Document 180086

93
Sensors md Actuators A, 37-38 (1993) 93-105
Capacitive sensors: when and how to use them*
Robert Puers
Katholreke Unrversltelt Leuven, Departement Elektrotechmek, ES A T -M I CA S , Kardmaal Merclerlaan 94,
B-3001 Heverlee (Belgmm)
Abstract
Capacitive sensors for the detectIon of mechamcal quanhtles all rely on a displacement measurement The movement
of a suspended electrode with respect to a fixed electrode establishes a changmg capacitor value between the electrodes
This effect can be measured and if the mechamcal quantity controls the movable electrode, a sensor 1s reahzed Smce
the value of the capaator IS dmzctly related to Its size, and a small capacitor means high noise susceptlblhty, capantlve
sensors should be as large as possible Capacltlve pressure sensors have been developed mth success for mdustrlal
applications, where large membrane sizes are not a cntlcal Issue However, most centres of expertise m s&con sensors
show an interest m exploltmg &con technology to produce capacltlve pressure sensors as well From the above, this
mmlaturlzatlon trend appears to be an unsound Idea On the other hand, the prmclple of capacitive sensors allows
the reahzatlon of measuring systems with so far unknown performance Indeed, the capacltlve sensor reveals dlstmct
advantages when compared to Its plezoreslstlve counterpart high sensltivlty, low power consumption, better
temperature performance, less senatlve to drift, etc Nevertheless, only a minor fraction of the market for pressure
sensors 1s taken up by capaatlve-type sensors When observing the characteristics of capacitive sensors, It may seem
surpnsmg to encounter so few devices m real-world apphcatlons The reasons for the lack m breakthrough can be
found m the design complexity and the requirements for a matched sensing circuit This paper will extensively discuss
the Justlficatlon of the choice for this research effort, and will elaborate on the techniques to fabricate the devices
based on electromc manufacturmg procedures BasIcally, slhcon capacitive sensors differ from plezosensors m that
they measure the displacement of the membrane, and not its stress! This has important lmphcatlons on the final
assembled device less package-induced problems can thus be expected However, a far more Important Issue 1s their
extremely high sensltlvlty, together with a low power consumption These ISSUES
make them especially attractive m
biomedical implant devices, or m other telemetry applications, where power 1s not randomly available So far,
this 1s the only field of success for these sensors However, due to the mterestmg detectIon prmclple, new fields
of apphcatlon emerge, offering unique and superior performance when compared to avallable sensors Umaxlal
accelerometers are a good example, where extremely high cross-sensitivity reduction can be obtained
Introduction
sensors ~11 generate an electrical
signal as a result of the elastic deformation of a membrane, as 1s the case for other sensors, such as the
plezoreslstlve types However, It 1s not the bulk-up
stress m the membrane that causes the stgnal, but rather
Its displacement
This phenomenon 1s the essential
difference from other sensors and results m unique
properties, which will be illustrated later on The basic
structure of a capacitive sensor always consists in a set
of plates of surface A separated by a distance d (Fig 1)
One obtains for the capacitance value at rest
d
Capacitive-type
Co=&
0a
where E 1s the dlelectrlc constant (pernuttlvlty) of the
medium between the plates (usually air) One can see
Rg 1 SchematIc representation of a capacitor, consistmg of two
plates and a dlelectnc v&h thickness d To the nght, the non-parallel movement of the plates IS represented
that to change the capacitance value, one can either
change the surface of or the distance between the plates
Most capacitive sensors rely upon the change of the
distance d rather than the change m surface A, although
the latter phenomenon IS mtnnslcally linear The reason
has to be sought m a more elaborate structure and
hence a more difficult assembly [l] If one IS able to
deform the dlelectnc layer over its entire surface with
the distance Ad, such that the plates remam parallel,
one has
*Invited paper
09244247/93/$X500
@ 1993 -
Elsewer Sequoia All nghts reserved
94
From cqns ( 1) and (2), and for Ad c<d, one obtams
c
Ad
,=‘-Jwhich reveals a linear relatlonshlp
From eqn (1) one can also obtain the expression for
the sensltlvlty of this sensor
A
AC
dd=-5
which illustrates that the sensltlvlty will increase with
the square of the distance d between the electrodes For
a small gap a high sensltlvlty can thus be expected
However, most sensors cannot be constructed such
that expression (2) 1s valid, that is, the plates often do
not move m parallel In most cases the capacitance
change will be caused by a deformation of one of the
