼ OP ¼ exp 1 ; Eeff Nð!OP Þ Eeff kT q ðNdpl þ NS Þ; Eeff ¼ "SiC SR ¼ ð5Þ ð6Þ ð7Þ where NT is the total number of trapped charges, which include fixed and interface-trapped charges, T is the absolute temperature, NS is the surface inversion carrier concentration, Eeff is the effective field,27,30) N(ωOP) is the phonon occupation factor, ħωOP is the optical phonon energy, k is Boltzmann’s constant, εSiC is the permittivity of SiC, Ndpl is the surface concentration of the depletion charge, η is taken to be 1=2 for the electron mobility, and ΓC, Nscr, ζC, δ, and C are empirical parameters. The empirical parameters were the same among the conditions considered. In relation to Eeff, η was defined on the basis of the Si mobility model.30) μOP is assumed to be proportional to E1 eff . Under all conditions, an excellent agreement with experimental data was confirmed, and the errors between the calculated and experimental effective mobilities were less than 6% in the range of −40 to 250 °C. In this study, we investigated the impact of surface morphology on these three mobility factors, i.e., μC, μSR, and μOP. 3. Results and discussion 3.1 Surface morphology at trench sidewalls Figures 2(a)–2(c) show AFM images of samples fabricated under conditions A, B, and C. A striation can be seen under conditions B and C. The corresponding line profiles parallel to the direction of the drain current are shown in Figs. 2(d)– 04FR02-2 © 2018 The Japan Society of Applied Physics 1.5 ¼10-4 1.0 ¼10-4 K. Kutsuki et al. 12 2.0 ¼10-5 Effective Mobility, µeff (cm2V-1s-1) Drain Current, ID (A) 2.0 ¼10-4 Transconductance, gm(S) Jpn. J. Appl. Phys. 57, 04FR02 (2018) 1.5 ¼10-5 1.0 ¼10-5 gm 5.0 ¼10-6 0 0 5 Ra: 0.40nm 0.66nm 1.4nm 10 15 20 VG (V) 5.0 ¼10-5 VD=0.1V measured at 25°C 0 0 5 10 15 20 Gate Bias, VG (V) Table I. Ra values of line profiles are measured both parallel and perpendicular to the current direction for each sample. Vth and Dit are also listed. Ra values of line profiles (nm) Ra: 0.40nm 0.66nm 1.4nm 8 6 4 2 VD=0.1V measured at 25C 0 0.6 0.8 1 1.2 1.4 1.6 Effective Field, Eeff (MV/cm) Fig. 3. (Color online) ID–VG characteristics of isolated one-cell trench MOSFETs with different surface morphology. The measurement temperature was 25 °C and the drain voltage was 0.1 V. The inset shows the transconductance properties. Parallel Perpendicular Vth (V) Dit (×1012 cm−2 eV−1) A 0.40 0.95 3.4 2.8 B 0.66 2.6 3.3 2.8 C 1.4 2.4 3.5 2.9 Sample conditions 10 2(f), and those perpendicular to the drain current are shown in Figs. 2(g)–2(i). The arithmetic average roughness (Ra) values of line profiles are also described in the figures. It is demonstrated that the surface roughness could be controlled at Ra values parallel to the drain current in the range of 0.4 to 1.4 nm, judging from Figs. 2(d)–2(f). The line profiles perpendicular to the current flow also revealed that the average peak-to-valley distance of striations, as shown in AFM images, [Figs. 2(b) and 2(c)], was about 10 nm. The trench shape of each sample was confirmed to be almost the same and the trench angle in the channel region was 87° by cross-sectional TEM analysis (data not shown). 3.2 Effects on channel mobility Next, ID–VG characteristics of isolated one-cell trench MOSFETs were evaluated at temperatures in the range of −40 to 250 °C. In this section, to discuss the relationship between channel mobility and surface roughness, we focused on the surface roughness parallel to the direction of the drain current. Figure 3 shows ID–VG curves with different surface morphology measured at 25 °C. The drain voltage (VD) was 0.1 V. The inset shows the transconductance (gm) properties. The values of Vth and Dit for each sample are summarized in Table I with Ra values of line profiles both parallel and