Leveraging GaN for DC-DC Power Modules for Efficient EVs: A Review

Limitations of Silicon (Si)-based devices have compelled us to use alternative devices for modern power electronics applications. Material attributes of wide-band-gap devices (such as GaN and SiC) possess the ability to bridge these gaps. They can be used for higher power applications and provide high power-density, high efficiency and better thermal performance. GaN-based devices in electric vehicle power modules make the vehicle more efficient, achieving an extended range for the same battery size. Worldwide, researchers and engineers are working on GaN-based power-electronics modules. On the one hand, the utilisation of GaN switches in power modules eliminates a few existing design concerns. While it also introduces new challenges for designers. Mere replacing Si with GaN doesn’t give the stipulated result. This article identifies the works related to GaN-based DC-DC converter modules to determine how researchers address the circuit design issues with GaN. This paper presents a detailed description of the material benefits of GaN and paves the path of future work by identifying the research gap in the field of GaN-based DC-DC converters.


I. INTRODUCTION
Alarming concerns of global warming, climate change, deplenishing oil reserves, and mounting oil prices have forced intellectuals, researchers and scientists to take concrete steps to save mother earth. Elevating the usage of alternate energy sources (like solar, wind, tidal etc.) and electric vehicles (EVs) are a few of them. Conventional vehicles, especially road transport ones, critically affect the environment by blowing out greenhouse gases (74.5% of global CO 2 emission by various transport sectors is accounted for by road transports [1]). Hence gas-guzzling vehicles are one of the major sources of global warming. In this regard, EVs are emerging as an option to replace conventional fuel vehicles. EVs are more efficient, safe, and reliable and have lesser maintenance and operating costs than conventional ones. However, accelerating towards EV adoption demands a few challenges to be addressed. Higher charging time, more capital cost (mainly because of high battery price), shorter driving range, reliability and lack of charging infrastructure are the challenges The associate editor coordinating the review of this manuscript and approving it for publication was Jahangir Hossain . before automakers. To make EVs viable for customers, these challenges should be taken care of [2]. Charging time, for example, can be lowered by raising the power level of the on-board charger (OBC). The current trend is to increase the power level of OBC from 3.6 kW to 22 kW [3]. However, an increment in power level will add cost, weight and volume. In contrast, we require to have an OBC that must accommodate the given physical envelope of the vehicle with minimum to no increase in weight, volume and cost without impacting the driving ranges. So we need to increase the power-density to get a compact OBC.
Similarly, driving range can be increased by using large and bulky battery packs (which will put an economic burden) or improving the vehicle's energy efficiency. Better energy efficiency means lower energy consumption per km, i.e. lower losses. It is important to note that while we are trying to increase the power-density and get better energy efficiency, Si (silicon) cannot deliver the technical exigency because of its material limit. Here wide band-gap (WBG) devices come into the picture. Figure 1 demonstrates the comparative analysis of the energy efficiency of various EVs available in the market, their battery capacity and range [4], [5]. We can  see that EV manufacturers are attempting to improve energy efficiency and reduce battery capacity for the same vehicle range. The values shown are based on the standard testing condition as per various testing agencies, which may vary depending on multiple factors, including driving style, route and weather conditions, vehicle equipment and payload [6].
Power electronics modules, essentially based on switchmode power conversion, are an integrated part of EVs. Figure 2 shows the various power electronics modules used in a battery-electric-vehicle (BEV) powertrain. DC-DC converter finds its application in three main areas. The first one is a bi-directional high-power module (∼80 kW), which converts the battery voltage (typically 200-400 V) to 650-800 V (to feed high-power traction inverter as well as to store back the regenerative energy to the battery). Without this converter, we would require a high-voltage battery with a rating that matches the traction inverter, which would raise the price of the battery. Hence DC-DC converter can be considered the core of energy management of EVs as it improves its performance and efficiency [7]. Another DC-DC converter module is low-power, which supplies 12 V EV loads from the battery. Finally, OBC, a medium-power module (∼4 kW), is another place where DC-DC converters are used. A typical EV's OBC contains a power-factor-correction AC/DC stage followed by a DC-DC converter. This converter can be bi-directional in case, a vehicle to grid connectivity is required [8].
Passive components, which include inductors, capacitors, filters, heat sinks, and transformers, constitute a large chunk of power converters. Higher switching frequency reduces the size of passive components. However, it fundamentally increases the switching losses (as the number of times switches ON and OFF increases). Reduction in the size of power modules demands a decrement in power loss as well to keep the temperature of components constant [3]. Because the existing surface area is now tiny for heat ejection, this shows that higher power-density requires a simultaneous reduction in power losses. This is where Si stumbles to deliver the technical requirement. Si-based devices suffer from efficiency issues when used for switching frequencies beyond a few kHz, especially for high-power applications [9]. WBG devices such as Gallium nitride (GaN) and Silicon carbide (SiC) can go beyond the limit of Si and are capable of providing higher switching frequency, higher voltage operation, better thermal performance and lower losses.
