IT is among the hardest substances known to science, only diamond and boron carbide are harder. It can withstand temperatures of 2000 掳C . And it is a semiconductor more resilient than silicon will ever be. Its name? Silicon carbide (SiC), a ceramic material with such amazing qualities that it is equally at home as an abrasive, an armour plating for tanks, a thermal protection coating for re-entry vehicles and as a strong but light structural support for huge lenses and mirrors.
But the most promising new applications for SiC lie in the world of semiconductors. Its combination of properties means that sensors or electronic switches made from SiC will be able to operate in conditions that would destroy silicon chips. SiC devices will be capable of working in the high temperatures inside car and aircraft engines, in geothermal drill holes and even on Venus, the Solar System鈥檚 hottest planet. SiC devices will also be able to operate at high voltages, making it possible to generate high power microwaves using solid-state devices and to control power grids with smart electronic circuits rather than physical circuit breakers.
So why all the fuss about SiC now? After all, the material has languished in laboratories for 30 years or more. The answer is that researchers have only recently discovered how to make large wafers of SiC in a way that can be mass produced. Alongside this progress, they have developed techniques for growing SiC crystals with fewer defects leading to even better crystals.
Advertisement
The evolution of SiC wafers is set to revolutionise all kinds of industry. Practical SiC devices will crop up everywhere from the power generating and distribution industry to the science of planetary exploration. The first large-scale applications, in broadcasting high-definition television (HDTV) and measuring car emissions, are already being tested.
So what makes SiC so special? The key lies in the way electrons behave within the material. Semiconductors are crystalline materials in which atoms are held closely together by atomic bonds. This close proximity allows the orbits of electrons in neighbouring atoms to overlap, leaving electrons free to move from one atom to the next. The overall effect is the creation of energy bands for electrons in the material.
Semiconductors are unique because their crystal lattices contain just enough electrons to exactly fill the lowest energy band, known as the valence band. This creates a kind of electronic gridlock鈥攏o electrons can move so no current flows. This gridlock occurs for all semiconductors at absolute zero, when the material acts as an insulator.
However, if electrons are given extra energy, they jump up to the next band, called the conduction band, where they can move and create a current. So semiconductors can act as insulators and conductors. The difference in energy between the valence band and the conduction band is called the band gap, and it is this that determines the most important properties of the material.
Semiconductors can also be made to carry current in another way: by adding dopant atoms that contain either too many or too few electrons to fill the valence band. For example, an atom of nitrogen has one too many electrons to fill this band. Inside a semiconductor, this extra (negatively charged) electron acts as a charge carrier and results in current flow. So SiC doped with nitrogen is known as a negative or n-type semiconductor. Aluminium, on the other hand, has one too few electrons, and so creates a 鈥渉ole鈥� in the gridlocked electron structure. Holes act like positive charges, moving against the flow of electrons like the spaces between cars when gridlocked traffic begins to move. That is why SiC doped with aluminium is known as a positive or p-type semiconductor.
The amount of charge that n or p-type semiconductors can carry depends on the number of dopant atoms, and this can easily be controlled during manufacture. This ability to fine-tune the conducting properties makes semiconductors very flexible materials. And n and p-type materials are the building blocks of the transistors used to make everything from microchips and radios to light-emitting diodes and flat-panel displays.
But in certain circumstances the special properties of semiconductors break down, making them useless as diodes and transistors. If the temperature increases, for example, the electrons eventually gain enough energy to jump the band gap from the valence band to the conduction band and the material turns from a semiconductor into a conductor.
This is the reason why silicon chips and sensors cannot operate at high temperature鈥攎ost fail above 125 掳C. Silicon components used in car or aircraft engines have to be insulated against high temperatures or actively cooled.
Demolition derby
The performance of semiconductors is also limited by voltage. Applying a voltage sets up an electric field inside the material鈥檚 lattice, which accelerates the conducting electrons. Conduction is a kind of demolition derby鈥攖he electrons accelerate forward, smash into an atom, give up their energy, accelerate forward again, collide and so on. The macroscopic effect of this reckless driving is called resistance.
