Wireless Circuits and Technology

Center Director

Keren Bergman, Columbia University

Theme Leader

Harish Krishnaswamy, Columbia University

Task Leader

Elaheh Ahmadi, University of Michigan

Researchers

Ahmadi, Mishra, Chowdhury, Rodwell

Primary Anticipated Result

N-polar GaN HEMT devices with high mobility, gm, and gain at low current densities around 10% of Imax. This will be an enabler for higher efficiency, >25% at 300 GHz, and 70% at 30 GHz, achieved through fundamental improvements in materials and interfaces, and suppression of 2D/1D transport.

Task Background

The mobility at low charge density in N-polar GaN HEMTs is reduced due to two factors: (i) relatively high density of threading dislocations in GaN HEMT structures grown on foreign substrates like sapphire, SiC, and Si and (ii) 2D to mixed 2D/1D transport because of the quantum confinement of electrons near pinch-off along the step edges of the miscut substrates commonly employed for N-polar epitaxy. At high charge densities, these dislocations are screened by the accumulation of electrons around the dislocation, however, at low charge densities (corresponding to deep class AB bias), the presence of insufficient electrons to screen these dislocations results in high dislocation scattering. This dislocation scattering leads to poor electron mobility. This forces the quiescent bias for max PAE of N-polar devices, when operating at a substantial fraction of fmax, to be approximately 30% of Imax rather than the 10% preferred for high efficiency operation reducing PAE. Both the problems can be simultaneously addressed using by growing on on-axis GaN substrates with lower TDD.

Research Description

Improved mobility as a function of 2DEG density on a sample grown on semi-insulating bulk GaN substrate with a TDD of ∼106 cm−2 versus another sample grown on a foreign substrate with TDD ∼ 108 cm−2 was first demonstrated in 2019 by Ahmadi and Mishra. This observed higher electron mobility near pinch-off is despite the poor quality of N-polar GaN substrates’ surface due to lack of optimized surface polishing technique at the time. The quality of Npolar GaN substrates have since significantly improved. At Michigan, Ahmadi’s group have focused on growth on Npolar GaN substrates by MBE. Record electron mobility of 2050 cm2 V-1s-1 was demonstrated in N-polar GaN.

In this program, N-polar GaN HEMT structures will be grown on bulk GaN substrates to mitigate mobility degradation at low charge bias conditions, due to miscut 1D quantum confinement and threading dislocation scattering. Highly scaled devices will be fabricated on this platform, then characterized and modeled.

  • Year 1: Develop devices with AlGaN backbarrier, grown by MBE and fabricate demonstration devices.
  • Year 2: Develop devices with InAlN backbarrier for higher 2DEG charge density and fabricate demonstration devices. Adapt large signal models from Task 1 to these devices.
  • Year 3: Further optimization of growth, supply epi wafers to UCSB/Mishra for scaled device fabrication, supply epi wafers to Stanford for diamond deposition (Front and back). Continued model optimization.
  • Year 4: Further optimization and supply of Gen 2 epi wafers to UCSB and Stanford, and modeling
  • Year 5: Final optimization and supply of Gen 3 epi wafers to UCSB and Stanford, and modeling.

Graduate Students

Oguz Odabassi ([email protected])

One more student will be hired for this task.

Liaisons

Dan Dennighoff ([email protected]), HRL