This paper deals with WBG-GaN (devices and related power modules) only. The journey of GaN embarked in 1940 when Juza and Hanh passed ammonia over hot gallium [10]. Later the industry witnessed its applications in optoelectronics, such as blue and white light-emitting diodes (LEDs), LASER, ultraviolet-emitter, and microwaves [11], [12]. In the quest for higher voltage and frequency, GaN's application extended to power electronics. Figure 3 highlights the various milestones in the development of GaN [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23], [24]. Although a long way has been covered, GaN is still an immature device. More advancements and significant efforts are still required to deal with the technical challenges in future.
This paper does a comprehensive analysis of the benefits of the material properties of GaN for efficient DC-DC converters and emphasises the design challenges of using GaN. This review article is composed of 7 sections. Following the introduction, section II of this paper describes why GaN is significant for today's power electronics converters. Section III presents the details of the material benefits of GaN. Section IV deals with the challenges of circuit-design on using GaN switches. Section V offers some illustrations of GaN's benefits in DC-DC converters. Section VI identifies and summarises those papers where GaN has been used for DC-DC converter. Finally, section VII gives a design example of a GaN-based buck converter and its superiority analysis over its counterpart. devices even after enhancement in their structure or fabrication process [25], [26], [27]. As we approach high-power applications, Si-based devices (like MOSFET and IGBT) can't be switched ON or OFF beyond a few kHz because of their thermal limits or low permissible junction temperature. In other words, the switching frequency of Si based devices is limited because of the higher power losses associated with them at increased frequency.
Higher power loss results in higher device temperature, which they cannot bear due to the low thermal limit. This limitation forces the use of bulky cooling components, such as water-filled copper jackets [28]. The thermal management becomes more complex in the harsh ambient environment of EVs drivetrain, especially in hybrid electric vehicles (HEVs) where the temperature goes beyond 120 • C in the surroundings of the engine and in the traction inverter, where the power level is more than 100 kW [29]. Moreover, extensive cooling pays off with vehicle weight, size and cost.
Si-based switches, hence, are primarily used at low switching frequencies resulting in increased size of passive components and filters; not desired for EVs application. Using GaN, the necessities of today's power electronics converters, which are characterised by higher operating frequency, higher power-density (lower weight and volume), lower losses, and lower cost, can be fulfilled [30].
To understand how power modules can leverage GaN for efficient EVs, we need to understand the material properties of GaN. Table 1 summarises the various material properties of GaN, SiC and Si [12], [31], [32], [33], [34], [35] and Figure 4 shows the resulting benefits because of the fundamental material properties of GaN.
As Table 1 shows, in comparison with Si, GaN possesses more than 10 times the breakdown electric field, almost 3 times band-gap energy, more than 33% electron mobility, and 10 times more electron drift velocity. These attributes effectively make GaN devices capable of operating at high breakdown voltage (because of wider energy-gap and higher breakdown electric field) and high switching speed (because of higher electron mobility and saturation drift velocity) (figure 4). The high electron mobility of the GaN device is because of the formation of two-dimensional electron gas (2DEG) at its hetero-junction. Higher mobility and breakdown field allow GaN to shrink the die size for a given current and voltage capability [36], [37]. Smaller die size results in lower gate and output capacitance, which further contributes to high switching frequency.
Moreover, the higher breakdown field helps GaN to be optimised with a thin drift-region, which gives lower ON-state resistance (Rds (ON) ). Higher mobility also helps to achieve lower Rds (ON) and hence increased frequency. Table 1 also indicates that although GaN excels Si-devices, it lags behind SiC in thermal performance. SiC boasts more than three times better thermal conductivity than GaN, making it suitable for high-temperature and high-power applications (in the range of 100 kW), such as traction inverters and three-phase converters [38]. Moreover, the temperature coefficient of Rds (ON) of GaN is more than that of SiC, i.e. with the rise in operating temperature, GaN shows more variation in Rds (ON) . This increases the losses in GaN devices at high temperatures. Better thermal performance (of SiC) positively impacts the size of cooling components and hence the cost. The thermal performance of SiC also indicates that, theoretically, it can serve at a higher power-density than GaN or Si [39], [40]. In this aspect, the relatively poor thermal performance of GaN devices challenges their heat management design.
However, GaN outperforms SiC devices with higher electron mobility (which makes GaN 10 times faster than SiC), higher band-gap energy, lower Rds (ON) (lower conduction loss), and lower junction capacitance (lower switching loss) [41], [42]. In addition, the GaN device has zero reverse recovery (because of no body diode) and lower dead-time loss than SiC devices [38]. The high switching frequency of GaN also helps to reduce passive components' size and hence to gain high power-density, making the converter scalable. It means that for the given power rating, we need a smaller footprint to design a converter. Reducing size demands decrement in power losses to keep the converter temperature 95876 VOLUME 11, 2023 Authorized licensed use limited to the terms of the applicable license agreement with IEEE. Restrictions apply.   constant. Again, GaN meets the expectation with its lower switching and conduction loss, which further benefits with less cooling requirement and an efficient converter. With an efficient converter, we can have a longer driving range for the given battery capacity. Ultimately it saves the cost. Figure 5 pictorially summarises the benefits of employing GaN devices in the DC-DC converter of EVs. The use of GaN devices can satisfy the necessity of higher voltage, better efficiency, faster switching speed, and higher powerdensity, enhancing their performance and, subsequently, the performance of the whole vehicle.

