Silicon carbide (SiC) wide bandgap power electronics are being applied in hybrid electric vehicle (HEV) and electrical vehicles (EV). The Department of Energy (DOE) has set target performance goals for 2025 to promote EV and HEV as a means of carbon emission reduction and long-term sustainability. Challenges include higher expectations on power density, performance, efficiency, thermal management, compactness, cost, and reliability. This study will benchmark state of the art silicon and SiC technologies. Power modules used in commercial traction inverters are analyzed for their within-package first-level interconnect methods, module architecture, and integration with cooling structure. A few power module package architectures from both industry-adopted standards and proposed patented technologies are compared in modularity and scalability for integration into inverters. The current trends of power module architectures and their integration into inverter are also discussed. The development of an eco-system to support the wide bandgap semiconductors-based power electronics is highlighted as an ongoing challenge.
Power electronics are essential components for every electronics application. Applications range from simple laptop switches to power plants and electrical grids to provide energy. Power modules have historically been silicon-based until recent decades when new applications such as power management for renewables, data center, and electric vehicles/hybrid electrical vehicles (EV/HEV) demanded better performance.
The demand for better power modules is further fueled by climate change, as EV/HEV are poised to replace existing technologies in heavy fossil-fuel consumption sectors such as transportation. The Department of Energy (DOE) has been one of the major drivers for energy consumption reduction as well as development of low carbon emission technologies. According to Monthly Energy Review on energy consumption by sectors in United States of America in 2018, transportation was one of the largest energy-consuming sectors as of 2018, consuming 28% of all energy in the U.S. as shown in Fig. 1 . Improvement of energy conversion in EV and HEV has thus been one of the major focus of DOE. Targets for inverter set by the DOE aim to reach 13.4 kW/L power density for small electric vehicles by 2020 and 100 kW/L for heavy duty electric vehicles in 2025 . Reaching these 2025 goals is heavily dependent on the success of wide bandgap power electronic technology.
Research initiatives in silicon carbide (SiC) power electronics are not new. The Defense Advanced Research Projects Agency (DARPA) has had wide bandgap semiconductor technology high power electronics programs since 2004 for high power military and aerospace solid-state power substations at 2.7 MVA . Naval Research Laboratory continues to focus on wide bandgap devices, processing, and packaging improvements . Army Research Laboratory has programs aiming for efficient power transfer systems for military electrical vehicles . National Renewal Energy Laboratory and Oak Ridge National Laboratory also have aggressive research programs directed toward the improvement of wide bandgap power electronics [6,7]. PowerAmerica Institute, a consortium funded by the DOE for the advancement of SiC and Gallium Nitride (GaN) technologies, also has a roadmap for the development of wide bandgap electronics . Although goals and targets vary, the fundamental research findings and technology advancements are in agreement.
The implementation of SiC technology into traction inverters in commercial vehicles is a recent development [9,10]. Industry wide adoption requires mature SiC devices, processing technologies, new packaging materials, and packaging architectures to enable system integration into inverter. HEV and EV are expected to gain market share at 35% of total new vehicle sales in 2045 . Higher demand would propel an economy of scale and more investments in R&D to allow SiC power electronics to gain momentum for cost reduction and performance improvement to compete against Si technology . Other wide bandgap semiconductors are also under considerations.
This study focuses on issues surrounding power modules for both silicon and silicon carbide power electronics. The learning curve of SiC technology can be accelerated by the experiences established by silicon technology. Targets set by DOE for EV/HEV will be discussed first. Case studies of power module designs adopted by EV/HEV commercial car manufacturers will then be examined. Next, power module architectures from both industry standards and patents will be analyzed in the context of modularity and scalability. Finally, entry barriers for suppliers and paths for cost reduction will be discussed.
