Demonstration of Single Crystal InN & GaN/InN Heterostructures by MOCVD

Abstract

III nitride semiconductors have recently disrupted several industries and have remarkable potential to address unmet needs in electronics and optoelectronics. While gallium nitride (GaN) is the champion nitride material in solid state lighting and high power and high frequency electronics, there is abundant untapped potential in other nitrides that are otherwise harder to grow and to gain control over their electronic properties. Among these, Indium nitride (InN) that promises the highest saturation velocity in the nitride group poses two difficult challenges to realize its potential: (1) The epitaxial growth of high-quality InN films or structures is very challenging due to the competitive decomposition of InN at the temperatures for which efficient cracking of nitrogen sources occur, and (2) the difficulty in growing a larger bandgap barrier layer at the surface which requires even higher temperatures than the InN decomposition temperature, and with the absence of such layer, surface Fermi energy pinning challenges the modulation of the electron charge density to make functional transistor devices from any grown InN layers. Yes, there has been outstanding progress in the growth of InN layers by molecular beam epitaxy or plasma assisted metalorganic chemical vapor deposition (MOCVD), but the growth of smooth InN surfaces and the growth of larger barrier layers on InN has not been accomplished before: This work demonstrates by standard pulsed MOCVD growth the ability to grow defect free InN structures with smooth surfaces and the ability to grow with control GaN barrier layers and to demonstrate excellent transport characteristics in these structures. We hypothesized that coalescence of InN islands in constrained geometries such as lateral Fins can accelerate the formation of void-free and defect free InN growth within the first few nanometers of grown InN. We further hypothesized that we can raise the effective V/III ratio by pulsing the tri-methyl-indium source to avoid droplet formation, ripening and segregation which otherwise leads to the growth of defective and porous InN layers. We validated these hypotheses by studying growth parameters and Fin widths in the range of 60 nm to 2000 nm (and control planar samples) and demonstrated InN Fins with defect-free surfaces and single crystal structures. Unannealed Ti/Au metal contacts on packed InN Fin arrays resulted in effective (un-corrected) specific contact resistance of approximately 6Q¢um, which is much smaller than that obtained for current GaN ohmic contacts, with an extracted contact transfer length less than 1pm. To leverage these structures for high electron mobility applications and un-pin their surface Fermi energy, we successfully grew in-situ GaN/InN heterostructure without compromising the smooth InN surface. Transmission electron Microscopy showed clear interface between GaN and InN and single crystal structure. We then demonstrated that GaN can be selectively etched in the source/drain regions of the heterostructure and that unannealed Ti/Au metal contacts retain their low specific contact resistance on these structures. We are now investigating the fabrication of GaN/InN and AIN/INN HEMTs and their DC/RF performance.