Numerical Simulation Analysis of Switching Characteristics in the Source-Trench MOSFET’s

: In this paper, we compare the static and switching characteristics of the 4H-SiC conventional UMOSFET(C-UMOSFET),doubletrenchMOSFET(DT-MOSFET)andsourcetrenchMOSFET(ST-MOSFET) through TCAD simulation. In particular, the e ﬀ ect of the trenched source region and the gate trench bottom P + shielding region on the capacitance is analyzed, and the dynamic characteristics of the three structures are compared. The input capacitance is almost identical in all three structures. On the other hand, the reverse transfer capacitance of DT-MOSFET and ST-MOSFET is reduced by 44% and 24%, respectively, compared to C-UMOSFET. Since the reverse transfer capacitance of DT-MOSFET and ST-MOSFET is superior to that of C-UMOSFET, it improves high frequency ﬁgure of merit (HF-FOM: R ON-SP × Q GD ). The HF-FOM of DT-MOSFET and ST-MOSFET is 289 m Ω · nC, 224 m Ω · nC, respectively, which is improved by 26% and 42% compared to C-UMOSFET. The switching speed of DT-MOSFET and ST-MOSFET are maintained at the same level as the C-UMOSFET. The switching energy loss and power loss of the DT-MOSFET and ST-MOSFET are slightly improved compared to C-UMOSFET. on C GD . The extracted Q GD values are 114.8 nC / cm 2 , 80.6 nC / cm 2 , 89.2 nC / cm 2 for C-UMOSFET, DT-MOSFET, and ST-MOSFET, respectively. In addition, the extracted total gate charge ( Q G ) values are 839 nC / cm 2 , 805.1 nC / cm 2 , 812.7 nC / cm 2 for C-UMOSFET, DT-UMOSFET, and ST-UMOSFET. This result is proportional to the extracted results of C GD . The HF-FOM ( R ON-SP × Q GD ) of DT-MOSFET and ST-MOSFET is 289 m Ω · nC, 224 m Ω · nC, respectively, which is improved by 26% and 42% compared to C-UMOSFET. The switching energy loss and power loss of the DT-MOSFET and ST-MOSFET are slightly improved compared to C-UMOSFET.


Introduction
4H-SiC MOSFETs are widely considered to be the leading next-generation power semiconductor devices due to their superior material properties, such as high critical electric field, high thermal conductivity, and ability to operate at high temperatures [1,2]. Of the various SiC MOSFET structures, gate-trench MOSFETs (UMOSFETs) typically have lower on-state resistance compared to planar MOSFETs (VDMOSFETs). In addition, UMOSFETs have higher channel density and mobility than VDMOSFETs due to their ability to form vertical channels on the trench sidewalls and their ability to reduce cell pitch [3][4][5][6][7]. However, there are two main drawbacks to UMOSFETs. First, they have a relatively large reverse transfer capacitance (C rss ). For high-frequency applications, devices must have a small Miller plateau (Q GD ) and C rss [8]. Since Q GD and C rss are determined by the overlap between the gate and the drain, the gate-trench structure has a relatively large C rss . As such, it is necessary to reduce the device C rss to ensure high power efficiency and low energy loss in high frequency operations. The second problem with UMOSFETs is the appearance of gate oxide reliability issues that arise from the gate oxide at the bottom of the trench when UMOSFET operates in the off-state. Because SiC, a wide bandgap material, has a small offset between the conduction band and the valance band with respect to the SiC and the gate oxide, Fowler Nordheim tunneling (FN tunneling) will occur in the electric field (generally over 3 MV/cm) smaller than Si (~6 MV/cm). This FN tunneling current leads to oxide degradation [9][10][11]. Therefore, in the case of SiC UMOSFETs, it is very important to suppress electric field crowding at the gate oxide edge. To address this problem, a structure which includes a gate trench bottom P+ shielding region (BPR) has been proposed [12][13][14][15]. Various other structures have been proposed additionally to reduce the electric field