RESUMO
We investigated the effect of a sacrificial AlN layer on the deep energy level states of 4H-SiC surface. The samples with and without AlN layer have been annealed at 1300 °C for 30 minutes duration using a tube furnace. After annealing the samples, the changes of the carbon vacancy (VC) related Z1/2 defect characteristics were analyzed by deep level transient spectroscopy. The trap energy associated with double negative acceptor (VC(2-/0)) appears at Ë0.7 eV and was reduced from Ë0.687 to Ë0.582 eV in the sacrificial AlN layer samples. In addition, the capture cross section was significantly improved from Ë2.1×10-14 to Ë3.8×10-16 cm-2 and the trap concentration was reduced by approximately 40 times.
RESUMO
4H-SiC nano-channel field effect transistors (FETs) with various widths of 3 µm-50 nm have been fabricated by "top-down" approach using electron-beam lithography process. It has been demonstrated that the gate controllability of the SiC FETs is improved with decreasing channel width. In the fabricated devices the threshold voltage V(th) for the 50 nm-width nano-channel FETs shows a positive shift (ΔV(th) = 1.4 V) with respect to that of the reference FETs. The on-current degradation of the SiC nano-channel FETs is found to be 1.5 times lower than that of the reference FETs at elevated temperatures up to 450 K. This attributed to the improved heat dissipation of the nano-channel structure with a large surface to volume ratio.
RESUMO
In this study, nano-scale honeycomb-shaped structures with anti-reflection properties were successfully formed on SiC. The surface of 4H-SiC wafer after a conventional photolithography process was etched by inductively coupled plasma. We demonstrate that the reflection characteristic of the fabricated photodiodes has significantly reduced by 55% compared with the reference devices. As a result, the optical response Iillumination/Idark of the 4H-SiC photodiodes were enhanced up to 178%, which can be ascribed primarily to the improved light trapping in the proposed nano-scale texturing.