Process Simulation vs. Process Emulation: is SEMulator3D really TCAD?

I was interested to note that Silvaco has recently listed SEMulator3D as a competitor for their VICTORY Process Cell software on their website. It’s great to be mentioned as a contender in the TCAD process simulation space. But I’d like to take the opportunity to examine the following question – are SEMulator3D and VICTORY Process Cell really direct competitors?

On the surface, both SEMulator3D and VICTORY Process Cell can do some similar things. Both tools are fast, layout driven process modeling engines that are designed to build 3D models of MEMS and semiconductor devices. Both tools can model individual process steps or entire process sequences, and can model a variety of process and device types. And both tools can create meshes suitable for further physics simulation.

Left - Silvaco Process Cell (image is from the Silvaco website and is the property of Silvaco).

Right - SEMulator3D

Figure 1: A 3D MEMS Actuator, fabricated using the SCREAM process [1]

But despite the superficial similarities, there is an important distinction to be made between process simulation and process emulation. Process Cell and its direct competitors (Synopsys Sentaurus Process and similar) are process simulation tools and are based on TCAD simulation technology. The word “simulation” is key here because these tools simulate process steps using physical process models, driven by physical input parameters (implant energies, etch times and temperatures, etc). The models generated by these tools are well suited for detailed simulation of transistor electrical performance. But due to the complexity of the models, the chip area that can be modeled is relatively small – usually a single transistor, or at most a single layout cell. And because of their TCAD heritage, process simulation tools are often driven by a scripting language (powerful but can be difficult to learn).

SEMulator3D, on the other hand, is a process emulation tool. SEMulator3D creates models of process steps using geometric parameters – parameters that describe the shape, thickness, or depth of individual processing steps. While not as fundamental as physical parameters, geometric parameters are easier to determine (for example, can be extracted from a SEM of a device). SEMulator3D is driven by an easy to use graphical interface, freeing users to think about processing issues rather than script syntax. And SEMulator3D is uniquely capable of modeling large areas of silicon, measured in microns rather than nanometers.

SEM image reproduced with permission of the author (Arjun Selvakumar)

SEMulator3D model

Figure 2: A MEMS threshold accelerometer, fabricated using a dissolved wafer process and Silicon/glass wafer boding [2]

So what does this all mean? From my perspective, it means that both process simulation and emulation are useful, but for different tasks.

3D process simulation (TCAD) is ideally suited for detailed device simulation – and for detailed electrical performance simulation – of individual devices or perhaps single cells.

In contrast, 3D process emulation (SEMulator3D) is ideally suited for fast 3D model generation, visualization, process/layout verification, and especially communication. Since SEMulator3D is fast and easy to use, it’s a great way to communicate about processing topics within the fab. Process engineers and integration teams can save wafers by validating their process and layout before fabbing. And since SEMulator3D supports document creation in a number of standard formats, it’s a great tool for process documentation.

So when you look at these two tools in a bit more detail, they are really quite different. Both are useful and capable, but in different ways. Perhaps both tools will find a niche in the process development cycle and combine to make process development more efficient and profitable.

More information about SEMulator3D is available on the main SEMulator3D web page.

Silvaco VICTORY Process Cell and Synopsys Sentaurus Process are the trademarks of their respective owners.

References:

[1] MacDonald N, 1996 “SCREAM MicroElectromechanical Systems”, Microelectronic Engineering 32 49-73.

[2] Selvakumar A,Yazdi N, Najafi K, 2001 “A wide-range micromachined threshold accelerometer array and interface circuit”, J. Micromech. Microeng. 11 118-125.

This entry was posted in Coventor Blog and tagged , , , , , . Bookmark the permalink.

Comments are closed.