perpendicular to the current direction. The correlation between effective field and channel mobility is also shown in Fig. 4. The effective mobility in this study was found to be lower than that expected from lateral MOSFETs. In addition, the effects of surface roughness on ID–VG characteristics and extracted parameters appear to be very small at 25 °C. For further investigation, μeff was broken down into three factors, namely, μC, μSR, and μOP, on the basis of the Fig. 4. (Color online) Correlation between effective field and effective mobility at 25 °C at different surface roughness. proposed mobility model, and then the limiting factors of channel mobility for each sample were analyzed. Figure 5 shows the correlation between Ra values from line profiles and mobility factors measured at each temperature. The mobility factors are plotted at Eeff = 1 MV=cm. The results clearly indicate that the limiting factors of the total mobility (μTOTAL) were μC at low temperatures (≤25 °C) and μOP at high temperatures (≥150 °C). In addition, μSR had a small contribution to μTOTAL, and it had no correlation with Ra in the range of 0.4 to 1.4 nm. The result about correlation between Ra and surface roughness scattering is similar to the case in planar MOSFETs.33) If a much smoother face is achieved at trench sidewalls with Ra less than 0.4 nm, Ra will correlate with μSR. In terms of the evaluation method for surface roughness, however, further investigation is still required. Compared with the mean free path of electrons in the inversion layers of 4H-SiC MOSFETs (∼15 Å), the interface roughness measured by AFM on the length scale of 100 Å and larger is large enough not to affect electron scattering. From the viewpoint of drastic improvement in channel mobility, the results shown in Fig. 5 suggest that μC and μOP should be increased. To increase μC, interface states located at the SiO2=SiC interface, fixed charges near interface, and charge traps in the SiO2 layer should be reduced. On the other hand, the origin of μOP is a lattice vibration; thus, the crystal structure of SiC near SiO2=SiC interface is considered to affect μOP properties. How to evaluate this phenomenon physically and=or electrically and how to control μOP are next key issues that should be resolved. If both μC and μOP increase, μSR will have strong impact on the total mobility. 3.3 Effects on reliability in gate oxides Finally, we investigated the effects of surface roughness on reliability in gate oxides. Figure 6 shows IG–VG characteristics of trench MOSFETs with different surface morphology. AFM images and Ra values from line profiles perpendicular to the current direction corresponding to each IG–VG curve are also described in Fig. 6. There was no correlation between IG–VG properties and the surface roughness parallel to the direction of the drain current. On the other hand, the surface roughness perpendicular to the drain current shows strong correlation with IG–VG properties. With the increase in Ra, the gate leakage current started to flow at a lower gate 04FR02-3 © 2018 The Japan Society of Applied Physics Mobility at Eeff=1MV/cm (cm2V-1s-1) Jpn. J. Appl. Phys. 57, 04FR02 (2018) K. Kutsuki et al. -40ºC 1000 25ºC (b) (a) 250ºC 150ºC (c) µC (d) µOP 100 10 µSR µSR µOP µC µC µTOTAL µC µSR µSR µOP µTOTAL µTOTAL µTOTAL µOP 1 0.0 0.5 1.0 1.5 2.0 0.5 1.0 1.5 2.0 0.5 1.0 1.5 0.5 2.0 1.0 1.5 2.0 Ra values from line profile (nm) Gate Leakage, IG (A) 1¼10-1 Charge to breakdown, QBD (C/cm2) Fig. 5. (Color online) Correlation between Ra values from line profiles and mobility factors measured at each temperature. The mobility factors are plotted at Eeff = 1 MV=cm. 2.4nm 1¼10-3 200nm 1¼10-5 2.6nm 1¼10-7 1¼10-9 [ ] [ Current direction ] 1¼10-11 1¼10-13 0 0.95nm VD=VS=0V measured at 25C 20 