ISSCC€2012€Tutorial Getting€In€Touch€with€MEMS: The ... · Aaron€Partridge Getting€In€Touch€with€MEMS: 7 of€70 The€Electromechanical€Interface Example€Applications

ISSCC 2012 Tutorial

Getting In Touch with MEMS:The Electromechanical Interface

Dr. Aaron [email protected], Corp.

February 19, 2012

mailto:[email protected]

Aaron Partridge 2 of 70Getting In Touch with MEMS:The Electromechanical Interface

Overview

These slides accompany the 2012 ISSCC Tutorial,Getting In Touch with MEMS: The ElectromechanicalInterface.

The tutorial is written for practicing IC engineers andstudents. No MEMS background is needed.

The goal is to expand the attendee’s potential role fromcircuit designer to system designer. From “Here is theMEMS device, design the interface circuit.”into “Here isthe problem, define an optimal solution.”


Outline

o MEMS Materials, Processes, and Example Applications

o Electrical Interfaces

o Scaling Laws

o Packaging is Critical

o CMOS Integration

o How to Succeed

o References




o Scaling Laws


o CMOS Integration

o How to Succeed

o References


Materials

o Standard Semiconductor Materialsn Silicon (single crystal and poly).n Oxide (thermal and deposited).n Nitride.n Aluminum.

o Unusual Materialsn Gold, various other metals.n Piezoelectrics (AlN mostly).n Plastics (e.g. SU8).n And then just about anything else.


Processes

o Early in MEMS many unusualetches were common.

o Now standard fab processare preferred when possible.

o A few special processesn Bosch etch.n HF vapor etch.n Oxide plasma release.n Xenon difluoride (XeF2)

release.

o Deep etches are common.

Tuning fork resonator, Bosch 2003

S. Pourkamali, F. Ayazi, 2004


Example Applications

o MEMS will find its way into practically every application.

o Right now, it is strong inn Automotive (pressure, acceleration, rotation).n Consumer (acceleration, rotation, time).n Industrial and Military (pressure, acceleration).n Medicinal (pressure, biological sensors).

o Future hot apps will ben Medical, for diagnostic tools.n Timing, to replace quartz.n RF Filters, switches, etc.n Inertial, to sense motion of all types.


Pressure Sensors

Sensimed intraocular pressure sensor in contact lens


Accelerometers & Gyroscopes

Freescale accelerometer


Microphones

Akustica microphone


Light Modulators & Projectors

Two pixels in a TI DLP mirror array


Resonators & Oscillators

SiTime oscillator


RF Switches

G. Rebeiz UCSD, RF switch




o Scaling Laws


o CMOS Integration

o How to Succeed

o References


Capacitive Overview

o Capacitive transduction is used in 90% of MEMSinterfaces.

o Good Points:n Easy to build, no need for special materials.n With the Bosch etch we can make beautiful cap structures.n Can move small to large distances.n Can move inplane and outofplane.n Can sense tiny displacements.

o Bad points:n Often will not deliver as much force as desired.n Needs bias voltage, sometimes large.n Output signals can be very small.


Capacitive Drive

o How does capacitive drive work?Take a parallel plate example:

o Were F=force, 0=permittivity,w=width, h=height,g=capacitive gap, V=voltage.

o The voltage squared givesattractive forces and drivenonlinearity.

o The gap squared givesdisplacement nonlinearity.

222

VgwhF cε

=

V

g

h

w

F


Capacitive Drive

o Want bipolar force?

o We can offset the attraction (pullmore and pull less) with DC biasand AC drive.

o Set Vbias >> Vdrive and we get abipolar offset drive.

o As bias is increased and drive isdecreased the linearity improves.

22 )(

2 drivebiasc VVgwhF +=

ε

drivebiasc

offset VVgwhF 2

ε≈

g

h

w

Vbias+Vdrive

F


Capacitive Drive

o Second option for bipolar isdifferential (pull one way,pull the other),

o Offset and differential can becombined,

o Typical bias and drive are 5Vand 0.5V.

