Distributed Charge-Trap Dynamics in Graphene Transistors Revealed by RF Impedance Dispersion

Abstract Charge trapping at the graphene/dielectric interface is a major source of instability in graphene transistors, yet its signature in the radio-frequency electrical response remains poorly understood and characterized. This work reports that trap-related charge exchange along the graphene channel produces a distinct, frequency-dependent dispersion in the transistor output impedance, observable from standard microwave measurements without dedicated stress or pulsed-bias protocols. This dispersion primarily affects the reactive part of the output impedance, particularly at low frequencies, and gradually diminishes toward the GHz range, consistent with a distribution of trap relaxation times along the channel. To capture this behavior, a non-uniform transmission-line model is developed that treats channel trapping as a spatially distributed process. An equivalent lumped circuit, consisting of a single RC branch in parallel with the channel resistance, reproduces the distributed response accurately, and explicit analytic relationships between both representations enable extraction of local distributed-model parameters from the lumped circuit. The equivalent capacitance converges to a stable value as the number of distributed trapping sections increases, whereas the effective time constant remains dependent on trap position, preserving information about the distributed nature of the phenomenon. Using the converged values, the extracted interfacial trap density is consistent with reported values for graphene/AlOx interfaces, and the extracted trap dynamics depend systematically on channel aspect ratio, identifying device geometry as a means of tuning the frequency range over which trap-induced dispersion is observable. Excellent agreement between model and experimental data is demonstrated across the full characterization frequency range, and the resulting trap-aware description further improves prediction of the device input reflection coefficient. This work establishes high-frequency electrical characterization, interpreted through a distributed-to-lumped framework, as a non-destructive, quantitative route to probe the density, spatial distribution, and dynamics of traps in graphene channels, with direct relevance to interface engineering and defect characterization of graphene devices.