The performance advantages of Silicon Carbide (SiC), a wide bandgap material, are well known to designers of high-voltage power systems. The drawbacks of the products and the supply chain that supported them, however, have in the past appeared sufficiently serious to dissuade some designers from taking the risk of using SiC components in switching power converters.
A distributor such as Future Electronics is in a unique position to see the rapid evolutions as they happen in a growing market, such as that for SiC MOSFETs. And it is clear now that the drawbacks of standard SiC products that might have discouraged some designers in the past have for all intents and purposes been eliminated.
It is true that the successful use of a SiC MOSFET in power circuits which historically would have used a silicon MOSFET or IGBT calls for a rethinking of the approach to the design, as this article will explain. But the availability, reliability and affordability of SiC MOSFETs have all improved dramatically compared even to two or three years ago. Adventurous designers who are prepared to think differently about high-voltage system components can now gain a substantial advantage over systems that fail to move on from the familiar silicon MOSFET or IGBT.
SiC: Time to Break Out of Its Niche?
To date, usage of SiC MOSFETs has generally been confined to end product types in which an unusually high value is placed on efficiency, power density and light weight, or high-temperature operation. For these reasons, SiC MOSFETs are widely used in:
- Inverters in renewable energy generation systems
- Deep-hole drilling equipment
- Induction heating systems
The distinctive requirements of these applications reflect the particular strengths of SiC power semiconductors (see Figure 1):
- Extremely low switching and conduction losses – a SiC MOSFET’s zero reverse-recovery charge helps markedly improve conversion efficiency. SiC MOSFETs also have lower on-resistance than silicon equivalents, and on-resistance is more stable over temperature.
- High switching frequency – the ability to operate at maximum frequencies as much as ten times higher than that of an IGBT with the same voltage rating means that a SiC MOSFET requires much less inductance and capacitance, helping the designer to reduce system size and weight.
- High thermal conductivity, which allows the use of a smaller heat sink, again helping to reduce size and weight.
- High-temperature operation – this gives the designer greater headroom in the design of the thermal management system, and can again enable the use of a smaller heat sink.
Fig. 1: comparison of the construction of a silicon (left) and a SiC (right) MOSFET
Despite the substantial physical superiority of the SiC material over silicon, however, SiC’s share of the market for power transistors remains small, and this is probably due to designers’ fears over the three factors of availability, reliability and affordability. These fears in the past had a reasonable foundation; SiC MOSFETs did indeed suffer from these drawbacks until relatively recently. On all three counts, however, the situation today is very much improved or indeed eliminated altogether.
Broader Supply Base
Before 2010, there was only one high-volume source of SiC wafers: Cree. Naturally, this meant that the supply chain for packaged SiC MOSFETs was tight and volatile. Today, the situation is very different; several semiconductor manufacturers now perform in-house SiC wafer fabrication, including, in Europe, ROHM Semiconductor and STMicroelectronics, as well as ON Semiconductor and Littelfuse.
At the same time, supply volatility has been greatly reduced. In the past, one market sector – inverters for solar power-generation equipment – sucked up almost all of the available supply, leaving a tiny surplus for the wider market. This meant that it was very hard to predict in any given month whether any product would actually be available to buy.
With more suppliers in the SiC market, there has been a huge increase in the rate of production of SiC wafers, and so the inverter market consumes a smaller proportion of total production. There is now a reliable unallocated supply at any time of available-to-buy standard SiC MOSFETs from distributors such as Future Electronics that are committed to the technology.
Another concern that has discouraged designers from the use of SiC MOSFETs has resulted from reports of poor reliability compared to equivalent silicon devices. In the early years of SiC production, this was justified. Over the past five years, however, SiC device manufacturers have spent heavily to implement sophisticated and effective testing and validation processes which enable them to identify potentially weak devices and to prevent them from entering the supply chain.
In fact, SiC device manufacturers now make detailed quality test reports freely available on the web. These provide users with assurance that the quality and reliability of SiC MOSFETs are now equal to that of any equivalent silicon device.
Unit Cost vs System Cost
The third concern that designers have had, affordability, has to some extent declined in importance as the supply chain has broadened. Market forces have worked their usual magic, and a better balance between supply and demand has helped to put downward pressure on the unit cost of SiC MOSFETs.
