▎ 摘　　要

Tight-binding (TB) Hamiltonian is widely used in the simulations of many low-dimensional devices, such as graphene nanoribbon (GNR)-based FETs, and usually, it is an empirical parameterized matrix. In this paper, we propose a novel strategy to construct the edge TB parameters of armchair GNR (a-GNR) under different edge saturations. Different from the traditional way's only edge bonds analysis, this method also considers the diversity of three a-GNR families and the saturation atoms impact on the subedge bonds. The effects of several common elements or groups on the edge of GNRs are studied numerically and among them, hydrogen (H), fluorine (F), and hydroxyl group (OH) show better saturation properties. Through elaborate verifications of these three types of saturations, the TB fitting errors of the proposed new strategy are drastically reduced, in comparison with the traditional parameters. These verifications, which are subjected to the ab initio results, go from an energy-band level to a device-performance level. Simulation results also show that saturated by different atoms, the band structure of GNR obviously varies. Thus, edge saturation can be another effective method of tuning GNR device properties, such as bandgap, current on/off ratio, and carrier mobility.