M E M S Tr e n d s
M a g a z i n e
ISSUE n°4
o n
e d itorial
CMOS MEMS,
hybrid
integration,
wafer-level
bonding…
what will
be the future for MEMS
production?
In this issue of MEMSTrends we are giving
some attention to MEMS Technologies
evolution. One big question that arises in the
MEMS world: How will the manufacturing
technologies evolve? Particularly to reduce
cost and provide more integrated devices for
consumer applications. ...
To be continued on page 2
R E V E R S E E N G I N E E R I N G
Three ways of finding magnetic north - a look inside electronic compass chips
Three-axis electronic compasses, along with accelerometers and gyroscopes,
are now being integrated into smart phone and other portable consumer
electronics devices.
The compass provides a definite magnetic reference
direction vector for the accelerometer and gyroscope
sensors. The combination of these three sensors
provides nine degrees of freedom (9DoF) sensing
for applications such as dead reckoning for GPS
positioning during times when the satellite signal
EVENTS
MEMS Technology Summit: Vision for the future
is healthcare and connectivity
P
ioneers of the MEMS industry told the usual
war stories of how it took much longer than
expected to make MEMS devices in volume,
how the competing incumbent technologies
improved faster than the MEMS products could
be developed, and how they got seduced by cool
platinum partners:
M E M S T e c h n o l o g i e s
& M a r k e t s
O C T O B ER 2010
analysis
CMOS MEMS: Next generation integration looks for
simpler build or bond on top, or lower cost TSVs
Like most everything else in MEMS, it sometimes seems every device needs a
different solution for integration with CMOS. But the view from some of the folks
on the forefront is that simpler build or bond on top, and lower cost TSVs, are the
keys to improving next generation cost and performance.
Analog Devices opts for diversity technology. “For us, it’s the way we started. We’ve eked
of solutions every drop we can out of the process and can keep re-
selling product. But I’m not convinced full integration
The pioneer of monolithic integration questions the is the way to go for everyone. We ate, breathed and
industry’s faith that one-chip solutions will necessarily slept all the related issues in the past. I think folks
be lower cost. “There’s a tendency to believe that have underestimated the problems. I still haven’t seen
integrated processes yielding for low margin
2
monolithic could be cheaper, and everyone is talking
about going back to mono,” says Rob O’Reilly, Analog business. Where average selling prices are going,
Devices, Inc. staff engineer for MEMS sensors and there’s no margin to recoup all that R&D.” ...
Printed on recycled paper
R1 R2
R2 R4
is lost, sophisticated gesture recognition and
R1 R3
authentication, gaming, image stabilization and
pedometer functions. The benefits of 9DoF sensing M3 coil
M4 coil
have been recently discussed in an
InvenSense Whitepaper. ...
14 R3 R4
LSM303DLH XY AMR Sensor Die Photograph
C O N T E N T S
Chipworks
ANALYSIS 2
COMPANY v ision 5
MEMS technology to introduce products the market
didn’t want, on the way to creating some of the
sector’s most successful businesses.
ANALYST CORNE R 1 2
But the industry luminaries gathered at the recent
MEMS Technology Summit in Palo Alto, CA,
organized by the founders of NovaSensor for that R e v erse
company’s 25th anniversary, also see a maturing
E n g ineerin g 14
industry that can now get products to market faster,
and that has finally become a real systems business,
16
poised for significant growth in networks
and biomedical applications. ... E V ENTS 1 6
Free registration on
www.i-micronews.com
 OCTOBER 2010 i s s u e n ° 4
M E M S Tr e n d s
M a g a z i n e o n M E M S T e c h n o l o g i e s & M a r k e t s

E D I T O R I A L analysis

We will review several different opinions.

Although they have pioneered the CMOS
MEMS integration, Analog Devices Inc.,
today opts for diversity of solutions. For them,
the question is not only to choose between
monolithic or hybrid integration but to rethink
the partitioning of different functional blocks
among the chips in a multichip module. Here
interconnects are really important and TSV
cost might be an issue.
On the other hand, Dalsa, currently producing
CMOS MEMS, strongly believes wafer-
level integration is a cost efficient solution.
Dalsa uses TSVs technology but also tries
to use the efficiencies of the CMOS into the
MEMS world. They propose standard process
modules that speed up the development of
MEMS devices. SVTC also believes MEMS
on CMOS approach ease the management of
the supply chain.
Analog Devices has a very interesting
perspective on this discussion. Testing is
an important factor that is often over looked
as part of the MEMS development process.
For example, a high- g shake test is required
for qualification in the automotive market,
but the MEMS industry has not developed
the equipment to produce such a test. This
obviously puts a lot of strain on the growing
plug and play products used for stability control,
cruise control and pedestrian protection. All
of which have a lot of MEMS, with a variety
of specs, which in turn creates the need for
MEMS testing at high volume. Is this possible?
Dr Eric Mounier,
Project Manager,
Yole Développement
e v ents
CMOS MEMS: Next generation integration
looks for simpler build or bond on top,
or lower cost TSVs
From page 1
ADI intends to keep using its well established process
for making its MEMS sensors, a relatively complex
interweaving within the CMOS flow, primarily
because it offers a predictable cycle time, as well
as small size and good performance, especially for
automotive applications. But the company is now
making even some of its automotive inertial sensors
as hybrid devices, with two chips in a package,
for more flexibility and faster time to market. The
separate ASIC can be more easily replaced to
add new features or updates, without having to
“We [Analog Devices] ate, breathed and slept all the related
[CMOS MEMS] issues in the past. I think folks have underestimated
the problems,” says Rob O’Reilly.
redesign the whole unit. And the more different
chips in a module, the more useful this flexibility.
O’Reilly argues that hybrid solutions of multiple
chips in one package are especially good for mixed
signal applications that combine multiple different
chip technologies and devices from different fab
locations, as the way to add more features and
functions for more specialized applications.
Besides trying to choose the best monolithic or
hybrid integration approach for each product, ADI is
also focusing on re-thinking how best to partition the
The 4 evolutionary steps for MEMS manufacturing
(Yole Développement – October 2010)
functional blocks among the chips in the multichip
modules, re-distributing the functions for the system
across multiple die in non-traditional ways, by, say,
putting all the high voltage circuitry on a separate chip,
to optimize the system design for cost and performance.
One function it’s including in these systems is self
test, so users don’t have to do their own motion
testing, aimed particularly at industrial markets
where users just expect long term reliability without
special treatment for MEMS. ADI is aggressively
modeling the behavior of its accelerometers and
gyroscopes under all kinds of conditions and
temperatures, then testing them on its homegrown
equipment to verify each axis and quantity and the
interactions. Then the on-board test system keeps
checking performance in use, and lets the user
know if there’s a problem.
How are all the chips in these multi chip modules going
to be interconnected? Alternatives that shorten the
connections to reduce noise and improve performance
and reduce size are needed, but most solutions

 re-use of semiconductor manufacturing lines

 cost can be lower compared to hybrid
 increased performances
 BUT process compatibility is an issue
 XeF2 use

• MEMS Executive Congress,

Scottsdale, AZ, USA:
 Higher design freedom.
November 3 to 5, 2010  More sophisticated devices.

• Electronica, Munich, Germany :  Integration with high-voltage

CMOS (reduced parasitances)
November 9 to 12, 2010  Structure increased vertical
deflection (micro mirrors);
higher design freedom

 New design freedom

• SEMICON Japan, Tokyo, Japan:  imves elecprotrical perf.

 improves yield
 CMOS standard
December 1 to 3, 2010

gold partners:

2
O C to B E R 2010 i s s u e n ° 4
Example of Monolithic Integration
have yet to be demonstrated in serious volumes at
reasonable cost. “That’s a whole blank page,” says
O’Reilly. “Interconnect work is just starting to catch up.
It’s all up for grabs.” He argues that costs of most of the
alternative interconnection technologies, like TSVs, are
too high, and quality and reliability still too unproven in
high volumes, for wide adoption anytime soon, though
ultimately costs will come down. “I’m still looking to
see a viable product released that meets quality and
reliability standards,” he says.
Dalsa sees more complex chips
limiting usefulness of monolithic
integration
Increasingly complex chips are also impacting yields
and costs of monolithic integration, making wafer-level
bonding look a more attractive alternative. “More
and more the trend for the next generation is wafer-
Examples of Package Level Integration
level bonding to combine the MEMS with the control
circuits,” argues Dalsa Corp.’s Donald Robert, VP of
sales and marketing. While the company currently
manufactures a significant portion of its portfolio with
monolithic integration of the MEMS and CMOS, for
“More and more the trend for the next generation is wafer-level
bonding to combine the MEMS with the control circuits,” argues
Dalsa Corp.’s Donald Robert.
the next generation more customers are moving to
more advanced CMOS. That means 25-30 CMOS
mask layers, and adding another 2-25 MEMS mask
layers on top of that quickly becomes unwieldy for
cycle time and yields with anything more than a very
limited number of MEMS mask layers. “Thirty mask
M E M S Tr e n d s
M a g a z i n e o n M E M S T e c h n o l o g i e s & M a r k e t s
layers are too much,” says Robert. “We see more and
more that wafer-level bonding is more cost efficient.”
Dalsa aims to reduce the cost of closer integration
of the MEMS and CMOS wafers by wafer stacking
with lower cost copper TSVs. It’s working with
Alchimer and its wet processes for via filling and
isolation that use fewer process steps than vacuum
deposition to reduce capital costs and increase
throughput. Alchimer has reported work with Korean
tool maker KPM on a tool for use with its materials.
“We believe we will have a cost advantage over
other TSV approaches,” says Robert. Dalsa is now
building a customer product with the process, and
plans commercial launch later next year, offering
the TSV interface as a standard process module.
More standard process modules
Like most producers, Dalsa is trying to incorporate
more of the efficiencies of the CMOS world, by offering
re-usable standard processes whenever possible.
Currently it offers standard process modules, like low
stress film, that enable it to run different products, like,
say, mics and pressure sensors, on the same tool at
the same time. Each customer gets a unique platform
of his particular sequence of process modules,
for which he gets a kit that enables much faster
development of the next generation product.
But more re-usable modules would ease development.
“We often do the same function for different customers
in different ways, but the end result is very similar for
cost and performance,” notes Robert. “If we could
standardize that, it would be much more efficient.”
So the company is also developing standard
building blocks of microvalves and pumps for
microfluidic devices. And it’s been looking at
standard processes for accelerometers, gyros and
pressure sensors, to allow all to be processed in the
same flow, or integrated on the same die. While it is
of course not possible to make all MEMS devices
with a standard process, Robert argues, it is
probably possible to develop a technology platform
that would work for a family of related products, and
half a dozen of these family platforms could suffice
for most types of MEMS devices.
More standard manufacturing means that more of
the value of the device would move from the MEMS
component up the chain to the system supplier,
a change not in the interests of many MEMS
component makers, and thus likely to happen only
slowly. But IC makers entering the field are already
driving this change, asking the foundry for, say, a
quick, cheap MEMS device to replace quartz for
the timing function to add value to their system,
for which a generic product made on a standard
process platform is the ideal option. “They’d like
to get a process design kit from the foundry for the
3
 OCTOBER 2010 i s s u e n ° 4
Wafer Level Packaged MEMS device
with Through Silicon vias. (Courtesy of Dalsa)
MEMS function to add to their IC,” says Robert.
“It’s doable, but we need to look at the economics.
It’s a lot of effort for such a small market.”
SVTC sees more MEMS built
or bonded on CMOS
With its roots in the CMOS world and its customer
base that includes a wide range of new silicon
devices, development foundry SVTC reports
increasing interest among its customers for putting
MEMS on top of CMOS wafers. “What I’m seeing is
more MEMS on top of CMOS, whether built directly
or bonded on top of ASIC wafers from the foundry,”
says Wilbur Catabay, SVTC VP of technology
and engineering. “High volume needs integration
because of cost. If it’s all made in one fab, the
supply chain is easier to manage.”
“What I’m seeing is more MEMS on top of CMOS, whether built
directly or bonded on top of ASIC wafers from the foundry,”
says Wilbur Catabay, SVTC.
Catabay argues that SVTC has been doing this
kind of integration for some time, building new
technologies like magnetic memory, phase change
memory, or 3D anti-fuse memory devices on top
of foundry CMOS wafers. It has been regularly
building MEMS on top of CMOS wafers from
TSMC, UMC and Global Foundries, particularly for
resonator makers, including SiTime, who need the
performance of the shorter circuits, for micromirror
arrays that need pixel by pixel addressing, and
for lab-on-a-chip devices. Complex sensors like
multiaxis inertial sensors and magnetometers units,
however, remain more challenging to integrate.
If there is a major die size mismatch, customers
are choosing die-to-die or die-to-wafer stacking
on the CMOS. This also of course allows the use
of known good die to improve yields, so makes
sense for complex 6-axis devices, especially for
the development and qualification stages. Though
SVTC is using TSVs to connect CMOS to CMOS or
memory to CMOS, it has yet to see much demand
for it for MEMS.
SVTC has developed technologies for aligning its
first lithography steps to the foundry wafers, and
for adding barrier/seed layers to integrate tungsten
interconnect, says Catabay. It coordinates alignment
marking with the foundry, and matches SVTC
scanners with those at the foundry to adjust offsets.
4
Wafers come from the foundries with a top layer of
TaN sealing off the copper, then SVTC adds TiN to
get good adhesion and electrical performance with
the subsequent tungsten or aluminum.
Equipment trends: New demands
for automation, test and etch
As the MEMS sector moves towards higher volumes
it’s making some new demands on its equipment
suppliers, for more automation and for new testing
equipment. More new materials getting into volume
production may start to mean more demand for
more selective etch technology as well.
Rakesh Kumar, director of MEMS for Global
Foundries, suggests higher volume markets will
mean equipment makers will need to develop
MEMS tools with more automation and other
productivity improvements more suitable for high
volume production, as well as tweak their legacy
their 200mm IC tools to make them more suitable
for MEMS. Though MEMS equipment is not a
huge market overall, it is a big part of the market
for 200mm tools, he argues, and says he sees
increasing commitment from tool makers to invest
in MEMS development. Kumar aims to work with
equipment makers to solve some of the challenges
for MEMS automation, like the variability in wafer
thickness, wafer deformation, and wafer handling.
Analog Devices’ Rob O’Reilly challenges MEMS
equipment makers to offer the equipment for high-g
shake tests required for qualification for automotive
markets. “There are no high-g production shake test
systems out there,” he says. “They’re all going into
flipping.” But with the automotive trends to plug and
play, and the burgeoning of units for stability control,
cruise control, and pedestrian protection, there are
lots of MEMS and lots of specs involved. Volumes
are significant and reliability is needed. “Somehow we
need to test all these things,” he notes. “How do we do
that in large volume and still make money?”
The plethora of new materials beyond silicon finally
moving into volume production are starting to boost
demand for the XeF2 etch tools, claims David
Springer, president of Xactix. The highly selective
etch compound does not release its fluorine until it
touches silicon, so doesn’t attack Al, SiN, or metal-
dielectric stacks, making it potentially useful for
applications that use more non-silicon materials,
like RF MEMS, optical and reflective devices,
microbolometers and microphones. Springer also
reports seeing work in piezoelectric and ferroelectric
devices for resonators, pumps and little actuators.
The cost of xenon may be coming down, to ease the
adoption of XeF2 etching, he argues. Air Products
has announced it will supply the compound, as well
M E M S Tr e n d s
M a g a z i n e o n M E M S T e c h n o l o g i e s & M a r k e t s
as offering its xenon recovery service, and a new
supplier in Japan has reportedly announced plans
to offer material as well, for an increased supply
that could bring competition to help bring down
costs and give credibility to the technology.
The technology is still largely used in R&D labs, and
some users suggest it still needs to mature some
more, but there are now more than 100 tools in the
field, and Yole has identified several customers now
using or planning to use it in production of displays
and RF devices.
Paula Doe for Yole Développement
Donald Robert VP, of Sales
& Marketing at DALSA’s
Semiconductor Division
In this role, he is responsible
for the strategic and tactical
activities related to the
development of DALSA’s
semiconductor products &
foundry services. He holds a degree in electronics
from Sherbrooke College and a degree in Business
Management from the University of Sherbrooke and
has over 29 years of industry experience. His areas
of expertise include business development and
semiconductor manufacturing.
Rob O’Reilly,
Member Technical Staff
MEM Sensors,
Analog Devices, Inc.
Rob O’Reilly began his career
at Analog Devices in 1993
and is the former Director
of the Product Test and
Measurement groups for the
Micromachined Products Division. Rob currently leads
the Advanced Test needs for inertial sensors and
as well as developing unique business and market
opportunities. Rob joined Analog Devices in the very
early days of MEMS and has spent the last 17 years
creating test and characterization capabilities for
MEMS accelerometers, gyros and optical devices. A
former flight engineer in the US Navy, Rob attended
Northeastern University with focus in the area of Shock
and Vibration analysis.
Wilbur Catabay,
Vice President Technology,
SVTC Technologies
Wilbur Catabay is a veteran
of the semiconductor industry
with more than 23 years
of experience. Recently,
Mr. Catabay was President
of Silicon Integrated Solutions, Inc. He also was
Senior Director for LSI Logic’s Foundry Engineering &
Integration organization and Director of the Advanced
Process Module Development in the R&D organization.
Mr. Catabay has submitted more than 100 invention
disclosures and has been awarded more than 60
patents with more than 26 patents pending during his
tenure. He has a BS degree in Industrial Technology
from San Jose State University with a minor in
Business Management.
O C to B E R 2010 i s s u e n ° 4
M E M S Tr e n d s
M a g a z i n e o n M E M S T e c h n o l o g i e s & M a r k e t s

company v ision

EVG launches new high performing L3 Bonder

By separating the bonding process into three distinct steps, EVG innovates their bonding
equipment with increasing throughput for high temperature or high vacuum bonding.

Yole Développement: Can you explain to our

readers why EV Group decide to develop such
equipment?
Eric Pabo: In keeping with our Triple I (invent,
innovate, and implement) philosophy EVG decided
to develop an automated wafer bonding system
capable of increased throughput while performing
high temperature and high vacuum bonding.
Also, this wafer bonding system needed to be
capable of being fully integrated into the EVG
Gemini fully integrated cassette to cassette wafer
bonding systems.