But if the voltage is high enough, an accelerated electron crashing into an atom passes on enough energy to knock a valent electron into the conduction band. These new conducting electrons are also accelerated. And they dislodge yet more valent electrons, starting an avalanche. The overall effect is a switch from a semiconducting state into a conducting state.
Obviously, the smaller the band gap, the more vulnerable the semiconductor is to higher temperatures and voltages. For silicon at room temperature, the energy that electrons need to jump the band gap is 1.1 electronvolt (eV). Gallium arsenide, another semiconductor, has a band gap of 1.43 eV. But in SiC the gap is 2.9 eV, which makes the material far more resilient. SiC semiconductors work at temperatures above 600 掳C and with electric fields that are ten times higher than those silicon can stand.
Growing crystals
In practice, making SiC crystals that display these amazing properties has proven difficult. The breakthrough came in the mid-80s when researchers at North Carolina State University in Raleigh perfected a way of growing SiC crystals on a large scale. Their technique is to heat a powdered mixture of silicon and carbon to about 2500 掳C so that it sublimes. The gas is then allowed to condense on a seed crystal of SiC. The resulting large crystal can then be sliced into wafers that are used as the substrates for SiC-based devices.
One of the biggest problems is defects in the crystal structure, known as micropipes. These are micrometre-size holes in the crystal that short-circuit the material, destroying its useful electronic properties. Growing SiC crystals without micropipes is a difficult task, so researchers have concentrated on reducing their number to a level where their effects are small. In the early 1980s, SiC crystals had at least 1000 micropipe defects per square centimetre.
Since then, the methods have been fine-tuned and the latest SiC crystals have around 10 micropipe defects per square centimetre. They are also larger. The crystals are semitransparent wafers of blue or green ceramic a few centimetres across鈥攁bout the size of large coins. They range from a few to several hundred micrometres thick.
But manufacturers of SiC wafers will still have to double the size of their wafers and reduce the number of defects to about one or two defects per square centimetre before practical, cost-effective SiC devices can be built. This is perhaps the biggest challenge facing the SiC industry and, naturally, manufacturers are reluctant to share their solutions.
Once the SiC substrate has been manufactured, it has to be turned into a useful device. This requires researchers to grow thin layers of SiC on top of the wafer. It is on these layers that the desired pattern of electrical components must be created by selective doping and etching. Almost all electrical activity is confined within these layers.
Using this method, scientists have been developing prototype semiconducting devices from SiC with 100 defects per square centimetre. One of the first applications has been the production of high-power microwaves, which will be used for broadcasting the next generation of HDTV pictures. Powerful microwaves are usually produced by a klystron, a vacuum tube in which accelerated electrons emit microwaves. But klystrons are inefficient, wasting about half their energy as heat. They are also expensive, costing up to $30 000 each.
This cost is one of the biggest problems facing TV stations which want to upgrade from their current transmitters to ones capable of handling HDTV. According to media mogul Rupert Murdoch, the cost is 鈥渁bsolutely terrifying鈥�. His FOX network which broadcasts in the US faces a bill of $100 million to make the change.
One way to bring down the cost would be to replace the vacuum tubes with solid-state devices that can be mass produced and so are much cheaper to make. But conventional semiconductors, such as silicon and gallium arsenide, cannot cope well with the voltages required to produce microwaves powerful enough for TV transmissions.
Enter SiC. Last year, at a convention of the National Association of Broadcasters in Las Vegas, the American engineering companies Westinghouse Electric and Northrop Grumman demonstrated the first live broadcast of HDTV using solid-state SiC transmitters. Because the SiC devices are 80 per cent efficient, they use less power than klystrons. Moreover, each module costs only about $50. Researchers at Westinghouse believe that the SiC modules should become commercially available this year.
