Efficient ab initio analysis of quantum confinement and band structure effects in ultra-scaled Si1−xGex gate-all-around and fin field-effect transistors for sub-10 nm technology nodes

This paper analyzes the sub-10 nm ultra-scaled Si 1− x Ge x fin field-effect transistor (FinFET) and gate-all-around transistor (GAAFET), based on an improved method by combining Slater partial occupation and ab initio density functional theory (DFT). Compared to the existing generalized gradient approximation (GGA) DFT, which is computationally efficient but inaccurate, and the established hybrid functional DFT, which is accurate but computationally demanding, the proposed method achieves both hybrid functional-like high accuracy and GGA-like high computational efficiency concurrently. By using this improved methodology, batch simulations are performed to investigate the size- and composition-dependent quantum confinement and band structure effects in the Si 1− x Ge x FinFET and GAAFET. It is shown that the quantum confinement in ultra-scaled FinFET and GAAFET significantly increases the bandgap of Si 1− x Ge x channel in the sub-10 nm scale. It is quantitatively demonstrated that, due to its two-dimensional quantum confinement, the ultra-scaled GAAFET has larger confinement-induced bandgap increase, lower direct source-drain quantum tunneling, lower off-state transmission and low-bias conductance, and smaller drain-induced barrier lowering, compared to the FinFET which is confined only in one-dimension. These differences jointly indicate that the ultra-scaled GAAFET could offer better device performance than the ultra-scaled FinFET. It is quantitatively shown that, by reducing the Ge composition of the Si 1− x Ge x channel in ultra-scaled FinFET and GAAFET, the bandgap and the threshold gate voltage can be significantly increased; and the transmission coefficients, the low-bias conductance, and the source-drain current can be significantly reduced. These trends can be utilized to optimize device performance by tuning Ge composition in different technology nodes.