of the gate oxide [16][17][18][19]. Infineon's CoolSiC Various other structures have been proposed additionally to reduce the electric field of the gate oxide [16][17][18][19]. Infineon's CoolSiC and Rohm's Double Trench structure were also commercialized at a 1.2 kV class [20][21][22]. Among them, the double trench structures have been most actively studied in recent years with regard to their dynamic properties and reliability [23][24][25]. In the case of the double trench structure, the source region and gate are both trenched. As a result, the electric field is not concentrated in the gate trench region. As a variant of the double trench, a source trench structure for distributing the electric field through thermally grown oxide has been proposed [26]. However, the dynamic characteristics of the 1700 V source trench MOSFET structure have not been actively discussed. In this study, we researched the switching characteristics of the 1700 V UMOSFET, Double Trench MOSFET (DT-MOSFET), and Source Trench MOSFET (ST-MOSFET) structures. Figure 1 shows the three structures considered in this paper. Figure 1a is the conventional UMOSFET (C-UMOSFET), and Figure 1b,c are DT-MOS and ST-MOS, respectively. The device parameters of each structure have been optimized for static characteristics while the BPR is grounded. In addition, a depletion stopping layer (DSL) [27], also known as a current spreading layer (CSL) [28,29], is added. The DSL improves the on-state characteristics by suppressing the expansion of the depletion region in the JFET region. The cell pitch of each structure is 5.55 μm, and the total thickness of the epi-layer is 18 μm. The thickness of the BPR is 0.3 μm, and the gate trench width and depth are 1.55 μm and 1.5 μm, respectively. In the case of DT-MOS, the source trench width is 0.5 um, the depth is 1.5 μm, and the thickness of the P+ shielding region of the source trench is 0.3 μm. In the case of ST-MOS, the source trench width is 0.5 μm, the depth is 2.4 μm, and the source oxide thickness is 0.1 μm. In addition, the concentration of the epi layer of each structure is 3 × 10 15 cm −3 , 5 × 10 15 cm −3 , and 5 × 10 15 cm −3 for C-UMOS, DT-MOS, and ST-MOS, respectively. Additional device parameters are listed in Table 1.

Device Structures
In the next section, the characteristics of each structure are described. Static characteristics were simulated using Synopsys TCAD, and dynamic characteristics were analyzed through mixed mode simulation [30]. Electron/hole continuity equations and Poisson equation are solved with Shockley-Read-Hall recombination and Auger recombination model. The doping dependency, high field velocity saturation and mobility degradation are included in the mobility model. In particular, the Lombardi model was considered for the interface that affects the channel mobility [31]. Bandgap narrowing, anisotropic material properties, and incomplete ionization effects of each structure were considered [32,33].

Static Characteristics
First, in order to facilitate normalization, the active area was assumed to be 1 cm 2 , and simulation of static characteristics was performed. Figure 2 shows the off-state characteristic curves of the three structures. The breakdown voltages are 1699 V, 1706 V, and 1724 V for each C-UMOSFET, DT-MOSFET and ST-MOSFET, respectively. Figure 3 shows the breakdown voltage when the doping concentration of the epi-layer of each structure is varied. Each structure is designed to have the breakdown voltage close to 1700 V by controlling the doping concentration of the epi-layer. Figure 4 shows the on-state characteristic curves of the three structures when the gate voltage is 15 V. Figure 4a is the on-state characteristic curves of the linear region with the drain voltage range of 0-25 V, and b is the overall on-state characteristic curves with the