The interplay of blocking properties with charge and potential redistribution in thin carbon-doped GaN on n-doped GaN layers

Abstract

In carbon-doped GaN (GaN:C) buffers used in a GaN-on-Si technology, the buffer is embedded in between transition and channel layers. This makes the analysis of buffer properties difficult due to e.g., carrier injection from or potential drop at these adjacent layers. Here, we analyze capacitance- and current-voltage characteristics of 200–300 nm thick GaN:C ([C] = 1019 cm−3) layers which are embedded between a top metal electrode and bottom n-doped GaN (n-GaN). Such structures allow a better potential control in GaN:C and thus determination of the band diagram quantitatively. The accumulation of negative charge (concentration up to 6 × 1017 cm−3) with bias is observed in GaN:C at both polarities. For biases Vappl < +1.7 V at the top electrode, negative charges accumulate in GaN:C near to its interface with n-GaN so that GaN:C exhibits no potential drop and blocks leakage current. For Vappl > +1.7 V, accumulated negative charges in GaN:C raise an energy barrier of ∼1.1 eV for electron injection from n-GaN to GaN:C. This causes a potential drop in GaN:C leading to a significant leakage current increase. The Fermi level pinning in GaN:C at a commonly referred acceptor at EV + 0.7(±0.2) eV is extracted only from electrostatic considerations. The occupancy change of carbon acceptors is attributed to trapping processes where the dislocation-related conductive paths are supposed to be involved in carrier transport from the top metal electrode to the carbon defect.