( )2222 rightleft

cdif VV

gwhF −=

ε

( )leftrightbiasc

dif VVVgwhF −≈ 2

ε

g

h

w

Fdif

Vleft

g

Vright


Capacitive Drive

o Interdigitated fingers(combs) can move furtherand are more linear.

o N is the number of fingers.

o Since p does not effect Fcit is linear in displacement.

o Pairs of fingers can beused differentially tolinearize V and pushpull.

2VghNF c

difε

=

V

g

h

p

Fc

V


Capacitive Sense

o For capacitive sensing, weneed to think about charge,

o We all learned that,

o But for MEMS sensing wesometimes care more about,

o And we use a bias Vdc, oftenabout 5V but can be 100’s!

dtdQiCVQ /, ==

)/( dtdVCi =

)/( dtdCVi dc= i

cap C

Any structures with dC/dxwork. Fingers are common.


Capacitive Sense

o What you will need to do as an engineern Because capacitances are small, sense currents are small.n Design the lowest noise sense amps possible.n If noise is not critical then shrink the MEMS.n Always push the circuits, always simplify the MEMS.

o Drive Circuitsn For AC system (gyros, vibrometers, oscillators) we need to

sense AC current.n For DC systems (accelerometers) we need to modulate a

carrier.n Clasic accelerometer drives a differential signal on plates

and measures current with a lockin amplifier.


Capacitive Sense

o Differential lockin sense amp for accelerometers.

OUTA

phasetrim

diffosc

sensor

Aaron Partridge 23 of 70Getting In Touch with MEMS:The Electromechanical Interface M. Dugger, Sandia Labs


Piezoresistive Overview

o Transduces strain to resistance.o One of the earliest MEMS interfaces and still important.

o Good points:n There is mechanical gain, typically about 30x.n The common sensor structure is a Wheatstone bridge.n Siliconfriendly fabrication, doped resistors work well.

o Bad points:n Main problem is temperature sensitivity –moderately

doped silicon resistors change about 0.5% per C or more.n 1/f noise and drift can be problematic.n Only senses, does not drive.


Piezoresistive Sense

o A simple idea with may coefficients–strain changes resistivity.

o =change in resistivity, arepiezo coefficients, are stress.The form a sparse 6x6 matrix.

o For specific cases the equation canbe simplified to,

∑=

=∆ 6

1λλωλ

ω σπρρ

R

axialeffectiveRR σπ=

∆



o Typical sense circuits are bridges.n This minimizes the temperature sensitivity that can swamp

signals.n Input offset often vital –use switched caps, diversity, etc.n Often must minimize 1/f noise –use switched topologies.n Temperature compensation of offset and gain variation is

often needed.

o See: A.A. Barlian, WT. Park, J.R. Mallon Jr., A.J.Rastegar, and B.L. Pruitt, “Review: SemiconductorPiezoresistance for Microsystems”, Proceedings of theIEEE, v.97, n.3, March 2009.



o Bridge amp with temperature offset and gain correction

OUTA

OUT

Vbias

offsettrim

gaintrim

tempsense

sensor

Aaron Partridge 28 of 70Getting In Touch with MEMS:The Electromechanical Interface A. Partridge, Stanford


Piezoelectric Overview

o Transduces force to voltage.o Aluminum Nitride (AlN) is the most common material.

o Good points:n Moderately easy to fabricate, available in MEMS foundries.n Can provide low impedance and tremendous power

handing for RF.n Works well at high frequencies.

o Bad points:n Displacements are tiny, so not generally used for motion.n Rather low Q’s when used in resonators.n Does not transduce DC signals.


Piezoelectric Drive and Sense

o For sense and drive, think interms of S21 and S12.

o In the simplest form,

o Where t=thickness,e=dielectric permittivity, anddbar is the piezoelectriccharge coefficient (a functionof material and orientation).

Fewh

tdV =

V

h

w

F

t

Aaron Partridge 31 of 70Getting In Touch with MEMS:The Electromechanical Interface G. Vigevani, UC Berkeley


Other Transducers

o Thermal sensors, particularly thermistors, are often usedfor IR imaging bolometers.

o Chemical sensors of all kinds are used in biology.

o There are lots of optical transducers, and this is animportant area for displays. Optical forces can even beused to drive MEMS, but that is rare.

o Magnetic transducers are common in macrosystems butdon’t work well in micro. They are rare.

o There are endless other ways to transduce signals.