2 Traction Inverter
The function of a traction inverter is to convert DC current to AC current using semiconductor devices to drive an electrical motor. Typical power electronics in power modules include diodes, IGBTs, or MOSFETs. For a simple three-phase inverter, one diode combined with one IGBT or one MOSFET forms one switch, and a pair of high and low switches in an inverter circuit form one-phase leg. Si technology is still dominating the market, though some commercial applications are now using SiC technology. To enable variable speed in the motor, pulse-width modulation is often used in inverter design. Inverter designs tend to minimize switching losses and conduction losses for efficient power conversion. The characteristics of high breakdown voltage wide bandgap SiC enable thinner devices, lower conduction loss, and higher power compared against comparable silicon-based devices. Likewise, the lower switching loss of SiC allows devices to run at higher frequency for electric vehicle applications . These advantages result in lower cooling requirements and potential benefits of package size reduction and less energy loss to waste heat for SiC power electronics.
2.1 General Requirements for Electric Vehicle/Hybrid Electric Vehicle Inverters.
A few key requirements and targeted goals in 2020 from the DOE for small HEV/EV inverters [14,15] include: low cost ($3.3/kW) , small inverter size (>13.4 kW/L), light weight (>14.1 kW/kg), continuous power output at 30 kW and peak power output at 55 kW, maximum switching frequency at 20 kHz, low conduction and switching losses and high efficiency (> 93%), robustness against vibration, shock, and extreme temperatures, and lifetime of 15 yr and 150,000 miles, and 5000 h for key vehicle component.
The impact of inverter efficiency improvement is significant. For electric vehicles running at 80 kW or 350 kW, 96% efficiency translates to a sizeable 3.2 kW and 14 kW power loss. Inverter designs  and matching driving patterns [17,18] also play roles in optimizing inverter operating efficiency.
The inverter must be less than 4.1 l and 3.9 kg to meet the minimum requirements for a small electric vehicle. For the limited battery capacity and small space, electrical vehicles must seek light weight and compact size options for inverters. A comparison of Si-IGBT and SiC MOSFET devices under similar operating conditions shows a much smaller temperature increase by using SiC MOSFET, especially at higher frequencies contributed by the low switching losses and less heat dissipation requirements. .
A detailed analysis of inverter components in an R&D system with SiC power modules shows that cooling systems comprised a significant portion of the weight of an inverter [14,15]. SiC power modules and thermal interface materials occupy 7 wt % out of total of 6.09 kg (14.52 lbs) of an inverter. The cold plate occupies 20 wt %, the bussing system with heat sink and current sensor occupies 20 wt %, film capacitor and printed circuit board occupies 8 wt %, and the aluminum enclosure, ports, and fan occupy 29 wt %. Based on component weight distribution, major areas for technology improvement include: high efficiency inverters to reduce energy losses in form of waste heat, high efficiency power module architecture heat transport paths, and compact inverter cooling design to reduce weight of enclosure and accessories.
2.2 Inverter 2025 Targets.
As most requirements of the DOE 2020 goals have been met, a new set of higher expectations has been announced as goals for 2025 . The target goals set by the Electrical and Electronics Technology Team (EETT) for electric drive systems and high voltage power inverters are listed in Table 1. As shown in Table 1, power density is expected to increase by six to eight times. Cost reduction is expected to come from technology innovation and improvement as well as the development of an economy of scale with the expansion of the EV market to 35% of new vehicles sales in 2045.
Expectations on performance, safety, and reliability for vehicle components are much higher than typical consumer electronics. Continuous improvement in efficiency and safety are required. The most challenging goal is to extend lifetime of vehicles and their major components to last 300,000 miles for the most energy and cost savings .
3 Power Modules in Hybrid Electric Vehicle/Electric Vehicle Inverters
A few models of inverters with publicly available information from either manufacturer reports or reverse engineering reports are examined here. Five case studies on existing designs of working inverters can help to understand the state-of-the-art design methodology. The within-module first-level interconnect methods, power module architectures, cooling system designs, and supplier collaboration are discussed. Both Si and SiC technology inverters are included.
