Chipworks: Inside Analog Devices’ New MEMS Strategy

St. J Dixon-Warren, Manager, Process Analysis, Technical Intelligence, Chipworks Inc.

Analog Devices has recently changed their MEMS strategy. Since the introduction of the ADXL50 in 1991, Analog has built their MEMS inertial sensors using their iMEMS technology, which integrated the micromechanical structures and ASIC circuitry on a single die. The development of this technology culminated, in some respects, with the ADXL330, which was launched in 2006. The ADXL330 provided three sense axes in a 4 mm × 4 mm × 1.45 mm LFCSP package. An X-ray of the ADXL330 package, which contains a single integrated chip, is presented in Figure 1.

Figure 1 ADXL330 Package X-Ray

Figure 2 shows that the ADX330 MEMS sensor was fabricated as a single chip, with the MEMS structure in the centre of the die, beneath a hermetic cap, and the ASIC circuitry around the outside edge. The ASIC circuitry uses a single metal, single poly 3 µm BiCMOS process, while the MEMS is fabricated using three layers of polysilicon, with the top 4 µm thick poly 3 being used to form the MEMS structures, as shown in Figure 3.

Figure 2 Decapsulated ADXL330 Chip

Figure 3 ADXL330 MEMS Structures

The integration of BiCMOS and MEMS process technology onto a single die represented a significant technological achievement; however, the price of this integration was significant limitation on the complexity of the circuitry available to device designers, and limitations on the range of MEMS processing possible.

After more than twenty years, Analog has decided to abandon their integrated iMEMS technology, and adopt the more common strategy of using a separate MEMS die and ASIC die wire bonded together in a single package. The ADXL345 represents one of the first examples of Analog’s new strategy.

Figure 4 shows an X-ray of the ADXL345 3 mm × 5 mm × 1 mm LGA package. The X-ray clearly shows the presence of a separate ASIC die and MEMS die, with a hermetic cap. The internal structure of the device is more clearly seen in the SEM micrograph of the decapsulated device, Figure 5.

Figure 4 ADXL345 Package X-Ray

Figure 5 Decapsulated ADXL345 MEMS and ASIC Die

Abandoning their integrated iMEMS technology brings a number of benefits for Analog Devices. It allows the use of more advanced process technology in the ASIC, thus enabling greater functionality, including SPI and I2C digital outputs. It also allows them to more easily adopt a foundry strategy, with the possibility of separate foundries for the MEMS and ASIC die. Apparently, Analog has already started using TSMC to provide some MEMS foundry services.

Chipworks completed a full analysis of the ADXL330 in 2006. We are presently undertaking a detailed analysis of the ADXL345, which will include information on the process used to fabricate the MEMS and ASIC dies, plus a discussion of the MEMS architecture. For further information, please contact the author.


  • Analog Devices ADXL330 Three-Axis ±2 g MEMS Accelerometer Process Review (PPR-0602-801)
  • Analog Devices ADXL330 3-axis Accelerometer ICWorks Surveyor (MEMS) (ICS-0807-802)
  • Analog Devices ADXL345 Digital Accelerometer MEMS Process Review (MPR-0907-802)
  • Analog Devices ADXL345 3-Axis Accelerometer ICWorks Surveyor Report (MEMS) (ICS-0905-801)

G20: The world comes to Pittsburgh, home of MEMS Industry Group

Monica Takacs, Director of Marketing and Membership, MEMS Industry Group

After much preparation, the G20 is finally underway in Pittsburgh. MEMS Industry Group is proud that our fair hometown will be showcased to the world. Hopefully we’ll get rid of our inaccurate reputation as a dirty steel town, an industry that has been gone since I was a little kid in the seventies.