Remember This:

You will usually need to design thelowest noise circuits possible. If theMEMS is producing extra signal thenit should be simplified.




o Scaling Laws


o CMOS Integration

o How to Succeed

o References


Scaling Laws

o Most things scale against us, not for us.

o Mechanical structuresn Volume and mass: x3 (simple)n Mechanical stiffness: x (extensional)n Resonant frequency: 1/x (extensional)

o Transducersn Piezoresistance: (no scale)n Capacitance: x (voltage to force)n Piezoelectrics: x (voltage to force)n Magnetics: x4 (current to dipole torque)n Optics: (wavelength limits)


Why Does Scaling Matter?

o Most MEMS things get worse when made smallern Mass goes down cubically, so inertial sensing gets tougher.n Capacitive and piezoelectric transducers get worse linearly.n Magnetic transducers scale terribly.

o A few things get bettern Circuits can be mounted closer, so Cstrays decrease.n Resonant frequencies increase.n Reliability improves.n And (the most important) unit costs decrease.

o Poor scaling is counterintuitive for circuits engineers.


Remember This:

Know how your system scales andleverage things that work for youand not against you.




o Scaling Laws


o CMOS Integration

o How to Succeed

o References


Packaging is Critical

o In circuits we don’t think about packaging, except forn Size.n Lead inductance and resistance.n Heat dissipation.

o In MEMS, packaging is the single most important thingafter the transducer selectionn How do we protect the parts in operation?n How do we handle the parts in packaging?n Can we dice the parts from the wafers?n How do we connect to the sense/drive medium?n How do we isolate from the environment?n A million problems happen here.


Backend Packaging

o Sometimes just put the MEMSand CMOS into a package.

o Don’t touch it!

o Production complicationsn How to dice?n How to pick & place?n Need a clean room?

o For some apps, like chemicaldetectors, it can work well.

ADI ADXL50 circa 1995


Bonded Covers

o Wafer bonded covers protectthe MEMS elementsn Can use frit glass to glue the

wafers together.n These covers can take 80%

of the die area.n Building the covers can be

expensive.

o Much less expensive thenhandling naked MEMS wafers.

o The dominant technologytoday.

Bosch Gyroscope circa 1999


Deposited Thin Film Covers

o Save space by depositingrather than bonding covers.

o Harder than it looksn How to empty it out inside?n Thermal mismatches.n Contamination.n Need strength for plastic.n Limits MEMS designs.

o Development is expensiven Only makes sense for high

volume applications.J.L. Lund, Hilton Head, 2002

B.H. Stark, JMEMS 2004


Deposited Encapsulation Example

o SiTime (my company) buriesresonators under the wafer surface.

SiTime Resonator


Stacked Die in Plastic

o When the MEMS iscovered or encapsulatedn It can be diced like a

CMOS wafer.n Pickplace, wire bond,

mold, etc, all normal.

o A low cost option when itworks for the app.

o Applications like opticalswitches can’t use this. SiTime Oscillator Construction


Remember This:

Design from the outside to the inside,and do the packaging first.




o Scaling Laws


o CMOS Integration

o How to Succeed

o References


Homogeneous Integration

o Only three waysn MEMS on CMOS.n MEMS in CMOS.n MEMS under CMOS.

o Usually not the best approachn Do you have the time and money?n Do you feel lucky?

o Just a few of the problems…n The MEMS process can change the CMOS behavior.n The CMOS process may need to be adjusted.n Price leverage of CMOS at risk.n MEMS process limitations are severe.n Wasted area increases costs for each process.n Yield issues multiply (literally).


Homogeneous Example

3Axis Accelerometer, Sandia


Heterogeneous Integration

o Usually a better approachn Build the MEMS and CMOS die into one package.n Can be wire bonded or bump bonded.n Requires the MEMS be precovered for most apps.

o Usually cheaper, faster, and more versatilen MEMS+CMOS can be developed in parallel.n Fewer material and temperature restrictions on the MEMS.n Shorter development times.n Many fewer surprises.

o Be prepared to argue against a homogeneous approachn It seems obvious, everything is getting integrated right?n Actually, heterogeneous integration is very common in RF.


