Full text loading...
(Color online) C-V curves of the studied devices: (a) reference device without Gd2O3 capping layer (poly-Si/TaN/HfSiO/SiO2) [VFB = −0.6446 V], (b) device with poly-Si overlayer (poly-Si/TaN/Gd2O3/HfSiO/SiO2) [VFB = −0.9952 V], and (c) device without poly-Si overlayer (TaN/Gd2O3/HfSiO/SiO2) [VFB = −0.5270 V]. A negative shift [ΔVFB = −350.6 mV] in the flatband voltage is observed for the poly-Si capped device, while a positive shift [ΔVFB = 117.6 mV] is observed for the uncapped one. Extracted EWF values from plots of VFB vs. EOT, for EOTs from 2 to 4 nm and capacitor area of 2 × 10−5 cm2 are (a) 4.45, (b) 4.13, and (c) 4.54 eV, support the observed trend.
(Color online) XPS depth profiles of the device (a) with and (b) without poly-Si overlayer. Oxygen diffusion into the TaN metal gate can be observed, while the dielectric layer seems to behave the same.
(Color online) Comparison of (a) Gd, (b) O, (c) Ta, and (d) N depth profiles for the capped and uncapped devices. Oxygen downward diffusion oxidizes the TaN metal gate in the upper part but does not reach the TaN/high-K interface, where more nitrogen has accumulated (d). Gd exhibits the same profile regardless of the presence of the poly-Si overlayer layer (a).
(Color online) Change of the average electrostatic potential between the ideal and nitrogen-substituted TaN/HfO2 interfaces. It is clear that such substitution of nitrogen on O sites in the HfO2 lattice can produce a positive shift in the electrostatic potential, consistent with our experimental observation for TaN devices without Si overlayer.
Article metrics loading...