3.1 Case 1: Nissan Leaf Inverter.
The 2012 power inverter is assembled with three diodes and three IGBTs in each switch and total of 18 diodes and IGBT in six switches. This simple inverter design keeps the planar assembly concept with liquid cooling at the bottom as shown in Fig. 2(a). The diodes and IGBTs are soldered to thick copper molybdenum (CuMo) spacers for coefficient of thermal expansion (CTE) matching and then soldered on thick copper plate for heat spreading as shown in Figs. 2(b) and 2(c). This copper plate is sometime called baseplate interchangeably. Wirebond methods are used for the electrical interconnects. The insulation layer is a thin silicone layer with thermal grease on both sides placed between the copper plate mounted with power modules and the cold plate of liquid cooling casing. The casing with cooling structure is sturdy but heavy and bulky to keep good contact with planar copper plate for heat exchange between the bottom side of power modules and coolant. The coolant flow is guided by grooved serpentine structures on the top surface areas of the bottom casing in a traditional inverter design. Nissan plans to put inverter components closer to the motor to reduce energy loss from power transfer between the inverter and motor in the future .
3.2 Case 2: GM Generation-2 VOLT Inverter.
The second-generation VOLT inverter released in October 2015 improved in performance and weight over the first generation. The inverter technologies are results of close collaboration between General Motor and Delphi Electronics and Safety . There are two traction power inverters at 48 kW and 87 kW using 12 pairs of silicon diodes and IGBTs as shown in Fig. 3(a). The voltage rating is 360 V at nominal and 450 V at peak value. The power modules have silicon diodes and IGBTs bonded between two direct bonded copper (DBC) using metal spacers to enable double-sided cooling as shown in Fig. 3(b). Interior sides of DBCs have patterned traces serving as circuit interconnects.
These power modules are assembled in two parallel linear arrays on a planar power board. The heat sink structures connected to power module substrates cover extensive areas of inverter electronics including electronics accessories. The design of dual-pipe cooling structures integrated with heat sinks creates cooling passages on both sides of the power modules as shown in Fig. 3(c). These straight dual pipes diverge at one end and merge together at the exit ends with coolant flowing across power modules in the linear arrays arrangement. This makes the coolant-in temperature of each power module the coolant-out temperature of the previous power module in the array. Architectures of power modules and cooling design become more flexible using dual-pipe structures than conventional single-side cooled inverter. The total contact area between the power modules and the coolant is also doubled for the double-sided cooling as compared to single-sided cooling.
3.3 Case 3: Toyota Prius Fourth Generation Inverter.
The 2016 Prius fourth generation inverter has noticeable changes in design from the third-generation inverter. The third-generation inverter has one diode and one IGBT pair in each power module mounted on baseplate with liquid cool underneath for single sided cooling. The fourth-generation inverter has a power module in a card-shape with two pairs of diode and IGBT in each power card as shown in Fig. 4. The reverse engineering report indicates that these power modules use thick patterned copper plates to form electrical circuits on both sides of the chips . Copper spacers are placed between chips and copper plates to adjust the height difference between different chips. The molded power cards having exterior sides of copper plates exposed to enable the flexibility of double-sided cooling, as shown in Figs. 4(b) and 4(c).
The cooling structure has a half-looped pipe with channels connected in-between as shown in Fig. 4(a). Power cards are inserted between channels using electrical insulation layer with thermal grease on both sides. The parallel placement of power cards between cooling channels enables compact double-sided cooling structure design . The coolant flows in from one side of the pipe, crosses through channels to cool power modules, then merges to the exit side of the half-looped pipe. This allows the power modules to receive almost uniform coolant-in temperatures. The newer design achieves 20% improvement in efficiency and 33% volume reduction by reducing switching loss and conduction loss and improving cooling structure . Furthermore, Toyota announced a collaboration with Denso to build its in-house SiC technology capabilities in 2014 . More SiC power electronics are expected to be implemented in their EV/HEV.