III. DETAILED ANALYSIS OF THE MATERIAL BENEFITS OF GaN
GaN has superior material characteristics over Si and SiC, as mentioned in section II. This section gives a thorough explanation of those characteristics and how they contribute to ameliorating the circuit design.
The time required to charge the parasitic capacitors, essential to turn-ON the MOSFET at a given operating voltage, is small if the gate charge is low. This reduces the switching loss and VOLUME 11, 2023 contributes to achieving high switching capability. Smaller die sizes of GaN devices allow them to have lower input and output capacitances [36]. Hence, GaN devices possess lower gate charges compared to Si and SiC devices enabling them for faster-switching transients (faster rise and fall time) and lower switching losses.
However, there is a trade-off between the total gate charge and ON-state resistance. Usually, a smaller die size gives smaller Q G , but it increases Rds (ON) . In other words, since Rds (ON) has direct implications on conduction loss, there will be a conflicting relation between conduction and switching losses because of the smaller die size of GaN [43], [44].
Here come the excellent material properties of GaN devices. Because of the high breakdown field, GaN devices can be optimised with relatively thin drift regions, resulting in lower specific ON resistance. Moreover, its higher electron mobility also contributes to lower Rds (ON) [36]. Thus GaN switches possess lower gate charge as well as lower ON-state resistance resulting in reduced switching, conduction and gate driver losses along with higher switching speed. At a given breakdown voltage, GaN possesses the lowest Rds (ON) among Si, SiC and GaN.
Rds (ON) ×Q G , also known as gate charge figure of merit (FOM), is a common FOM that considers both on-state and switching behaviour. While analysing and comparing the switches of different semiconductor materials, this FOM is an essential tool for it is directly connected with system losses (This is to be noted that lower FOM is considered better) [45]. Figure 7 shows a FOM comparison of different semiconductor devices based on the datasheets of their manufacturers [46], [47], [48], [49], [50], [51]. As the figure depicts, GaN switches have ultra-low FOM showing faster switching and lower gate-drive loss.   devices (most notably, e-mode GaN high electron mobility transistors (HEMTs)) don't have intrinsic body diodes and parasitic BJT between drain and source [36], [52], [53]. However, they have the reverse conduction property the same as Si MOSFETs (as shown in Figure 8) when the gate-tosource voltage (V GS ) is zero or negative. The reason behind it lies in the symmetrical behaviour of this device. When the drain to source voltage (V DS ) is positive, the 2DEG channel gets turned ON if V GS exceeds the threshold limit (V GS_TH ). Likewise, when V DS is negative, the channel turns ON if the gate to drain voltage (V GD ) reaches its threshold limit (V GD_TH ). The negatively biased drain terminal causes a potential gradient in the channel. This, in turn, generates a negative electrical potential in the depletion region below the gate electrode, which is responsible for turning the device ON for reverse conduction [52]. This phenomenon is also called ''self-commutation in reverse-conduction''. This is to be noted that V GD_TH is approximately equal to the specified V GS_TH . If the device is carrying the drain current I D , the voltage drop in the reverse conduction mode (D T ) is given by 95878 VOLUME 11, 2023 Authorized licensed use limited to the terms of the applicable license agreement with IEEE. Restrictions apply.
where R DS_REV represents the channel resistance during reverse conduction and is typically greater than Rds (ON) . Reverse conduction without a body diode gives some real advantages to GaN devices. No body-diode implies no reverse recovery charge (Q rr ). Zero Q rr results in lower switching losses. It helps in fast switching as well as provides lower EMI noise (no noise of turning ON the body diode). This makes GaN fit for half-bridge applications with hard and soft switching (Zero Q rr lowers the dead-time, necessary to ascertain zero voltage switching (ZVS)) [54]. There are added benefits of high dv/dt ruggedness and reliability. Si MOS-FETs show a failure mechanism during reverse recovery [55]. In the absence of a body-diode, GaN HEMT is safe from such failure.
However, the downside is that GaN HEMT experiences a high voltage drop in reverse conduction (D T ), which can be as high as 3-5 V (greater than an equivalent drop in Si MOSFETs). Higher D T can reduce efficiency by increasing losses in the dead time of a particular circuit. Reducing the dead time can suppress these losses.