Heterogeneous Example

Oscillator, SiTime


Waferscale Integration

o MEMS wafer bonded to CMOS wafern The wafers are bonded faceface at the MEMS fab.n Various proprietary metallurgies at MEMS fabs.n Common are gold compression and eutectics.

o Has some good pointsn Electrically close –low capacitance.n Many interconnects possible.n Usually a decent and hermetic lid.n Built in parallel, a full wafer at a time.

o Not always a good idean Enforces a 1:1 size constraint on MEMS:CMOS die.n Usually wastes space on one or both die.n Yield losses multiply.


Waferscale Example

Rich Ruby, Avago, IFCS 2011

MEMS Wafer

Circuits Wafer


Remember This:

Never integrate MEMS on, in, or underCMOS without a very compelling reason.




o Scaling Laws


o CMOS Integration

o How to Succeed

o References


Problems to Manage

o You will need to build something earlyn The MEMS components will not be finished when you start

your IC design.n You will not have solid data for your analysis.n If prototypes have been tested, the new parts (the real

parts!) will be different.n When the MEMS is tested it will show “surprises”.

o You will need to make your own circuit modelsn The MEMS components will not modeled for you.n At best you will have theory or Matlab code.n Spice? VerilogA? CppSim? You will need to build them

yourself.


Problems to Manage

o Specs will change in the middle of developmentn There are three, maybe four things in flux.n Circuits, MEMS design, MEMS process, and software.n On the applications side, marketing and sales learn too.n Rates of flux are higher than you may be used to.

o To improve the chances your circuits workn Understand the system in detail.n Understand the MEMS as much as possible.n Design circuits with as much flexibility as possible.n Insist on rapid prototyping cycles with test chips.n Don’t go for product too early, it can waist design effort.


Important Points

o Be sure you have a MEMS expert on the teamn Hire one or consult with one, but get one.n Amateurs who think they are smart enough are wrong.

o Understand the application driversn How does the scaling work?n Does it require integration?

o Consider the packaging carefullyn Design from the outside to the inside.

o Be sure an IC engineer (you) are involved earlyn System design requires balancing circuit tradeoffs.n Does a good circuits person know the details?


Important Points

o Build flexibility in your ICn The MEMS will not work as expected, so roll with it.n The MEMS will be improved later, and the earlier IC should

support that.

o Insist on many small learning cyclesn Use test chips to prove out the concepts.n Don’t go into product design too early.

o Take a bigger role than IC designern You must be the guardian of your success.


Aaron’s Complexity Paradox

o The “Simple”solutionsare difficult.

o The “Complex”solutionsare easy.

o Move toward easysolutions wheneverpossible, this meansincreasing complexity.

o Embrace complexity tosucceed.

Material

Device

Analog

Digital

Microcode

Firmware

Software

Sim

ple

Com

plex

Easy

Difficult


Aaron’s Complexity Paradox

o For system designers this meansn Push the problem out of MEMS whenever possible.n Don’t ever make the MEMS designers do something that

one can possibly do elsewhere.

o For IC designers this meansn You will be (or should be!) asked to push your circuits to

the limit to ease the work in MEMS.n Working with MEMS will never be easy –you will always

need to bring your best game.


Complexity Management Example

o MEMS oscillator researchers (including me!) tried to buildfrequencyaccurate temperatureflat resonators.

n That didn’t work.

o Then my colleagues and I moved the complexity intoCMOS and managed the frequency with fracN PLLs.

n That worked!

o Then all sorts of great things started happening –wecould include new functions in that CMOS.


Complexity Management Example

o SiTime Architecture.o Will step through this

block by block.

CMOS Oscillator

MEMS Resonator

GND

ProgOE/ST

Vcc

ClkTemperatureSensor

ConfigurationPROM

FrequencyControl

I/O Regs

FracNPLL

SustainingAmp

Drive


Choose the Right Products

o Must have a Starving Market. (Kurt Petersen)o Must have an Unfair Advantage. (Arno Penzias)

o Why discuss this in a circuits tutorial?n Because we engineers and scientists can almost always

make our stuff work.n But failure happens when people don’t want our stuff.n Good engineering, when not needed, is wasted effort.

o We don’t have time to waste, we are very expensive forour companies.

o And besides, we have better things to do than engineerstuff that people don’t need.


