mechanical engineering August 2008 25

Research and development (Grade: A-):R&D spending for MEMS has remained relatively con-stant at a favorable grade since 1998, never dropping be-low A-. In earlier times, the Defense Advanced ResearchProjects Agency funded many activities focused onMEMS device development. Recent DARPA funding hasincluded reliability of RF MEMS and packaging. TheNational Science Foundation continues to provide sup-port to MEMS R&D in the form of Small Business Inno-vation Research grants for early phase design and devel-opment activities. A good indicator of funding is thenumber of graduate students graduating from MEMS-rich academic programs.

This number appears to be holding constant. There hasbeen a great deal of “small R” and “large D” in the com-mercial sector. Funds appear to be directed to fine-tuningprocesses to increase yield. This is especially true in areaswhere large-volume MEMS products are being pro-duced—for example, automotive and consumer applica-tions that use accelerometers, gyros, and displays. Therehas been a significant increase in the R&D funding levelsin emerging-technology countries, including India andChina. Areas of expanded MEMS funding on the rise in-clude energy harvesting. The explosion of interest in nan-otechnology has diminished the funding levels of MEMS.

Marketing (Grade: C+):Marketing grades have languished at the C level from theinception of the report in 1998. Efforts to date have been atechnology push versus a market pull. Only recently, with

the introduction of MEMS into consumer applications—such as the optical micromirror in Texas Instruments’ Dig-ital Light Projector, and accelerometers by Analog Devicesand ST Microelectronics in the Nintendo Wii game sys-tem—has MEMS been exposed to the mass market.

Most recently, MEMS has been introduced into portableelectronic systems—MEMS accelerometers and gyros inGPS and mobile phones, microphones in portable com-puters and mobile phones, and gyros in digital cameras.

There still exist several serious voids on the part ofMEMS suppliers to understand customers’ needs andtheir products’ unique competitive advantages. Mostcompanies attempt to sell their products on specificationsand fall far behind semiconductor companies on market-ing expertise.

Marketing of MEMS is quite challenging and expensive,since application sectors are diverse and fragmented, andrequire application engineers who are well informedabout customers’ needs and requirements. Another majorproblem is that there is a serious lack of product differen-tiation among suppliers. Marketing communications re-sources tend to be inadequate. In my opinion, the bestmarketing efforts have been turned in by companies in-cluding Analog Devices and Freescale, who are majorMEMS players. It’s interesting to note that these compa-nies are also veterans of the semiconductor industry andperhaps MEMS has benefited from the cross-pollination.The study shows that European companies lag behindU.S. companies with respect to expertise in marketing.

Market research (Grade: B):Market research for MEMS continues to be provided by anumber of organizations worldwide (including ours).Markets are analyzed according to device types, such asaccelerometers or gyros, and by market sector—automo-tive, military, and so on. In addition, specific market re-ports on “hot” application topics, including inertial ratesensors, are made available by multiple research groups.

Conducting accurate market research in the MEMS mar-

ket is quite difficult since there areso many suppliers (most of themprivate companies). As such, salesvolumes are not public information and typically are heldin high confidentiality. As a result, market numbers fromdifferent organizations are prone to vary from report to re-port. As MEMS has become a larger business opportunity,traditional large semiconductor research firms are enteringthe field.

24 August 2008 mechanical engineering

he fir st manifestation of microelectro-mechanical systems was in 1954, whenCharles Smith of Bell Laboratories publisheda paper in Physical Review that discussed hisobservations of the piezoresistive property ofsilicon. This property of the mater ial to

change its resistivity with applied mechanical stress cameto form the basis of many MEMS sensors, especially pres-sure sensors. Thus, we can say that the MEMS industry isover 50 years old.

The semiconductor industry had its founding less than10 years earlier by scientists from the same lab. Consid-ering that today semiconductors outsell MEMS by 20 to1, we must ask ourselves: Why has MEMS underper-formed its semiconductor cousin? In other words,where are the problems?

A number of MEMS devices have been commercial-ized, beginning with the pressure sensor in 1990. Roger

Grace Associates has conducted a study to determinethe commercialization timetable of some of the moresignificant MEMS products that have been developed inthe last 50 years. Based on our findings, we estimate thatit takes approximately 25 years from discovery to fullcommercialization. We expect that, as people look tolessons learned from other commercialization efforts,they may reduce the timeline necessary to achieve fullcommercialization.

Another emerging engineering interest, nanotechnolo-gy, is already eating into the funds available for MEMS re-search. Like MEMS, it is aimed toward commercializationin areas ranging from materials to medicine. As a busi-ness, it has farther to go than MEMS. So the question is,what can the practitioners of nanotechnology learn fromthe example of MEMS technology?

