MEMS sensors are a hot topic for us, and one interesting aspect is how companies commercialize this technology. Stephen explores an angle we don’t usually see, with an exclusive look inside a sophisticated handheld spectrometer made possible using MEMS.

Microelectromechanical Systems, now known affectionately by the MEMS acronym, have been showing up in all kinds of places: air bag deployment, automobile suspensions, mountain bike altimeters, cell phones, video cameras, and even computer games. The canonical MEMS component is a silicon chip with some micromechanical parts designed to respond to some physical variable, such as pressure, acceleration, flow, sound, and radiant energy. Depending on the particular device architecture, the sensing electronics reside on the same chip as the MEMS or the MEMS are copackaged with a separate Complementary Metal Oxide Semiconductor (CMOS) chip. Product volumes for such MEMS components are measured in millions of units per year. The emphasis today is on achieving low cost at acceptable levels of performance, interchangeability, and reliability. Customers buy these MEMS for incorporation into larger systems, such as automobiles or video games.

Market volumes for another type of MEMS product, the MEMS-enabled product, are measured in thousands of units per year instead of millions of units per year. This market is too small to permit a successful business based on the manufacture and sale of the MEMS component alone. MEMS of this type can only be sold in combination with the full product into which it is incorporated.

MEMS inside

The traditional example of a MEMS-enabled product is a measurement instrument with a unique MEMS chip embedded in its core. One can think of a MEMS-enabled product as a large inverted pyramid (the instrument) supported by a very tiny but very strong base (the MEMS). MEMS sensors have functionality that supports a novel way of doing something useful (the tip region of the inverted pyramid), which supports the next broader level – a new instrument architecture that has some important advantages over extant technology. The new instrument competes successfully because of MEMSí capability. But it doesnít compete in the MEMS market; it competes in the instrument market and is judged by its performance as an instrument. The MEMS component disappears inside, but it is essential.

A few significant challenges hinder the creation of successful MEMS-enabled products. The first is the need to codesign MEMS sensors and the system into which they will be embedded. The Polychromix PHAZIR handheld, near-infrared spectrometer for materials analysis illustrates this challenge.

The spectrometerís architecture is based on the modulation capabilities of an electrically programmable diffraction grating built as a MEMS chip. Figure 1 shows a schematic version of this chipís operation and depicts a microphotograph of a fully packaged part, complete with optical window. Think of it as a set of long parallel piano keys that can be positioned up and down with electric signals. In the all-up position, the chip is a mirror. With alternate keys depressed, light impinging on the chip is diffracted away at an angle that depends on the ratio of the wavelength of light to the width of the piano key. The change from all-up to alternate-down states can be achieved in less than one millisecond.

(click graphic to zoom by 2.2x)

Design follows function

To take advantage of this technology, an instrument architecture (starting up the inverted pyramid) was developed around the MEMSí high-speed modulation capability. Incoming light is dispersed by wavelength using a fixed grating so that each wavelength range of interest hits a different set of piano keys, called a pixel. The MEMS component has 100 pixels. Light reflected from the chip is sent to a detector. If a pixel is set in diffraction mode, the light hitting that pixel is diffracted away and does not reach the detector.

The MEMS codesign has three requirements:

• The length of the piano key must match the spot size produced by the overall optical design


• The voltage required to drive the piano key from up to down must be within the voltage range of available low-power electronics chips


• The MEMSí mechanical structure must create a diffraction angle large enough to permit exclusion of the unwanted light from the optical path to the detector

Thus, the MEMS chip has design details tightly linked to both the optical and electronic design of the spectrometer as a whole (shown in Figure 2). It is not a separate commodity.

(click graphic to zoom by 2.2x)

Based on the MEMS chip, this architecture supports a sophisticated form of Digital Transform Spectroscopy (the next tier of the inverted pyramid), in which the chip is driven through a coded sequence of 100 different up-and-down pixel combinations, creating a unique time signature at the detector. A computer analyzes this time signature to create the full spectrum. Because of the MEMS chipís intrinsic speed, the entire operation of collecting and displaying the spectrum takes less than one second. Also, because of the particular architecture, the spectrometerís operation is insensitive to stray light and drifts in the detector, creating a wide dynamic range of sensitivity fully competitive with conventional detector-array spectrometers.