Remember This:

We are valued not by what we do, butby what we do that makes a difference.


Your Work Does Make a Difference

o Technology hasfundamentally helpedHumanity.

o Engineers make smallcontributions that aremultiplied countlesstimes.

o What you do matters toBillions of people!

Dorothea Lang, “Migrant Mother”1936


Thank You!

Questions?




o Scaling Laws


o CMOS Integration

o How to Succeed

o References


References

o A complete background to MEMS and thorough basic references

n Gregory Kovacs, “Micromachined Transducers Sourcebook”, McGrawHill Science/Engineering/Math, ISBN 0072907223 , 1998.

o A good general introduction of scaling and technology options

n Marc Madou, “Fundamentals of Microfabrication, The Science ofMiniaturization”, CRC Press, ISBN 0849308267, 2002.

o A deep dive into RF MEMS and systems

n Gabriel Rebeiz, “RF MEMS: Theory, Design, and Technology”, WilleyInterscience, ISBN 9780471201694 , 2004.


References

o A solid theoretical underpinning of common MEMS devices

n Ville Kaajakari, “Practical MEMS: Design of microsystems,accelerometers, gyroscopes, RF MEMS, optical MEMS, and microfluidicsystems”, Small Gear Publishing, 0982299109, 2009.

o The primary conference proceedings and journals

n Hilton Head, “SolidState Sensors, Actuators, and Microsystemsworkshop”, Transducers Research Foundation.

n Transducers, “International Conference on SolidState Sensors,Actuators and Microsystems”, IEEE.

n MEMS, “International Conference on Micro Electro MechanicalSystems”, IEEE.

n JMEMS, “Journal of Microelectromechanical Systems”, IEEE.


Additional References

Capacitive transduction

o W.C. Tang, TC.H. Nguyen, M.W. Judy, R.T. Howe, “ElectrostaticComb Drive ofLateral Polysilicon Resonators”, Sensors and Actuators A: Physical, v.21, pp.328331, 1990.

o A. Selvakumar, K. Najafi, “Vertical Comb Array Microactuators”, J.Microelectromechanical Systems, v.12, pp.440–449, 2003.

o H. Hammer, "Analytical Model for Comb Capacitance Fringe Fields", J.Microelectromechanical Systems, v.19, pp.175182, 2010.

o L. Prandi, C. Caminada, L. Coronato, G. Cazzaniga, F. Biganzoli, R. Antonello, R.Oboe, “A LowPower 3Axis DigitalOutput MEMS Gyroscope with Single Driveand Multiplexed Angular Rate Readout”, ISSCC 2011, pp.104106, 2011.

Piezoresistive transduction

o A.A. Barlian, WT. Park, J.R. Mallon, A.J. Rastegar, B.L. Pruitt, “Review:Semiconductor Piezoresistance for Microsystems”, Proceedings of the IEEE, v.97,n.3, 2009.

o Y. Kanda, “A Graphical Representation of the Piezoresistance Coefficients inSilicon,”IEEE Transactions on Electron Devices, vol.29, n.1, pp.6470, 1982.



o F.N. Hooge “1/f Noise Sources,”IEEE Transactions of Electron Devices, v.41,n.11, pp.19261935, 1994.

o J.A. Harley, T.W. Kenny, “1/F Noise Considerations for the Design and ProcessOptimization of Piezoresistive Cantilevers”, J. Microelectromechanical Systems,v.9, pp.226235, 2000.

o L.M. Roylance, J.B. Angell, “A Batch Fabricated Silicon Accelerometer,”IEEETransactions on Electron Devices, ED26, n.12, pp.19111917, 1979.