To better understand the progress MEMS has made asan industry and to identify some of the obstacles thatremain, we have developed what we call the commer-cialization report card. The first commercialization re-port card was issued in 1998 and has been updated year-ly to reflect the changes in the MEMS industry thataffect its performance.

The results of the study are provided in a report cardformat. The grade scale ranges from D to A. We grade 14separate areas, but the ones we consider the most signifi-cant are discussed below.

T

MEMS: As the latest report card shows, the microelectromechanical systemsbusiness falls short in many key areas. Can nanotechnology learn from the example? By Roger H. Grace

MEMS COMMERCIALIZATION TIMETABLE

PRODUCT DISCOVERY PRODUCT COST FULL EVOLUTION REDUCTION COMMERCIALIZATION PRESSURE SENSORS 1954-1960 1960-1975 1975-1990 1990ACCELEROMETERS 1974-1985 1985-1990 1990-1998 1998GAS SENSORS 1986-1994 1994-1998 1998-2005 2005VALVES 1980-1988 1988-1996 1996-2002 2002NOZZLES 1972-1984 1984-1990 1990-2002 2002PHOTONICS/DISPLAYS 1980-1986 1986-1998 1998-2005 2005BIO/CHEMICAL SENSORS 1980-1994 1994-2000 2000-2010 2010RADIO FREQUENCY (R.F.) 1994-1998 1998-2001 2001-2009 2009RATE SENSORS 1982-1990 1990-1996 1996-2006 2006MICRO RELAYS 1977-1993 1993-1998 1998-2010 2010OSCILLATORS 1965-1980 1980-1995 1995-2010 2010

lessons for nano

FEATURE FOCUS MICRO- AND NANOTECHNOLOGY

Roger H. Grace is president of Roger Grace Associates,a marketing consultant to the sensor and semiconductorindustries based in Naples, Fla. He can be reached byemail at rgrace@rgrace.com.

Ever more compact: Shrinking packages show the downward sizemigration over time of accelerometers from ST Microelectronics.

ST

MIC

RO

ELE

CTR

ON

ICS

AN

ALO

G D

EVI

CE

S

The die of a MEMSgyro from AnalogDevices.

mechanical engineering August 2008 27

and manufacture of MEMS. The main infrastructure ele-ments are software tools, manufacturing and test equip-ment, and manufacturing facilities, a.k.a. foundries.

Numerous companies have a solid history of provid-ing excellent design, analysis, and simulation tools toMEMS designers. Included in this are Coventor, Intel-lisense, and SoftMEMS. All of these providers have theirown specialties.

Manufacturing and test equipment has come a longway from the retrofit days of semiconductor processingequipment. Today, a broad spectrum of equipment madespecifically for MEMS is being offered by companies, in-cluding EVG, Jenoptic, and Suss MicroTec. These firmshave done an excellent job in their Marketing 101 tasksand have developed equipment well suited to the re-quirements of the industry.

The major area of interest in infrastructure is in MEMSfoundries. These organizations produce MEMS wafers fromdesigns provided by their customers. Our research showsthat there are more than 60 foundries worldwide, and thereis an oversupply of services. Major players include Asia Pa-cific Microsystems, Colibrys, IMT, and Micralyne.

The newly adopted “fabless” or “fablite” MEMS busi-ness model was taken from the semiconductor industry.Most venture capitalists prefer to fund companies usingthis model, since their investment is in ideas and peopleinstead of brick and mortar. The major drawback here isthat each foundry has its unique set of processes and tools.

If development of the early design is conducted at a uni-versity, there’s a good chance it will be necessary to under-take process modifications or device design changes to ef-fect a smooth transition to manufacturing. Also of note isthe need to convert 4-inch and 6-inch wafer sizes to 8-inch in order to take advantage of the new technologies as-sociated with 8-inch production equipment, as well as thelower cost per device resulting in using the larger formatwafer. SVTC is an organization focused on providing thisservice. Here again, process and design changes may beneeded, which could lead to additional cost and a time-to-market delay.

In summary, there is more infrastructure in place than ispresently required for the successful commercializationof MEMS. It is critical that MEMS organizations fully un-derstand their short-term and long-term manufacturingrequirements early on in order to judiciously select theright manufacturing partners.

Industry associations (Grade: B+): A number of industry associations exist worldwide to sup-port the commercialization of MEMS. Chief among themare the Micro and Nanotechnology CommercializationEducation Foundation, the MEMS Industry Group in theU.S., IVAM in Germany, and Nexus in the EuropeanUnion. All of these organizations conduct numerous meet-ings yearly and provide a valuable forum for their members.