3.4 Case 4: Cadillac CT6 Inverter.
The Cadillac CT6 released in March 2016 has further improvements over the VOLT's inverter. The CT6's inverter was a joint development between General Motor and Hitachi Automotive Systems . The inverter design based on the concept of multimode split power gives rises to three voltage source inverters. Among three inverters, two of them function as motor drives and one functions as an oil pump drive. Part of the inverter with the power module assembly and main chassis is shown in Fig. 5(a). The sandwiched power module architecture has diodes and IGBT power electronics soldered between two copper leadframes using copper spacers for height adjustment as shown in Fig. 5(b). The predesigned copper leadframes also function as circuit interconnects.
The power electronics bonded by leadframes are then placed between two aluminum pin fin heat sinks with isolation sheets and molded together into a popsicle-shaped power module. No DBC is used for these power modules. The power module has all external electrical connections going out from the top side of power module as shown in Fig. 5(c). Although the power electronics are connected to two leadframes on both sides for double-sided cooling, the architecture at the power module level enables all-sided cooling, except the top side. The design of cooling structure shown in Fig. 5(a) has the coolant channel in main chassis. The power modules are assembled in parallel. These power modules can serve as coolant dividers when they are immersed into coolant channel in main chassis. When these molded power modules are directly immersed in coolant as the dividers of coolant flows, they become structural elements of the cooling system. It would be greatly beneficial to inverter cooling system designs for efficient volume and weight reduction. This power module design is a patented technology by Hitachi .
3.5 Case 5: Tesla Model 3 Inverter.
Based on a reverse engineering report, the 2018 Tesla model 3 inverter uses 24 SiC MOSFET power modules assembled on a planar structure shown in Fig. 6(a). It is the first all SiC power module adopted by a commercial sedan. The power module design is a result of collaboration between STMicroelectronics and Advanced Packaging Center based in the Netherlands . The inverter is assembled with 24 650 V and 100 A rated SiC MOSFET molded modules. Copper ribbon bonding is the within-module interconnect method and chips are mounted on DBC as shown in Fig. 6(b). The power module is designed by Advanced Packaging Center using current automotive industry practices but is not quoted as an industry standard packaging module. The side view of the teardown report shows that heat sink is connected at the substrate side of the power module for single-side cooling with a pin fin heat sink . It has a better form factor than the Tesla Model S inverter. The Tesla Model S uses Si IGBTs in TO-247 packages. The inverter has 14 Si IGBTs in each switch and 84 Si IGBTs in the three-phase inverter. Each phase leg having 2 × 14 IGBTs is assembled on one board, and three boards form a prism-shape inverter .
4 Survey of Standards and Patents
The following is a survey conducted over power modules in patented technology from the past 20 years. Popular power electronic packaging standards are also included for comparison. The projected applications for EV/HEV are in the voltage range of 600–1200 V and power range at 80–500 kW . The intent is to understand current technology trends as well as future directions for power module architectures and inverter design.
4.1 Power Electronics Package Standards.
Power electronics packaging standards such as TO-220, TO-247, and their variations are well known in the industry. TO-220 and TO247 are JEDEC standards with package outlines . Their variation forms include differences in number of leads, shapes of leads, hole locations, and corner geometries, and are tailored for different applications or different mounting methods. The basic package architecture of these package standards is a substrate at the bottom of the package where power electronics can be soldered on. Wirebonds, typically aluminum wires, between stamped copper leads or leadframes and the top of power electronics form the first level of the within package interconnect. It is then molded into the package using molding compounds to protect the power electronics from external environment. The exposed bottom metal surface provides paths for heat dissipation from the bottom side of the package for single-sided cooling. Isolation foil is commonly recommended for electrical insulation to metal heat sink if leadframe is the substrate of choice .
4.2 Surveyed Patents.
Patented technologies from an industry often indicate the issues of interest at the time of the patent. Based on the patent application procedures, these submissions must be filed in a timely manner even when the concepts are barely developed. A survey of related patents is therefore summarized here as an indication of industry trends in power modules and inverter assembly. Since many patents are in concept stages, no detailed dimensional analysis is conducted. The parameters considered include interconnect methods to electrodes of power devices, package or module architecture, integration of power module and cooling structures, and inverter assembly with emphasis on cooling system. Not all patents include all parameters listed. Assessable parameters in the sampled 20 patents are included for comparison as summarized in Table 2 [31–50].