drain voltage range of 0-800 V. In Figure 4a, the specific on-resistance at a low V D of the C-UMOSFET, DT-MOSFET, and ST-MOSFET are 3.37 mΩ·cm 2 , 3.57 mΩ·cm 2 , and 2.52 mΩ·cm 2 (at V G = 15 V and I D = 20 A), respectively. From Figure 4b, the quasi-saturation current [34] was highest for C-UMOSFET and lowest for DT-MOSFET. Due to the trenched source region, the width of the JFET region of the DT-MOSFET and ST-MOSFET is smaller than that of the C-UMOSFET. Since the current path decreases in proportion to the JFET width, the quasi-saturation current of DT-MOSFET and ST-MOSFET is reduced compared to the C-UMOSFET. The FOM (BV 2 /R on-sp ) [35], which represents the trade-off relationship of static characteristics, was calculated as 856.6 MW/cm 2 , 815.2 MW/cm 2 , and 1179.4 MW/cm 2 for the three structures, C-UMOSFET, DT-MOSFET, and ST-MOSFET, respectively. Figure 5 shows the off-state gate oxide electric field distributions of each structure when V G = 0 V and V D = 1200 V. The maximum gate oxide electric field (E ox-max ) is 1.1 MV/cm for the C-UMOSFET, 0.6 MV/cm for the DT-MOSFET, and 0.7 MV/cm for the ST-MOSFET. The grounded BPR applied to all three structures effectively blocks the electric field at gate oxide, reducing the E ox-max . In the case of the DT-MOSFET and ST-MOSFET, not only the BPR but also the trenched source region disperses the electric field applied to the gate oxide, further reducing the E ox-max . Figure 6 shows the transfer characteristic curves of the three structures at V D = 20 V. As shown, at the same given gate voltage, the current handling capability varies with the JFET width. The wider the JFET width, the larger the drain current at the same gate voltage. Furthermore, the threshold voltages (Vth) calculated at V D = 20 V and I D = 1 A are 5.74 V, 5.75 V, and 5.75 V for each structure. Since the doping concentration and depth of the P-base were kept constant, there was little Vth shift. The overall static performance of the three structure is  Table 2. Figure 7a shows the change in the threshold voltage of each structure as the temperature varies, and the threshold voltage shows the negative temperature coefficient [36]. As the lattice temperature increases from 300 K to 500 K, the threshold voltage decreases due to the increase in intrinsic carrier concentration [37]. In all three structures, the threshold voltage at 500 K decreases by nearly 24% compared to the value at 300 K. Figure 7b shows the specific on-resistance versus temperature. As seen in Figure 7b, the DT-MOSFET is the least temperature dependent, increasing by approximately 140% at 500 K, with respect to its on-resistance at 300 K. For the ST-MOSFET and C-UMOSFET, the on-resistance at 500 K increases by 160% and 180%, respectively, compared to the on-resistance at 300 K.        Figure 5 shows the off-state gate oxide electric field distributions of each structure when VG = 0 V and VD = 1200 V. The maximum gate oxide electric field (Eox-max) is 1.1 MV/cm for the C-UMOSFET, 0.6 MV/cm for the DT-MOSFET, and 0.7 MV/cm for the ST-MOSFET. The grounded BPR applied to all three structures effectively blocks the electric field at gate oxide, reducing the Eox-max. In the case of the DT-MOSFET and ST-MOSFET, not only the BPR but also the trenched source region disperses     decreases due to the increase in intrinsic carrier concentration [37]. In all three structures, the threshold voltage at 500 K decreases by nearly 24% compared to the value at 300 K. Figure 7b shows the specific on-resistance versus temperature. As seen in Figure 7b, the DT-MOSFET is the least temperature dependent, increasing by approximately 140% at 500 K, with respect to its on-resistance at 300 K. For the ST-MOSFET and C-UMOSFET, the on-resistance at 500 K increases by 160% and 180%, respectively, compared to the on-resistance at 300 K.