C. LOWER OUTPUT CAPACITANCE (C OSS )
Output parasitic capacitance (C OSS ) of a MOSFET is given by C OSS = C DS + C GD , where C DS is the capacitance between the drain and source and C GD is the capacitance between the gate and drain. It is an important parameter that directly impacts the soft-switching performance of a converter. While turning-ON the MOSFET, in the case of ZVS transition, C OSS must be discharged before the switch is turned ON, bringing the drain to source voltage to zero [56]. Owing to the smaller die area, GaN devices offer smaller C OSS. Lower C OSS discharges quickly, leading to lower losses and better soft switching.
The time required to achieve ZVS is given by t zvs = Q oss I zvs , where Q OSS is the output charge of MOSFET and I ZVS is the current required to get ZVS [57]. Lower Q OSS of GaN provides a shorter time to attain soft switching. This also leads to having a shorter delay time and helps in achieving a higher switching frequency. This is to be noted that with the help of GaN, the switching frequency can be scaled up to MHz. But increased switching frequency comes with increased switching losses. This increases the operating temperature, and consequently, the heat sink size also increases. To overcome this barrier, softswitching gets essential. Soft-switching provides an apparent reduction in switching losses. Moreover, the severe EMI issue during fast commutations at high frequency is suppressed by soft switching [58]. Which otherwise requires large filters. Lower C OSS of GaN helps to attain better soft switching. So at the same time, we can have a higher frequency, lower switching losses and better EMI immunity.

D. HIGHER FREQUENCY
As discussed in section II of this article, GaN devices enable high-frequency operation, which is the key to shrink the passive components' size and to attain high power-density. Figure 9 shows that GaN devices can be used for frequencies up to 10MHz, which surpasses all other devices [38]. We can also observe that there is an overlapping region below the 10 kW application where Si, SiC and GaN can all be used. However, the maximum switching frequency is limited by losses associated with the devices. Si shows higher losses at higher frequencies; hence, GaN is the best choice for medium-power applications such as on-board chargers, DC-DC converters, etc. From Figure 9, it is also clear that keeping the power constant at around 10 kW, the switching frequency in GaN can be scaled up to MHz. This provides a significant reduction in the volume of transformers, inductors and capacitors saving system cost and space. At high frequencies, a smaller filter size can be achieved and the transformer can be replaced by PCB-based planar transformer with integrated leakage inductance [59]. It means at higher frequencies, the series inductor needed to achieve ZVS (such as in resonant converters) is very small. So by giving adequate air-gap between primary and secondary winding, the leakage inductor can be used as a series inductor [60]. Such integration reduces the volume of magnetic components as well as magnetic losses. Table 2 summarises the eminence of GaN over Si and SiC based on the data sheets of 650 V switches taken as an example.