4.3 First-Level Interconnect in Packages.
Several packaging methods for forming the first-level interconnect to chip are identified. Wirebonds have been widely used to make these electrical connections to sources and gates on the top side of chips [31,33,48]. Soldering or bonding to ceramic substrates or insulated metal is a common method to form electrical interconnect to drains at the bottom side of power chips. Most research and development work focus on interconnects at these top source and gate areas.
Alternatively, solder balls are used to create interconnection between power devices when using wiring board, metal traces on DBC, or patterned insulated metal [32,36,45]. Direct bonding to a predesigned leadframe is another common method [37,42,49]. Recent patents suggest direct bonding or soldering to ceramic substrates or insulated metals on both the top and bottom sides of power electronics [34,37,38,50]. Although soldering or bonding is commonly used in patent wordings interchangeably, more recent patents indicate preferences of transient liquid-phase sintering , sintered silver [46,50], and other specialty metals .
Thermomechanical issues are another concern. Some patent specifies the corner rounding of metal patterns for stress reduction . One such example of patterned posts with rounded edges as a result of stress optimization by finite element analysis is shown in Fig. 7(a). Spacers are used to adjust chip height differences and package height. Copper, copper molybdenum, or other spacer metals are often materials of choices for spacers [46,48]. Molybdenum is suggested to be an alternative to leadframe materials for better CTE matching . Likewise, pillars and posts are alternatives to spacers with design variations [41,48]. Posts with different heights are fabricated on DBC and then soldered to semiconductor chips to form interconnects as shown in Fig. 7(a) . Alternatively, pillars are fabricated on top of chips as well as metal trace on DBC to form interconnects as shown in Fig. 7(b) .
Multichip interconnect methods are also proposed. Interconnect can be formed by patterning and metal deposition on an dielectric layer laminated on top of multichips [35,40,43]. An example is power overlay technology (POL) . A multilevel copper interconnect is built on laminated polyimide dielectric with copper vias connecting to the device-level metal pad as shown in Fig. 8 . This finer design allows lower parasitic and impedance to reduce power loss [51,52].
Overall, the advantages of direct bonding to top and bottom DBCs are larger area for electrical connections to reduce current density which could otherwise be crowded in wirebonds. The current crowding issue in copper ribbons or other metal ribbons is also less severe than that in wirebonds .
4.4 Package or Module Architectures.
Package or module architectures are strongly related to within package first-level interconnect methods. Almost all packages or modules using wirebonds or metal ribbons for source and gate electrodes interconnects are molded on top to protect power electronics against the environment [31,33,44,47]. The substrate or insulated metal at the bottom side usually serves as heat spreader. The TO-247 package method and standard developed in 1994 is still in use today for the simplicity of assembly and reliability . More recent patents have focused on sandwiched architectures with power electronics bonded between two metalized ceramic substrates (DBCs), insulated metals, or predesigned and fabricated leadframes to achieve double-sided cooling [34,36–38,41,42,49]. The interconnect methods discussed in Sec. 4.3 with examples shown in Figs. 7 and 8 can allow DBCs or insulated substrates on both sides of multichip modules [35,40,43]. However, the example of the power module using POL technology uses compliant thermal interface materials such as adhesives, greases, liquid metals, or compressive metal between POL interconnect layer and DBC with heat sink assembly .
The placement of power semiconductor chips within the power module has also been explored. Besides placing chips on the same side of the substrate, alternative methods have been discovered. A third substrate placed at the center of the sandwiched module has been recently proposed . Architectures with third substrates would allow the drain side to be bonded to metalized ceramic substrates, and the source and gate side to be bonded facing the third substrate at the center as shown in Fig. 9. These designs allow pairs of devices mounted in symmetrical positions on both sides of the third substrate for module size reduction . The third substrate can be either ceramic substrates with patterned metal traces or laminated insulators with surface wiring .
