As transistor scaling makes leakage power increasingly important, engineers need device structures and models that describe switching behavior beyond the limits of conventional MOSFETs. A negative capacitance FET model can help evaluate how a ferroelectric gate stack interacts with a semiconductor channel. For back-gated transistors built with two-dimensional (2D) materials such as MoS2, that analysis must also account for parasitic capacitance and the device’s depletion-to-accumulation operation.
An analytical model developed for long-channel, back-gated 2D negative-capacitance FETs provides a continuous current-voltage description across these operating regimes. Its value is not limited to predicting on-state current: the framework also helps explain unusual behaviors, including negative drain-induced barrier lowering (DIBL) and negative differential resistance (NDR), that can affect device and circuit design.
Why model back-gated 2D negative-capacitance FETs?
In a conventional MOSFET, the subthreshold swing is limited by the thermal behavior of carriers at room temperature. Negative-capacitance devices seek to improve gate control by incorporating a ferroelectric material into the gate stack. The ferroelectric response can amplify the potential coupled to the channel, creating the possibility of a subthreshold swing below 60 mV/dec at room temperature.
Two-dimensional transition metal dichalcogenides are of interest as thin channel materials. Their small body thickness can support electrostatic control, while MoS2 negative-capacitance FETs have demonstrated steep switching over a limited gate-voltage range. Interpreting and designing these devices, however, requires models that reflect their actual structure and operating mode.
The devices considered in this work are junctionless transistors: source, drain and channel have the same dopant type and concentration. They operate through depletion and accumulation, rather than the depletion-inversion behavior assumed in some conventional transistor models. The distinction matters because a model built around the wrong operating regime may not represent measured device characteristics reliably.
What the negative capacitance FET model includes
The modeled structure uses a metal-ferroelectric-metal-insulator-semiconductor (MFMIS) gate stack. It treats the device as a baseline 2D transistor in series with a ferroelectric capacitor. Channel electrostatics are derived from Poisson’s equation, while drift-diffusion transport provides the basis for calculating drain current. The resulting analytical formulation uses the Lambert W function to express the channel potential and supports current calculations across the device’s operating range.
A key contribution is the inclusion of parasitic capacitance, represented by overlap and fringe capacitances between the interfacial metal gate and source/drain regions. In a back-gated device, these parasitics can be significant. They combine with the channel capacitance and influence the total capacitance seen by the ferroelectric layer. Since capacitance matching affects the gate-stack response, omitting these terms can lead to an incomplete picture of device behavior.
The model also calculates channel charge and incorporates it into the ferroelectric voltage relationship using the Landau-Khalatnikov description. This connects the electrical behavior of the channel to the polarization response of the ferroelectric. Crucially, the current-voltage relationships remain continuous from depletion to accumulation, making the model better suited to time-efficient simulation than an implicit formulation that requires solving for surface potential during each calculation.
What the model and measurements reveal
The model was compared with measurements from back-gated MoS2 devices using Hf0.5Zr0.5O2 (HZO) as the ferroelectric material. For one reported device, measured minimum subthreshold swings were 52.3 mV/dec on the reverse sweep and 57.6 mV/dec on the forward sweep. The simulated transfer characteristics showed good agreement with experimental results, supporting the model’s ability to describe the measured behavior.
The comparison also highlighted two effects that may appear counterintuitive to engineers accustomed to conventional MOSFET operation:
- Negative DIBL: The threshold voltage increases as drain voltage rises, rather than decreasing as it commonly does in conventional MOSFETs.
- Negative differential resistance: In subthreshold operation, drain current can decrease as drain voltage increases.
In the model, both effects are linked to the interaction between the ferroelectric gate stack and parasitic charge. As drain voltage increases, the charge associated with the drain-side parasitic capacitance decreases. This changes the average gate charge density and, in the depleted region, the interfacial metal-gate potential. For a fixed gate voltage, the potential can fall as drain voltage rises, elevating the channel’s conduction-band energy in subthreshold operation. The resulting higher electron barrier can reduce channel electron density and drain current.
For a long-channel baseline transistor in subthreshold, drain voltage has little effect on current compared with the change in interfacial potential. Consequently, that potential shift can dominate the negative DIBL and NDR response of the complete negative-capacitance device.
Using the model in device and circuit design
For engineers evaluating emerging device options, the practical lesson is that the ferroelectric layer cannot be assessed in isolation. Back-gate geometry and parasitic capacitance influence capacitance matching, while channel electrostatics and charge determine how the gate stack responds across operating regions. A model that accounts for these interactions can support device optimization and help explore implications for integrated circuits.
The reported formulation has important boundaries. It targets long-channel devices and simplifies some effects, including polarization damping, series resistance and field-dependent mobility. The authors identify these omissions as possible contributors to differences between simulated and measured output characteristics. The devices’ design parameters were also selected to balance steep subthreshold swing with negligible steady-state hysteresis; the results should not be taken as a guarantee for other structures or operating conditions.
Conclusion
A continuous analytical negative capacitance FET model offers a practical way to connect 2D channel transport, ferroelectric response and back-gate parasitics. Its comparison with MoS2 measurements supports its use for studying subthreshold behavior, negative DIBL and NDR. For device designers and circuit teams, the central implication is clear: accurate analysis of 2D negative-capacitance transistors depends on modeling the full gate-stack interaction, including parasitic capacitance, rather than treating the ferroelectric and channel as independent elements.