Access Official Technical PDF

Share Summary:LinkedInEmail

Accurate power semiconductor simulation depends on more than matching a device’s typical current-voltage curves. Engineers also need models that represent nonlinear capacitance, temperature effects, switching behavior, and the influence of physical design choices. Scalable SPICE modeling addresses these needs by linking circuit-level models to device process and layout parameters.

Instead of waiting until a device has been fabricated and characterized to extract a model, a physically based approach can support evaluation earlier in technology development. It connects device physics, layout, process technology, and circuit simulation, giving designers a way to examine how changes in these areas may affect application behavior.

Why physical, scalable SPICE models matter

Power devices—including SiC MOSFETs, trench IGBTs, Super-Junction MOSFETs, trench MOSFETs, GaN HEMTs, and SiC diodes—have different structures and operating characteristics. A simple subcircuit may not capture all the relevant current-voltage, capacitance-voltage, transient, and thermal behavior. Behavioral models can represent observed responses, but may lack a direct connection to the physical layout and process parameters that shape those responses.

A model that responds to process and layout changes can make simulation part of device development rather than only a final design-checking step. Engineers can explore technology options and circuit sensitivities in simulation, potentially reducing reliance on repeated fabrication iterations during early assessment.

Scalability is central to this approach. When layout dimensions or process parameters change, the corresponding model quantities can be recalculated rather than refitted from scratch. This makes it possible to investigate how a design variant affects device behavior while keeping the simulation linked to its physical basis.

How the SiC MOSFET model represents device behavior

A SiC MOSFET model combines several SPICE elements to represent the main regions and effects of the device. The channel uses the BSIM3v3 MOSFET model, which can describe operation from subthreshold through weak and strong inversion. Temperature behavior can also be tuned to reflect the device’s characteristics.

The region between the p-wells, often described as the JFET region, is represented with a SPICE JFET model. Its gain and pinch-off parameters are derived from physical quantities such as doping, p-well spacing, junction depth, and mobility. This matters because the drift region’s behavior is nonlinear; representing it only with a fixed linear resistor would not capture the same effects.

Other parts of the model address effects that influence circuit simulation:

  • Body diode: A physical diode model accounts for reverse recovery and can be scaled for the device layout.
  • Nonlinear capacitance: The gate-to-drain capacitance depends on oxide thickness, doping, geometry, and depletion conditions. Its behavior changes as the JFET region approaches and passes pinch-off.
  • Layout parasitics: Active area, gate runners, gate pads, cell pitch, and other layout features inform component multiplicities and parasitic elements.
  • Electrothermal behavior: Device power drives a thermal impedance network, allowing junction temperature and self-heating to be solved during simulation.

Together, these elements allow the model to represent more than static conduction. Capacitance and temperature behavior are also important inputs to switching and application-level analysis.

Modeling trench IGBTs and process trade-offs

The same physical and scalable principles can be applied to trench IGBTs, while accounting for their different structure and carrier transport. The model described for this device family uses a BSIM3v3-based channel representation, physical scaling equations, nonlinear dynamic capacitances, and electrothermal effects. Its trench-related capacitance model includes the effects of depletion and pinching between trenches.

Process parameterization can also support design trade-off studies. For example, the IGBT model allows anode doping to be varied to examine its relationship to turn-off energy (EOFF) and saturation voltage (VCEsat). A designer can run parameter sweeps to assess application sensitivity to a process choice, rather than treating each process derivative as an unrelated device model.

Verification across operating conditions

Model validation needs to cover the conditions that matter in power circuits. The described verification process compares simulation with measured current-voltage behavior across temperature, including transfer characteristics, output characteristics, threshold voltage, and on-resistance. Pulsed simulations are used to reproduce test conditions and account for junction temperature during measurement.

Capacitance-voltage and gate-charge comparisons test whether the model captures nonlinear behavior across bias. Double-pulse switching simulations then assess turn-on and turn-off waveforms, including the influence of Miller effects and reverse recovery. These checks connect device-level model behavior to the switching circuits engineers need to evaluate.

Scaling is tested by applying changed layout parameters to a base model without further tuning. The reported examples include changes in cell pitch and gate-poly length for a SiC MOSFET, as well as scaling between trench IGBT device sizes. Such checks assess whether the model’s physical relationships remain useful when the device geometry changes.

Portability and practical use

Power electronics teams use a range of SPICE-compatible simulators. A simulator-agnostic implementation therefore favors broadly available SPICE elements and syntax. Physical equations for currents, charges, and capacitances can be implemented with controlled sources when suitable primitives are unavailable. Avoiding reliance on a single simulator-specific language can help maintain consistent model use across different platforms.

For engineering teams, the practical value is a more connected workflow: device physics informs the SPICE model, layout and process inputs drive its scaling, and circuit simulations help assess application implications. The model still needs to be validated against the relevant device and application conditions; simulation does not remove that responsibility.

Conclusion

Physical, scalable SPICE modeling gives power semiconductor design teams a way to examine device behavior and design variation within circuit simulation. By representing nonlinear regions, capacitances, thermal effects, and layout-dependent parasitics, these models can support earlier technology evaluation and more meaningful application analysis. For engineers and decision-makers, the key implication is that SPICE can serve not only as a final verification tool, but also as a link between device development and power circuit design.

Access Official Technical PDF

About The Author