YD: How long does it take to develop it? And

did you collaborate with an industrial partner?
EP: We did the development internally based on
the feedback of multiple high volume customers
looking for increased throughput high yield wafer
bonding.

YD: What are the main characteristics of the

L3 Bonder?
EP: The primary advance of the EVG L3 bonder
is the separation of the process into three distinct
steps. The first chamber heats the wafers and
allows pump down of the chamber or pump and
purge to fill the chamber with an inert gas. When
the first step is completed the wafer pair is passed
through a gate valve into the primary chamber.
This primary chamber can be kept pumped
down to the 1e-6 mbar range. In the primary
chamber the bonding process is completed using
force, temperature and time. After this process
is completed the bonded wafer pair is moved
through another gate valve to the third location
where it is cooled prior to unloading.
This process separation can result in a three to
five fold increase in throughput compared to a
single bond chamber. Of course this increase is
dependent on the process being performed.
YD: According to you, what is the added-value
this new bonder?
EP: The added value of the EVG L3 bonder is
increased throughput for wafer to wafer bonding
processes that require high temperature or high
vacuum.
EVG 520 L3 (Courtesy of EVGroup)
YD: Which markets do you expect to address?
EP: EVG expects the EVG L3 bonder to address high
volume production of products requiring vacuum
sealing using metal bonding technologies. We see
that many MEMS manufacturers are switching
to metal bonding (both thermo-compression and
eutectic) from glass frit and anodic bonding. The
EVG L3 bonding system is designed to address
volume production of MEMS devices which have
high hermeticity requirements such as micro-
bolometers, accelerometers, gyroscopes, and other
devices. Integration of the L3 bonding system into
the field proven fully integrated EVG Gemini system
will provide a high throughput, high yield, cassette
to cassette fully automated wafer bonding system.

www.evgroup.com

Eric Pabo is the business development

manager for MEMS for EVGroup,
prior to accepting this position he was
the bonding applications engineer
for North America for EV Group.
Before joining EVG he spent 5 years
working on wafer level packaging and assembly
processes for Agilent Technologies. He has over
20 years experience in electronics manufacturing,
is a professional engineer registered in the State
of Colorado, is finishing his Six Sigma Black Belt
Certification and earned a Bachelor’s Degree in
Mechanical Engineer from Colorado State University.
EVG 520 L3 (Courtesy of EVGroup)

5

company v ision
CMOS players eye MEMS system business
Maturing process technology and tighter integration of MEMS and CMOS
systems are attracting a range of CMOS players to the MEMS business, with
strategies as varied as Global Foundries’ CMOS-compatible standard processes
to Silicon Laboratories’ monolithic integration using SiGe.
A
re MEMS veterans right that the CMOS
folks don’t realize how hard it will be to ramp
MEMS to good yields in high volumes? Or
is the value moving up to the system, with some
MEMS just a commodity component in the IC
makers’ system?
GlobalFoundries targets more
CMOS-like approach to production
GLOBALFOUNDRIES argues MEMS is poised for
big volumes, and for more CMOS-like approaches to
high volume manufacturing, including more standard
process modules. Rakesh Kumar, director of MEMS
for Global Foundries, is emphatic that the company
sees high volume business in MEMS. “We’ve
done a lot of work looking at the market and feel
that this is a turning point for the MEMS business,”
he says, noting all the products just entering the
consumer market and starting growth, with multiple
sensors in all those millions of cell phones and
most other consumer electronics products. “As
volumes increase, costs will decrease, and spur
further growth in volume. That’s a huge number of
wafers that will become a real business for CMOS
foundries.”
The company targets applications that can run
1000 wafers a month or more, starting with
accelerometers, gyroscopes and RF devices, and
focusing on bulk MEMS with engineered SOI wafers
to simplify manufacturing. It’s finishing installing
its infrastructure this year, aiming at starting risk
production with two customers in 3Q 2011. And it
recently announced a partnership with SVTC for
a window in the US for development and to align
manufacturing for porting to the volume fab.
6
Though it has been reported that GlobalFoundries
plans CMOS MEMS, Kumar clarifies that what
the company means by that is CMOS-compatible
manufacturing, by making the MEMS process as
close to CMOS as possible. That means sharing
tools with CMOS where possible to reduce capital
spending, making most of the front end on CMOS
processes first, doing the contaminating MEMS
processes separately afterwards. It also means
making sure the MEMS is compatible with the
CMOS and the WLP, looking at the CMOS design
to understand the issues like thermal mismatch that
impact MEMS system performance.
But it doesn’t mean monolithic integration. Kumar
notes that hybrid integration makes most sense
for low cost consumer inertial MEMS, though high
frequency RF devices may need short interconnects
to avoid parasitics, probably with TSVs, though the
process may be too expensive. Wafer-to-wafer
bonding looks promising if die sizes are similar. It’s
developing a wafer-level packaging module and
may start to go to TSV technology in perhaps 2013,
if customers require it.
Coming from the CMOS world, however,
GlobalFoundries counts on standard process
modules to make production more efficient and
reduce time to market. Kumar argues that most
MEMS production processes are in fact already
becoming fairly standard, and most MEMS makers
already develop a few process modules and then
try to build all new products on that same platform.
“MEMS is not like CMOS where you can go from
design to product in 3-6 months. It won’t happen that
way,” he says. “But new entrants will find it too costly
to enter the business unless they develop on standard
processes, and CMOS foundries can do that.”
(Courtesy of GlobalFoundries)
M E M S Tr e n d s
M a g a z i n e o n M E M S T e c h n o l o g i e s & M a r k e t s
GlobalFoundries is working closely with
Singapore’s Institute of Microelectronics for
development, and getting support from the
government in Singapore, who aims to develop a
local MEMS industry, with research grants and
support of IME’s development fab. A consortium
has also formed to try to bring in standards in design
and packaging to help develop the local industry
infrastructure. Singapore already claims 12% share
of worldwide MEMS manufacturing, though almost
all of that is from the STMicroelectronics plant there.
Silicon Laboratories opts
for monolithic integration
with SiGe MEMS
Coming down firmly for monolithic integration is
Silicon Laboratories, who aims to make a higher
performance MEMS timing product, to add to the
product lineup in its fast growing ~$60 million timing
business. The company has never made a MEMS
device, but prides itself on solving similar integration
problems in building its complicated mixed signal
products in mainstream CMOS processes.
“MEMS timing hasn’t lived up to its promise so far,”
argues Mike Petrowski, GM of the SiLabs timing
products business. “There’s room for improvement
in performance, and customers find comfort in an
established supplier with high volume capacity….
We looked at the technology and considered the
other MEMS structures being used, and decided
the easiest possibility was just to add a few more
layers on to the CMOS wafer, as long as the recipe
worked to grow the structures.” So the company
acquired Silicon Clocks last spring for its relatively
straightforward low temperature SiGe MEMS on
CMOS process to eliminate the bonding wires with
their parasitic capacitance. Monolithic integration
also made sense because the timing products
could potentially be very high volume, and the
programmability already offered in the CMOS
meant there wasn’t much need for flexibility in
switching out the parts.
Petrowski argues that SiGe has good mechanical
properties, but more importantly can be deposited
without high heat on top of the CMOS wafer,
with existing equipment in a CMOS foundry with
some retrofit, but without major additional capital
investment. CTO Emmanuel Quevy has reported in
presentations that the LPCVD deposition process
stays under 425°C and produces a polycrystalline
material similar to polysilicon, with aging of less
than 1ppm per year. He said the process used gold
SiGe bonding with getters to prevent out gassing.
O C to B E R 2010 i s s u e n ° 4
M E M S Tr e n d s
M a g a z i n e o n M E M S T e c h n o l o g i e s & M a r k e t s

SiLabs sees potential for making other products

with SiGe MEMS on CMOS going forward, and
Petrowski notes that the extendablity of the
technology was a major factor in the acquisition.
“We have microprocessors with good analog
converters for access to the analog world—that
start to look a lot like things that use sensors,” he
says. The company also recently acquired CMOS

www.silabs.com
www.globalfoundries.com
Silicon Clocks CMOS MEMS monolithic integration

Michael Petrowski, General
Manager, Timing Solutions - Silicon
Laboratories Inc.
Michael Petrowski serves as general
manager for Silicon Laboratories’
timing solutions, responsible for the
company’s clock generators, oscillators, jitter-attenuating
clocks and other timing products. Prior to joining Silicon
Laboratories, Mr. Petrowski served as a product manager
for Crystal Semiconductor where he was responsible for
the strategic direction of the T1 Line Interface Units and
Ethernet PHY products. Mr. Petrowski also spent more than
11 years at Harris Semiconductor where he served in senior
engineering, applications and marketing roles for DSP,
wireless communications and multimedia audio products.
Mr. Petrowski has co-authored numerous technical articles
and presented at technical conferences on digital signal
processing, frequency synthesis and receiver architectures.
Mr. Petrowski has a bachelor’s and master’s degree in
electrical engineering from North Carolina State University.
Rakesh Kumar, Director of MEMS
program, GlobalFoundries
He received his B.S. (Hons.) and Ph.D.
degrees in electrical engineering from
Punjab Engineering College, India and Nanyang
Technological University, Singapore, respectively. He
is currently Director of MEMS program (200mm
Business Unit) at GlobalFoundries, Singapore.
Prior to this, he was deputy director of Semiconductor
Process Technology Lab at Institute of Microelectronics,
Singapore where he was responsible for MEMS process
development and technology transfer. His areas of
interest include advanced copper interconnects, 3D
wafer level packaging and MEMS technologies. He
has authored and co-authored more than 90 research
publications in journals and conferences.

