Multiphysics Challenges For EDA Tools

Cost and performance are the main drivers for scaling of integrated circuits. However, some applications do not scale as easily as others. This is particularly true for analog circuits and everything related to high voltage and high power.

Still, the demand for these kind of applications is growing rapidly due to new emerging markets such as Industry 4.0, IoT, and e-mobility. In the automotive sector, almost any innovation relates to some sort of analog electronics. In order to stay competitive in those markets, suppliers are forced to scale their products to reduce cost, while increasing performance and reliability.

To meet those tough specifications, the efficiency of the IC needs to increase by integrating more and more functions on a single die, i.e., monolithically integrating different blocks of the circuit. One of the most prominent application areas involves power management circuits (PMICs), where monolithic integration leads to significant savings in area and cost. Although conservative technologies are commonly used, physics will introduce new proximity effects, which prevent a straightforward scaling and feature density increase.

New types of substrates have emerged to deal with these effects, including SOI, BiCMOS, and BCD. A good electrical insulator usually will be a good thermal insulator. In addition, there are advanced packaging technologies on the horizon that promise to facilitate high-density integration while keeping physical effects under control. But the laws of physics still restrict the number of possible technological solutions.

With more complex technologies, the number of requirements such as design rules and parameter limits will increase. The space of possible solutions for a specific problem will become more complex. The creation of products that meet the product specification becomes more and more challenging, let alone the ability to get this right the first time. However, the laws of physics also can help to predict future bottlenecks in IC design.

Among the major challenges to be tackled by EDA tools:

• Electrothermal coupling. One prominent application that is already subject to electrothermal coupling issues is RF Power Amplifiers. Self-heating leads to a shift in the electrical characteristics. If this is not taken into account during the design process, there is no way to apply a compensation. The biggest challenge in this area involves vast differences in the characteristic time-scales between the electrical and the thermal domain. This can be solved only with automated thermal model generation, which can be used inside various analysis types such as steady-state, transient, envelope, and harmonic balance.
• Substrate coupling. With increasing frequencies and decreasing separation between the different parts of a circuit, dielectrically insulated substrates become increasingly transparent. Parasitic extraction in the metal stack is already a standard requirement, while substrate coupling is often neglected. However, if the circuit area is scaled down, substrate coupling effects will become more and more significant. Novel EDA tools will be required, which extend extraction capabilities into the substrate. There is no standard tool today that can predict latch-up effects due to parasitic substrate transistors. It is safe to predict that for e-mobility applications, such verification tools will become a technical as well as legal requirement.
• Electromechanical coupling. Mechanical stress is another hot topic that is not supported by standard EDA tools today. Due to more complex packaging technologies, there will be increasing stress due to CTE mismatch or transmission of vibration into the die. Such effects can be shown to affect sensitive electronics, such as those developed for sensory applications or memory cells.

In summary, the verification process of analog circuits turns into an increasingly multi-physics challenge. This requires new simulation capabilities that can stay ahead of the available resources. New EDA tools therefore need to be much more efficient when processing physical data. It is vitally important to have EDA tool flows that are more homogeneous than those that exist today. New multi-physics challenges must be met with novel EDA paradigms.