40 60 Fig. 6. (Color online) IG–VG characteristics of trench MOSFETs with different surface morphology. AFM images and Ra values from line profiles perpendicular to the current direction for each IG–VG curve are also described. bias. This result indicates that the main factor affecting reliability in gate oxides was the surface roughness perpendicular to the drain current. Subsequently, the CCS-TDDB tests were performed at 25 °C. Figure 7 shows the results of charge to breakdown (QBD) as a function of Ra. The constant current was 10 µA during the CCS-TDDB tests. The surface roughness was found to affect QBD directly. In the case of the sample with a Ra value of 0.95 nm, QBD is of the same level as the reported value.6) With the increase in surface roughness, however, QBD drastically degraded. It is assumed that the electric field locally concentrates at the rough SiO2=SiC interface, which locally increases the amount of charge passing through gate oxides and shortens the time to reach the dielectric breakdown. For further consideration, we focused on the relationship between electric field concentration and surface roughness based on Gauss’s law. Gauss’s law is a general law applied to any closed surface and described as q 1 ; 4" r2 15 10 5 IG=10µA 25ºC 0 0.0 80 Gate Bias, VG (V) E¼ 20 ð8Þ where q is the elementary charge and r is the curvature radius at the SiC surface. At SiO2=SiC surfaces, a large Ra value 0.5 1.0 1.5 2.0 2.5 3.0 Ra from line profile (nm) Fig. 7. (Color online) Correlation between QBD and Ra extracted from line profile perpendicular to the current direction. The applied constant current was 10 µA and the measurement temperature was 25 °C during CCS-TDDB tests. means that the average curvature radius is small as shown in Figs. 2(g)–2(i). According to Eq. (8), a small curvature radius results in a large electric field. This is the reason why the local electric field concentration occurs. In previous reports, the electric field concentration at the corner of stepbunching on SiC wafers also increased the Fowler–Nordheim current and shortened the lifetime of gate oxides in planar MOSFETs.34,35) Judging from these results and previous reports, in terms of reliability of gate oxides, it is especially significant to reduce surface roughness in trench sidewalls. 4. Conclusions In this study, we focused on the surface morphology of trench sidewalls and its effects on electrical properties. In terms of channel mobility, the limiting factors were Coulomb scattering and optical phonon scattering. Surface roughness scattering mobility had a small contribution to the total mobility, and it had no correlation with Ra parallel to the drain current in the range of 0.4 to 1.4 nm. For further improvement in channel mobility, it is essential to control Coulomb scattering and optical phonon scattering. On the other hand, IG–VG characteristics and CCS-TDDB test results revealed that the surface roughness directly affected the reliability of gate oxides. 04FR02-4 © 2018 The Japan Society of Applied Physics Jpn. J. Appl. Phys. 57, 04FR02 (2018) K. Kutsuki et al. 1) H. Matsunami, Jpn. J. Appl. Phys. 53, 011001 (2014). 2) S. M. Sze, Physics of Semiconductor Devices (Wiley, New York, 1981) 2nd ed., p. 440. 3) T. Hosoi, S. Azumo, Y. Kashiwagi, S. Hosaka, K. Yamamoto, M. Aketa, H. Asahara, T. Nakamura, T. Kimoto, T. Shimura, and H. Watanabe, Proc. Int. Symp. Power Semiconductor Devices and ICs, 2017, p. 247. 4) G. D. Wilk, R. M. Wallace, and J. M. Anthony, J. Appl. Phys. 89, 5243 (2001). 5) S. Nakazawa, T. Okuda, J. Suda, T. Nakamura, and T. Kimoto, IEEE Trans. Electron Devices 62, 309 (2015). 6) T. Nakamura, Y. Nakano, M. Aketa, R. Nakamura, S. Mitani, H. Sakairi, and Y. Yokotsuji, IEDM Tech. Dig., 2011, p. 599. 