MEMS foundry processing compared to IC foundry processing holds fundamental differences that are necessary to understand and manage in order to achieve efficient and competitive time to market for customized MEMS products. Where IC process development often rests on decades of process legacy, applying only minor tweaks to a well-known fundamental baseline, MEMS process development is usually done from a far less established starting point.
One way the industry has tried to solve this issue is to make custom MEMS more “IC like” by offering standard process platforms. This works well in the IC space where the foundry owns the process platform, supporting the customers with robust design rules, and the foundry customer will create and compete against its peers by developing a product design and stack of 2D masks within the limitations provided by the foundry. The challenge such approach puts on MEMS is that MEMS designs are very often competing not only in the 2D top view of the design but actually much more in the integrated 3D stack comprising also the vertical axis of the mechanical design with a far wider range of process parameter dependency than any process simulation approach can properly manage today.
Building a standard process platform in MEMS very often ends up being a very poor compromise versus the true capability limitations. The most competitive product solutions need to be at the limitation of what is physically possible, which, given the inherent nature of MEMS, is very difficult to accommodate in a standard platform.