Physical and Technological Limits of Silicon Oxide
The gate oxide is the first device dimension reaching the atomic scale (~10 Å). The atomic scale gate dielectric has become the most critical constraint for further downscaling of the MOS transistors. Thirty years ago, the gate oxide thickness was 120 nm and now it is 1.2 nm in production, which is much thinner than the direct tunneling limit for SiO2 (about 3 nm). Power consumption is a primary concern for high-performance logic integrated circuits with the tunneling gate oxide. For 1.2 nm thick silicon oxide, the gate leakage current density reaches 100 A/cm2 at 1 V (see Fig. 3). Such large currents would not be acceptable in most applications. In the 70 nm technology node, the required gate oxide thickness is about 0.7 nm, which is only two atomic layers of silicon oxide and is the ultimate limit of bulk silicon oxide.4 However the actual technologically feasible thickness is well above this physical limit because of the large leakage current as well as the nonsalable features of the film properties and fabrication technology. In the nanometer thick dielectric film, the interface layer is so thick (compared to the total oxide width) that any no uniformity in chemical composition and even the surface roughness can cause pronounced fluctuations of the device characteristics. If we allow the oxide thickness to fluctuate with a half monolayer, which means 5% of the oxide thickness, then the smallest oxide film thickness to meet this tolerance is 1.6 nm (Fig. 3), which is very close to the oxide thickness used in our current complementary (C)MOS technology. If one could relax the fluctuation to 8%, then the minimum thickness would be 1 nm and could be used in the 80 nm technology node in these years. Introducing physically thicker high-k materials, while maintaining the same value of the capacitance required for controlling the current flow in the channel, can resolve all the aforementioned problems.
Physical and Technological Limits of Silicon Oxide
The gate oxide is the first device dimension reaching the atomic scale (~10 Å). The atomic scale gate dielectric has become the most critical constraint for further downscaling of the MOS transistors. Thirty years ago, the gate oxide thickness was 120 nm and now it is 1.2 nm in production, which is much thinner than the direct tunneling limit for SiO2 (about 3 nm). Power consumption is a primary concern for high-performance logic integrated circuits with the tunneling gate oxide. For 1.2 nm thick silicon oxide, the gate leakage current density reaches 100 A/cm2 at 1 V (see Fig. 3). Such large currents would not be acceptable in most applications. In the 70 nm technology node, the required gate oxide thickness is about 0.7 nm, which is only two atomic layers of silicon oxide and is the ultimate limit of bulk silicon oxide.4 However the actual technologically feasible thickness is well above this physical limit because of the large leakage current as well as the nonsalable features of the film properties and fabrication technology. In the nanometer thick dielectric film, the interface layer is so thick (compared to the total oxide width) that any no uniformity in chemical composition and even the surface roughness can cause pronounced fluctuations of the device characteristics. If we allow the oxide thickness to fluctuate with a half monolayer, which means 5% of the oxide thickness, then the smallest oxide film thickness to meet this tolerance is 1.6 nm (Fig. 3), which is very close to the oxide thickness used in our current complementary (C)MOS technology. If one could relax the fluctuation to 8%, then the minimum thickness would be 1 nm and could be used in the 80 nm technology node in these years. Introducing physically thicker high-k materials, while maintaining the same value of the capacitance required for controlling the current flow in the channel, can resolve all the aforementioned problems.
การแปล กรุณารอสักครู่..