Buzz about the promise of wide-bandgap semiconductors abounds, but engineers are only just starting to look to them as real solutions, not mere hype. There are still birthing pains, but mainstream acceptance of the technology appears imminent.
Still, it is relevant to ask why wide-bandgap is such a disruptive technology. Isn’t a better MOSFET just an evolutionary step in embedded electronics? Why should you care about this?
First things first: There are two current players in the wide-bandgap space — Silicon Carbide (SiC) and Gallium Nitride (GaN). They differ fundamentally, but both outperform Silicon, operating at higher voltages, frequencies, and temperatures. The band gap is the energy required to dislodge a valence electron bound to an atom so that it becomes a conduction electron that can move within the crystal lattice and serve as a charge carrier. “Wide bandgap” is a reference to the higher energy levels required to initiate semiconducting actions in SiC and GaN. Silicon and other common non-wide-bandgap materials have a band gap on the order of 1 to 1.5 electronvolts (eV). Wide-bandgap materials. in contrast, typically have band gaps on the order of 2 to 4 eV.
In the words of Thomas Naher of On Semiconductor at the recent PCIM show, “it’s playtime!” When pressed to elaborate, he said, “Wide-bandgap is an enabling technology. It isn’t a one-to-one replacement for Silicon, you have to drive it differently and address the thermal issues and EMI, but once you do, you are out there and will be beating your competitors.”
The development issues of SiC and GaN are directly related to their properties. Many legacy packages for Silicon like the TO-220 and TO-240 have serious parasitic issues, which significantly reduce the benefits of wide-bandgap devices in a system. A lot of first-generation solutions went this route, which weakened the value-add proposition and slowed acceptance. Second-generation packages have been developed that look and install like TO packages, but without the parasitic issues, helping adoption.
Packaging is probably the most critical of the development issues facing GaN and SiC, because not only are parasitics a performance issue, proper packaging will also maximize the material attributes of wide-bandgap devices. For example, GaN uses piezoelectric stress between its doped layers to create a near-superconducting environment within the device. This means that vias can only come in through the side of the package, not the top or bottom. SiC doesn’t have this issue, but due to its high voltage and thermal capacity could use better packaging to exploit its benefits.
The recent advances in GaN packaging are examples of solutions now available to the design industry, bringing GaN development into the mainstream. GaN Systems’ newest GaNPX packaging for extreme speed and current leverages the technology in a near-chipscale embedded package, able to handle high current density with optimal thermal performance and extremely low inductance, with no wirebonds. Transphorm uses a Cascode topology in their devices to ease adoption by engineers who are not familiar with GaN.
The advent of these new packages and a greater educational effort by the wide-bandgap industry to educate designers on the issues and advantages of these new materials is making wide-bandgap less of a disruption and more of an evolution, but there is still more to be done; the technology is still in its adolescence. However, the migration to these materials is not only inevitable, it is desirable.