Characterisation and Modelling of Graphene FETs for Terahertz Mixers and Detectors

University dissertation from Chalmers University of Technology

Abstract: Graphene is a two-dimensional sheet of carbon atoms with numerous envisaged applications owing to its exciting properties. In particular, ultrahigh-speed graphene field effect transistors (GFETs) are possible due to the unprecedented carrier velocities in ideal graphene. Thus, GFETs may potentially advance the current upper operation frequency limit of RF electronics. In this thesis, the practical viability of high-frequency GFETs based on large-area graphene from chemical vapour deposition (CVD) is investigated. Device-level GFET model parameters are extracted to identify performance bottlenecks. Passive mixer and power detector terahertz circuits operating above the present active GFET transit time limit are demonstrated. The first device-level microwave noise characterisation of a CVD GFET is presented. This allows for the de-embedding of the noise parameters and construction of noise models for the intrinsic device. The correlation of the gate and drain noise in the PRC model is comparable to that of Si MOSFETs. This indicates higher long-term GFET noise relative to HEMTs. An analytical power detector model derived using Volterra analysis on the FET large-signal model is verified at frequencies up to 67 GHz. The drain current derivatives, intrinsic capacitors and parasitic resistors of the closed-form expressions for the noise equivalent power (NEP) are extracted from DC and S-parameter measurements. The model shows that a short gate length and a bandgap in the channel are required for optimal FET sensitivity. A power detector integrated with a split bow-tie antenna on a Si substrate demonstrates an optical NEP of 500 pW/Hz^1/2 at 600 GHz. This represents a state-of-the-art result for quasi-optically coupled, rectifying direct detectors based on GFETs operating at room temperature. The subharmonic GFET mixer utilising the electron-hole symmetry in graphene is scaled to operate with a centre frequency of 200 GHz, the highest frequency reported so far for graphene integrated circuits. The down-converter circuit is implemented in a coplanar waveguide (CPW) on Si and exhibits a conversion loss (CL) of 29 ± 2 dB in the 185-210 GHz band. In conclusion, the CVD GFETs in this thesis are unlikely to reach the performance required for high-end RF applications. Instead, they currently appear more likely to compete in niche applications such as flexible electronics.