Cree gets set for MOSFET launch

More-efficient power devices deliver many benefits. Some of these are obvious, such as a reduction in the carbon footprint. However, others, which are application specific, are more subtle and can easily be overlooked.

Hybrid electric vehicles using more-efficient power electronics have the obvious benefit of improving the efficiency of conversion from the DC battery to the AC source that powers the engine, leading to reduced gasoline consumption. However, improved electronics also generate less heat and consequently reduce the size, weight and complexity of the cooling system. This cuts production costs, frees up space under the bonnet and leads to a further increase in the miles-per-gallon figure through a reduction in the vehicle s weight.

Improvements in device efficiency also provide a variety of benefits for military aircraft and alternative energy systems. Power electronics are used to control the flight path of military aircraft and efficiency gains can save weight through a reduction in cooling requirements, leading to a greater flying range. In another example, more-efficient electronics can increase the output of fuel cells, wind turbines and solar panels, by providing more-efficient conversion of their electrical output into a form that is suitable for grid connection.

These benefits are driving substantial efforts to create more-efficient power modules. The workhorse of today s power systems is the insulated gate bipolar transistor (IGBT), a mature silicon device that has little headroom for improvement. But efficiency gains will come from the deployment of SiC transistors, which are undergoing a transition from development to commercial release.

SiC devices outperform silicon equivalents on many fronts, because of their superior intrinsic properties. A wider bandgap produces devices that can perform at high voltages and temperatures, handle current density spikes, and deliver low conduction and switching losses. These attributes allow systems that employ SiC power electronics to deliver high efficiencies, and be lighter and smaller, due to reduced cooling requirements. In addition, systems can operate at high frequencies, which cuts the passive component count.

SiC JFETs, BJTs and MOSFETs are all under development at Cree, which is headquartered in Durham, NC. Currently our main focus is on the MOSFET because it is likely that the SiC market will evolve along a similar path to that of silicon. Initially, silicon JFETs and BJTs dominated the power-switching market, but sales plummeted when a viable MOSFET was launched and its rivals were reduced to serving niche applications. This shift in market share was spurred by the ease of operation offered by the MOSFET, a normally off, voltage-controlled device. In comparison, JFETs are normally on and BJTs are current-controlled devices.

High-voltage devices
Our SiC MOSFET, which is now being targeted for commercial release, will provide customers with a versatile power electronics device. Silicon MOSFETs only serve applications employing voltages below 600 V, because higher voltages produce unacceptably high device resistances, so IGBTs are used beyond 600 V. SiC MOSFETs, however, can operate at up to 6–10 kV.

Our SiC MOSFETs will also have to compete with an emerging technology – GaN transistors. These two devices share many attributes that result from a wide bandgap. However, SiC is a more mature technology and devices can be built on native substrates with high thermal conductivities that improve surge and avalanche ratings. In comparison, nitride transistors are grown on foreign substrates and have to employ a lateral device geometry that complicates packaging and module assembly.

SiC MOSFET performance hinges on the oxide quality of the gate in the transistor. Although SiC shares silicon s native oxide, techniques that silicon developers use to improve SiO2 quality have not worked to date. Instead we have devoted many years of research to developing a thermally grown SiO2 layer and improving its quality. We fabricated our first power MOSFET on the 6H form of SiC in 1993 and four years later we replicated this device on the 4H polytype that we use today. This produces transistors with lower on-resistance thanks to the substrate s higher vertical electron mobility.

More recently, we have added an annealing process in nitrogen oxide gases, which boosts channel mobility through a reduction in interface state density by almost an order of magnitude. In 2002 we fabricated our first nitrided SiC MOSFET and further efforts over the last few years have enabled us to deliver a gate oxide with a mobility of more than 20 cm2/Vs on DMOS structures. This reliable and stable oxide is the key to the fabrication of a commercially viable MOSFET.

Our oxide s mobility is not the highest reported value – researchers at Chalmers University of Technology in Gothenburg, Sweden, have produced oxide mobilities of more than 100 cm2/Vs on low-doped, epitaxial-grown layers. However, this Swedish process is not compatible with vertical structures, making it unsuitable for MOSFET production.

We have assessed the reliability of our MOSFET s oxide with time-dependent dielectric breakdown testing, which revealed a projected life of more than 100 years at fields below 6 MV/cm (figure 2). This testing accelerates the failure rate by increasing the field on the oxide.

The final hurdle for our MOSFET is gate oxide stability. This is being assessed by high-temperature gate bias tests, which heat the MOSFET to its maximum temperature while it is biased at its maximum gate voltage. To pass, the MOSFET characteristics must remain stable for 1000 hours. Data relating to these tests have not yet been published, but the initial results are encouraging.

Military backing
Many SiC developers have benefited from the backing of the US Army Research Lab and the US Air Force Research Lab. These institutions have managed programs focusing on improvements in gate oxides and those that investigate the potential of many types of SiC device. We have been involved in several of these, including a recent project to develop large SiC switch devices, which spurred our development of 1200 V MOSFETs operating at 20 and 50 A.

It is possible to build MOSFETs capable of handling even higher currents by making bigger devices. However, production yields suffer because larger transistors are more difficult to process and the likelihood of a material defect in the device is higher. A better, well known approach to higher currents is to build circuits that share the current load between multiple transistors by connecting them in parallel. SiC MOSFETs exhibit ideal behavior for this – if one transistor has a slightly lower on-resistance than the others it will take more current and heat up, leading to an increase in its on-resistance that shifts current to other devices. We have used this approach to build power modules featuring several SiC MOSFETs for the Air Force Research Lab. These can be used in inverters that drive the flight-control actuators on combat aircraft (see box "Modules for the US Air Force").

SiC MOSFETs can also improve the efficiency of inverters that connect the DC output from a solar-power system to the AC grid. The benefit of SiC MOSFETs in this application has been revealed by Bruno Burger s team at Fraunhofer Institute for Solar Energy Systems (ISE) in Germany, which compared the performance of inverters employing silicon IGBTs and our SiC MOSFETs.

Comparisons were made with a 7 kW, 16.6 kHz three-phase grid-connected solar inverter designed for 400 V AC operation. Voltage peaks will be near 600 V, so 1200 V IGBTs are required in the inverter, which were swapped with our 1200 V SiC MOSFETs during testing.

The wide-bandgap transistors increased the maximum efficiency by 1.9% to almost 98%, which equates to a halving of the inverter loss. SiC MOSFETs also cut the heat-sink temperature from 93 to 50 °C, at an ambient temperature of 25 °C.

Our SiC transistors can provide a significant financial benefit in European countries with feed-in tariffs for renewable energies. In Germany, 7 kW photovoltaic systems are producing 7 MWhr a year with a €0.49 ($0.63)/kWh feed-in tariff. The tariff generates €3430 for the solar-system s owner. A switch to SiC MOSFETs would produce an additional income of €81, assuming the efficiency gains calculated by Fraunhofer ISE. Greater financial benefits can be expected in sunnier climates, such as southern France and Spain, and here there is a stronger case for using SiC MOSFETs in inverters. This cash benefit is one of the less obvious advantages of SiC, but one that complements the reductions in greenhouse gas emissions that stem from greater device efficiencies.

It is clear that SiC MOSFETs are already delivering benefits over silicon IGBTs, even though these wide-bandgap devices are still to be optimized. Inevitably, improvements will follow that will deliver further increases in energy efficiency, which will carve out a bigger share of the overall power semiconductor market for the SiC MOSFET.

Further reading
B Burger et al. 2007 EPE.
B Hull et al. 2007 ISDRS.
B Burger et al. 2008 CIPS.   

View pdf of article