At the Semicon West 2011, 3D IC was all the rage –so much so that there were several simultaneous technical sessions
on that at the TechXPOT North One, TechXPOT North Two, TechXPOT South and Extreme Electronics that I felt a dire need for concurrent polymorphism to take advantage of those! And the booths of many equipment vendors were awash with displays of 3D capabilities including specific processes for through-silicon via (TSV) formation, micro bumping & micro-joining technologies and so on.
In sharp contrast and with much less fanfare, the MEMS product companies have been contending with 3D challenges since inception. For more than a decade, with traditional chip packages as the basis, we have witnessed MEMS technology produce hermetic packaging, multi-die stacking, wafer capping, to mention a few which are on track to become mature technologies. In yet another sharp contrast and in my opinion quite remarkable, many MEMS products companies are continuing to innovate by adding more functionality to the traditional semiconductor packages. There is of course the option to integrate more functions at the Silicon-level but adding more features at the package-level is much less time-consuming and the existing OSAT ecosystem can fully support it.
The automobile industry has been an early adopter of MEMS-based sensors. From seat-belt activation to rollover sensing, a typical automobile on the road today will have few tens of sensors, many of them MEMS-based. In high end automobiles, this number can be as many as 150 although not all of them are MEMS-based. The market forecast for MEMS-based products in this sector is expected to cross $2B by 2012 and twice as much by 2015.
Figure 1. An Overview of Automotive Sensor Applications (Source: Infineon AG)
The MEMS package is in general the sensing interface to the physical, chemical and biological inputs –most common ones are pressure, temperature, acoustics, light, magnetic and electric fields, gas, humidity, frequency, acceleration, tilt/angular, mass, DNA (e.g., many countries are indeed considering steering-integrated breathalyzers!). In addition to simply translating sensory inputs into equivalent electrical signals, the package can also be leveraged to sense without compromising linearity, reproducibility, reversibility, accuracy, noise and stability of the sensed inputs. Clearly there is a benefit to adding more non-Silicon-based sensing capability exploiting the package topologies, materials and manufacturing processes with marginal increases in cost and package complexity. Additionally, there is another distinction between the automobile MEMS products and their 3D IC counterparts – miniaturization is not the primary driving force for automobile sensors – reliability and cost are the major drivers. These two drivers are indeed triggering a renewed look into older technology packages for innovative new functionalities at minimal cost additions.
Figure 2. Packaging Needs for Automotive Sensor Applications (Source: Infineon AG)
Using afore mentioned strategy, Infineon's automobile products group is actively working on enhancing existing packages to add more functions. At Semicon West 2011's TechXPOT North One (North Hall), Infineon presented more details on their approach to realize new products with sensing capabilities for tomorrow's automobiles. Some examples presented include: (a) Using the package lead frame for current sensing, (b) Hall-effect sensing with package providing the magnetic bias field, and, (c) Package with integrated socket for optical interconnection. (on the last one, I had worked on a similar concept
Figure 3a. Packaging enhancement for current sensing (Source: Infineon AG)
Figure 3b. Packaging enhancement for Hall-effect sensor (Source: Infineon AG)
Figure 3c. Packaging enhancement for fiber optic interconnection (Source: Infineon AG)
Infineon has successfully demonstrated a packaging / product vehicle (eCUBES) with sensory capabilities for pressure-acceleration integrated with transceiver ASIC, sensor ASIC, BAW-Resonator, antenna, matching circuit (passives), blocking capacitors and batteries. eCUBES, as a packaging platform has already demonstrated tire pressure monitoring system (TPMS) function for automobiles. Along with innovations in an older technology like molded interconnect devices (MID), eCUBES or similar packaging vehicles can be leveraged to add many more functions bypassing the long and costly Silicon-integration route.
More about eCUBES can be found at the followings URLs:
The opportunities for integrating non-Silicon based sensory functions in lead frame and substrate packages are only limited by one's imagination. To that end, even wafer-level packages using back-end of the line (BEOL) and/or back-end fabs also offer potential lower-cost alternatives to Silicon-integrated sensing using front end fabs.
For More-Than-Moore to succeed, newer packaging technologies are indeed driving 3D & 2.5D IC innovations. However, there is still a lot left for innovation in older packaging technologies for MEMS!
Acknowledgments: Dr. Horst Theuss
for providing information on the examples discussed here.
-- MP Divakar is a technologist in the Silicon Valley, specializing in semiconductor backend, packaging, thermal management and test. In addition to juggling two startups, he manages to contribute at IEEE Communication and Power Electronics societies. He is a regular commentator at www.eetimes.com.