More significantly, SiC device manufacturers are starting to migrate production to larger wafer sizes. Today, 70% of the world’s SiC devices are fabricated in small 4” wafers. Some manufacturers, such as ROHM Semiconductor and STMicroelectronics, have already begun SiC production on 6” wafers; this gives a huge reduction in production costs, and this will feed through into lower prices for packaged devices.
The fall in SiC device costs relative to silicon device costs is set to continue as well, since manufacturers’ road maps forecast production on 8” wafers in 3-5 years, promising another big fall in production costs.
Nevertheless, SiC is a more expensive material than silicon, and the unit cost of SiC MOSFETs is noticeably higher than that of silicon MOSFETs or IGBTs, and will likely remain so. But when total system cost, rather than MOSFET unit cost, is compared, the SiC alternative can be preferable in high-voltage power-system designs – provided the user is prepared to rethink the circuit design to make the most of the attributes of SiC devices.
This can mean for instance:
- Reducing the size, weight and cost of the heat sink, and allowing a SiC MOSFET to run hotter than the equivalent Si MOSFET or IGBT. ST’s SiC MOSFETs such as the 1,200V SCT50N120 are rated for a maximum junction temperature of +200°C, compared to +150°C for a typical silicon MOSFET and +175°C for an IGBT.
- Increasing the switching frequency and reducing the size of the inductor and capacitors in the power-converter circuit.
- Including transportation costs in the system cost budget. A system based on SiC technology should be both lighter and smaller than the silicon-based alternative, and this can result in a substantial reduction in shipping charges.
Care Required in Board Layout
As part of the process of rethinking the power system, the designer must also consider the special factors that affect the way a SiC-based circuit is laid out on the board.
In particular, high di/dt and dv/dt ratios during switching operations call for careful design of all switching loops and nodes. It is important to keep a close eye on parasitic inductance and capacitance in the design to keep current/voltage spikes at a minimum and to reduce the risk of exceeding EMI emissions limits.
A successful SiC layout requires smaller and more carefully designed switching paths than is the case with designs using silicon devices. There are however tools available to guide the designer through the process of optimising the layout; the Dynamic Characterization Platform, a tool developed jointly by Littelfuse and Monolith Semiconductor, is a good example.
Fig. 2: Comparison of attributes of SiC, silicon and gallium nitride (GaN) semiconductor materials
Figure 2 illustrates in a diagram the size of the advantage that SiC devices (and gallium nitride, another wide bandgap material) have over silicon equivalents. When implemented successfully, these devices can produce dramatic results. Examples of successful designs implemented by customers include:
5kW LLC DC-DC converter:
75% smaller than the equivalent siliconbased design, with 85% weight saving. 60% more efficient.
Total power dissipation reduced by 75%; lower component count.
6-20kW EV battery charger:
Up to 50% higher power output with the same physical dimensions as the equivalent silicon-based design.
Magnetic bearing amplifier:
80% reduction in filter inductor size and weight, 25% output power increase.
New Opportunity to Improve System Cost and Performance
The description above shows that SiC devices are now ready for use in any high-voltage power-switching application. SiC offers lower on-resistance for higher efficiency, and higher switching frequency for smaller passive components and reduced size and system cost. Lower power losses at high temperature and support for a higher maximum junction temperature help to reduce the cooling requirement and enable the use of a much smaller heat sink.
In addition, SiC MOSFETs are easier to drive, resulting in a lower component count and giving reduced bill-of-materials cost and improved reliability.
As Tables 1 and 2 show, there is a large and healthy supplier base today for SiC diodes, MOSFETs and modules. The problems of availability, reliability and affordability have been eliminated by the SiC device manufacturers’ investments in production, design and testing.
For high-voltage system designers, the time has come for experimentation with the latest standard SiC devices in new designs.
Table 1: SiC packaged device manufacturers’ current and planned product portfolios
Table 2: typical specifications for SiC power ICs available as packaged parts from STMicroelectronics, ON Semiconductor, Littelfuse, ROHM Semiconductor and Microsemi