According to the Pittsburgh G20 web site, “The reason the White House chose Pittsburgh was that today it serves as a model for economic and environmental transformation in the United States and abroad. The city has reinvented itself by building a balanced, innovation-driven economy based on its strengths in advanced manufacturing, financial services, information and communications technologies, health care and life sciences, education and research, and energy and environmental solutions.”  This list of innovators includes a hand full of MEMS manufacturers and users.

To kick off the G20, the Pittsburgh Technology Council brought in Eric Schmidt, CEO and Chairman of Google to speak to area technology leaders and to inspire all of us to continue to innovate. Google opened an office in Pittsburgh in 2006. Mr. Schmidt covered many topics that hit home for any successful company, including MEMS companies. He echoed some of the points that were made by the companies that we interviewed as part of the preliminary research for METRIC 2009, in fact.  Mr. Schmidt shared these tidbits of knowledge and recommendation:

  • “Empower your employees” – Mr. Schmidt talked about how employees should love their work. Innovators don’t like being told what to do. Let your employees work on what they are passionate about; you’d be surprised about the innovative new ideas they bring to the table.
  • Know your corporate culture – There is a corporate culture within every company. If you ask the CEO about the culture, you’ll get a different response than if you ask a member of the staff. If you are the CEO, make sure to ask your staff about the corporate culture and learn from them.
  • Transparency – Let your customer know what is going on with your organization. Post everything to your web site, you never know who’s loooking.
  • Advice to young people – Work harder. And older folks should turn off the TV and go talk to the young people around them. Then go work for them.
  • Acquisitions – Look for small innovators to acquire. These companies may enhance your technology, but don’t come with all of the baggage of a larger company. For example, Google recently acquired reCAPTCHA, a CMU spin-off company here in Pittsburgh.

I found his talk to be very timely with the G20 in town. The G20, the economic summit of the world’s largest economies in the world, convenes in Pittsburgh, September 24-25, to talk about the global financial crisis and economic policies.  It’s apparent, that innovators, like those in MEMS and other technologies, will play an important role in improving the world’s economic situation and will continue to move society as we know it forward. Keep up the good fight!!!!

You can read more about the event with Eric Schmidt in the Pittsburgh Business Times. You can learn more about Pittsburgh’s G20 summit at

Thoughts on TSV Technology

Paul Werbaneth, Vice President – Marketing and Applications, Tegal Corporation

The ink was almost dry on the TSV etching chapter I am contributing to a project called 3D Technology Platforms when I opened the newspaper today to read about very interesting news from Elpida regarding TSV-stacked DRAM (here: ).

It’s some kind of harmonic convergence for me, as I sit here thinking about everything I wrote over the last month on TSV etching (which already becomes dated with today’s news), as I sit, thinking, about the nascent DRAM recovery, sit, thinking, about the magnificent Elpida site in the green wooded hills outside Hiroshima, hills set with large farming houses sporting freshly tiled roofs last time I visited, sit thinking about 300mm TSV processing challenges, sit, thinking, about the novel (and movie) Black Rain, and sit, remembering, how I would change trains at the Shin-Hiroshima station in order to reach the Hitachi factory in Kudamatsu after a long day of travel from San Francisco.

You had to be quick:  Shin-Hiroshima to Tokuyama (Kudamatsu) could be a tough connection to make, and you would be stuck for the night if you didn’t.

Through-silicon via technology for 3D IC applications has deep roots in other commercial through-substrate via processing, most notably in compound semiconductor fabrication for components found in your friendly cell- and smart-phone handsets.  And we all recognize how much TSV processing owes to MEMS fabrication, where deep structures in silicon are the name of the game.

Many eyes are now watching for the commercial launch of TSV technology beyond its use today in CMOS image sensors.  DRAM, NAND Flash, memory stacked onto processor chips, all seem fair game for 3D integration, but, with recent and all-too-painful revenue woes being felt from Boise to Seoul, the investment budgets for TSV in memory applications just haven’t been there much this year.

Encouraging, then, to catch a pale view of recovery and progress today over Hiroshima hills.