A more recent design using two substrates places power devices between two DBCs with one switch face-up and one switch face-down in opposite directions. This placement method gains additional benefits of shorter interconnections between two switches, even heat dissipation, and reduction of module size as shown in Fig. 10 [46,50].
4.5 Cooling Structure and Power Module Integration.
Power packages usually use metalized substrates as heat spreaders or heat sinks. Standalone packages or modules using ceramic substrates, leadframes, or insulated metals as heat spreaders can often be assembled to cold plates, pin fin heat sinks, coolant microchannel structures, or impingement cooling structures [31–34,36, 42,48]. Packages and modules could use one-sided cooling or double-sided cooling depending on available exterior surfaces for heat dissipation. Material of choice is also of consideration. Metal graphite composites are explored for cooling structures for better heat dissipation with higher thermal conductivity . Power module designs with heat exchangers , pin fin structures [44,45,49,50] or microchannels [38,46] usually target for specific coolant passages and inverter structures.
The cooling structure is the crucial component of the inverter which integrates with both power modules and inverter. Power modules using wirebond interconnects are often mounted on a planar casing with coolant chambers at the bottom side for single-sided cooling. To reduce stresses and warpage in planar power inverter assembly, one patent proposed four divided coolant chambers . Likewise, another patent had power electronics assembled on planar casing structures with separate top covers for each switch or phase leg. The isolated structural elements reduce thermomechanical stresses . A strong planar baseplate or casing is usually required to provide sturdiness, reduce potential warpage, sustain vibration, incorporate cooling structures, and meet reliability requirements. The disadvantage of these planar casing is bulkiness. This places constraints on inverter size and weight reduction [33,44,47].
Power modules using two ceramic substrates or insulated metal with two available surfaces for cooling can allow designs with flexible cooling structures for double-sided cooling [37,38,41]. Power modules are mounted on planar structures with coolant flow through both sides separately [38,41], or cross flow with coolant fluid merging between power switches . One patent proposes to have power modules and associated cooling structures in a separate unit .
Inverter assembly is evolving with more flexible designs that do not require large planar casing. Moreover, power modules designed as individual units such as the Hitachi power module shown in Fig. 5(c) or the planar-bond-all power module as shown in Fig. 11(a) enable more flexible coolant flow designs [46,49,50]. The abovementioned power modules can be coolant dividers in cooling structure designs and be stacked in parallel as shown in the example in Figs. 5(a) and 11(b) . As these power modules become cooling structural elements of inverter, compact designs are feasible for inverter size and weight reduction [49,50].
5 Analysis on Modularity and Scalability
Based on case studies of commercial power modules and surveyed patents, the analysis is conducted on both Si technology and SiC technology. Trends, modularity, scalability, and other considerations are discussed as indications of the current direction of research and the technology evolution.
To make a qualitative comparison, categories are assigned to twenty patents and five commercial power modules. Categories are based on assessable cooling surfaces on power packages or power modules. Category 1 represents discrete packages or power modules with single-sided cooling. They are typically mounted on the planar casing of the inverter with heat sink and coolant flow underneath. Category 2 represents discrete packages or power modules capable of double-side cooling. These could be mounted on the wiring board of the inverter to connect to heat sink and cooling structures for coolant passages on both sides. Category 3 represents discrete packages or power modules that are feasible to have greater than three-side cooling and to be independent structural components. They could be assembled by parallel stacking or be immersed in coolant. The category assignment is shown in Table 2 with trends of power module architectures and cooling methods for cases and patents analyzed here shown in Fig. 12.
The trend indicates the migration of power module architectures from category 1 (single-sided cooling) to category 2 (double-sided cooling) and most recently to category 3 being independent structural components capable of more than three-side cooling or immersion cooling. The associated cooling structures have also shown the major design differences and increased flexibility as described in Secs. 3 and 4.5. The only exception to the trend is the Tesla model 3 using category 1 power module design. The anticipated differences are derived from the SiC MOSFET technology which might bring the beginning of another learning curve for SiC technology.