Terminal Capacitance Characteristics
In this section, the simulation conditions for capacitance extraction are set such that the active area is 1 cm 2 , the ac small signal is 1 MHz, the gate voltage is fixed at 0 V, and the drain voltage sweeps from 0-600 V. Before analyzing the terminal capacitance of the three structures, the effect of BPR on the capacitance was analyzed. Figure 8 shows the capacitance for both ground and floating BPR and for DSL in the C-UMOSFET. The grounded BPR in contact with the source increases the overlapping area between the gate and the source. Thus, in the case of the grounded BPR, the input capacitance (Ciss:Ciss = CGS + CGD) is higher than that when no BPR is used or when a floating BPR is applied. However, the floating BPR causes dynamic degradation due to its charge storage mechanism [38], and when no BPR is applied to the C-UMOSFET, there is increased degradation of the dynamic characteristics due to hot hole injections that arise from the high electric field in the gate oxide [24]. On the other hand, in the case of grounded BPR, Crss, which plays the most important role in switching energy loss, is effectively reduced. The grounded BPR blocks the electric field between the bottom

Terminal Capacitance Characteristics
In this section, the simulation conditions for capacitance extraction are set such that the active area is 1 cm 2 , the ac small signal is 1 MHz, the gate voltage is fixed at 0 V, and the drain voltage sweeps from 0-600 V. Before analyzing the terminal capacitance of the three structures, the effect of BPR on the capacitance was analyzed. Figure 8 shows the capacitance for both ground and floating BPR and for DSL in the C-UMOSFET. The grounded BPR in contact with the source increases the overlapping area between the gate and the source. Thus, in the case of the grounded BPR, the input capacitance (C iss :C iss = C GS + C GD ) is higher than that when no BPR is used or when a floating BPR is applied. However, the floating BPR causes dynamic degradation due to its charge storage mechanism [38], and when no BPR is applied to the C-UMOSFET, there is increased degradation of the dynamic characteristics due to hot hole injections that arise from the high electric field in the gate oxide [24]. On the other hand, in the case of grounded BPR, C rss , which plays the most important role in switching energy loss, is effectively reduced. The grounded BPR blocks the electric field between the bottom gate and the drain and screens the charge coupling [38]. Therefore, only the capacitance between the side gate oxide and the drain contributes to the C rss [12]. When the DSL is added, the C rss is nearly identical as when the ground BPR is applied, excluding a slight increase at low drain voltages.
Electronics 2020, 9, x FOR PEER REVIEW 7 of 16 gate and the drain and screens the charge coupling [38]. Therefore, only the capacitance between the side gate oxide and the drain contributes to the Crss [12]. When the DSL is added, the Crss is nearly identical as when the ground BPR is applied, excluding a slight increase at low drain voltages. In the UMOSFET structure, the gate-drain capacitance (CGD or Crss) is the series connection between the gate oxide capacitance and the depletion capacitance. According to a previous study on the modeling of gate-drain capacitance in UMOSFET structures, the equation for CGD is as follows [39,40]: where t'D is the trench depth not overlapping the p-base region, tW is the trench width, tcell is the cell In the UMOSFET structure, the gate-drain capacitance (C GD or C rss ) is the series connection between the gate oxide capacitance and the depletion capacitance. According to a previous study on the modeling of gate-drain capacitance in UMOSFET structures, the equation for C GD is as follows [39,40]: where t' D is the trench depth not overlapping the p-base region, t W is the trench width, t cell is the cell pitch, C ox is the gate oxide capacitance, and C D is the depletion capacitance. However, in the case of a structure in which the source region is trenched, such as DT-MOSFETs and ST-MOSFETs, it is difficult to intuitively analyze the C GD due to the geometric complexity of the structure. Recently, X. Luo et al. [40] illustrated that the C GD operates as a serial connection between the gate-source capacitance and the drain-source capacitance. The gate-drain capacitance equation in Double Trench MOSFET Structure (without BPR) claimed in his paper are: where C GS is the gate-source capacitance, C DS is the drain-source capacitance and C GD,bottom is the gate bottom-drain capacitance. In DT-MOSFET with BPR, C GD,bottom is negligible, so C GD = (C GS −1 + C DS −1 ) −1 . However, this mechanism may contradict the results shown in Figure 9. Figure 9 plots the C GD as function of the trench depth of the source region in the DT-MOSFET. In Figure 9, C GD decreases as L (source trench depth) increases. When L increases, the overlapping area between the gate and the source increases. As such, C GS increases and at the same time, the distance between the drain and the source decreases, leading to an increase in C DS . So, according to (2), as L increases, C GD should increase. This contradicts the simulation results in Figure 9, where C GD decreases as L increases.
