While silicon has almost reached its theoretical limit, silicon carbide power devices have achieved a high degree of reliability and maturity, offering fast switching and an unprecedented efficiency level in the automotive sector. This article, based on an analysis conducted by PGC Consultancy, explains how SiC will play an essential role in the transition from 400-V to 800-V (and beyond) electric-vehicle systems.

In a previous article, “SiC Power Devices: Lowering Costs to Drive Adoption,” the costs of a SiC device were analyzed, justifying why a SiC MOSFET costs 2× to 3× as much as an identical Si IGBT and attempting to anticipate how the price of these will decrease over time.

Mapping power devices

The 2021 generation of SiC MOSFETs, from different manufacturers and with voltage ratings of 650 V, 1,200 V, and 1,700 V, have been mapped in Figure 1, in which their specific on-resistance has been plotted against the rated voltage. Silicon-based MOSFETs and IGBTs have also been included for reference.

The diagonal lines in the plot represent the unipolar limits of each material — that is, the lowest resistance theoretically possible for a given voltage rating. This limit dictates that a 2× increase in the breakdown voltage of a MOSFET will result in about a 4.5× increase in the resistance of the device. As already pointed out in the previously mentioned article, a lower specific on-resistance is an essential factor for reducing the cost of SiC power devices. When assessing these graphs, keep in mind that even though a device is far from the ideal unipolar limit, it can still be a good device. However, because its die size will be larger, its cost will be higher. That is the reason why, to maximize the yield, manufacturers will aim to shrink their technology to get as close as possible to the unipolar limit.

SiC vs. Si power devices

As shown in Figure 1, the latest generation of Si devices are nearly optimal, as most of them are very close to the unipolar limit. These devices start to exhibit their lower limits below 100 V, where fixed resistances due to the substrate, JFET, and channel begin to dominate the total device resistance. While the two Si IGBTs examined are below silicon’s unipolar limit, they involve significant switching losses when compared with the other unipolar devices. Hence, before the advent of SiC, designers had two silicon-based options: fast-switching MOSFETs operating at low voltage or slow-switching IGBTs operating at medium and high voltages.

When looking at the SiC devices in Figure 1, we observe that there is significant dispersion within each voltage class. What the devices with the lowest resistivity have in common is their high current ratings. As a result, they are large chips with large active areas, and a smaller proportion of the chip is reserved for non-current–carrying portions, such as the gate pad and termination regions. As expected, devices that have been launched most recently, since mid-2021, achieve even better performance than those mapped.

The factors that contribute to the device’s separation from the unipolar limit include the overdesign of the drift region and the fixed resistances originating from the substrate and channel region. First, there is a clear trend indicating that some manufacturers are overdesigning their devices, meaning they would be able to support considerably higher voltages than they are rated for. This is done for reasons of reliability and robustness, protecting the oxides therein from the high electric fields when operated near their limit. Hence, a plot with the actual breakdown rather than the rated breakdown would look somewhat different, with the devices fractionally closer to the unipolar limit. This will be explored further in a future article.

The fixed resistances mean that the SiC MOSFETs rated for 650 V in particular are not very close to the unipolar limit. Their resistance is high enough that today’s 650-V IGBT variants, trench-gated field-stop technology, can achieve conduction losses lower than these SiC MOSFETs. However, when both devices are operating at the same switching frequency, this small increase in conduction losses is negligible compared with IGBTs’ large switching losses.

This is what allows Tesla to demonstrate a faster-switching, higher-efficiency solution at 650 V, despite the fact that its 2018 Model 3 inverter is roughly 40% the weight of that of a Nissan Leaf’s inverter that delivers half the power of Tesla’s. The efficiency advantages will have allowed for a reduction in the quantity of expensive and heavy batteries required in the vehicle, thus recouping the component costs.

At 1,200 V, SiC MOSFETs prove their superior characteristics, with the specific on-resistance closer to the SiC unipolar limit. These devices, at 1,200 V, are 14× to 33× above the unipolar limit, compared with 35× to 90× at 650 V. As shown in Figure 2, a 650-V SiC MOSFET has 2× the power density and 6.5× lower switching losses than a 650-V Si IGBT. These differences are amplified at higher voltages, with a 1,200-V SiC device having 16× greater power density and 11× lower switching losses than a 1,200-V Si IGBT.

SiC enables 800-V EV batteries

There are several advantages to doubling the 400-V system standard to 800 V in the EV market. These advantages stem from the simple idea that doubling the voltage allows for twice the power to be delivered through the same cabling, or that maintaining the same power would allow for half the current to be carried (or a tradeoff position between the two). On one hand, greater power delivery results in faster charging of the rearranged battery stacks. On the other hand, a reduced current would allow for a reduction in the copper windings surrounding the motors, reducing its size and weight, with a similar reduction possible in cable weight. These benefits have a significant impact on overall system efficiency, range extension, and/or system cost reduction.

It is clear from the analysis of today’s state-of-the-art SiC devices that SiC is a wide-bandgap semiconductor playing a key role in maximizing the potential of a transition to 800 V bus in EVs. At 1,200 V, SiC begins to show off, leaving little but reliability concerns, design legacy, and/or conservative application spaces as reasons for sticking with Si IGBTs. This is an open question, therefore, for the recent Porsche Taycan, in which a Si IGBT solution won out despite the transition to 800 V, while simultaneously, the 900-V Lucid Air opted for SiC MOSFETs.

Conclusion

In summary, Si technology is about as mature as it is possible for it to get, with MOSFETs on the horizon and IGBT technology well established. SiC is a relative newcomer, but after only a decade and three generations of devices, SiC MOSFETs have proven to be competitive with Si IGBTs. The research backs up what PGC Consultancy has said for a long time: While SiC is unquestionably good at 650 V, it keeps getting better at 1,200 V and beyond.

By: DocMemory
Copyright © 2023 CST, Inc. All Rights Reserved