Future Power Electronics with GaN and Diamond

Current trends in power conversion demands systems with more efficiency, higher temperatures of operation and higher power density. Switches, an integral part of every power converter, directly impact the efficacy of the converter, and indirectly its cost.   Widebandgap semiconductors like Silicon Carbide (SiC) and Gallium Nitride (GaN) have shown promising performance owing to their superior material quality (compared to Silicon (Si)) that make them the leading materials for next generation power electronics. While SiC and GaN are making impressive progress wider bandgap semiconductors like Aluminum Nitride and Diamond are attracting research interest. This presentation will focus on a developing technology, the vertical GaN technology and its current challenges, and elucidate some recent developments on an emergent one - diamond.

Lateral GaN based HEMTs [1] have been designed into power converters successfully that allowed us to study the outcomes at all levels, all the way up to the system. The two important factors that benefited power converters due to the inclusion of a GaN-switch are 1) higher frequency of operation 2) higher temperature of operation, compared to what Si can offer as it reaches its material’s limit. High frequency operation leads to the reduction in electrical form-factor due to reduced size of passive components (capacitors and inductors). Higher temperature of operation allows reduction of cooling accessories hence reducing the mechanical form factor.

Higher power (>20KW, although the limits are yet to be determined) applications require vertical device design to maximize the power density. Vertical topology allows higher On-state currents and off-state voltages over a smaller chip area, where the lateral design become unattractive in both cost and performance due to large chip area. Our recent understanding of vertical GaN devices along with the cost of producing bulk GaN substrates portrays a sustainable solution exists for the high power market. Early results from CAVETs [1] have established the fact that bulk GaN devices offer breakdown electric field almost 3 times higher compared to lateral HEMTs. CAVETs have also demonstrated dispersion less output current without the need of complex field plate structures (an integral part of the lateral HEMT design). Our recent research has identified vertical device designs that warranties single chip normally off devices, which will further improve switching performance by eliminating bond wire and PCB trace related inductances. GaN substrates with defect densities lower that 10⁴cm^-2 and allowing electron mobility >1100cm²V^-1s^-1 [2] and recent reports on successful activation of [Mg]-implantated GaN [3] create a very encouraging picture for bulk GaN based power devices.

Energy bandgap above 3.4eV is very attractive to continue the roadmap of power electronics towards higher power application (3kV and up), with an increasing power density. With a bandgap of 5.5eV, a critical electric field over 10 MV/cm, and thermal conductivity superior to Si, SiC and GaN, diamond offers the most appropriate combination to serve very high voltage power electronics. One of our recent research achievement demonstrated excellent contacts to n-type diamond, which has been challenging due to very low work function of diamond. Phosphorous doped diamond achieved by Koeck et al. [4] on homoepitaxial diamond substrates led to excellent ohmic contacts to n-type diamond. Applying the novel contact technology on a homoepitaxialy grown p (bottom)-i-n(top) structure, diode operation was achieved with reverse blocking over 150V held over a 1μm thick  i-layer. The forward current density was >500A/cm² around 4V.

These early results obtained in diamond clearly show an extremely promising path beyond GaN and SiC.

Reference:

[1] S.Chowdhury and U.K Mishra, IEEE Transaction on Electron Devices, 60 3060(2013)

[2] P Kruszewski et al. The International Workshop on Nitride Semiconductor (2014)

[3] T. J. Anderson et al, Electronics Letts. , 50, 197(2014)

[4] F.A.M. Koeck et al, Diamond Relat. Mater. 18, 789(2009)