Electronics 2020, 9, x FOR PEER REVIEW 8 of 16  Figure 10 shows the capacitance distribution of the three structures. Depletion capacitance is considered to be the serial connection of the JFET capacitance (CJFET) and the Drift capacitance (CDrift). The modified gate-drain capacitance equation is given follows: Using this model, the capacitance of the trench MOSFET can be analyzed intuitively. First, the gate oxide thickness of the three structures is the same, so Cox is the same. Before the depletion region is fully extended to the drift region, at a low drain voltage, the depletion regions of the DT-MOSFET and ST-MOSFET extend further than that of the C-UMOSFET due to their trenched source regions. Therefore, the CDrift of DT-MOSFET and ST-MOSFET is smaller than that of C-UMOSFET. Indeed, the DT-MOSFET has the smallest CDrift of the three structures because its depletion region is wider than that of the ST-MOSFET due to the P-shielding region below the source region. In the case of the CJFET, it is proportional to the JFET width. The charge in the JFET region is proportional to the JFET width, and the CJFET is proportional to the charge in the JFET region. Therefore, CJFET is smallest in DT-MOSFET and largest in C-UMOSFET. In addition, at a high drain voltage, the drift region is fully depleted and the CDrift of the three structure is nearly the same. Thus, CGD is determined by the CJFET.  Figure 10 shows the capacitance distribution of the three structures. Depletion capacitance is considered to be the serial connection of the JFET capacitance (C JFET ) and the Drift capacitance (C Drift ). The modified gate-drain capacitance equation is given follows: Using this model, the capacitance of the trench MOSFET can be analyzed intuitively. First, the gate oxide thickness of the three structures is the same, so C ox is the same. Before the depletion region is fully extended to the drift region, at a low drain voltage, the depletion regions of the DT-MOSFET and ST-MOSFET extend further than that of the C-UMOSFET due to their trenched source regions. Therefore, the C Drift of DT-MOSFET and ST-MOSFET is smaller than that of C-UMOSFET. Indeed, the DT-MOSFET has the smallest C Drift of the three structures because its depletion region is wider than that of the ST-MOSFET due to the P-shielding region below the source region. In the case of the C JFET , it is proportional to the JFET width. The charge in the JFET region is proportional to the JFET width, and the C JFET is proportional to the charge in the JFET region. Therefore, C JFET is smallest in DT-MOSFET and largest in C-UMOSFET. In addition, at a high drain voltage, the drift region is fully depleted and the C Drift of the three structure is nearly the same. Thus, C GD is determined by the C JFET . Figure 11 plots the terminal capacitance of the three structures. In Figure 11a, the C iss of the three structures are almost the same. In the DT-MOSFET and ST-MOSFET, the trenched source region for improving the static characteristics increases the overlapping area between the gate and the source. However, as shown in Figure 12, the capacitance between the bottom gate and the BPR is the largest of all C GS components, and BPR is applied to all three structures, so the C iss of the three structures is not significantly different. The output capacitance (C oss :C oss = C DS + C GD ), which depends on the distance between the drain and source, is also not significantly different, though the C-UMOSFET has the smallest C OSS . In the case of C rss , according to the previous analysis, the C rss of the DT-MOSFET is the smallest, and the C rss of the C-UMOSFET is the largest. At V D = 600 V, the C rss of DT-MOSFET and ST-MOSFET decreases by 44% and 24%, respectively, compared to the value of C-UMOSFET. The capacitance simulation results for each structure are shown in Table 3. Table 3. Dynamic performance of each structure.     Figure 13a shows the gate charge curves of the three structures. The test circuit is shown in Figure 13b and a constant current of 100 mA is used to charge the gate. The active area of the device under test (DUT) for gate charge simulation was set to 1 cm 2 . In addition, test conditions were set so that the supplying voltage (VDD) was 1200 V and the load current (ID) was 20 A. The gate-drain charge (QGD or Miller plateau) is one of the key parameters that can determine the switching speed of the device and is dependent on CGD. The extracted QGD values are 114.8 nC/cm 2 , 80.6 nC/cm 2 , 89.   