electrodes Itself (see Rg 1, nght) Expressions (2) and
(3) no longer hold Equation (2) should be wntten as
a
a
x=oy=o
where w, a function of x and y, represents the local
deflection as a consequence of the pressure One can
define u as the mean value of the displacement of the
entire electrode
a 0
1
o=w dx dy
ments must be done mslde the patlent, they require
extremely stable, temperature msensltlve and accurate
devices Sohd-state strain-gauge devices have already
been on the market for a long time [3] and have been
widely used for arterial pressure measurement for more
than 20 years [4] Whereas these systems are very useful
and accurate for measuring dynamic pressure changes,
they show deficlenaes when It comes to long-term
measurements of low-pressure signals This 1s because
these sensors dnft about 100 Pa per day If the pressure
signal 1s in this range, they become unsrutable A posslble solution 1s the use of fluld-filled catheters, which are
coupled to precise manometer systems outside the patient However, this approach IS generally accepted to
be cumbersome, may cause inaccuracies m dynamic
signals and restrains the patient It 1s unsuitable for
application m telemetry devices Moreover, for such
apphcatlons, the power consumption of the sensor and
its signal-processmg network become Important, since
telemetry system have to rely on a battery The limited
amount of power makes the use of plezoreslstlve
devices difficult, especially for long-term contmuous
momtormg This 1s especially the case for implantable
systems, where the size of the battery must be kept to
an absolute mmlmum, or where even no battery can be
tolerated [5] These conslderatlons gave nse to the
development of capacitive-type pressure transducers to
achieve an improved static accuracy [6,7] and to
provide low-power operation of the en&e system
A
Construction methods
so eqn (5) becomes
AC=C&$=C,,(;)
(7)
The sensltlvlty to pressure can be obtained from this
expression There 1s a non-hnear relatlonshlp If, however, the displacement IS small compared with the membrane thickness, v equals about one quarter of the
maxnnal deflection at the centre of the membrane [2],
allowing an estlmatlon of the performance of the sensor
to be made
In contrast to piezoresistlve or any piezosensitive
transducers, for the capacitive type of sensor the stress
in the membranes 1s of no direct relevance to the
transduction phenomenon From a constructlon vlewpomt, this offers distinct advantdges, for example, a
much larger tolerance can be accepted with respect to
the deposited structures on the membrane In fact, the
membrane itself acts as a transducllon medium, and not
the deposited structures (e g , resistors have to be
placed very accurately to obtam maximal sensitlvlty,
and are very susceptible to mlsahgnment)
Justdicatioo of capacitive sensors
Complete dwcrete construction
In order to understand the underlying reasons for the
growing research m this field, one has to consider the
hrmtatlons of the so-called well-established techniques,
such as the plezoreslstlve devices One of the fields that
has always been looking ahead for more and tighter
speclficatlons for sensor systems 1s without doubt me&cal engmeenng Many pressures can and must be monstored m the human body m order to allow proper
dlagnosls and therapy Smce some of these measure-
Given its mtrmslc slmphclty, the capacmve sensor
construction 1s not that obvious First there are the
specific requirements with respect to the tolerances m
the construction, especially concerning the electrode
separation Moreover, the choice and homogeneity of
the dlelectnc layer are not evident Discretely built
sensors have been realized m commercial apphcatlons,
for apphcatlons requumg a high sensltlvlty or for smallsignal differential pressure measurements (e g , m flow-
95
Differential
Fig 2 Example of a commermal chfferentral capacltw
pressuremeasuring deuce The diameter of the membranes IS about 10 cm,
and the output 1s dehvered under the 4-20 mA standard
measuring devices) Figure 2 Illustrates a cross section
of such a device [8] Since parasltlc effects should be
reduced, these sensors are generally large constructions,
because the capacitive value (at rest) 1s directly proportional to the surface, and the parasitic capacitance
remains approximately constant For mdustrlal control
systems the large size IS usually acceptable
Figure 3 Illustrates two capacitive sensors, built for
medical Implant devices The one on the left 1s a sensor
built to cope with the drift problems of strain-gauge
sensors It was intended to monitor mtracramal pressure [9], and 1s a good example of what the state of the
art was some 15 years ago The size of the sensor 1s
7 mm by 2 mm, with a spacing of 25-75 pm between
the diaphragm and Its base The conductive metal
surface, which forms the fixed electrode, 1s fired on a