Power module designs for power electronics are based on various considerations. Some designs include electrical components and gate drive circuits along with the power semiconductors in one discrete power module . Some small-scale inverters mount all power electronics together without clear discrete modules . In the case 1 of Nissan Leaf inverter, power electronics are mounted on copper plate in arrays . Power modules in the case 2 of GM VOLT Gen-2 inverter  demonstrate the modularity with discrete modules mounted on power board. Modularity is well defined in the case 3 of power card in Prius Gen-4 inverter , the case 4 of Hitachi power module in Cadillac CT6 , and the case 5 of discrete power module in Tesla model 3 . For the scope of this study, it is found that commercial EV/HEV power modules have been migrating to simple discrete power module architectures with well-defined modularity.
Power modules with discrete modularity are typically friendly for performance and functional optimization, manufacturability improvement, cost reduction, and flexible system integration. For example, power module designs can be optimized at the power module scale for thermomechanical stability, electrical functionality, compactness and light weight, and efficient cooling paths. Each power module becomes a functional block for system integration into the inverter.
In line with goals from the DOE, electrical vehicle technology will be expanded to heavy duty vehicles in the coming years. EETT indicates a list of scalability measures on power modules and inverter . This scaling can be achieved in many ways. Without the weight and volume constraints set for EV/HEV, scalability is not a concern. When volume and weight are constrained due to limited space in each EV/HEV, scalability becomes a big concern. To reach 100 kW/L power density, many potential improvements are listed as scalability goals .
The current study places emphasis on power module architecture and system integration of cooling structures. The importance of a qualitative measure for cooling efficiency associated with cooling surface area to power module volume ratio is also highlighted. Extensive review of cooling methods is covered in the referenced articles [53–55]. Ultrahigh efficiency cooling methods might be needed to meet the requirements. However, inverter cooling requirements are derived from the amount of waste heat generated during the DC–AC conversion processes. Improving energy conversion efficiency should be encouraged, since cooling is meant to handle the waste heat generated during energy conversion. If the efficiency of power module and inverter operation can be improved by just 2% from 96% to 98%, for example, the waste heat to be removed would be reduced by 50%. It directly reduces the burden on the cooling system and hence the volume and weight of inverter.
SiC and other wide bandgap semiconductor materials are under considerations for their superior properties in power electronics applications. Many materials, devices, and processing issues in SiC power electronics still need to be resolved [56,57]. The potential use of higher junction temperature at 180–250 °C also creates materials and packaging challenges. Continued research and development of novel materials, innovative power module architecture, cooling structure design, and system integration are anticipated .
Research is also ongoing in parallel to improve efficiency by improving pulse modulation using multilevel voltages for a closer representation of sinusoidal wave to reduce switching losses. For example, three-level voltage inverters use power electronics with half the voltage rating but double the number of power switches. Power density is less concentrated and less energy loss is expected for waste heat dissipation . In another example, as mentioned in Sec. 3, inverters in the Cadillac CT6 are split-mode power inverters consisting of three inverters. Split power inverters improve efficiencies by leveraging multiple inverters with more design parameters.
6 Challenges in Eco-System
An eco-system for power electronics and inverters currently exists for silicon technology. The expansion of the eco-system is anticipated due to the increasing demands and potential implementation of power electronics using new semiconductor materials such as silicon carbide, gallium nitride, and gallium oxide. The intrinsic properties of SiC are expected to meet the designated technology requirements, but the success of SiC technology implementations still must overcome many device, materials, processing, and manufacturing issues. Likewise, gallium nitride and gallium oxide power electronic technologies are facing similar challenges. Supply chains for silicon carbide wafers, processing, and packaging exist, but at a very small percentage of the power electronics market share . This is mainly due to the high cost and high defect density of SiC wafers and processing. The rough price estimation of Si IGBTs and SiC MOSFETs listed in Table 3 of the Appendix shows the significant difference in cost. Although these are rough estimations based on the discrete packages of the respective semiconductor, it provides some insight on the current status of the technology. The cost of a SiC MOSFET is about 10 times that of a Si IGBT. Furthermore, cost of SiC MOSFET is far more expensive at high voltage and current rating. Challenges in SiC technologies and reliability remain to be overcome. New device structures are under development for better efficiency and reliability . As a result, only a few companies show strong activity in the silicon carbide market. Demands have just increased over the past year after a full silicon carbide inverter was implemented in one model of commercial electrical vehicle. It is believed that higher demands will help to increase R&D investment and drive further increase of market penetration.