Figure 13a shows the gate charge curves of the three structures. The test circuit is shown in Figure 13b and a constant current of 100 mA is used to charge the gate. The active area of the device under test (DUT) for gate charge simulation was set to 1 cm 2 . In addition, test conditions were set so that the supplying voltage (V DD ) was 1200 V and the load current (I D ) was 20 A. The gate-drain charge (Q GD or Miller plateau) is one of the key parameters that can determine the switching speed of the device and is dependent on C GD . The extracted Q GD values are 114.8 nC/cm 2 , 80.6 nC/cm 2 , 89.2 nC/cm 2 for C-UMOSFET, DT-MOSFET, and ST-MOSFET, respectively. In addition, the extracted total gate charge (Q G ) values are 839 nC/cm 2 , 805.1 nC/cm 2 , 812.7 nC/cm 2 for C-UMOSFET, DT-UMOSFET, and ST-UMOSFET. This result is proportional to the extracted results of C GD . The HF-FOM (R ON-SP × Q GD ) of DT-MOSFET and ST-MOSFET is 289 mΩ·nC, 224 mΩ·nC, respectively, which is improved by 26% and 42% compared to C-UMOSFET. Figure 13a shows the gate charge curves of the three structures. The test circuit is shown in Figure 13b and a constant current of 100 mA is used to charge the gate. The active area of the device under test (DUT) for gate charge simulation was set to 1 cm 2 . In addition, test conditions were set so that the supplying voltage (VDD) was 1200 V and the load current (ID) was 20 A. The gate-drain charge (QGD or Miller plateau) is one of the key parameters that can determine the switching speed of the device and is dependent on CGD. The extracted QGD values are 114.8 nC/cm 2 , 80.6 nC/cm 2 Table 3 summarizes the results including the values for terminal capacitance, gate charge, and HF-FOMs, which are significant parameters for high frequency performance [8,37,41].

Dynamic Charateristics
Finally, the switching performance parameter of the device was extracted through a doublepulse test (DPT). The active areas of all the DUT were set to 0.3 cm 2 [42], which is similar to that of commercial devices [43]. Figure 14a plots the full waveform of the ST-MOSFET. Figure 14b shows the test circuit in the three structures. The gate resistance is set to 10 Ω, and the gate voltage switched between 15 V and −3 V for the on-and off-states. The stray inductance was assumed to be 10 nH. The load inductance was set to 300 μH, and the first gate voltage pulse lasted 5 μs so that the load current flow was 20 A. The body diode of the same device as the DUT was used as a freewheeling diode, and the supply voltage was 1200 V.  Table 3 summarizes the results including the values for terminal capacitance, gate charge, and HF-FOMs, which are significant parameters for high frequency performance [8,37,41].
Finally, the switching performance parameter of the device was extracted through a double-pulse test (DPT). The active areas of all the DUT were set to 0.3 cm 2 [42], which is similar to that of commercial devices [43]. Figure 14a plots the full waveform of the ST-MOSFET. Figure 14b shows the test circuit in the three structures. The gate resistance is set to 10 Ω, and the gate voltage switched between 15 V and −3 V for the on-and off-states. The stray inductance was assumed to be 10 nH. The load inductance was set to 300 µH, and the first gate voltage pulse lasted 5 µs so that the load current flow was 20 A. The body diode of the same device as the DUT was used as a freewheeling diode, and the supply voltage was 1200 V. In this paper, turn-on time (TON) and turn-off time (TOFF) are defined as follows [43][44][45].
where TD-ON is the turn-on delay time (from 10% of VG to 90% of VD at the rising edge), TR is the rise In this paper, turn-on time (T ON ) and turn-off time (T OFF ) are defined as follows [43][44][45].
where T D-ON is the turn-on delay time (from 10% of V G to 90% of V D at the rising edge), T R is the rise time at the turn-on transient (from 90% of V D to 10% of V D at the rising edge), T D-OFF is the turn-off delay time (from 90% of V G to 10% of V D at the falling edge) and fall time at the turn-off transient (from 10% of V D to 90% of V D at the falling edge). Figure 15 shows the switching waveforms of each structure, and Table 4 summarizes the detailed switching performance data. The switching speed (T ON and T OFF ) of the device is most affected by C iss , and T R and T F are most dependent on C GD [37,41].