Piezoelectric transduction

o W. G. Cady, “Piezoelectricity; An Introduction to the Theory and Applications ofElectromechanical Phenomena in Crystals”, McGrawHill, 1946.

o R. Ruby, P. Bradley, J. Larson, Y. Oshmyansky, D. Figueredo, “UltraMiniatureHighQ Filters and Duplexers Using FBAR Technology”, ISSCC 2001, p.120121,2001.

o G. Piazza, P.J. Stephanou, A.P. Pisano, “One and Two Port Piezoelectric ContourMode MEMS Resonators for Frequency Synthesis”, SolidState Device ResearchConference ESSCERC 2006, pp.182185, 2006.

o R.L. Kubena, F.P. Stratton, D.T. Chang, R.J. Joyce, T.Y. Hsu, M.K. Lim R.T.M’Closkey, “MEMSBased Quartz Oscillators and Filters for OnChip Integration”,International Frequency Control Symposium, p.6, 2005.



Examples of less common drive and sense technologies

o C.H. Liu, A.M. Barzilai, J.K. Reynolds, A. Partridge, T.W. Kenny, J.D. Grade, H.K.Rockstad, “Characterization of a HighSensitivity Micromachined TunnelingAccelerometer with Microg Resolution,”IEEE Journal of MicroelectromechanicalSystems, v.7, n.2, pp.235244, 1998.

o A.M. Leung, J. Jones, E, Czyzewska, J. Chen, B. Woods, “MicromachinedAccelerometer Based on Convection Heat Transfer”, International Workshop onMicro Electro Mechanical Systems, pp.627630, 1998.

o A. Rahafrooz, S. Pourkamali, “Fully Micromechanical PiezoThermal Oscillators”IEEE Int. Electron Devices Meeting, pp.7.2.17.2.4, 2010.

o M. Lutz, W. Golderer, J. Gerstenmeier, J. Marek, B. Maihofer, S. Mahler, H.Munzel, and U. Bischof, “A Precision Yaw Rate Sensor in Silicon Micromachining,”International Conference on Solid State Sensors and Actuators, , v.2, pp.847850, 1997.

o Y. Li, J. John, X. Zhang, J. Zhang, J.A. Loeb, X. Xu, “3D Neural Probes withCombined Electrical and Chemical Interfaces”, SolidState Sensors, Actuators,and Microsystems Workshop, Hilton Head, pp.134137, 2010.



MEMS waferlevel packaging

o Y.T. Cheng, L. Lin, and K. Najafi, "Localized Bonding with PSG or Indium Solderas Intermediate Layer," Twelfth IEEE International Conference on Micro ElectroMechanical Systems, pp. 285289, 1999.

o C.H. Tsau, S.M. Spearing, and M.A. Schmidt, "Fabrication of WaferLevelThermocompression Bonds," Journal of Microelectromechanical Systems, vol. 11,pp.641647, 2002.

o C.M. Mastrangelo and R.S. Muller, "VacuumSealed Silicon MicromachinedIncandescent Light Source," Proceedings of the International Electron DevicesMeeting, pp.503506, 1989.

o M. Esashi, S. Sugiyama, K. Ikeda, Y. Wang and H. Miyashita, “VacuumSealedSilicon Micromachined Pressure Sensors,”Proc. IEEE, v.86 pp.1627–1631, 1998.

o K.S. Lebouitz, A. Mazaheri, R.T. Howe, and A.P. Pisano, “Vacuum Encapsulationof Resonant Devices Using Permeable Polysilicon,”Twelfth IEEE InternationalConference on Micro Electro Mechanical Systems. MEMS'99, pp.470475, 1999.

o B.H. Stark, K. Najafi, “A LowTemperature ThinFilm Electroplated Metal VacuumPackage,”Journal of Microelectromechanical Systems, v.13, pp.147157, 2004.

o A. Partridge, A.E. Rice, T.W. Kenny, and M. Lutz, "New Thin Film EpitaxialPolysilicon Encapsulation for Piezoresistive Accelerometers," 14th IEEEInternational Conference on Micro Electro Mechanical Systems, MEMS 2001,pp.5459, 2001.

o M. Esashi, “Wafer Level Packaging of MEMS”, J. of Micromechanics andMicroengineering, iopscience, v.18, 2008.


End

Documents

ISSCC€2012€Tutorial Getting€In€Touch€with€MEMS: The ... · Aaron€Partridge Getting€In€Touch€with€MEMS: 7 of€70 The€Electromechanical€Interface Example€Applications