IV. CHALLENGES IN THE CIRCUIT DESIGN WITH GaN SWITCHES
Although unique features of GaN devices provide a lot of benefits to circuit design engineers in terms of high powerdensity, lower losses etc., various design challenges arise because of using these new class of devices. There are multiple trade-offs and limitations between the achievable benefits and new drawbacks. These limitations have been classified here and summarized in Figure 10.

A. DEVICE-LEVEL CHALLENGES
The first limitation of the GaN device is regarding its voltage rating. GaN devices are commercially available up to 650 V VOLUME 11, 2023  only [36], [64] and therefore, a 650V e-GaN device can't be applied easily in the conventional DC-DC converters of higher voltage rating. For a higher voltage rating, we have to go with the SiC device. This is to be noted that today, GaN transistor up to 1700 V and diode up to 4000 V breakdown voltage is available [65]. However, they are only for research purposes and their techniques are not suitable for large-scale production.
Another major limitation of GaN devices is the commercial non-availability of their vertical structure. The available lateral structure of GaN requires a large chip area [36], [64], [66], [67] if used for converters of high current rating, which also makes the overall area of the converter large. And therefore, it is cumbersome to manufacture the lateral GaN topology for high current and power demand.

B. CHALLENGES WITH THE GATE-DRIVER DESIGN
The high switching frequency and low gate-to-source threshold voltage (typically 1.5V) of the GaN device make the gate voltage of GaN HEMT extremely sensitive to ''miller current'', parasitic inductance, and noise. Hence, the gate circuit design for high frequency remains an addressable issue in converter design with GaN [68]. There is a small margin between the recommended and maximum V GS of GaN. For example, the typical V GS to turn-ON a GaN FET is 5V, while the maximum V GS is 6V. This shows that we need very precise and accurate V GS , which is difficult to obtain for high-side switches by traditional bootstrapping supply [69], [70]. One of the ways to achieve a clean gate drive signal is to adjust the gate resistance for critical damping.
Another issue is, when GaN switches are used under high dv/dt, which causes unwanted charging of gate-source capacitance (C GS ) because of the 'Miller effect'. If the charges in C GS goes beyond the threshold value, this may lead to a spurious turn-ON of the GaN device [71], [72]. Spurious turn-On can also happen if the voltage drop through the resistances of gate-drive loop exceeds the threshold limit of V GS [73]. Therefore the selection of external gate resistance and minimization of the gate-loop parasitic inductance are crucial. A gate driver co-packaged with the switch will have positive impact in this case.

C. HIGH EMI NOISE
Owing to having a low gate charge (Q G ), a GaN FET has a shorter turn-ON time and hence its switching frequency is higher than that of a Si-based device. However, because of the shorter turn-ON time, the GaN devices face intense di/dt and dv/dt transition, which leads to severe EMI noise. EMI can be suppressed using an input filter, but this increases the overall system cost and volume [74]. Moreover, with the increase in switching frequency, the converter gets smaller and the components of the converter get closer. This equally creates EMI, a significant challenge for engineers to tackle.

D. CHALLENGES WITH THE DESIGN OF MAGNETICS
The design of an isolated DC-DC converter using GaN face the major difficulty in the design of a high-frequency transformer. At megahertz, phenomena like the skin effect and proximity effect come into play, which increases the AC losses [8], [54]. This can be mitigated by using improved core and winding materials and by the integrated design of the magnetic components.

E. PRINTED CIRCUIT BOARD (PCB) LAYOUT DESIGN AND DEVICE PACKAGING
Because of the high switching frequency of GaN devices, PCB layout and device packaging demand proper 95880 VOLUME 11, 2023 Authorized licensed use limited to the terms of the applicable license agreement with IEEE. Restrictions apply.
consideration while designing a power module to lower the associated parasitics. Otherwise they can create overshoot, ringing and EMI issues, overstressing the GaN switches. Good layout practice and technique are essential to maximize the benefits of GaN and converter performance. Moreover, a device with low package inductance enables low-inductance power-loops in PCB.
Common-source-inductance (CSI) is of the most critical parasitic elements. CSI is the summation of the source terminal's inductance (inside the package) and the inductance of lead of the package itself [75]. CSI shares the main current path (which carries the drain-source current) and loop of the gate driver (which carries the gate charging current). The voltage induced in CSI because of high di/dt in the drain-source path alters the effective V DS and V GS , impacting the switching performance, spurious switching and switching losses [76].
The ringing of V GS can even cause gate-breakdown because of its low safety margin. In Si, we increase the gate resistance to dampen the ringing. However, in the case of GaN, such practice is not suggested, for the inherent fast switching of GaN will compromise. So a through-hole (like TO-220) package is not ideal for GaN. Surface-mount-device (SMD) package is better choice for GaN devices. However, thermal management is challenging, for SMD for they rely on PCB for heat transfer [77].

