The power electronics (PE) market is growing rapidly, driven by the accelerating demand of EV and HEV vehicles. Power devices lend themselves to design and manufacturing innovations at the transistor-level to improve device performance and reduce development and production costs. Silicon-carbide (SiC), gallium-nitride (GaN), and other wide bandgap materials have started to replace silicon in high-voltage power devices. Anyone designing or manufacturing silicon, SiC, or GaN technologies for the power device market should use TCAD simulations as part of their R&D efforts to understand their devices in greater detail and improve their key figures of merit. Silvaco TCAD is a market leader in simulation for power devices. Its products have been used by foundries and fabless semiconductor companies worldwide for over 25 years.

TCAD modeling of power technologies allows device engineers to make virtual changes in operating conditions, the device technology (for example, planar versus trench-based structure), or in semiconductor technology (such as doping and layer thicknesses). This exploration builds qualitative and then quantitative understanding of devices. Iterative simulation then allows engineers to optimize device performance and operating area.

Figure 2
Figure 3
Figure 1. Planar vs. trench cell in an IGBT device
Figure 2. Process simulation of key steps to
fabricate a split-gate UMOSFET

With optimized devices, engineers can produce better semiconductor products and reduce the time needed to reach volume production by decreasing the number of prototype wafers that need to be manufactured and characterized. Another benefit for fabless product engineers is the ability to suggest improvements in the manufacturing process to their chosen foundry to improve yield and performance.

Typical device types for each kind of material are:

  • Silicon: LDMOS, IGBT, Bipolar, DMOSFET, Schottky Diode, Thyristor
  • SiC: Diode, MOSFET, DMOSFET, IGBT
  • GaN: Diode, Lateral HEMT, Vertical HEMT

Silvaco TCAD process and device modeling solutions allows users to virtually prototype realistic devices in 2D or 3D, and to explore device performance in DC, AC, and transient simulations. Additionally, the capability to pair physics-based TCAD devices with a surrounding SPICE circuit allows users to explore the performance of their devices in realistic circuits.


2D/3D Device Simulation

The drift-diffusion engine of TCAD device simulation allows users to explore device performance with physics accuracy. The designer can consider temperature effects and perform a mixed device-SPICE simulation using specific modules:

  • The Giga module allows for device self-heating to be simulated. It self-consistently solves the temperature locally, considering the high current flows within the device.
  • The Mixed Mode module allows users to take a physics-accurate TCAD device and use it in a SPICE simulation to see how the device performs in a larger circuit. Users can include parasitic, passive, and active circuit elements to simulate and optimize device performance.
plot 6
plot 4
plot5
Figure 3. a) Mixed-mode 3D simulation of a short circuit test of an IGBT, b) Short-circuit waveforms, c) Electron current density distribution at device failure point t = 5.57 μs.

The performance of power devices can be greatly influenced by the edges or corners of their physical layout. For example, in most power devices the first location to experience high voltage breakdown is at the corner of the design. The ability to switch easily between 2D (for faster-running simulations) and 3D (for higher accuracy) is a major benefit to tackle the hardest TCAD problems while taking a practical engineering approach.

2D/3D Process Simulation

Process simulation is most generally used to construct 2D and 3D TCAD structures based on GDSII layout formats, per user specifications of layer thicknesses, composition fractions, doping, etc.

Realistic process models are physics-based, especially those with an accurate doping representation. Depending on the materials used, the process models are at different level of maturity:

  • Silicon: Process simulation using implant, diffusion, oxidation, etc. is very mature because of decades of CMOS R&D efforts by the semiconductor industry.
  • SiC: Through investments in partnerships and research sponsorships, Silvaco supports simulation of Monte Carlo implant, dopant activation/diffusion, orientation-dependent oxidation, and epitaxy. The Silvaco solution is mature and collaborations with customers and research institutions continue to enhance this analysis.
    plot 4
    plot5
    plot 6
    Figure 4 Process simulation for physical oxidation of 4HSiC technology during diffusion in dry O2 or H2O ambient conditions.
  • GaN: Process simulation in GaN is rapidly evolving. Process simulation of GaN does not require oxidation which is usually one of the most complicated process steps to simulate. However, epitaxy simulation is a key process step for GaN and will require a significant R&D effort to achieve maturity. As of today, dopant implantation and activation are available, and other process steps are being explored.
plot 4
plot 6
Figure 5. Activation efficiency of dopants, after implantation into GaN, predicts reduced electrical activation for low implanted concentrations and annealing temperatures in Victory Process.
Figure 6. TCAD device simulation of I-V curves of GaN JBS rectifiers showing annealing temperature TA strongly affects the forward current density JD and the on-resistance Ron

Design Technology Co-Optimization

Design Technology Co-Optimization (DTCO) can be described as a software-based methodology for developing new semiconductor process nodes that take into consideration how technology elements impact circuit performance. This iterative co-optimization is automated through Silvaco’s virtual wafer fab tool and the DeckBuild run-time control environment.

A typical Silvaco TCAD flow has the following steps:

  • Read in layout files (GDSII)
  • Use Victory Process to do process simulation in 2D or 3D of device under test (FEOL)
  • Use Victory Mesh to remesh accurately device structures
  • Use Victory Device to simulate in 2D and 3D the electrical performance of the device simulation
  • Use Utmost IV to extract a TCAD-aware compact model for SPICE simulation based on Victory Device generated I/V curves
  • Use Clever to extract parasitic capacitance and resistance (BEOL) using a high-accuracy field solver
  • Use SmartSpice to simulate a back-annotated SPICE netlist including active device and parasitics
  • Measure performance and re-run earlier steps until desired performance is achieved
plot 4 plot 6
Figure 7. Components of the DTCO flow.
Figure 8. Utmost IV modeling is supported for a wide range of silicon, SiC and GaN devices.


plot 4
Figure 9. Example of TCAD device simulation generating I-V characteristics with and without self-heating effects using the ASM-GaN HEMT compact model in Utmost IV.

With Silvaco TCAD, engineers can produce better semiconductor products, reduce the time to reach volume production, and attain the insights to suggest process improvements to their chosen foundry to improve yield and performance.