Assessment of High-Frequency Performance Limit of Black Phosphorus Field-Effect Transistors

Abstract

Recently, gigahertz frequencies have been reported with black phosphorus (BP) field-effect transistors (FETs), yet the high-frequency performance limit has remained unexplored. Here we project the frequency limit of BP FETs based on rigorous atomistic quantum transport simulations and the small-signal circuit model. Our selfconsistent nonequilibrium Green's function (NEGF) simulation results show that semiconducting BP FETs exhibit clear saturation behaviors with the drain voltage, unlike zerobandgap graphene devices, leading to >10 THz frequencies for both intrinsic cutoff frequency (f <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</sub> ) and unity power gain frequency (f <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">max</sub> ). To develop keen insight into practical devices, we discuss the optimization of f <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</sub> and f <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">max</sub> by varying various device parameters such as channel length (L <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ch</sub> ), oxide thickness, device width, gate resistance, contact resistance, and parasitic capacitance. Although extrinsic f <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</sub> and f <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">max</sub> can be significantly affected by the contact resistance and parasitic capacitance, they can remain near THz frequency range (f <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</sub> = 900 GHz; f <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">max</sub> = 1.2 THz) through proper engineering, particularly with an aggressive channel length scaling (L <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ch</sub> ≈ 10 nm). Our benchmark against the experimental data indicates that there still exists large room for optimization in fabrication, suggesting further advancement of high-frequency performance of stateof-the-art BP FETs for the future analog and radio-frequency applications.