Status of the MEMS Industry 2010

Growth is back, but only a limited number of players benefit

MaRKEt tREnDs
MEMS market forecast 2009-2015 in $M
“the growth is back, but the industry infrastructure has changed: a limited $18 000

number of companies are taking full benefit of the growth of the market. $16 000

Industry restructuration is now about to happen…,” says Jean-christophe

$14 000 Others (microstructures, micro tips, flow meter …)
Micro speakers
Oscillators

Eloy, Yole Développement.

$12 000 RF MEMS
Microdispensers (microfluidics)
Microfluidics for IVD
$10 000 Microfluidics for Research
Other optical MEMS
Projection systems
$8 000 Micro displays
Microbolometers
Digital Compass
$6 000 Gyroscopes

KEY FEatuREs
Accelerometers
Microphones
$4 000
Pressure Sensors
InkJet Heads

the objectives of this report are to provide: $2 000

TOP 30 MEMS Company ranking 2009-2008

• The expected evolution of the MEMS markets applications by
$0
2009 2010 2011 2012 2013 2014 2015

application
• A long term vision of MEMS markets 2008-2020
• Technology and business trends at equipment and materials level
• The evolution of the MEMS foundry business, the MEMS production
equipment business and the MEMS packaging YOLE DÉVELOPPEMENT

• An analysis of the strategies of the TOP 30 MEMS manufacturers

• An analysis of the business trends in the major 15 applications areas of
MEMs devices
Y O L E D É V E L O P P E M E N T

contact us
For more information, feel free to contact David Jourdan:
tel: +33 472 83 01 90, Email: jourdan@yole.fr Y O L E D É V E L O P P E M E N T

7
 W o r ld ,
r th e e d
All ov e
o n n e c t
s ta y c r b u s i n e s s . . .
rs, you
re a d e
Ou r

YOLE DÉVELOPPEMENT

In-depth Market & Technology Analysis

Quarterly Magazines
Combined reach to over 15,000 subscribers
Y O L E D É V E L O P P E M E N T

Become a sponsor today!

Contact: Editorial, Advertising & Subscriptions leroy@yole.fr / www.i-micronews.com Y O L E D É V E L O P P E M E N T
OCTOBER 2010 i s s u e n ° 4
M E M S Tr e n d s
M a g a z i n e o n M E M S T e c h n o l o g i e s & M a r k e t s

company v ision

The MEMS pioneer VTI relies on its proprietary

3D MEMS technology
VTI Technologies can be considered as a pioneer in MEMS for the past 20 years.

T
he company is a leading supplier of acceleration,
inclination and angular motion sensor solutions
for transportation, medical, instrument and
consumer electronics applications. VTI develops and
produces silicon-based capacitive sensors using its
proprietary 3D MEMS (Micro Electro-Mechanical
System) technology.
In 2009 VTI was the first MEMS company to adopt
Wafer Level Packaging in the world’s smallest and
least power consuming three-axis acceleration
sensor, the CMA3000, and the company has already
announced that it will launch new MEMS solutions at
Electronica 2010.
Mr Anssi Korhonen, VTI Chief Technology Officer,
was interviewed for the MEMS Trend Magazine.
CMA3000 (Courtesy of VTI)
Yole Développement: VTI is one of the very few
MEMS companies using a Through-Glass Vias
technology for its 3-axis accelerometer. Why
using glass wafers instead of Si?
Anssi Korhonen : “We are actually using a silicon
wafer and molten glass material for isolation of
TSVs. Benefits of the VTI cap wafer technology
include good insulation and very low parasitic (stray)
capacitance. Glass, on the other hand, provides
planar surface and reliable bonding interface to the
structural wafer. Also, glass is very inexpensive
starting material”, Mr. Korhonen explains.
YD: There are different ways to do TGV. What
makes the VTI technology specific?
AK: “The process is VTI proprietary technology. We
avoid using plating processes in forming the vias. It is
compatible for wafer level processing although needs
some specific equipment. Currently we are satisfied
with the via resistance in the tens of ohms range.”
YD: Is VTI Technologies planning to use its TGV
technology to other MEMS devices? Moving to
different wafer size?
AK: “The technology in its initial form (planar
isolation and one via) has been in use since 1984.
In the late 90’s due to requirements by multi-axis
accelerometers and gyros we added the capability
for a multitude of vias. More recently this technology
has been developing for finer pad pitch and size by
utilizing dry etching of silicon instead of mechanical
machining. The process is scalable for larger wafer
sizes. It is used for all VTI MEMS designs.”
YD: VTI has recently achieved the smallest
accelerometer on the market (2x2 mm²). Do you
plan to go even smaller?
AK: “Smallest size components can be achieved with
the Wafer Level Packaging (WLP) technology, which
is close to WLCSP technology that has received
wide acceptance in the market. VTI WLP goes one
step further by flip chip attaching ASIC on the MEMS
sensing element.”
“Further size reduction is possible and restricted to
specific MEMS or ASIC design requirements, not so
much on packaging technology”, Mr. Antti Korhonen
concludes.
www.vti.fi
Mr. Anssi Korhonen, M.Sc.
in electrical engineering, has
worked as Chief Technology
Officer for VTI Technologies
since 2008.
He has worked for electronics
manufacturing services industry
since 15 years.
He has held various management positions in
manufacturing, account management, business
development and general management. Early in
his career Mr. Korhonen worked as a development
engineer in packaging and assembly as well as a
research scientist of MEMS technologies.

APM on MEMS manufacturing and production

High volume MEMS mass production has been the mission for APM since it was founded in 2001.

A
PM currently ships tens of thousands of wafers
to customers in a variety of MEMS devices.
APM, as an independent pure-play foundry,
accomplishes high productivity with two success
factors: responding to customers’ manufacturing
needs and investing resources on the development
of versatile process modules and platforms.
SOI process platform, for example, has long been
extensively recognized advantageous to many MEMS
devices both on performance and manufacturability.
In response to this increasing demand for SOI
MEMS, a great deal of APM’s internal resources has
been placed into the area in order to bring varieties
to SOI platform. APM has accumulated extensive
process experiences on making complex structures
on the SOI of very thin to thick device layer, multiple
device layers, or the one with embedded cavities.
Fine combs and flexures, two-step structures, multi-
step well-aligned structures have been achieved in
APM and qualified by numerous customers. Today,
the SOI platform is being expanded into the Silicon-
on-Glass process platform, which makes this
platform more considerate for different customer
requirements.
Temporary wafer bonding and debonding are the
other technologies under development in APM.
They are considered to greatly enhance the flexibility
of manufacturing sequences, and allow handling
thin wafers through processes. APM also see their
huge potential in wafer packaging technology. Other
techniques that catch our eyes are, for example, dry
release and monolayer coating. They play a grate role
when talking about device reliability. APM prioritizes
these techniques in the roadmap while working closely
with customers on their needs.
APM believes it can stand out in the MEMS
manufacturing competition through proactive mindset
and rigorous production discipline on customers’
needs, and, as a result, assists them in staying
competitive in their MEMS product races. Deep Silicon Ecth on cavity SOI wafer
with high verticality
www.apmsinc.com

9
 OCTOBER 2010 i s s u e n ° 4
M E M S Tr e n d s
M a g a z i n e o n M E M S T e c h n o l o g i e s & M a r k e t s

company v ision

Compugraphics continues to build on its

expertise in making MEMS photomasks
Compugraphics has been making photomasks for over 40 years and today provides top-class photomask solutions
for MEMS companies and research institutes all over the world.

P
hotomask technology has been used since
manufacture began of the earliest semi-
conductors over 40 years ago. It was then
that Compugraphics made its first photomask in
Aldershot, UK. In 1970 the company moved north
to Glenrothes, Scotland, to be closer to the multitude
of technology companies emerging in the area that
became known as "Silicon Glen". Compugraphics
has continued to expand and is today one of the
most highly regarded and established Photomask
names in Europe and USA. It has an experienced
team of photomask professionals to rival any of its
global competitors with similar sized operations in
Glenrothes, Scotland and Los Gatos, California.
Photomasks are high precision plates made out of
quartz or soda lime glass with a chrome layer on
one side, usually containing microscopic pictures of
electronic circuits etched into the chrome. Whilst
the technology has advanced dramatically since
the start, the principal remains the same with the
photolithography process now also an integral part
of the production of many MEMs and Optoelectronic
devices. Customers send their design data to
Compugraphics, who then write the masks using
sophisticated laser and e-beam exposure tools and
finish them in a Class 1 clean room environment.
With the growth in MEMs and Optoelectronic devices
over the past 10 years, Compugraphics has seen its
customer base expand to include these market sectors.
This has presented just a few challenges. MEMs and
Optical applications often use structures where smooth
curves and accurate radius is critical for example.
Conventional writing tools used to make photomasks Quartz photomask during the production process
are predominantly designed to write Manhattan or
straight-line geometries for the IC sector. on meeting the needs of emerging markets like
MEMS, not leading edge IC market needs. We have
Dave Muir, Customer Interface Manager, explained a reputation for quality, reliability and flexibility as
the key to Compugraphics’ success in this area : well as a wealth of experience that can only benefit
our customers."
"Our people have been trained in MEMs design to
ensure a better understanding of our customers’ Compugraphics is owned by OM Group and
needs. We know what is important to ensure that their certified to ISO 9001 and ISO 14001.
design will translate into an application that works.
Our business model is unique: The MEMs market
place is our market place, our best and most
experienced people have been steadily working www.cgi.co.uk

IMU & High Performance Inertial MEMS 2011

Complete review of inertial sensors market 2009-2015

MaRKEt tREnDs Application Space for IMU

Application Space for IMU & High Performance Inertial MEMS
& High Performance Inertial MEMS
… “the market for IMus is estimated to be $1.55B in 2009, dominated by
defense and aerospace applications. Yole sees this market growing at a Industrial,

9% annual growth rate, to reach $2.60B in 2015…” says Laurent Robin, • Agriculture
• AUVs
• Freight transport ship
civil naval
and offshore
Defense
• Defense ships
• Defense transport aircraft

Market analyst, MEMs, Yole Développement.

