SiC BJTs promise to cut the cost of greener driving

Oil prices are rocketing. Black gold shot past $100 a barrel this spring, and many speculate that it could hit twice this figure by the end of the decade. This may be great news for oil traders, but for the rest of us it will just mean higher prices at the pump.

Against this backdrop, alternative vehicle fuels look more attractive. One option is biofuels, such as ethanol, which have been touted as promising alternatives to oil. However, closer inspection reveals the folly of going down this route. Fuel manufacture comes at the expense of global crop production, and this can t be the right thing to do when many of the world s poor have barely enough food to eat.

On a personal level, it is possible to minimize the effects of higher gas prices – and simultaneously reduce your carbon footprint – by switching to a more efficient automobile. This might mean the purchase of a small diesel, but more radical alternatives also exist. Although fuel-cell cars will probably be far too expensive for the foreseeable future, it is possible to go out today and buy competitively priced hybrid electric vehicles (HEVs) from the likes of Honda, Toyota and Lexus.

HEVs deliver more miles per gallon by using an intelligent control system to link a combustion engine with an electric motor. Combustion engine efficiency increases, which slashes fuel consumption by up to 50%. Further improvements are possible, thanks to plug-in HEVs. Charging from the grid cuts airborne pollutants and could deliver an incredibly low carbon footprint if the primary energy source is renewable.

Although HEVs are enjoying some commercial success, further improvements will spur sales. Alongside better batteries, one area needing attention is the power electronic system that drives the motor. This features silicon PN diodes and insulated gate bipolar transistors (IGBTs) to convert the DC source from the battery to an AC form that can power the motor. The systems in today s HEVs are too bulky, so a slimmed-down version would free up space and boost automobile efficiency.

A big impact could be made by eliminating the separate water-cooling system currently needed for the power electronics, which maintains the temperature of silicon devices at less than 150 °C and allows these components to operate reliably. Replacing the silicon chips with SiC equivalents would make this water-cooling system redundant because the combustion engine s system could be used instead.

A switch to SiC would involve the deployment of SiC Schottky diodes and transistors in the boost converter, the DC/AC inverter and the step-down (buck) converter. The boost converter s job is to up-convert the voltage from the DC battery to around 400 V DC. This output is fed into the DC/AC converter. A three-phase AC output from this module produces variable amplitude and frequency that can control the speed of the electrical motor. Step-down converters are also required to transform the high input voltage into lower ones that are suitable for other electronics in the car.

The transistor s role is that of a switch, which can block high voltages in its off state and conduct high currents with low-power losses in its on state. The current alternates between the transistor and a diode that protects the transistor from inductive voltage overshoots. Both of these SiC chips can operate at temperatures well beyond 300 °C, and the maximum operating temperature tends to be limited by their packaging technology. Thanks to the wider bandgap, these devices also enjoy lower conduction and switching power losses than silicon equivalents.

In the past, poor-quality substrates have prevented SiC power devices from fulfilling their promise. Micropipes were the major killer, but they have been eliminated in 100 mm and smaller substrates. Cree has also reduced the density of basal plane dislocations, which will aid device performance by enabling manufacture of reliable, large-area chips.

The breakthroughs in substrate quality have opened the door to the manufacture of high-quality, large-area power devices. SiC Schottky diode chips with 1200 V, 50 A ratings are available from a handful of manufacturers, and the race is on to produce and market SiC power transistors that can deliver all of the benefits associated with this material.

SiC developers have already demonstrated several species of SiC power transistor, including MOSFETs, junction field effect transistors (JFETs), bipolar junction transistors (BJTs) and IGBTs.

The JFET is the most developed technology. This transistor, which is manufactured by SiCED and SemiSouth Laboratories, delivers fast switching and benefits from a low on-resistance. However, the device is normally on, which raises concerns about reliability for HEV applications due to the risk of a driver supply voltage failure.

The MOSFETs that are being developed by SiCED and Cree are another option. However, the channel mobilities in this type of transistor are relatively low, and there are question marks over oxide reliability and threshold voltage stability.

Cree has also been looking at IGBTs. However, this transistor suffers from high on-state losses owing to an additional 3 V forward voltage drop resulting from a PN junction. The IGBT is better suited to high-voltage applications of more than 5 kV, such as the motors in trains and high-voltage DC power transmission.

This leaves the BJT, which is a very promising device for HEV deployment, thanks to its fast switching speeds and low power losses. At TranSiC – which is based in Kista, Sweden, and is a spin-out from KTH Royal Institute of Technology – we have been developing this device since 2005 through funding from Volvo Technology Transfer, Midroc New Technology, the Swedish Energy Agency and the Swedish Governmental Agency for Innovation Systems.

Cree and United Silicon Carbide are also developing the SiC BJT, but we are the first to market following our product launches in May. We now offer two forms of 1200 V BJTs – 6 and 20 A versions, which have active areas of 3.4 and 8.4 mm2, respectively. Customers can build their own modules by purchasing our single dies, buy our TO-220 packaged BJTs, or purchase BJTs mounted and wire-bonded onto a ceramic direct copper bonded substrate.

Our devices are manufactured on high-quality 3 inch SiC in the Electrum Laboratory, which is owned by KTH. A vertical NPN structure is employed, with dry etching used to create the emitter and base regions. Ion implantation forms low-resistance ohmic contacts with the base, along with high-voltage junction termination. The surface is passivated by thermal oxidation of the etched SiC. This also cuts defect concentrations at the surface and boosts current gain through a reduction in surface recombination. Pads with two metal layers are added over the active region for the bond wires.

Our BJTs produce low-conduction power losses because the collector-emitter voltages (VCESAT) at a collector current density of 100 A/cm2 are just 0.8–0.85 V (figure 1). In comparison, silicon IGBTs have a VCESAT of at least 2.5 V. This is significant because it slashes the conduction power chip losses by 70% (assuming that both types of chip are the same size) and cuts the chip s cooling demands.

Higher operating temperatures increase the on-resistance of SiC BJTs and reduce the current gain (figure 2). This is beneficial because it enables current sharing and ultimately the stable parallel connection of several transistor chips.

Our 1200 V BJT chips also have a wider reverse bias safe operating area than their silicon equivalents – the breakdown voltage is 1500 V across a wide temperature range (figure 3). This characteristic is a plus point for hard-switching application in motor-drive systems, where there are rapid changes from high currents to high blocking voltages.

Our BJTs can meet the requirements for fast switching at low power losses in the boost converter, thanks to minimal storage of charge and a high current gain (figure 4). They are capable of switching speeds of 100 kHz at DC voltages of 600–800 V, which are typically used for a 1200 V transistor.

We have demonstrated our BJT s ruggedness with 550 °C I-V tests that house our chip in a special high-temperature package. Its short-circuit capability was verified at room temperature by a 10 µs wide turn-on pulse with a very small series resistance at a VCE of 600 V, followed by successful turn-off. This implies that our device is unlikely to fail in motor-drive systems that have a mistimed transistor turn-on.

We are now scaling up production to 100 mm wafers to increase production capacity and reduce die-fabrication costs. Other targets on our agenda include a BJT gain of 100, which will reduce the current required to maintain the device s on state. Ongoing activities include further fast turn-off tests at high voltages, and efforts to cut the on-resistance and associated on-state power losses. The latter goal will be realized through optimization of the junction-termination technology, which will permit an increase in collector doping. All of these advances will make our BJTs even more appealing for both the HEV market and niche applications, such as high-temperature electronics.   

View pdf of article