Standards (Grade: C): The progress on standards for MEMS has been less thanexciting. Fewer than 10 standards currently exist for

MEMS, whereas in the semiconductor industry, more than700 have been published by the Semiconductor Equip-ment and Materials International organization. The reasonfor the lack of standards is the absence of standard process-ing for MEMS. Many companies use their processing asproduct differentiators. I believe that many items associat-ed with MEMS packaging, processing, and testing can havestandards, and the creation of standards will have the effectof reducing part cost in the industry.

Venture capital attraction (Grade: C): Venture capital attraction continues to focus on MEMScompanies that promise to be successful in the large-vol-ume consumer and medical markets. The grade has mi-grated to C from its high point grade of A at the heightof the high-tech bubble in 2001. Recent startups, includ-ing SiTime (system timing products) and Invensense (gy-

ros for digital cameras), have received advanced Round Band C funding in 2007.

Venture capitalists are always seeking good investmentopportunities characterized by good management, alarge and growing market, and a great idea with defensi-ble intellectual property. However, not many MEMScompanies are able to prove that they can deliver a 10-fold return on invested capital in five years (the goal ofthe VC business). Recently, venture capitalists have be-gun to focus their interest in energy, green, and Web 2.0opportunities, and have moved away from nanotechnol-ogy. New MEMS investments are taking the back burner,while earlier funded companies continue to be funded.

Cluster development (Grade: B-): Cluster development was added to the report card in2003. The creation of clusters has proven to be a major

26 August 2008 mechanical engineering

While pub-lished reports by

established organi-zations appear ade-quate to serve theindustry’s need, the

use of in-depth custom researchis woefully inadequate. Marketing

101 states that, before one enters amarket, one needs to know and under-

stand the size of the market, the competi-tion, a firm’s unique differentiated advan-

tage, and most important, the unfulfilled customerneed that the company will supply. Until MEMS compa-nies acknowledge and embrace this valuable tool, theycannot expect to be truly successful in the marketplace.

Design for manufacturing and test (Grade: B): DfM&T is one of the most important and critical successfactors for MEMS commercialization. It is at the heart ofdetermining the cost of manufacturing and the reliabili-ty of the produced part. Unlike the semiconductor in-dustry, where the device package plays a minor role inthe overall product solution, MEMS packages are oftenmore important and almost always more expensive thanthe devices themselves.

A commonly accepted rule of thumb establishes MEMSpackaging, assembly, and testing to be between 60 and 80percent of the total solution cost. The reason for this is

that many MEMS devicesare subject to the rigors of harsh

chemical media—as are, for example, sen-sors for automotive oil pressure—in which they

must make their measurement. As such, the silicon dieoften needs protection, and a costly mechanical packag-ing solution must be used. In addition, since most MEMSare electromechanical in nature, their performance is af-fected by changes in temperature, especially when theyare made of silicon and are mounted on a substrate thathas a different thermal coefficient of expansion, whichresults in inducing stress into the chip.

As one can see, the device and its mounting method mustbe considered as a system and properly modeled as such.An example is on page 28. In addition, many MEMS de-vices are interconnected with semiconductor devices, typi-cally application-specific integrated circuits, or ASICs. Toachieve an optimum design, connectivity and functionalpartitioning strategies of the system must be considered.

For a MEMS device to be optimized for cost and relia-bility, issues of manufacturing and testing must be con-fronted at an early stage in the design of a device and fol-low concurrent engineering principles. Unfortunately,not many MEMS manufacturers have seen the light.Large-volume suppliers have typically accepted this ap-proach to meet the price demands of the consumer mar-ketplace. Companies including Bennington Microtech-nology Center and Infotonics are resources available forMEMS developers to use to assist in the development ofpackaging and DfM&T process development.

Established infrastructure (Grade: A-): The most dramatic improvement of any grade, from C+ in1998 to an A- in 2007, has been achieved in the MEMS es-tablished infrastructure area. We define “infrastructure” asthe resources needed to support the design, development,

MEMS INDUSTRY REPORT CARD

SUBJECT 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

R&D A A A A A A- A- A- A- A-MARKETING C- C C+ C+ C+ C C C+ C+ C+MARKET RESEARCH C B- B- B- B B B+ B- B- BDESIGN FOR MANUFACTURING C+ B- B B B B B C+ B- BESTABLISHED INFRASTRUCTURE C+ B B+ A A A A A- A A-INDUSTRY ASSOCIATION INC INC INC B B+ B+ B+ B B B+STANDARDS INC INC INC INC C B- B- B- C+ CMANAGEMENT EXPERTISE C C C+ C+ C+ C+ C+ B- B- BVENTURE CAPITAL ATTRACTION C B- B+ A C C- C C+ C+ CCREATION OF WEALTH C B- B+ A C C- C- C- C- CINDUSTRY ROADMAP N/A B- B B+ A- A A B B- C+PROFITABILITY C- C- C- C- C- C- C- C C+ CEMPLOYMENT INC INC INC INC INC C C C+ C+ C+CLUSTER DEVELOPMENT INC INC INC INC INC B B+ B+ B B-

OVERALL GRADE B B B- B-

Good vibes: The instrument for RedOctane’s Guitar Hero video gameuses a three-axis MEMS accelerometer for controller tilt motion.