glass layer The zero-pressure capacitance was 8 pF,
and varied up to 16 pF for a 100 kPa signal On the
right-hand side of the Figure 1s a more recent version of
the same construction, Introduced for use m cardiology
Here, the prime concern was high sensltlvlty coupled to
small size and low power consumption, since the device
1s intended for long-term lmplantatlon [lo] A htanrum
barrel of 2 5 mm diameter IS filled with a glass slug,
fused to its inner wall A layer of gold 1s sputtered on
the concave surface of the glass, and the movable
counter electrode 1s welded consecutively One obtains
titanium
pressure
a capacitance value of a mere 1 5 pF A slgnal-modulatmg clrcult IS added to boost the output of the sensor so
that It can be transmitted over the long catheter wires
The entire system 1s encapsulated into a hermetically
sealed tltamum package for hocompatlblhty purposes
It IS developed by a commercial pacemaker company
(for automatic defibrillator triggering)
Though the fabrlcatlon
sequence IS extremely
difficult, these kmds of sensor constructions still seem
to be very attractive to designers, with slmphclty as the
key issue to Justify their use As another example, the
capaclhve prmclple has also recently been proposed for
the detectlon of angular position [ 111 Here, printed
clrcult board technology 1s used to form overlapping
patterns on two different boards, which change their
common surface with changing angular position
Hence, the total capacitor surface A 1s changed, and
thus C varies The conductrve surface (electrodes of the
capacitor) 1s shaped m order to Improve the linear
relatlonshlp with the angle
Thick-film technology
Since most mecharucal sensors are m a rather high
cost-price category, the relatively low fabrlcatlon cost of
thick-film clrcults seems an attractive altematlve for the
realization of sensors This has been extensively the case
for plezoreslstlve sensors, where for large-market apphcations, such as the automotive sector, successful designs
have been elaborated [ 121 Capacitive devices can be
realized m thick-film technology as well, although vu-tually no research has been performed m this directloo
Standard thick-film technology allows capacitors to be
realized usmg a multilayered structure Special pastes
with a high dlelectrlc constant are available to obtain a
reasonable capacitor value with a small geometry
However, the compliance of this dlelectnc, mostly consisting of a glass matnx, 1s virtually zero, rendermg the
structures lmpractlcal for sensor use Other matenals
should be adopted Slhcone rubber IS one of them This
has been exploited m a device [ 131 by usmg flexible
polylmlde film carriers Figure 4 shows a thick-film
vanant of this sensor, a force transducer It will gve a
better linear performance when fabricated m thick-film
membrane
port
Fig 3 Examples of &scretely built capacmve pressure sensors left, an early intracranial pressure sensor (7 mm diameter,
thickness = 12 pm), right,, a more recent vewon (2 5 mm chameter) for apphcatlons m carrhology
membrane
;c
Rg 4 Capacltwe
ant d~electnc
electrode
thrck-film force transducer
matrix
substrates
Sllastlc
382
based on a comph-
technology, since the ceramic substrates give more
nglQty to the mdlvldual electrodes Another possrblhty
IS to make use of the mechanical properties of the
ceramic substrate itself, and measure the deflection of a
suspended membrane by the capacitive principle This
comes close to the device of Fig 2, but there the
ceramic electrode Itself does not flex The only apparent
drawback, I e , the low mltlal capacltlve value, typically
between 1 and 20 pF for this structure, can well be
overcome by rncorporatmg some impedance converting
network on the same thick-film substrate The field IS
still open for some good (and inexpensive) research’
Sdtcon technology
The use of silicon as a constructlon
material has long
been demonstrated and has been exploited by many
centres [ 141 Some 20 years ago, the first major efforts
were undertaken to fabricate mlmaturlzed sensors
based on exlstmg processing sequences used m electronic clrcmt fabncatlon, and expandmg these to smt
the specific requirements for the realization of novel
devices These new techniques or processmg sequences
are often referred to as mlcromachmmg This mvolves
all kinds of three-dlmenslonal structunng of s&con to
fabncate mechanical sensors The first efforts were pnmanly aimed at the feaslblhty of the reahzatlon of
sensor devices as such, focusing prunanly on the performance of the transducmg mechamsm of these sensors,
and looking mto posslblhtles to realize specific mechanical structures [ 15- 191
However, sensors always have to be considered as
part of an overall system, where they act pnmarlly at
the input stage of a usually complex process Therefore,
and also because the (electrical) signal out of the transducer IS usually extremely weak, the sensors must be