• Defense UAVs
• Healthcare IMU & High • Guided munitions
• High speed train
• Inclinometers Performance • LAV/Artillery Guns
• MAV/Tanks
• Oil drilling heads
• ROV Inertial MEMS • Military & special mission helicopters
• Military fighters
• Satcom antenna stab
• Military submarines
• Stabilization of optical systems
• Nuclear missiles
• Survey instruments
• Short, medium and long range missiles
• UGVs
• Vibration monitoring
Commercial • Soldier
•…
•… Aerospace
• Business Jets
• Civil aircraft
• Civil helicopters
• Civil and paramilitary UAVs
• General aviation

2009-2015 global market for high-performance IMUs

• Satellites

KEY FEatuREs
• Spacecrafts & skyrockets
•…

Breakdown by field in $M (Sept. 2010 Yole Développement)

the objectives of this report are to provide:
• Market data on high performance Inertial Measurement Units and MEMS
accelerometers / gyroscopes: key market metrics & dynamics
• Application focus on key existing markets and most promising
emerging ones YOLE DÉVELOPPEMENT

• Deep understanding of inertial sensor value chain, level of integration

& players

Y O L E D É V E L O P P E M E N T

contact us
For more information, feel free to contact David Jourdan:
tel: +33 472 83 01 90, Email: jourdan@yole.fr Y O L E D É V E L O P P E M E N T

10
OCTOBER 2010 i s s u e n ° 4
company v ision
Plan Optiks next step in MEMS cap wafer technology:
LED packaging
Plan Optik AG, the leading Cap wafer producer for wafer level packaging applications has launched it’s fully
equipped 200 mm cap wafer production line for LED packaging wafers recently.
T
his new line is capable of making glass and
glass-silicon cap wafers mainly for high
power LED packaging as well as packaging
wafers for projection technology.
Light emitting diodes (LEDs) require hermetic
packaging which protects them from environmental
impacts such as humidity, heat and dust. Standard
packaging methods sometimes struggle with
these requirements since they are typically based
on polymer materials which do not fulfil long term
reliability requirements. Plan Optik’s LED packaging
wafers consist of glass and silicon only - both
materials are widely used in the MEMS industry
for packaging solutions. This makes these LED
packaging wafers suitable for automotive, medical
and projection applications since they work
fine under harsh environmental conditions. The
temperature resistance is up to 400 centigrade.
The used borosilicate glass provides an excellent
transmission of the LED light whilst the silicon
typically acts as a spacer. Anodic bonding of the
glass lid and the silicon spacer creates a cavity
for a single or multiple LEDs whilst an optional
double side anti reflection coating increases the
transmittance to more than 97%. Due to anodic
bonding spacer and lid glass are coupled in a non-
reversible way - even under extremely challenging
conditions such as permanent temperature and
humidity cycles as often found in the automotive
industry.
Plan Optik is already running a mass production
of such LED packaging wafers which have been
developed together with one of the biggest LED
manufacturers world wide and the packed LEDs
are already implemented in large executive cars.
Demans are dramatically increasing due to the fast
implementation of high power LEDs in cars.
Plan Optik produces cap wafers from glass as well
as glass silicon compound wafers since many years.
They can be bonded to device wafers in various
ways (mainly anodic and adhesive bonding). Core
M E M S Tr e n d s
M a g a z i n e o n M E M S T e c h n o l o g i e s & M a r k e t s
Plan Optik’s new 200 mm cap wafer line
processes used by Plan Optik are grinding, polishing The recent capacity expansion includes investments
(CMP), (ultra sonic) drilling, sand blasting, wet etching in latest grinding, CMP, anodic bonding equipment
and bonding glass and glass-silicon compound and automation of existing processes.
substrates. Materials used are borosilicate and
alkaline free glass as well as silicon and silica. Plan Optik, a public company located in Germany
produces wafers for MEMS applications such as tire
Combining the big variety of processing techniques, pressure monitoring systems and drug dispensing
Plan Optik also offers cap wafers with optical solutions since almost 20 years and can refer to a
(transparent) cavities, through holes for conductive large experience in providing such wafers to almost
connection as well as high accuracy blank glass all key players in the MEMS industry.
substrates for various MEMS packaging tasks.
These wafers are clean room suitable and are Plan Optik’s ISO TS 16949, ISO 14001 and ISO
already implemented in many high end sensor 9001 certifications ensure the quality all substrates
applications. are produced to.
All wafers produced by the latest production
technology are characterized by low ttv, thickness
tolerance, low roughness and high surface quality.
Plan Optik set the benchmark in respect to virtually
perfect glass wafer surfaces by introducing it’s
proprietary development MDF polishing to the
market. MDF (Micro Damaging Free) polished wafers
are suitable for wet etching processes and avoid
the well known issue sub surface damaging which
can lead to defects. Sub surface damaging leads
to small cavities and interconnections between the
etched structures, weakens the bonding interface
200 mm LED packaging (cap) wafer
and results in a low yield. The use of MDF polished
wafers leads to reliable wet etching results and
provides a high wafer processing yield. www.planoptik.com
By implementing this new LED packaging wafer
production line Plan Optik has increased it’s capacity
for cap wafers for wafer level packaging by 80% - in
addition to the already existing 150-300 mm blank Carsten Wesselkamp,
Sales Manager of Plan Optik AG
and drilled glass wafer production capacity.
Mr. Wesselkamp serves as the
international sales manager of Plan
Plan Optik is reacting to the rapidly increasing
Optik AG, the technology leader in
demand for MEMS cap wafers since late 2009 –
the production of structured cap
mainly for sophisticated glass-silicon compound
wafers for MEMS applications in various industries.
substrates for wafer level packaging of optical MEMS Plan Optik’s head quarter is based in Elsoff near
(MOEMS) such as CMOS image sensors, projection Frankfurt, Germany. The company is listed in the
technology and LED applications as well as carrier Entry Standard at the Frankfurt stock exchange under
substrates for semiconductor wafer handling and ISIN DE000A0HGQS8.
processing.
11
 OCTOBER 2010 i s s u e n ° 4
M E M S Tr e n d s
M a g a z i n e o n M E M S T e c h n o l o g i e s & M a r k e t s

A N A L Y S T C O R N E R

MEMS technology roadmap: Demand for smaller, lower cost

devices drives major technology trends for next decade
MEMS may still be largely a one product, one process business, but customers across diverse applications share a
common set of demands for smaller devices with better performance at lower cost.

A
nd that will drive the industry towards new applications where users are willing to But another technology also starting
assembly solutions for tighter integration pay the higher cost to get the better to change the integration game is
and more standard process modules over performance and smaller size from wafer-level bonding or active capping,
the next decade. Yole Développement sees most the shorter connections. Yole projects bonding the ASIC wafer to directly to
common applications shrinking to 1mm2 or 2mm2 demand for MEMS with 3D TSV will the MEMS wafer as a functional cap.
die or smaller by 2020, and price points continuing reach several hundred thousand InvenSense has led the way with its
on a sharp downward path, driving demand for 3D wafers a year by 2015. unique metal-to-metal bonding of the
TSV, active capping, thin film capping, and CMOS ASIC to the MEMS with AlGe, which
MEMS technologies going forward. Cost and Early uses of the TSV technology are efficiently and compactly makes both the
time-to-market pressures will also favor increased as varied as microphones from Sonion connection and the hermetic seal for its
reliance on standard process modules.... and fingerprint readers from IDTECK/ Dr Eric Mounier, consumer gyroscopes. But other sensor
STMicroelectronics. MEMS foundries Project Manager, makers are expected to develop wafer-
Yole Développement
Demand for smaller die, closer Dalsa, Silex and Xintec are running level bonding approaches as well.
integration drives new assembly the 3D process for various applications
that include gyros, microphones and oscillators Also coming into wider use to shrink device size
solutions
for customers like InvenSense, VTI, Epcos, and and cost is thin film capping, where the wafer
SiTime. Yole has indentified at least 20 companies with released MEMS devices is sealed under an
The biggest change coming down the pipeline is
developing TSV for MEMS, spread across the value encapsulation layer. Though the high temperature
chip stacking with through-silicon vias. This 3D TSV
chain from MEMS foundries, packaging houses, encapsulation process developed by Bosch with
technology is now starting production for those few
and IDMS. Stanford University was originally seen by some
as a step towards building CMOS on top of MEMS,
—the process has turned out to be most useful for
Yole Développement has identified at least 20 companies wafer-level encapsulation to shrink die size. The
developing TSV for MEMS, spread across the value chain thin film cap layer eliminates the need to devote
from MEMS foundries, packaging houses, and IDMS. wafer surface area to all the framing frit lines
usually used for bonding on the glass or silicon cap

