MEMS: how did we get here?

Kurt Petersen

MEMS Released Products

1975

  Pressure Sensors   Strain Gauges •  TI Thermal Print Heads

TODAY (to name a few)

  Pressure Sensors (everywhere !!)   Accelerometers (everywhere)   Oscillators (soon to be everywhere)   Gyroscopes   Ink Jet Print Heads   Microphones   FBAR Filters for Cell Phones   AFM Probe Tips •  Digital Light Projectors •  Wafer Probers •  Variable Optical Attenuators •  Tunable Lasers for Telecom •  IR Image Detectors •  Near IR Spectrometers •  Gas Chromatographs •  Microfluidics for BioTech •  Implantable Pressure Monitors •  Cochlear Implants

  High Volume, Low Cost

Growth of MEMS

80’s - 90’s

Automotive

2000’s

Consumer

80’s

Medical

60’s - 70’s

Industrial

109 108 107 106 105

Market Size - units

1010

2010’s

Bio-MEMS + Consumer + Environmental

Yearly Shipments

MEMS

Semi 3.5% of Semi

1% of Semi

1st HH Mtg

1st PC

1st DLP

Billions of MEMS Devices Sold

Avago >2B Cell phone filters As of Q2 2010 Robert Bosch >1B Automotive sensors As of Nov 2008

Knowles >1B Microphones As of Sept 2009

Texas Instruments >1014 Light modulators As of Dec 2009

GE NovaSensor >750M Pressure sensors As of Aug 2011

ST Micro >1B Accels and Gyros As of Dec 2010

HP Et

Cepheid ~$1B Microfluidic systems As of Q4 2010

SiTime >100M MEMS Oscillators As of Q2 2012

“Billions and Billions of ink jet nozzles”

MEMS allows HUGE size reductions

Inertial Measurement Unit 1960’s

Inertial Measurement Unit 2011

Which translates to HUGE cost reductions

The Road to Miniaturization

•  If the size of an object has no correlation with its function, technology will eventually make it shrink –  i.e. houses and cars and TVs will NOT shrink (much)

•  Why? •  Data is the perfect candidate for shrinkage •  Electronic circuits for data are perfect candidates •  The size of a sensor has little to NO correlation

with its function - so, sensors will shrink

History and economics tells us this.

Leverage from Semi Industry

•  During 70’s and 80’s; we bragged that we leveraged the Semiconductor Industry –  litho, implantation, oxidations, diffusion, wet etch, etc

•  However, MEMS manufacturers always used obsolete IC equipment, 2 generations back

•  Today, we are only at 1 generation back •  But, most specialty MEMS equipment, bonders,

DS aligners, KOH, EDP, used to be custom made •  Today, equipment suppliers compete to sell

specialty MEMS equipment

Economics

Starving Markets are desperate . . . .

A Few Examples :

• Ink jet printing HP 1985 • Accelerometer ADI 1992 • DLP TI 1995 • Gyroscope Bosch 1998 • Microphone Knowles 2002 • Oscillator SiTime 2007 • Variable Cap Array WiSpry 2011

What is the Competition?

•  Ink Jet Printing low resolution dot matrix printer

•  Accelerometer steel ball in tube

•  DLP transparent LCD •  Gyroscope expensive quartz assembly •  Microphone electret could not be soldered

•  Oscillator quartz with 50X higher dpm •  Variable Cap Array low Q CMOS capacitors

Rapid Adoption of Knowles Microphone

Mill

ions

of M

EMS

Mic

roph

ones

0

100

200

300

400

500

600

700

800

2003 2004 2005 2006 2007 2008 2009 2010

~2/3 of all cell phones

University/Corporate/Government Open Research Environment

•  Major universities led the way –  Basic research, particularly processes –  Many early basic device demos –  Dedicated Professors –  Classes and curriculum –  Continuing supply of new MEMS students

•  NSF and DARPA funded heavily –  NSF did the initial funding in the early 80’s; still funds –  DARPA has funded over ~25 years

•  Corporate funding of R&D has been extensive –  DRIE invented at Bosch –  Si Fusion Bond developed for MEMS at NovaSensor

MEMS Mechanical Capabilities

•  Anisotropic etching •  Double-sided lithography •  Piezo-resistive sensing •  Anodic bonding •  Sacrificial etching •  Electrochemical etching •  Capacitive sensing; electrostatic actuation •  Fusion bonding •  DRIE •  Vapor HF etching •  XeF2 etching •  SOI •  Wafer bonding with transfer of electrical contacts

Manufacturability

•  DS aligners became commercially available in the mid 80’s

•  DRIE became commercially available in the early-mid 90’s

•  Wafer bonders became commercially available •  Vapor HF etchers became commercially available •  XeF2 etchers became commercially available •  SOI wafers became commercially available •  Extensive modeling capabilities became available •  The semi packaging industry evolved to

accommodate MEMS requirements

MEMS Packaging

Gold thermosonic wire bonds Standoff Stitch Bonds

Kobe

Leadframe

Insulative Die Attach Conductive Die Attach

CMOS Chip

MEMS Chip

CMOS die

MEMS die 0.8mm

Maxim – 2 chips in one package

Maxim buys the chip from SiTime and licenses the signal conditioning circuitry

Will we really get to a trillion?

•  We will need a 100x increase ($) from today •  100x increase for semis took 35 years •  Medical monitoring?

–  7B units nowhere near 1T •  The combination of wireless and environmental

monitoring “could” get us to 1T –  buildings, bridges, planes, lawn-mowers, roads, every

machine and structure on the planet, uninhabited areas •  But, the transformation of the infrastructure

of the planet will take YEARS •  Thumb in the air - 25 years to 1T

MEMS: where are we going?

•  MEMS is now where semiconductors were when the first PC was introduced in 1981

•  In 1981, MEMS was 1.0% of semi shipments •  In 2010, MEMS was 3.5% of semi shipments •  MEMS $ shipments has grown at ~15%/year •  MEMS infrastructure is in place to fuel growth •  Environmental monitoring + wireless will drive

MEMS to a trillion units in 25 years

Thank You