The power package and power module architectures are key components in the inverter design and optimization for vehicle level cooling design. These factors make the choice of power package and power module architectures a process of codesign with inverter system integration and cooling structures. The collaboration between car manufacturers and power module and inverter design companies in the five cases described in Sec. 3 have shown the importance of this codesign phase to accelerate the development of EV/HEV technology [21,24,27]. Nonetheless, power modules in case studies are mostly customized designs and not the industry power package or power module standards. Aside from some car manufacturers with in-house power module design and manufacturing capability on both silicon and silicon carbide technologies , most companies are still using contract manufacturing business models. Supplier competition on research, development, and codesign would continue. The continuous collaboration between car manufacturers and inverter design companies would be highly encouraged. Industry standards for power modules specific for EV/HEV applications would be hard to achieve in the near future, and power module volume would be limited by each specialty model from each supplier selected. Without industry standards, there will be significant barriers to entry for newcomers.
Some examples to accelerate the design process are demonstrated by Infineon and On Semiconductor [62,63]. Their recent products of double-sided cooled discrete power modules for electrical vehicles come with design guidelines or evaluation kits. The power modules can be assembled between specialized heat sinks. These heat sinks are stackable. Power modules can be stacked in multilayer between multistacks of heat sinks. There is also progress in the power module by Hitachi Automotive System . The inverter using Hitachi Power module as described in Fig. 5(c) is adopted by Audi automotive for mass production in its e-tron model vehicles.
A reduction in cost will likely be seen for SiC wafers once demand and volume increases. The analysis on inverter components shown in Sec. 2 indicate that 45–55% of weight is associated with cooling and enclosures, followed by electronics parts. The availability of reliable SiC devices might determine the pace of implementation of compact power modules and inverter design. Without both mature SiC devices and packaging methods, innovation will continue to take place seeking interim and long-term solutions for efficient power modules. The near-term cost of the power package and power module would have to factor in the R&D cost of special design and manufacturing. Entry barriers could be eased once the technologies reach maturity and industry standards are set for power modules. Leveraging the applications of power modules in renewable energy sectors, such as solar energy and wind energy conversions, would also help to increase the total volume and the presence of an economy of scale.
This study conducted analysis on trends, modularity, and scalability on power modules from surveyed commercial inverters and patented technologies. The strong interactions among within-package first-level interconnect methods, power module architectures, and inverter cooling system integration are demonstrated. Three categories of power module architectures are proposed. They usher the trends of technology innovation in power modules and inverter designs for EV/HEV applications to meet the goals of energy and sustainability set forth by Department of Energy as well as consensus of industries and societies. Category 3 power module architectures have the highest cooling surface areas to volume ratio in general for efficient cooling. The designs in this category enable more than three-sided cooling and lead to using power modules as independent structural elements of cooling system. These changes enable versatile power module designs, compact system integration into the inverter, and efficient inverter cooling structures. Along with the various methods of efficient power conversion, inverter size, and weight reduction will become viable. Although the state-of-the-art power module and EV/HEV inverters continues to be dominated by Si technology, continuous research and development in SiC technology will gradually increase its market share. Once device performance, processing technology, manufacturability, and reliability meet expectations, reasonable prices can be achieved from the leverage of an economy of scale across multiple industries and applications. The establishment of a SiC technology eco-system will take time but is a strong possibility in the future.
The author is indebted to all who inspired and supported this work.
Appendix: Pricing of Power Semiconductors
Prices found on available catalog at market rate for comparison purposes only. There are other parameters for semiconductor products not considered in detailed herein.1