7) H. Yano, H. Nakao, T. Hatayama, Y. Uraoka, and T. Fuyuki, Mater. Sci. Forum 556–557, 807 (2007). 8) K. Ariyoshi, S. Harada, J. Senzaki, T. Kojima, K. Kojima, Y. Tanaka, and T. Shinohe, Mater. Sci. Forum 778–780, 615 (2014). 9) Y. Kagawa, R. Tanaka, N. Fujiwara, K. Sugawara, Y. Fukui, N. Miura, M. Imaizumi, S. Nakata, and S. Yamakawa, Mater. Sci. Forum 821–823, 761 (2015). 10) C. T. Banzhaf, M. Grieb, M. Rambach, A. J. Bauer, and L. Frey, Mater. Sci. Forum 821–823, 753 (2015). 11) N. Tega, H. Yoshimoto, D. Hisamoto, N. Watanabe, H. Shimizu, S. Sato, Y. Mori, T. Ishigaki, M. Matsumura, K. Konishi, K. Kobayashi, T. Mine, S. Akiyama, R. Fujita, A. Shima, and Y. Shimamoto, Proc. Int. Symp. Power Semiconductor Devices and ICs, 2015, p. 81. 12) S. Harada, Y. Kobayashi, K. Ariyoshi, T. Kojima, J. Senzaki, Y. Tanaka, and H. Okumura, IEEE Electron Device Lett. 37, 314 (2016). 13) H. Kitai, T. Hatayama, H. Tamaso, S. Kyogoku, T. Masuda, H. Shiomi, S. Harada, and K. Fukuda, Mater. Sci. Forum 858, 639 (2016). 14) T. Masuda, R. Kosugi, and T. Hiyoshi, Mater. Sci. Forum 897, 483 (2017). 15) S. Harada, Y. Kobayashi, A. Kinoshita, N. Ohse, T. Kojima, M. Iwaya, H. Shiomi, H. Kitai, S. Kyogoku, K. Ariyoshi, Y. Onishi, and H. Kimura, Mater. Sci. Forum 897, 497 (2017). 16) D. Peters, R. Siemieniec, T. Aichinger, T. Basler, R. Esteve, W. Bergner, and D. Kueck, Proc. Int. Symp. Power Semiconductor Devices and ICs, 2017, p. 239. 17) J. Wei, M. Zhang, H. Jiang, H. Wang, and K. J. Chen, Proc. Int. Symp. Power Semiconductor Devices and ICs, 2017, p. 387. 18) W. W. Mullins, J. Appl. Phys. 28, 333 (1957). 19) J. A. Cooper, M. R. Melloch, R. Singh, A. Agarwal, and J. W. Palmour, IEEE Trans. Electron Devices 49, 658 (2002). 20) K. Sudoh, H. Iwasaki, H. Kuribayashi, R. Hiruta, and R. Shimizu, Jpn. J. Appl. Phys. 43, 5937 (2004). 21) Y. Takeuchi, M. Kataoka, T. Kimoto, H. Matsunami, and R. K. Malhan, Mater. Sci. Forum 527–529, 251 (2006). 22) Y. Kawada, T. Tawara, S. Nakamura, T. Tamori, and N. Iwamuro, Jpn. J. Appl. Phys. 48, 116508 (2009). 23) A. Takatsuka, Y. Tanaka, K. Yano, T. Yatsuo, Y. Ishida, and K. Arai, Jpn. J. Appl. Phys. 48, 041105 (2009). 24) K. Kutsuki, Y. Murakami, Y. Watanabe, T. Onishi, K. Yamamoto, H. Fujiwara, and T. Ito, Ext. Abstr. Solid State Devices and Materials, 2017, p. 697. 25) K. Kutsuki, S. Kawaji, Y. Watanabe, S. Miyahara, and J. Saito, Mater. Sci. Forum 821–823, 757 (2015). 26) K. Kutsuki, S. Kawaji, Y. Watanabe, M. Tsujimura, T. Onishi, H. Fujiwara, K. Yamamoto, and T. Kanemura, Mater. Sci. Forum 858, 607 (2016). 27) K. Kutsuki, S. Kawaji, Y. Watanabe, T. Onishi, H. Fujiwara, K. Yamamoto, and T. Yamamoto, Jpn. J. Appl. Phys. 56, 04CR03 (2017). 28) J. Pernot, W. Zawadzki, S. Contreras, J. L. Robert, E. Neyret, and L. Di Cioccio, J. Appl. Phys. 90, 1869 (2001). 29) C. Lombardi, S. Manzini, A. Saporito, and M. Vanzi, IEEE Trans. Comput.Aided Des. 7, 1164 (1988). 30) S. Takagi, A. Toriumi, M. Iwase, and H. Tango, IEEE Trans. Electron Devices 41, 2357 (1994). 31) A. Pérez-Tomás, P. Godignon, N. Mestres, and J. Millán, Microelectron. Eng. 83, 440 (2006). 32) H. Naik and T. P. Chow, Mater. Sci. Forum 679–680, 595 (2011). 33) A. Frazzetto, F. Giannazzo, P. Fiorenza, V. Raineri, and F. Roccaforte, Appl. Phys. Lett. 99, 072117 (2011). 34) Y. Mori, M. Matsumura, H. Hamamura, T. Mine, A. Shima, R. Yamada, and Y. Shimamoto, Mater. Sci. Forum 821–823, 468 (2015). 35) T. Hosoi, K. Kozono, Y. Uenishi, S. Mitani, Y. Nakano, T. Nakamura, T. Shimura, and H. Watanabe, Mater. Sci. Forum 679–680, 342 (2011). 04FR02-5 © 2018 The Japan Society of Applied Physics
0
You can add this document to your study collection(s)
Sign in Available only to authorized usersYou can add this document to your saved list
Sign in Available only to authorized users(For complaints, use another form )