Thursday, December 15, 2011

Future material and devices for power electronics

Alexandre Avron, market analyst in power electronics, Yole Développement, provided a briefing on semiconductor material’s potential through an analysis of devices and systems for power electronics.

According to him, there is still a bright future for silicon. It will keep good market share until at least 2016 and even further, being cost competitive and very standard. On the other side, SiC is more applied to higher voltages. These are the smallest markets, but probably the one requiring SiC properties the most. PV inverters and EV/HEV are at intermediary voltage levels, they could both be targeted by SiC and GaN, this makes the predictions very difficult.

No technical aspects helps in knowing which material will be more used. They have their advantages and drawbacks, and both deserve their place. Prediction must be based on developments advancements. The points to watch about SiC and GaN devices include: samples availability is a main point for future integration, reliability is also a main concern, especially for SiC devices, voltage capability seems to keep GaN at smaller power, and cost: GaN appears to be potentially cheaper, as it is based on Si wafers and can be CMOS compatible.

Silicon, SiC and GaN materials and devices
The super junction (SJ) MOSFET market forecast - 2009-2016, the SJ MOS device market growth will stay around 11 percent until 2016. Recovery will push the market up to 12 percent annual growth in 2013-2014. The CAGR ‘11-‘16 is 12 percent.

Main added value of wide band gap semiconductor includes:
Higher switching frequency -- 20 kHz power converter and reduce the need for electric noise filtering.
Higher voltage operation -- 15kV is envisioned for train traction or energy T&D. It will divide by a factor 2 to 3 the size and weight of the overall power module.
Higher junction T° -- 200°C is envisioned. It will likely increase reliability and allow the integration of the inverter within the motor in traction applications. There is reduction in power losses -- 2 percent + more efficient.Source: Yole Développement, France.

Available devices in SJ MOS have specifications of 900V/20A, 0.8 A/mm², Ron: 190 mO and priced at $0.5 to 2/part. SJ MOS growth will be very high -- approximately 13 percent until 2016. It will enter new power range thanks to higher operating voltages. Available SiC devices have specification 1700V/20A, 1.8 A/mm², Ron: 80 mO and priced ˜$50/part. Cree’s SiC MOSFET bare dies are now available. Samples and production volume will accelerate integration in high-end systems.

Available GaN/Si devices have specification 200V/12A, 2 A/mm², Ron: 18 mO, and priced $5/part. Developments are very active. We expect GaN devices to be widely used in the 200V-600V range for high-end applications. Planar (lateral) MOSFETs will probably disappear of the HV MOSFET market in the next 15 years.

The SiC market will grow at 46 percent per year up to 2016. The GaN market will grow at 275 percent per year up to 2016. The SJ MOS market will grow at 17 percent per year up to 2016. There is caution advised. There is a limitation in the analysis as for some of the applications (EV/HEV, PV inverter, motor control or UPS), the three technologies are competing. Thus, we cannot exclude part of the business to displace from one to another.

Today’s solution of interest is still silicon, either with IGBT or Super Junction MOSFET. We envision planar MOSFET to nearly disappear, being replaced by SJ MOS which will become a standard. We already started to see this evolution, with SJ MOS being more and more proposed by foundries and smaller players (AOS, SMIC, Magnachip, Icemos).

GaN devices are now available at lower voltages, near 200V. System manufacturers start to think about DC/DC converters and smaller power supplies with GaN. The pending question is to now how they will increase in voltage, and which cost/performance capacity they will propose. We believe in 600V GaN competing with Si.

SiC switches are now available at 1200V and 1700V. They are still under field test, and are very expensive. They would have a greater interest at higher voltages (up to 10kV), but only a few device makers are targeting this area. At the moment, PV inverters and UPS are the market where SiC shows most potential, mainly because of diodes being already used. EV/HEV can adopt different scenarios between wide adoption and slow adoption. On the other side, high temperature niche applications will benefit from SiC as soon as packaging solutions are ready.

Potential applications: EV/HEV and PV inverters
SiC diodes are today already in production, mainly coupled with IGBT technology. The next step of large production for SiC-based devices will be for JFETs and later on MOSFETs. But today product availability, price and test results are not in line with inverter makers expectations. Particularity of GaN-based inverters will primarily target medium voltage applications (in the 200 – 600V range). GaN-based diodes will appear basically at the same moment as switches will be ready for industrialization.

Focusing on EV/HEV -- device roadmap for full/plug-in hybrid and EV, it is believed that the 1200V and 1700V will increasingly be used in EV and HEV. It will open opportunities for SiC devices. We believe GaN will have more difficulties in being used in powertrains, but it will probably take market shares in chargers. The need for high efficiency, smaller and lighter converters will push for wide band gap semiconductor.

Forecast 2010-2020 for EV/HEV
Hybrid type cars will represent most of the EV/HEV market, according to Yole. As the amount of power electronics is still consequent, we observe the same share at module level. SiC’s penetration in this market can follow different scenarios, between wide and rapid adoption and slow episodic use of SiC.

EV/HEV is a very promising market. It is forecasted to reach $1.5 billion in 2015 for the power module market only, it makes a lot of companies to prepare a strategy of entry. Future device landscape is blurry. Today, IGBT is the almost unique solution available for traction converters. The biggest players are working towards both SiC and GaN (Toyota, Delphi with GM), and SiC looks to be a better challenger for EV, as of today. The EV/HEV market players will have to think at system cost, to see that the extra cost of wide band gap can be smoothen by system cost savings. Device samples will also need to be widely available to see a penetration for SiC or GaN.

Focus on PV inverters
SiC and GaN switches are to enter PV inverter market starting from 2013. We expect IGBT to keep a large market share, as a cost competitive solution. There is still a bright future for silicon. It will keep good market share until at least 2016 and even further, being cost competitive and very standard. On the other side, SiC is more applied to higher voltages. These are the smallest markets, but probably the one requiring SiC properties the most. PV inverters and EV/HEV are at intermediary voltage levels, they could both be targeted by SiC and GaN, this makes the predictions very difficult.

No technical aspects helps in knowing which material will be more used. They have their advantages and drawbacks, and they both deserve their place. Prediction must be based on developments advancements. The points to watch about SiC and GaN devices include: samples availability is a main point for future integration, reliability is also a main concern, especially for SiC devices, voltage capability seems to keep GaN at smaller power, and cost: GaN appears to be potentially cheaper, as it is based on Si wafers and can be CMOS compatible.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.