New MEMS manufacturing processes drivers

(Yole Développement – October 2010)

12
OCTOBER 2010 i s s u e n ° 4
M E M S Tr e n d s
M a g a z i n e o n M E M S T e c h n o l o g i e s & M a r k e t s

wafers, making it particularly useful for ultra small
devices like oscillators and RF MEMS. SiTime is
using the thin film capping for its oscillators, and
Bosch is now using the technology in a least one
of its accelerometer products. MEMtronics uses a
different thin film capping process on its RF MEMS
devices, while RFMD, WiSpry, ASE and CEA Leti
are all also developing capping films.
More CMOS-MEMS and standard
process modules coming too
These alternative TSV and capping technologies get
some of the gains of shorter electrical connections
between the MEMS and ASIC of the CMOS MEMS
approach without having to deal with the many
complexities of monolithic integration. But real one-
chip solutions –with the MEMS layers made before,
during or after the CMOS layers—may further
improve performance and reduce size and cost, in
part by using the existing CMOS infrastructure and
avoiding the cost of separate assembly of the MEMS
with the ASIC. Though of course wire- bonded, two-
chip system-in-a-package solutions offer lower costs
MEMS ASIC Integration
(Yole Développement – October 2010)
Pros & Cons for the different ASIC MEMS integration approaches
(Yole Développement – October 2010)
Pros Cons
CMOS First • IC foundry can be used
• Thermal budget is an issue
• Chip area may be minimized
• Microstructure topography is an issue
• Electronics and MEMS cannot be
easily stacked
MEMS First • No thermal budget for MEMS
• IC foundries are worried to accept
pre-processed wafers
(materials constraints)
• More control over materials,
• Need your own fab
processes
• Cost of increased chip area
• Optimize/compromise mechanical
• Mechanical properties of CMOS layers
and electrical components
Interleaved MEMS compromised
• Economical, reliability
• Supply chain issue: the MEMS
and yield high
technology is linked to the CMOS
• Simple post processing step
technology (that evolve much quicker!).
releases MEMS
in other cases, especially for small volumes, as well to control each pixel individually will require CMOS
as greater design flexibility in using different chips. MEMS, as do micromirror digital light processors
Going forward, all MEMS array devices that need or microbolometerswhich build the MEMS directly
above the CMOS. The very small geometries
in nano MEMS chemical sensors will also likely
require monolithic integration. A lineup of startups
has also turned to some flavor of CMOS MEMS in
recent years in hopes of disruptive advantage in
lower costs and faster volume ramps, from Akustica
and MEMSiC to Silicon Clocks and now Baolab.

The growing use of MEMS’ own standard process

modules may, however, also start to challenge
CMOS-MEMS’ purported advantage in established
processes and faster ramp to volume. As the
industry has matured, both the IDMs and the
MEMS foundries have built up repertoires of well-
characterized process blocks that can be used again
in other devices. Back end processes modules for
things like wafer-level packaging and TSVs were
the easiest and first to be applied across products,
but the approach is also now being extended to
things like silicon membranes. CEA-Leti and Silex
are most vocal about the advantages of standard
modules, but we hear a similar story from most of
MEMS A&M, fund raising 2010 the other leading MEMS foundries, and from IDMs
(Yole Développement – October 2010) like STMicroelectronics, who forces almost all its
MEMS devices into one of two established process
Company Round type Acquired Amount raised
company USD flows.

Calient Technologies First round - $10M

Additional Funds
Mezmeriz - $50,000
phase 1
DelfMEMS VC funds - $1.9M
CardioMEMS Equity Financing - $60M
Dr Eric Mounier, Yole Développement
Microvisk Financing round - $3.95M*
Measurement Pressure Systems Dr. Eric Mounier has a PhD in microelectronics from
Acquisition $25M
Specialties Inc. the INPG in Grenoble. Since 1998 he is a cofounder
Microvision Equity Financing - $12M of Yole Developpement, a market research company
based in France. At Yole, Dr. Eric Mounier is in
Microstaq Series B - $10.2M charge of market analysis for MEMS, equipment &
Silicon Labs Acquisitio Chip Sensors - material. He is Chief Editor of Micronews, a bi-monthly
magazine dedicated to micro and nano technologies.
*£2.5M converted in USD on 27/10/2010

13
 OCTOBER 2010 i s s u e n ° 4
R e v erse E n g ineerin g
Three ways of finding magnetic north - a look inside
electronic compass chips
From page 1
A
variety of technologies are being used to
create electronic compass devices. Hall
Sensors constitute the dominant technology,
with AKM being the major manufacturer of
devices for consumer applications. MEMSIC and
STMicroelectronics offer devices based on the
anisotropic magneto-resistance (AMR) effect,
while Achi Steel offers devices based on the
magneto-impedance (MI) effect. At the present
time, Chipworks has seen AKM Hall Sensor based
devices in several downstream consumer products.
Chipworks has analyzed the AKM AK8974 from the
Nokia N97, the AK8973 from the iPhone 3G and the
AK8975 from the recent iPhone 4. We have done
analysis of samples of the Achi Steel AMI302 found
in a Ball It “blobo” gaming device. In addition, we have
analyzed samples of the MEMSIC MMC3120XM
and the STMicroelectronics LSM303DLH electronic
compass devices. In this article we will focus on
an example of each technology type, namely the
AK8975 Hall, the AMI302 MI and the LSM303DLH
AMR electronic compass devices.
AKM AK8975 Hall-Effect Sensor
The AKM devices are based on integrated CMOS
Hall Sensor technology. Figure 1 shows an X-ray of
the AK8975 three-axis electronic compass, found in
the iPhone 4. The device is packaged as a 2.0 mm
x 2.0 mm x 0.6 mm thick wafer level chip scale BGA
package. Creating a three-axis electronic compass
using planar CMOS technology is challenging,
since the Hall Sensor structures are only sensitive
to the vector component of the magnetic field
perpendicular to the die surface. Thus, a critical
feature of AKM’s three axis electronic compass
technology is the use of a “magnetic concentrator”,
which serves to bend field lines parallel to the die
surface to give a perpendicular component that
can be sensed by the integrated Hall Sensors.
The location of the 0.33 mm diameter magnetic
concentrator, among the array of solder balls, is
annotated on the X-ray.
CHIPWORKS
0.48 mm
0.33 mm
magnetic
concentrator
Figure 1 AK8975 Package X-ray
14
Chipworks
die
solder
ball
magnetic concentrator
500 µm
Figure 2 AK8975 Package Cross Section
The magnetic concentrator can be seen in cross-
section in Figure 2. The cross-section shows the
upside down AK8975 die and cross-sectioned
through three solder balls. The concentrator is
formed using a circular pad of FeNi alloy, known
as “permalloy”, which has very high magnetic
permeability. This pad was deposited and patterned
over a die coat layer.
Hall S ens or
Y logic
X X
magnetic
Y
concentrator
CHIPWORKS
Figure 3 AK8975 Die Photograph
The decapsulated AK8975 die is shown in Figure
3. The location of the magnetic concentrator is
annotated in pink. Separate Hall Sensors structures
are located beneath the four quadrants of the
magnetic concentrator. Operation of the AK8975
depends on deconvolution of the signal from
each of these four structures. A field vector in
the horizontal direction parallel to the die surface
will give maximum differential signal for the two
sensors labeled X, while a field vector in the vertical
direction will give maximum differential signal for
the sensors labeled Y. A field perpendicular to the
die surface (Z-direction) will result in an equal signal
from all four sensors. The direction of the magnetic
vector can thus be extracted from the signal level
for the four structures. The electronic compass
sensor on the AK8975 die also features a coil, which
likely serves as a solenoid to provide a reference
magnetic field, perhaps for calibration purposes or
for demagnetizing the magnetic concentrator.
A detailed optical view of the left X Hall Sensor
structure is provided in Figure 4. Each structure is
actually comprised of two separate Hall Sensors,
M E M S Tr e n d s
M a g a z i n e o n M E M S T e c h n o l o g i e s & M a r k e t s
C hipworks
Hall
S ens or
s olenoid
coil
50 µm
Figure 4 AK8975 Hall Sensor Device
each of which is formed using an N-well in the
P-substrate. Current flows diagonally across each
sensor. A magnetic field perpendicular to the die
surface will result in a measureable voltage between
the two contacts perpendicular to the current flow,
due to the Lorentz force.
Achi Steel AMI302
Magneto-Impedance Sensor
The three-axis Achi Steel AMI302 was found in
the “blobo”, a novel gaming device. It is the first MI
electronic compass that Chipworks has seen in a
consumer product. The “blobo” also contains an
accelerometer and a pressure sensor, both from
STMicroelectronics. Figure 5 shows an X-ray of the
AMI302 package. The device contains an ASIC plus
three MI sensors, one each for X, Y and Z sensing.
Decapsulation of the AMI302 revealed a signal
processing ASIC with RICOH die marks, indicating
that Ricoh fabricated the ASIC for Achi Steel.
Chipworks
Y
AS IC
Z
X
Figure 5 AMI302 Package X-ray
Figure 6 shows a cross-section through the Y-axis
sensor in the AMI302. The sensor is comprised
of an iron-cobalt amorphous ferromagnetic wire
surrounded by a copper coil. The operation of a
MI sensor is described on the Aichi Steel web site
and by K. Morie, et al. Essentially, a high frequency
alternating current is passed through the FeCo wire.
O C to B E R 2010 i s s u e n ° 4
M E M S Tr e n d s
M a g a z i n e o n M E M S T e c h n o l o g i e s & M a r k e t s