F. OTHER CHALLENGES 1) NECESSITY of SOFT-SWITCHING
A higher switching frequency of GaN devices helps to reduce the size of passive elements of the converter circuit. However, this leads to increased switching loss which is another major challenge. Moreover, the energy loss during turn-ON is more than that during turn-OFF, and hence soft switching is essential for GaN devices [77], [78]. In the absence of softswitching, rapid transition during the commutation creates a serious EMI issue also. This is to be noted that stored energy in C OSS in GaN FETs can be recovered easily compared to Si-based MOSFETs. So it is advisable to use soft switching in high frequency application.

2) DEAD-BAND OPTIMIZATION
GaN devices possess a higher voltage drop in reverseconduction in comparison with that of Si or SiC devices as discussed in section III of this article; the dead-band should be optimized while designing a converter.

3) DRAIN-VOLTAGE RATING
Practically, majority of GaN switches are avalanche-incapable by design [77]. So it is crucial to consider sufficient margin in device voltage rating. Even while transient, the maximum transient voltage should not exceed the rated device voltage. For example, a device with voltage rating 600V is typically sufficient for 480V bus voltage in H-bridge topologies.

V. ILLUSTRATIONS OF THE BENEFITS OF GaN SWITCHES IN DC-DC CONVERTER TOPOLOGIES
Application of GaN-switches in DC-DC converter circuits gives advantages to the design engineers. It overcomes a few design barriers which were not possible earlier with the help of Si-based switches. It also enables new topologies, such as totem-pole power-factor correction (PFC) circuits. This section presents few examples of DC-DC converter circuits where the use of GaN solves existing issues to get an efficient and improved converter.

A. SYNCHRONOUS BUCK CONVERTER (SBC)
Two phenomena happen simultaneously when switch S1 is turned ON and S2 is turned OFF at hard-switching. First, C OSS1 is discharged, as shown in Figure 11 and energy is lost. Second, a reverse current flows to recover switch S 2 . Hence during turning ON switch S 1, there is an overshoot in current flowing through S 1 , which creates additional power loss. The total energy lost during this particular period in switch S 1 is [E =V BUS (Q OSS1 + Q rr )], where Q OSS1 is the charge stored in the output capacitance of switch S 1 [79]. In the case of VOLUME 11, 2023  GaN switches, there is no reverse recovery and Qrr is zero. Hence the said loss is because of Q OSS only [80]. Moreover, GaN switches have lower Q OSS values. This provides less EMI noise and high efficiency. This is worth noting that Q rr worsens with higher temperature, current, voltage rating and faster switching. Hence, GaN switches are best fit for lowering the losses of reverse recovery charges.

B. BRIDGELESS TOTEM-POLE PFC (BTPFC) CIRCUIT
Excellent reverse recovery of GaN (Zero Q rr ) enables a new topology of power factor correction, known as a bridgeless totem-pole circuit. This circuit is more efficient and has less switches count than the conventional boost-PFC.
A BTPFC circuit is shown in Figure 12. For power factor correction, switches S 1 and S 2 must be operated at a higher frequency (in the kHz range), while switches SR 1 and SR 2 are operated at line frequency. For the positive and negative half of the AC input, switches S 1 and S 2 are operated respectively for a specific duty cycle (D). For the positive half, when switch S 2 is ON, current flows through L, S 2 and the body diode of SR 2 . For the same positive half, when S 2 is OFF, current flows through L, the body diode of S 1 , load and body diode of SR 2 . A similar operation happens during the negative half, except the switches S 2 and SR 2 are replaced by S 1 and SR 1 , respectively.
The inherent issue with the BTPFC circuit is that when the AC input changes from the positive to the negative half, the body diode of the high-side switch S 1 goes under reverse recovery. In the case of Si switches, slow reverse recovery of the body diode results in large current spikes. Therefore, in BTPFC, the high-frequency switches (S 1 and S 2 ) must have zero or very low reverse recovery [81], [82]. Otherwise, there will be shoot-through and associated power loss. Hence designers don't find Si-MOSFET fit for BTPFC circuits.   Given no body-diode and zero Qrr, GaN is the best fit for BTPFC.

