The adoption of Silicon Carbide (SiC) power semiconductors in everything from electric vehicles to solar PV and industrial motors is accelerating, but where does the material come from? What’s so special about it? And why has it taken so long for SiC to gain traction in the semiconductor industry, when it was first used as the basis for radio detectors over a century ago?
Stardust on your boots
The earth’s crust comprises about 28% silicon and 0.03% carbon, so you might think that you’d find enough Silicon Carbide (SiC) to make a few semiconductor die stuck to the soles of your boots after a long walk in the countryside. If the walk was over a meteor impact crater you might find a few specks – the only naturally occurring SiC is in the form of Moissonite, debris from a supernova or ejecta from carbon-rich, red giant stars picked up in space and ending up as micron-sized particles in meteorites. Stardust indeed.
We might have never noticed SiC’s existence but in 1891 American inventor Edward G Acheson was trying to find a way to produce artificial diamonds, by heating clay (aluminium silicate) and carbon. He noticed shiny hexagonal crystals attached to the carbon arc light used for heating and called the compound carborundum, thinking it was a form of crystallized alumina like corundum. He might have thought he’d hit second best, as rubies and sapphires are types of corundum, but he realized he had something new, a compound nearly as hard as diamond which could be made as chips or powder on an industrial scale with application as an abrasive.
SiC LEDs came before transistors
Early in the 20th century, experimenters were finding that crystals of various substances such as germanium could give ‘unsymmetrical passage of current’ or rectification as we would know it, which found use in ‘crystal’ radios. When silicon carbide was tried, a strange phenomenon occurred; the crystal glowed yellow, sometimes green, orange or even blue. The first LED had been discovered, forty years before the transistor.
As an LED, SiC was soon superseded by gallium arsenide and gallium nitride with 10-100 times better emission but, as a material, SiC still generated interest in the electronics world; it has a thermal conductivity 3.5 times better than silicon and can be heavily doped for high conductivity while still maintaining high electric field breakdown. Mechanically, it is very hard, inert and has a very low coefficient of thermal expansion and high temperature rating. SiC does not even melt – it sublimates at about 2700⁰C.
SiC makes good
SiC was known as a good candidate for a semiconductor device very early on, so what held it back and let silicon become the standard? The main problem was elimination of defects in the SiC crystals, the list is long: edge dislocations, screw dislocations of different types, triangular defects and basal plane dislocations. The effect of the less-than-perfect crystal was very poor reverse blocking performance, making the parts essentially electrically unusable. There were also problems interfacing SiC with silicon dioxide (SiO2) to fabricate the popular MOSFET and IGBT device types. Continuous development, however, has improved quality such that 6-inch wafers can give an acceptable yield and a breakthrough called nitridation or annealing in nitrogen dioxide or nitrogen oxide enables SiO2 films to be grown onto SiC reliably.
From rocks in space to rocks on your finger
SiC has evolved from an abrasive on your grinding wheel through a glowing electrical curiosity to the semiconductor technology enabling longer-range electric vehicles and inverters in planet-saving solar power. Oh, and by the way, Acheson’s dream was as good as realised – SiC or Moissanite gems are barely distinguishable from pure diamonds.
By Anup Bhalla, VP Engineering at UnitedSiC (www.unitedsic.com)