Chipworks of the sensor, H. If we consider a device located M2 "barber pole"

shorting strip
within an external magnetic field vector B, then the
Al M2 artifact

F eC o amorphous ferromagnetic wire

magnetizing field inside the material is given by H interconnect
90 µm
= B/µo – M, where µo is the permeability of free
930 µm
space and M is the magnetization of the material. AMR strip

410 µm
Y -s ens or
It is worth noting that in free space, outside of current
flow
9.3 µm
a magnetic material the B and H are the same,
differing only in their units and magnitude. 2.0 µm Al M2

current
The operation of AMR sensors has been nicely 10 µm flow

described by S. Andreev and P. Dimitrova. In AMR

Figure 6 AM302 Y-Sensor Cross Section
The impedance of the wire changes significantly in
response to an external magnetic field aligned along
the wire, due to the “skin effect”. The phenomenon
is known as the Giant Magneto-Impedance (GMI)
effect. The signal is picked up by the copper coil,
which apparently detects the imaginary part of the
MI effect.
STMicroelectronics LSM303DLH
Anisotropic Magneto-Resistance
Sensor
The STMicroelectronics LSM303DLH is based on
Honeywell AMR technology. STMicroelectronics has
publicly partnered with Honeywell for the fabrication
of the device. The device also incorporates an
ST accelerometer. Figure 7 shows a plan-view
package X-ray of the LSM3030DLH, annotated with
the location of the various chips. Decapsulation of
the device revealed the accelerometer to be the
same MEMS and ASIC die combination used in
the LIS331DLH. The package also contained two
AMR sensing chips (one for XY and one oriented at
90° for Z) and an AMR signal processing ASIC. The
AMR sensing chips had Honeywell die markings.
Figure 8 shows a photograph of the XY AMR sensor
found in the LSM3030DLH. The die features two
independent Wheatstone Bridge structures, one for
X and one for Y axis magnetic sensing, color coded
red and yellow in the annotations.
pin 1
XY-magnetometer
accelerometer
and ASIC
magnetometer ASIC
ferroelectric materials the resistance varies with a
cosine-squared dependence on the angle between
the direction of current flow and the direction of
the magnetizing field vector. The resistance has
a maximum value for current flowing, in either
direction, along the vector of magnetization.
Application of a transverse magnetic field B
perpendicular to a stripe of magnetized AMR
material will rotate the magnetization vector H
resulting in a reduction (ΔR) in the resistance of
the strip. The sensitivity can be increased by the
application of “barber pole” strips of high conductive
material onto the magnetized ARM stripe, which
rotates the direction of current to 45° with respect to
the magnetization direction to the point of maximum
slope on the ΔR curve. Furthermore, the sign of ΔR
then depends on the polarity of the field B.
R1 R2
R2 R4
R1 R3
M3 coil
M4 coil
R3 R4
Chipworks
Figure 8 LSM303DLH XY AMR Sensor Die Photograph
The “barber pole” strips used on the LSM3030DLH
can be seen in Figure 8 for each of the eight
Wheatstone Bridge resistor structures on the
sensor die. A more detailed SEM micrograph of the
AMR layer in the LSM303DLH is shown in Figure
9. The AMR strips are magnetized lengthwise by
a metal 4 coil on the die. Cross-sectional analysis
showed that the metal 1 AMR film to be composed
of FeNi permalloy, while metal 2 was composed of
Al and TiW.
C hipworks 10 µm
Figure 9 LSM303DLH XY Sensor “Barber Pole”
Strips on the AMR Film
Concluding Comments
The earth’s magnetic field strength ranges from 0.3
gauss to maximum of 0.6 gauss, in areas near the
magnetic poles, such as in northern Canada. Table
1 shows the range and resolution provided by the
three commercial magnetic compasses discussed
here. The compasses all appear to provide sufficient
sensitivity for measurement of the earth’s magnetic
field, with the AK8975 apparently providing the best
overall range and resolution. The AKM sensor also
has the advantage of being a single chip CMOS-
based solution, and hence likely is the lowest cost
to produce. The earth’s magnetic field strength is
weak compared to the overall sensitivity rangeof
these devices, which likely explains why the AKM
sensors do apparently required fairly regular
calibration, as described on the AKM Electronic
Compass Calibration web site.
Related Chipworks Reports
• AKM AK8973 and AK8974 3-Axis Electronic
Compass Process Review (MPR-0910-801)
• AKM Semiconductor AK8973 3-Axis Electronic
Compass Circuit Analysis of Analog Blocks (CAR-
1001-801)
•AKM AK8975 3-Axis Electronic Compass Exploratory
Report (EXR-1007-804)
•MEMSIC MMC3120M Tri-Axis Magnetic Sensor
Process Review (MPR-0907-801)
• STMicroelectronics LSM303DLH 3-Axis Accelerometer
and 3-Axis Honeywell Magnetometer Sensor
MEMS Process Review (MPR-1002-801).
For references please contact the author.
St.J. Dixon-Warren - Chipworks Inc.
sdixonwarren@chipworks.com

Z-magnetometer
Chipworks
Figure 7 LSM303DLH Package X-ray Manufacturer Part Number Type Range Resolution ADC
AKM AK8975 Hall Sensor ±12 gauss 0.003 gauss 13 bit
Understanding AMR compass devices requires STMicroelectronics LSM303DLH AMR ±8.1 gauss 0.004 gauss 12 bit
a more careful consideration of the physics of Achi Steel AMI302 MI ±10 gauss 0.01 gauss analog
magnetism. Magnetic sensors are sensitive to the
magnetic field strength vector inside the material Table 1 Magnetic Sensor Range and Resolution

15
 OCTOBER 2010 i s s u e n ° 4
e v ents
MEMS Technology Summit:
Vision for the future is healthcare and connectivity
From page 1
B
ut the industry luminaries gathered at the
recent MEMS Technology Summit in Palo
Alto, CA, organized by the founders of
NovaSensor for that company’s 25th anniversary,
also see a maturing industry that can now get
products to market faster, and that has finally
become a real systems business, poised for
significant growth in networks and biomedical
applications.
The hyper connected socially networked world,
connecting people for tasks in entirely new ways,
is one of the major technology trends driving all
disruptive innovation, argued Kaigham Gabriel,
deputy director of DARPA, the US defense
research agency whose job it is to foresee the
unexpected. “Every person and every object is
going to be connected in 3-5 years,” he noted.
The agency recently challenged groups to find ten
big red weather balloons tethered across the US,
and found that groups used social networks and
Twitter to find them quite efficiently, with the winning
team recruiting some 5000 members to locate all
balloons in about nine hours. The second major
disruptive technology trend is the democratization
of innovation, so many more people will be involved
in creating new things, much like open system
software or IC design tools enable complex systems
to be developed by many more designers in much
less time. “I am most excited by the potential of
applying semiconductor design and fab technology
to things other than semiconductors,” says Gabriel,
noting the potential of simplifying the design of
electromechanical systems from planes to toasters.
“We think the internet of things will be a huge market,”
concurred Horst Muenzel, regional president,
Robert Bosch LLC Research and Technology
Center, citing figures from UC Berkeley of 7 trillion
devices to serve 7 billion people worldwide in 2017.
Lots of enthusiasm for future
of bioMEMS
Asked what the likely topic for the 50th anniversary
of MEMS would be, serial entrepreneur Kurt
Petersen offered “implantable things.” Benedetto
Vigna, STMicroelectronics group VP and GM,
MEMS, sensors, and analog also saw bioMEMS
as the next wave “In my opinion, we are now at the
dawn of an new age,” argued Vigna, pointing to a
market for MEMS in, on and around the body, for
sensing, drug delivery, and cell and fluid handling.
ST is working with partners on a small, wearable
electrocardiogram sensor, a biochip that measures
lactic acid, nozzles for drug dispensing, and
disposable contact lenses with pressure sensors to
test for glaucoma.
16
M E M S Tr e n d s
M a g a z i n e o n M E M S T e c h n o l o g i e s & M a r k e t s
Others outlined a staggering array of the desired range for several days.
biomedical products coming out of the Patients monitored with the device
lab. Neural probes of micromachined had 38% fewer hospitalizations
arrays of silicon are in use for research, than those getting the current gold
while stents with sensors in their walls and standard of care. But Allen also
gas analyzers to identify breath markers warned of the hurdles involved in
for TB just need further integration and making devices stable enough to
miniaturization, reported U. of Michigan last with minimal drift in the very
professor Ken Wise. Janusz Bryzek, now harsh environment of the body for
CEO of his sixth startup Jyve, noted that a decade or more. “The average
Kurt Petersen,
the smart phone was emerging as an $40M in MEMS development costs is
President
enabling platform for potential MEMS- of KP-MEMS about right, but you have to double
based medical tools, from ultrasound that for medical approval,” he noted.
imaging to breath analysis for disease. However, this second company
product took only five years to
CardioMEMS’ wireless condition monitor develop, half the time of the first.
for heart failure is now out of clinical
trials and ready to be submitted for Major suppliers all also saw plenty
FDA approval, reported Georgia Tech of new applications for their core
professor and company co-founder Mark technologies. Besides HP’s MEMS
Allen. High pressure puts stress on the data storage technology turned
heart and leads to heart failure, but if it to ultrasensitive accelerometers
can be monitored, it can be controlled for wireless sensor nodes, Avago
with the proper mix of medication, Technology is looking at using its
Benedetto Vigna,
reducing the need for hospitalization and General Manager,
FBAR technology for oscillators and
potentially saving some $15 billion a year MEMS Prod. Div., nano particle detectors. Knowles
in direct medical costs. The company’s ST Microelectronics is looking at using microphones as
capacitive pressure sensor with wireless sensors for sounds other than voice,
antenna is inserted near the heart with though still sees a big run up in
a catheter --which takes on average only seven voice applications. “I’m not a futurist,” noted Pete
minutes--then monitors blood pressure and sends Loeppert, VP of R&D at Knowles Electronics. “Back
results to a wireless scanner. Doctors can be in 1988 I predicted I could make a microphone in a
notified by cell phone when data remains outside year.” After ten years he did create a product, but
Status of the BioMEMS industry
(Yole Développement, BioMEMS report, Oct. 2010)
O C to B E R 2010 i s s u e n ° 4
M E M S Tr e n d s
M a g a z i n e o n M E M S T e c h n o l o g i e s & M a r k e t s