C. DUAL ACTIVE BRIDGE (DAB) CONVERTER
The main downside of a DAB converter is the loss of soft-switching in light-load conditions. Under lightload, in case of higher C OSS , the leakage-inductor current is not sufficient to get the required charging and discharging to obtain ZVS in a given dead time. For DAB, the minimum current required to ascertain the full voltage-transition in C OSS within a specific dead time is given by I l,min = 2C OSS dV C dt [83]. Where dV C is the change in switch capacitance-voltage and dt is the dead time.
Possible solutions for this drawback are to use switches with lower C OSS or to extend and provide enough dead time letting inductor current ramp up. However, the latter results in increased conduction loss. GaN switches, because of their low C OSS , help to get better ZVS and the soft-switching range can also be extended.

VI. GaN-BASED DC-DC CONVERTERS
Replacing Si with a GaN device in the given circuit topology invites multiple challenges from a circuit-design point of view, as discussed in the previous section. Worldwide, many researchers are working on GaN-based DC-DC converters for EV applications. To address the design challenges, researchers are working on novel, innovative topologies, improved transformer and inductor designs (such as magnetic integration), and new magnetic materials. As well as they are working on a modified design of PCB, gate drivers, soft-switching circuits etc. This section of the article identifies those works to understand the existing approaches and solutions implemented by the various authors using GaN switches. The converters have been reviewed on the basis of their various features, such as topology structure, characteristics, operation, merits, and demerits. For the sake of clarity this work has been divided into isolated and non-isolated DC-DC converters in Table 3 and 4, respectively.

VII. DESIGN EXAMPLE OF A BUCK CONVERTER USING GaN SWITCHES
To authenticate the superiority of GaN switches over Si, a design example of a buck converter is presented in this section. A comparative analysis of power losses and efficiency is done after the design and simulation of a buck converter in LTspice. Specifications of both converters are presented in Table 5. An e-GaN FET (EPC2001) [84] and a Si MOSFET (BUK9219-55A) [85] are selected for the analysis based on the specifications. The switching frequency is taken 95884 VOLUME 11, 2023 Authorized licensed use limited to the terms of the applicable license agreement with IEEE. Restrictions apply. as 100 kHz and 350 kHz for Si and GaN switches, respectively. Table 5 clearly indicates the benefits of the higher switching frequency of e-GaN FET in terms of reduction in the size of the inductor and capacitor.
Losses and efficiency calculations of both the converters are done based on standard equations [86]. Loss-breakdown is shown in Table 6. While the Si-based buck converter shows an efficiency of 93.25%, the one based on e-GaN FET boosts the efficiency to 96.61%. The major power loss component in the converter is switching loss and losses in passive elements. Losses in passive components are because of their equivalent series resistance. Table 6 identifies the lower switch losses in e-GaN FET because of the lower on-state resistance and lower gate charge of GaN switches.

VIII. CONCLUSION AND FUTURE ASPECTS
The elementary motive of this review paper is to come out with the updated research status of the GaN-based DC-DC converter circuits and the research gap, primarily for EV applications. The use of electric vehicles is the need of today's time because of the threat of climate change and the hike in oil prices. Researchers and engineers are working towards the performance enhancement of converters, which is an essential part of EVs. Converters based on GaN devices inherently obtain higher power density and efficiency than converters based on Si devices. On the contrary, we have to compromise with the cost of GaN devices and there are numerous other challenges. Therefore, to have optimum benefits, we need to consider the topology, the devices, and the magnetic materials used in the converter altogether. Since the GaN power switch has not reached its maturity stage, future research work should be focused to optimize the characteristics of GaN switches. On the one hand, we need to continue exploring the new topologies of the converter, while on the other hand, there are opportunities in improving the performance of the converter by adopting new materials and designs for the transformer and inductor, including magnetic integration. Future research should also focus on power train integration in EVs in order to reduce the part counts. For this, one of the ideas is to develop such topologies which can have both AC and DC outputs or which can have DC voltages at two different levels to be used in EVs at two different voltage buses. EMI remains a major issue that arises because of the high frequency caused by the use of GaN switches, which need to be addressed thoroughly. As the number of accessories in EVs is increasing with time, power converters with high power ratings should be the desired goal. Eventually, as renewable energy is being integrated with EVs and EVs are being integrated with the grid, critical research work should be carried out to improve reliability, power quality, cost, and efficiency.