In the palm of a hand

The next tier of the inverted pyramid derives from the fact that the optical package is small and light and requires sufficiently low power to permit the entire spectrometer to be built as a handheld, portable, battery-operated instrument, a capability simply not available from laboratory-scale instrumentation.

The MEMS chip is buried inside the small optics module, which serves as the heart of the instrument. But the spectrometer also includes a light source, a sampling head that directs light onto the sample and collects reflected light, a battery pack, a display, control electronics for the MEMS, and a microcomputer that performs instrument control and analysis of the resulting spectrum to identify materials. Industry-specific applications can be uploaded to the instrument without any change in hardware. Thus, the spectrometerís MEMS-enabled hardware is really a product platform. Multiple products can be implemented on this platform entirely through software changes. This is the top level of the pyramid – multiple products directed at different markets, all based on a single custom MEMS chipís capabilities.

Because the spectrometer is handheld and battery-operated, measurements only possible in a laboratory setting can now be completed in the field. Consider, for example, the problem of carpet recycling. When carpet is removed from a building site, the fiber can be recycled only if the fiber type can be identified correctly. The spectrometer can be used to make this identification so that carpet samples are directed to the correct recycler depending on fiber type without requiring the costly step of transporting the carpet to a laboratory-equipped warehouse. This eases the carpet recycling process and helps reduce the amount of waste carpet destined for landfills, an endeavor championed by the Carpet America Recovery Effort (CARE).

Similar measurement efficiencies can be achieved for incoming raw material inspection in the pharmaceutical industry. When receiving a large drum of chemical reagent, present practice involves quarantining the container until identification can be confirmed by taking a sample to an analytical laboratory, a procedure that can take hours and lead to transcription errors. The spectrometer can make the confirming measurement in one second without requiring the package liner to be opened, thereby streamlining the time required for incoming materials inspection and improving the integrity of the process.

Where rugged and accurate coexist

For a MEMS-enabled product to be competitive, it must pass all the same calibration, stability, and reliability tests that other instruments in its market segment undergo. In the case of the PHAZIR, this means that the instrument must correct for variations in light source intensity, adapt to the varying conditions of reflected light collection from different types of samples, and implement a wavelength calibration standard to prevent any drift-induced errors.

These various calibrations depend on the MEMSí intrinsic repeatability and ruggedness as well as accurate voltage application to the chip. The polysilicon technology used for the MEMS is similar to what Analog[] Devices implements in its accelerometer products, which have a long history of mechanical reliability. The MEMS are also rugged; unpackaged MEMS chips have been subjected to shocks as high as 30,000 g without failure.

The repeatability chain made possible by a MEMS sensor is clear: As long as the voltage applied to the MEMS is repeatable, the piano key displacement is also repeatable, which means the modulation depth is repeatable, and, consequently, the scaling inherent in the Digital Transform Spectroscopy algorithm to convert the time signature into the spectrum is repeatable as well. Therefore, the final calibration of the wavelength assigned to each pixel can be achieved with confidence.

The PHAZIR is just one example of this important class of MEMS-enabled products. The ability to manufacture such products depends indirectly on the health of the MEMS commodity market, which supports the vendor ecosystem needed to manufacture specialty MEMS in low volume. But with the MEMS commodity market now growing at a healthy pace, MEMS-enabled products can thrive.

Stephen Senturia is chairman and CTO of Polychromix, based in Wilmington, Massachusetts. A graduate of Harvard (1961) and the Massachusetts Institute of Technology (PhD, 1966), Stephen spent 36 years as Professor of Electrical Engineering at MIT. He is a member of the National Academy of Engineering and an IEEE Fellow.

Polychromix
978-284-6000
sds@polychromix.com
www.polychromix.com