it was too expensive for the consumer market. The

first sale finally came in 2002. Now the company
has sold more than 1.5B units, and is shipping ~2M
units a day. Loeppert projects growing demand for
noise cancellation will push demand up fast from
the current 1.3 mics per smart phone, to a potential
market of perhaps 1 billion units.
All these growing applications will generating almost
a $50 billion MEMS market by 2020, supporting a
$1.7 billion market in equipment, suggested Yole
president Jeff Perkins. However, by 2015 very few
MEMS devices will cost more than a dollar, and
only some of the dozens of companies competing in
some of these niches will survive.
Paula Doe for Yole Développement
Heard at MEMS Technology Summit
MEMS needs to move beyond MEMS.
While DARPA used to put a significant budget
into funding MEMS development each year, it no
longer does so directly, but instead funds MEMS
as part of a lot of other programs, from navigation
to thermal management to 3D interconnect,
pointed out Stanford professor Thomas Kenny,
just retired from the US defense research
organization. “MEMS may be starting to hold us
back,” he suggested, noting the need now for
more interdisciplinary exchange and synergies.
“Maybe we should end some MEMS conferences,
then the contents of the meeting would diffuse to
other meetings,” he proposed to stir up discussion.
“It’s taken 40 years, but we have finally crossed
a tipping point to a systems business,” said
professor Ken Wise of the University of Michigan.
The new electronics takes us into health,
transportation, security and energy systems. I
always sort of wanted to be a DRAM designer
because they were mainstream, but today MEMS
is mainstream.”
MEMS needs to escape the trap of one process,
one product, one application-the negation of all
the experience of CMOS, argued ST’s Benedetto
Vigna, noting that the company worked three years
on a new gyro technology, but then scrapped it
because it made more sense to manufacture gyros
with the same process as the accelerometers. It’s
working on moving its pressure sensors to the
same process as well.
Cars with MEMS electronic stability controls
reduce fatalities by 40%-50% wherever
introduced. All thanks to a Swedish journalist
testing cars for that important feature of moose
avoidance who famously rolled a Mercedes,
creating the market for MEMS gyroscopes in ESC
systems, pointed out by just about everybody.
A MEMS-based spectrometer developed by
Polychromix, now Thermo Fisher Scientific,
analyzed the composition of a plume of debris
ejected from a crater on impact to show that there
is water on the moon. Details are out in the Oct 22
issue of Science.
There’s a corollary to the rule that the typical
MEMS company goes through 3-4 CEOs before
shipping its first commercial MEMS product:
The typical MEMS company founder starts 3-5
different companies.
Paula Doe for Yole Développement

BioMEMS 2010
Microsystems devices driving healthcare applications
MaRKEt tREnDs
Trends in IVD:
•Major trend towards point

“the microsystem technologies market for healthcare applications will of care testing through
integration of disposable
biosensors in portable

grow from $1.2 B in 2009 to $4.5 B in 2015, representing over 1 B units Pharmaceutical research
Industrial and environmental testing
devices
•Combination of
biosensors and dispensing

per year in 2015. In the Meanwhile wireless systems will exceed 50%
Analytical devices devices f or theranostics
•Increase sensor density to

Dispensing IVD Diagnostic enable multiplexing and

screening

market share,” says Frédéric Breussin, Project Manager, Microfluidics.

Kidney Dialysis Safety
Trends in medical:

KEY FEatuREs •Development of BioMEMS

lead to a technology shif t
towards “home Care”
through miniaturization and
the BioMEMs 2010 report is a robust analysis of the Micro Devices with integration, f rom portable
to implantable
•Combination of diagnostic
the most advances to develop solutions for vital bio-medical applications.
via Bio Sensors and
treatment such as drug
delivery systems will lead
to improved patient
Microsystem devices for healthcare applications
Monitoring market 2009-2015 in $M
the devices considered are: monitoring Respiratory
5000
Infusion devices Point of care
Comfort
Trends in Home Care:
•Home care consists today
Blood Monitoring
Surgery , imaging,
Camera pills
Patient monitoring
mainly in monitoring of
some basic parameters,
Home Care such as temperature, blood
Medical
pressure, glucose, comf ort
Cardiac Monitoring and activity.
•Development towards
Activity Monitoring portable technology will
lead to a transf er of post-
Oxygen concentrators operation monitoring
Capnographs Hearing aids
towards home care

• Pressure sensors 4500 Medical devices Ventilator and anesthesia

SEPTEMBER 2010 - BIOMEMS REPORT
•Improvements of sensors
and bio-sensors will lead to
improved patient

• Silicon microphones Home Care monitoring and prevention

4000 In Vitro Diagnostics

• Accelerometers
Pharmaceutical and biological Research
3500

• Gyroscopes 3000

• Optical MEMS and Image sensors 2500

• Microfluidic chips 2000

YOLE DÉVELOPPEMENT

• Microdispensers for drug delivery 1500

• Flow meters
1000

500

• Infrared temperature sensors 0

• Emerging MEMS (RFID, Strain sensors, Energy Harvesting) 2009 2010 2011 2012 2013 2014

Y O L E
2015

D É V E L O P P E M E N T

contact us
For more information, feel free to contact David Jourdan:
Y O L E D É V E L O P P E M E N T
tel: +33 472 83 01 90, Email: jourdan@yole.fr

17
 OCTOBER 2010 i s s u e n ° 4
M E M S Tr e n d s
M a g a z i n e o n M E M S T e c h n o l o g i e s & M a r k e t s

About Yole Développement

Beginning in 1998 with Yole Développement, we have grown to
become a group of companies providing market research, strategy
consulting, teardown analysis, media and finance services. With
a solid focus on emerging applications using silicon and/or micro
manufacturing Yole Développement group has expanded to include
more than 40 associates worldwide covering MEMS, Microfluidics,
Compound Semiconductor, Power Electronics, Photovoltaic,
Advanced Packaging and Nanomaterial. The group supports
companies and investors worldwide to help them understand
markets and follow technology trends to develop their business.

SERVICES
• Market research and market data
• Technology analysis

SOLUTIONS FOR
• Market workshop
• Strategy consulting
• Corporate Finance Advisory

PUBLICATIONS
• Collection of market & technology reports
• Players & technology databases and market data
MEMS
• Manufacturing cost simulation tools
• Component reverse costing analysis
More information on www.yole.fr
PROCESSES
MEDIA
• Critical news, Bi-weekly: Micronews, the magazine
• In-depth analysis & Quarterly Technology Magazines: MEMS Lithography, spray coating,
Trends Magazine – 3D Packaging Magazine – PV Manufacturing top/bottom alignment
Magazine - Efficien’Si
• Online disruptive technologies website:
www.i-micronews.com
• Exclusive Webcasts
Nano imprint lithography and
• Live event with Market Briefings hot embossing

CONTACTS
For more information about : 3D integration and wafer level
• Yole Développement: Jean-Christophe Eloy (eloy@yole.fr) packaging
• Publications: David Jourdan (jourdan@yole.fr)
• Media activities: Sandrine Leroy (leroy@yole.fr)

Editorial Staff
Managing Editor: Jean-Christophe Eloy
Editor in chief: Dr Eric Mounier
Editors: Docteur Eric Mounier, Laurent Robin,
Jérôme Mouly, Paula Doe
PR & Media Manager: Sandrine Leroy
Assistant: Camille Favre
Production: Atelier JBBOX

18