In and out of the box: A view of one of Freescale’s MMA73xxL familyof accelerometers with the die removed from the packaging.

RE

D O

CTA

NE

FRE

ES

CA

LE

28 August 2008 mechanical engineering

catalyst in creating new MEMS companies. In his book,On Competition (published in 1998 by Harvard BusinessSchool Press), Michael E. Porter writes that clusters werecreated to increase the competitiveness of the organiza-tions within the cluster. The first MEMS cluster was cre-ated in 1986 in Dortmund, Germany. Since then, morethan 35 MEMS-specific clusters have been createdworldwide. (There is a paper, “Technology clusters andtheir role in the development of the microsystems indus-try,” available on the Roger Grace Associates Web site,www.rgrace.com, originally presented at COMS 2003 inAmsterdam, the Netherlands.)

There are successful clusters in Washington State;Hsinchu, Taiwan; Edmonton, Alberta, and Grenoble,France. MEMS clusters currently in the ramp-up stageare in Manaus, Brazil, and Paseo del Norte, Mexico.While existing clusters are tending to grow, and achievetheir financial objectives, funding activity for MEMS hasgiven way to the funding of nanotechnology clusters.Here, federal and local governments are making signifi-cant investments. Hopefully, some of the R&D undertak-en in these nanotechnology clusters will have MEMScontent or applicability.

how we collected the informationRoger Grace Associates used a Delphi method to create

the information that forms the basis of its commercializa-

tion report card.

The Delphi method uses interviews with a small number

of experts (in this case, 55) in the field of MEMS taken

from a broad range of applications and companies in the

U.S. and in several countries of Asia and Europe. It’s im-

portant to note that the interview universe represents

MEMS suppliers, MEMS users, and providers of MEMS in-

frastructure. As such, we attempted to closely represent

the broad MEMS universe. The most recent study was

largely conducted in April 2008, and was used as the ba-

sis of a presentation at Sensors Expo this past June. This

article is adapted from the presentation.

An expanded version of the report card is published on

www.rgrace.com/MEMSreportcard2007.

lessons and cautionsAs MEMS has learned a great deal from the

semiconductor industry, I submit that nan-otechnology should learn from its bigger

brother, MEMS. For those bringing radicallynew technology to market, I have four key cautions:

• Do not create technology for technology’s sake; un-derstand the unfulfilled market need by undertaking for-mal and well-planned market research.

• Take care to understand competitive offerings and tocreate a product that is defensibly different.

• Do not fall prey and oversell the ability of nanotech-nology to uniquely solve problems.

• Properly promote the product, especially its ability touniquely solve the customers’ application problems.

It’s fortunate that nanotechnology manufacturing hasbeen an area of significant funding from organizationsthat include the National Science Foundation. MEMSdevelopers were not so fortunate in the field’s earlydays, and that lack of support contributed to the slow-down of the commercialization timetable.

At this point, it appears that nanotechnology has been agood student in many of the critical success factors givenabove. The challenge will be to continue to fund re-search for nanotechnology and to support developmentin the infrastructure area, including the advancement ofmanufacturing and metrology tools. Nanotechnologystill will need venture capital money, which may proveespecially challenging to raise in the current financialmarket conditions. As an industry, it will need to createstandards and road maps to help guide the participants.

The MEMS report card has demonstrated a few signifi-cant advances in addressing the 14 critical success factors

that are important in achieving successful commercial-ization. While a number of the grades have changed forbetter or worse, the overall grade remained at B-, as itwas in 2006. Many grades still need major improvement,including those for marketing, standards, venture capitalattraction wealth creation, profitability, and employment.

I encourage individuals interested in the commercial-ization of nanotechnology to become students of theevolution of both the semiconductor and MEMS indus-tries. To quote the philosopher George Santayana in hisbook, The Life of Reason: “Those who forget the past arecondemned to relive it.” �

Package goods: A Tronics sensor (above) requires protective packagingto shield the device from a harsh operating environment. A look inside aMEMSIC accelerometer (above left) shows the die in the package.

ME

MS

IC (

FAR

LE

FT);

TR

ON

ICS

me.hotims.com/18875-11 or circle 11

me.hotims.com/18875-12 or circle 12mechanical engineering August 2008 29