Another area where SiC devices will cause a stir is in the monitoring and control of internal combustion engines. Such instruments will help engineers to improve fuel efficiency and reduce polluting emissions. As the silicon electronics and sensors that normally do this job can only cope with temperatures up to 125 掳C, these devices must be insulated and placed in contact with engine parts that do not rise above this temperature.
Measuring emissions
SiC devices, on the other hand, could be placed almost anywhere inside the engine. They could be in direct contact with cylinder heads or inside the exhaust pipe, providing more accurate measurements of emissions. With better measurements will come the control needed to improve efficiency and reduce emissions. This versatility also means that fewer wires and connectors will be needed, resulting in less maintenance and improved reliability.
鈥淵ou can cut down on the number of passive components normally needed when using silicon devices, increasing automotive electronics reliability levels,鈥� says Jayant Baliga, director of the Power Semiconductor Research Center at North Carolina State University.
High-temperature sensors could also dramatically change the way military aircraft operate. The US Department of Defense wants to reduce or even eliminate active cooling in future military aircraft. SiC devices could replace the cooling systems that are required to prevent the aircraft鈥檚 electronics from overheating.
The US Air Force believes that advanced SiC electronics on an F-16 fighter would allow the aircraft to shed almost 300 kilograms. This load could be replaced by extra fuel or weapons. And without cooling equipment, the aircraft would require less maintenance. Commercial aircraft could also benefit, saving their operators millions of dollars per plane in savings on fuel and maintenance.
At the Jet Propulsion Laboratory in Pasadena and the NASA Lewis Research Center in Cleveland, Ohio, researchers have developed a single-chip SiC microsensor capable of measuring how much hydrocarbon gas a car exhaust system emits. The JPL sensor contains a thin layer of porous SiC. The layer adsorbs hydrocarbon gases at high temperature and these change the layer鈥檚 conductivity. The change in conductivity gives a measure of how much gas has been adsorbed. Until now, only large and costly mass spectrometers were capable of detecting hydrocarbon gases directly.
鈥淎utomotive applications are just the tip of the iceberg for this microsensor,鈥� says Virgil Shields, a researcher at JPL. 鈥淭he degradation of industrial lubricants, remote sensing of human environments, and the decay of organic matter are other potential applications,鈥� he explains. Shields says his sensor could be produced for less than $20.
But there are challenges to overcome first. Any material coming into contact with SiC at high temperature tends to interact with it and change its properties. The biggest problem is with materials used to make contacts, such as aluminium and nickel. Shields believes that more work is needed to solve these and other problems. 鈥淲hat we really need is a source of funding to commercially produce such a sensor,鈥� he says.
Perhaps the most exciting application for SiC devices is as 鈥渟mart鈥� electronics for controlling output and distribution of electricity in power grids. Electricity companies find it difficult to measure and control demand in different parts of their grids because electronic switches do not operate reliably at high voltage.
Switching power
Instead, companies measure demand centrally at power stations with mechanical voltmeters. And to ensure a reliable service they generate more power than is needed. On average, power generating companies produce 20 per cent more electricity than is consumed鈥攁 huge waste.
SiC devices could change all this. They could operate remotely throughout the grid, measuring changes in demand and switching power from one area to another as needed. Because of this more efficient use of power the margins could be reduced. The Electric Power Research Institute in Palo Alto estimates that a mere 5 per cent reduction in the reserve-power margin would eliminate the need for some $50 billion worth of new power plants within the next five years. A similar kind of efficiency improvement could help to make electric vehicles viable for the first time.
While SiC devices on Earth could save billions, they could also help explore the Solar System. With no cooling airflow, satellites must find some other way of dissipating heat produced by onboard electronics. The usual method is thermal radiation or by allowing the heat to boil a liquid off into space. But high-temperature SiC devices would make this unnecessary. As a result, satellites would be lighter and cheaper to launch.
Perhaps the most exotic goal that SiC devices could make possible is the exploration of Venus, where the surface temperature is around 450 掳C鈥攈igher than the melting point of lead. Missions to Venus are some years away. But for life on Earth, the benefits of SiC should come a little sooner.