The ab initio microscope: on the performance of 2D materials as future field-effect transistors

Abstract

In this work, the potential of novel 2D materials for possible application as next generation ultra-scaled field-effect transistor (FET) is evaluated from an atomistic perspective. For this purpose, a first-principles simulation scheme based on density functional theory (DFT) and the non-equilibrium Green’s function (NEGF) formalism is employed. This approach can shed light on the device behavior at the nano-scale, where classical drift-diffusion models reach their limit. A DFT+NEGF simulator allows to investigate the observables of interest such as the charge density and the electric current in different device structures, for example in metal-oxide-semiconductor (MOS) FETs (MOSFETs) or band-to-band tunneling FETs (TFET) controlled by a single or multiple gates.First one-hundred potential 2D contenders for logic applications are examined in a single-gate (SG) MOSFET architecture for both n- and p-type configuration. The full I-V characteristics are simulated at a gate length of 15 nm to determine the potential ON-current at a fixed OFF-current.From this data, we identify 13 compounds that achieve electron and hole currents potentially outperforming those of future silicon FinFETs. The sub-threshold slope (SS) is studied down to a gate length of 5 nm to identify the scalability of all studied 2D materials. To analyze the obtained results, the concepts of transport and density-of-states (DOS) effective mas is generalized and systematically extracted for each 2D material.While these quantities partly explain the device behavior, they are not sufficient as the effect of narrow bands can strongly compromise the FET performance. A novel metric called pass factor is therefore introduced to quantify this phenomenon. Overall it is found that materials with a low transport effective mass, high DOS and a pass factor close to one yield excellent performance. Such materials are often characterized by a strongly asymmetric bandstructure. Black phosphorus (BP), the most promising candidate among all considered belongs to this category. It is used in a second study to explore the influence of a flake misalignment with respect to the source-to-drain direction on the ON-state current. The impact of misalignment is demonstrated using six different transport directions in a single gate MOSFET. Up to a misalignment angle of 20 degrees, the ON-state current remains almost constant. The current reduction does not exceed 30%for angles below 50 degrees before rapidly decreasing to around 60%of its maximum value in the worst-case scenario (90 degrees misalignment). This phenomenon can be explained by inspecting the dependence of the effective mass in transport direction on the misalignment angle. The ON-state current behavior between quasi-ballistic simulations and calculations where phonon- and charged-impurity scattering are present remains qualitatively equal. Consequently, the change in the transport effective mass can explain the observations and a misalignment tolerance of 20 degrees in experiments should be acceptable.In a third study, the potential of 2D materials as TFETs is evaluated. It is demonstrated that the well-known transition metal dichalcogenide(TMD) are not well-suited for TFET applications due to their large bandgap and effective masses a tunneling window is already open in the OFF-state and consequently the desired sub-thermionic SS cannot be achieved. Potential novel single-layer materials with a more favorable effective mass and band gap combination are shown to reach ON-state currents roughly two orders of magnitude higher than all of the TMDs, while also reaching sub-thermionic SS. In a final study, we explore the application of the DFT+NEGF approach to optoelectronic devices, photovoltaic cells in the present case. A dedicated self-energy is implemented for that purpose and the necessary inputs are derived from the ab inito level. A MoS2PIN-junction is then studied as a proof-of-concept structure to verify the implementation of our model. The approach produces physically meaningful results, for example, the Franz-Keldysh effect is captured by our method.