In hard switching, V D and I D are swept during T R and T F , so energy loss is dependent upon T R and T F . In the previous section, we observed that the C iss of DT-MOSFET and ST-MOSFET is almost the same as the value of C-UMOSFET. Therefore, the switching speed of DT-MOSFET and ST-MOSFET is maintained at the same level as C-UMOSFET. In addition, the C GD of DT-MOSFET and ST-MOSFET is superior to that of C-UMOSFET, which improves the switching energy loss. The switching energy loss of C-UMOSFET, DT-MOSFET, and ST-MOSFET are 827.1 µJ, 771.2 µJ, and 806 µJ, respectively. In addition, since the drain voltage sweeps from low to high at the falling edge of the turn-off transient, T OFF is relatively slower than T ON . Moreover, since the body diode of each device was used as FWD, E ON includes the diode reverse recovery. Therefore, E ON is relatively larger than E OFF . The total power losses P t of the device consist of conduction losses and switching losses and are calculated as follows [19,37,40]: where d is the duty cycle and f is the switching frequency. Figure 16 shows the power losses which vary with the switching frequency when the duty cycle is 0.5. As the switching frequency increases, the ratio of switching losses of total power losses increases. The power losses of C-UMOSFET, DT-MOSFET, and ST-MOSFET when operating at 200 kHz are 552.1 W/cm 2 , 514.9 W/cm 2 and 537.8 W/cm 2 , respectively.
where d is the duty cycle and f is the switching frequency. Figure 16 shows the power losses which vary with the switching frequency when the duty cycle is 0.5. As the switching frequency increases, the ratio of switching losses of total power losses increases. The power losses of C-UMOSFET, DT-MOSFET, and ST-MOSFET when operating at 200 kHz are 552.1 W/cm 2 , 514.9 W/cm 2 and 537.8 W/cm 2 , respectively.

Conclusions
In this paper, the static and dynamic performance of the 4H-SiC C-UMOSFET, DT-MOSFET and ST-MOSFET were compared through TCAD simulation. In static characteristics, the RON-SP of the ST-MOSFET is 2.52 mΩ•cm 2 , which is 25% lower than that of C-UMOSFET. Furthermore, the DC-FOM

Conclusions
In this paper, the static and dynamic performance of the 4H-SiC C-UMOSFET, DT-MOSFET and ST-MOSFET were compared through TCAD simulation. In static characteristics, the R ON-SP of the ST-MOSFET is 2.52 mΩ·cm 2 , which is 25% lower than that of C-UMOSFET. Furthermore, the DC-FOM of the ST-MOSFET is 1179 MW/cm 2 , which is 37% higher than that of the C-UMOSFET. At V D = 1200 V, the Eox-max of the DT-MOSFET is 0.6 MV/cm, which was 44% lower than that of the C-UMOSFET. To improve the static characteristics, both the DT-MOSFET and ST-MOSFET are trenched in the source region. For the same reason, the BPR is introduced in the three structures. The effect of the trenched source region and BPR on the capacitance was analyzed, and the dynamic characteristics of the three structures were compared. Due to the trenched source region, the overlapping area between the gate and source increases. Nevertheless, the capacitance between the gate and the BPR is largest of C GS components, so the C iss is almost identical in all three structure. On the other hand, the C GD of DT-MOSFET and ST-MOSFET is reduced by 44% and 24%, respectively, compared to C-UMOSFET. This is because the depletion region is expanded by the trenched source region. Since the C GD of DT-MOSFET and ST-MOSFET is superior to that of C-UMOSFET, it improves HF-FOM (R ON-SP × C GD ).