Major market drivers & technical answers in inertial MEMS & magnetometers
Over the last 3 years, inertial MEMS & magnetometers have been subject to dramatic market & technological evolutions. This has been driven by a large increase of the consumer market: mobile phones and tablets for accelerometers; gaming for gyros; mobile phones for magnetometers.
Along with “stand-alone” MEMS devices, inertial combo sensors, a combination of several inertial sensors into a single package, are also coming. Main applications are consumer (e.g. accelerometer with magnetometeror accelerometer with gyro) and automotive for ESC and rollover functions first.

Inertial MEMS & magnetometers are today driven by 4 major market trends:

1. Future generation of sensors will deliver functions and will become “solutions”
2. Fusion of sensors (combining data from different sensors) is starting to be widely used
3. New architectures are developed
4. Price pressure is still very strong (5% drop per quarter for consumer applications).

On the technical side, form factor is ever decreasing with reduced footprint and thickness. And power consumption has been reduced to a few microA while performances are still increasing.
The most successful type for inertial MEMS is based on capacitive transduction. Reasons are simplicity of the sensor element, no requirement for exotic materials, low power consumption and good stability over temperature. But will comb-drive architecture for accelerometers continue to be the main detection principle as MEMS die size keeps shrinking?
Regarding gyroscopes, most are falling into the categories of tuning vibrating fork/plate (STM, Bosch) or vibrating shells (Silicon Sensing Systems). This very common design gives ease of fabrication and possible integration in standard IC manufacturing industry.
For magnetometers, Hall Effect has been the dominant technology for a long time, but today it is changing as Magneto Impedance, Giant Magneto Resistance and Anisotropic Magneto Resistance are used. A new approach, Lorentz effect based on MEMS technology, is currently in R&D (VTT and others). This could bring easier integration in MEMS combo sensors.

MEMS testing will have to evolve

Testing has been also subject to strong evolution over the last years. For example, combo sensors will require new test solutions compared to “stand-alone” sensors. Beyond the usual wafer-level electrical test and package-level electrical and mechanical or functional testing, these sensor combos will need module level testing and calibration of the combined sensors. If they include an MCU in the package, the communication between the sensors and the MCU will also need to be tested. Solutions need to be cost effective with high throughput to test multiple axes of multiple devices, either in parallel or in separate modules, rather like separate chambers in IC equipment.
The world of MEMS testing has moved in the last several years from internal development at MEMS makers to co-development with test suppliers to commercial off-the-shelf equipment. So combo solutions that can test all axes of the module in a single tool for higher throughput will also likely be co-developed with the test equipment suppliers and available commercially. Assembly and test houses may also start to offer these test services on an outsource basis for fabless or fab-light MEMS makers. The Yole report will analyze the latest trends in MEMS testing.

A comparative analysis of 23 MEMS devices from 13 different manufacturers!
In order to understand the key evolutionary changes, a total of 23 different MEMS devices (9 accelerometers, 10 gyros, 3 combos and 1 magnetometer) - mostly consumer MEMS - have been disassembled, analyzed and cost simulations have been constructed for MEMS, ASIC and Packaging/Test. One of the key features of the reports is that ASICs have been analyzed as well. The MEMS have been analyzed and production costs have been simulated by System Plus Consulting, the reverse costing specialist company.

The teardown analysis results have been compared in terms of performance, total cost, MEMS size, ASIC lithography node, ASIC size, package size, year for market introduction.
From our analysis, we found there is a clear MEMS die size decrease over 2007-2011. For example, in 2008, the average size for an accelerometer (3-axis) was 4-5 mm². 3 years later, size is about 2 mm². ASIC size has been following the same trend with a lithography node in the range 0.18-0.35µ today. With latest ST announcement about the use of through silicon vias for inertial, we can expect even lower cost and size in the future.
The same analysis has been performed for gyros comps, combos and magnetometers.

Keys features of the report
The objectives of the reports are the following:

• Provide an understanding of the market drivers for inertial MEMS
  • Consumer
  • Automotive
  • High-end
• Give trends about packaging and tests strategies
• Provide in-depth analysis for 23 MEMS devices in terms of:
  • Production cost breakdown for MEMS, ASIC, Packaging & Test
  • ASIC litho nodes
  • Die & package size
  • Package type
  • Comparative analysis in terms of performance, production cost for MEMS, ASIC, sizes, year for market introduction.
• For each device, photos are depicting:
  • MEMS close-up structure
  • MEMS dimensions
  • ASIC dimensions
  • Package view
  • Specific process steps
  • Cost breakdown