essentially coupled to some electronic networks to enhance and process then signals Smce the sensor devices
are fabricated m s&con, the idea of integrating electronic networks on the same die 1s quite natural However, from the fabrication technology vlewpomt, this
approach IS far from natural Sensor fabncatlon requires dedicated processing sequences that conflict with
circuit processing, lmpedmg the merging of both
devices m the first attempt However, by careful processmg sequence design, one can achieve posltlve results Often, the circuit processmg will have to be
performed at the begmnmg, then the wafers are postprocessed m the mlcromachmmg foundry
With respect to the more usual plezoreslstlve devices,
the specdic ments of capacltlve devices have to be
found m their slmphclty, I e , they reqwre less extensive
processing (only an electrode 1s needed) on the membrane, and are less susceptible to mlsahgnment errors
[20] Once a process has been set up, a high degree of
reproduablhty can be expected This apparent ‘slmphclty’, however, should be considered carefully The mterelectrode distance IS a very important parameter m
the design and fabrication, and any lack of control will
be reflected m a loss of reproduablhty
Figure 5 shows
a common layout of such a capacitive sensor Mlsahgnment m the x or y direction will have mmor influence
on the performance of the device However, tight control of the distance z 1s mandatory for a reproducible
device z m fact reflects the distance between the two
electrodes This can be affected both by the cavity depth
and by all the intermediate layers between the glass
and the slhcon wafer Special care must be taken with
the feedthrough of the electrode connections to the
bonding pads (or the processing network) It 1s best to
make use of a dflused layer m order to mmlmlze
step coverage failure On the other hand, an ahunuuum
path 1OOnm thck can still be bonded hermetically
The Figure also reveals two possible mterconnectlon
schemes The one at the right 1s best with respect to a
parasltlc capacitance, smce the two bondmg pads are
on the highly lsolatmg glass layer The mterconnectlon
Rg 5 Usual construction for a bulk nucromachmed &con capacitive pressure sensor
on the sdlcon The arrows define possible nusahgnments in the assembly
Contacts
can be made on the glass substrate or
97
on the left 1s preferable when electromc processing
circuitry 1s added
No matter what method 1s selected, care must be
taken m dlcmg the bonded wafers one of the wafers,
usually the glass one, must be cut precisely m order not
to touch the underlying structure Also, tlus underlying structure should, evidently, not be bonded to the
glass, which requires special precautions m the design
stage
This means that, although mtrmsically much simpler
then the plezoreslstlve device, the biggest challenge of
the capacitive device hes m its packaging This also
forms the ongm of most of its problems
One of the speclficatlons directly influenced by the
packaging techmque 1s the temperature sensitivity of
the sensor depending on the sealing technique one can
realize absolute, relative or trapped-gas pressure sensors The first 1s sealed under vacuum, the second 1s
sealed such that the reference chamber has a connection
with the outslde world or any other medmm (to reahze
a differential measurmg set-up) In the case of entrapped gas, extra non-lmearlty 1s added for two reasons first, the gas Itself will be compressed for any
movement of the membrane This effect can be dlmmIshed by provldmg extra volume m the sensor reference
cavity, e g , at the edge of the sensmg membrane The
effect of thermal gas expansion, on the other hand, 1s
much larger, and cannot be reduced by any means It
can reach values of up to 1% of full scale per “C The
only solution 1s to use vacuum-sealed pressure sensors
Overmew of developed devices
An attempt 1s made here to give an overvlew of
capacitive devices currently under development or of
devices which have been demonstrated so far Table 1
gives the most important details of about 40 different or
related capacltlve-type sensors, all fabncated m slhcon
technology Although it would be mtngumg to comment on all of these papers, time and space do not
allow us to do so It 1s therefore left to the reader to
pick out relevant devices to assess his own state of the
art In this text, only a fraction of the Table will be
discussed This will be done to Justify the arguments
pven here and to illustrate general research pohcles
Pressure-sensmg dewces
First, It 1s apparent that the pressure sensor IS the
most successful device (600/o),at least at the early stages
of research m this field This has to match the field of
apphcatlon most of the devices are developed with low
power consumption m mmd Indeed, a capacitive
device consumes Just as much power as Its measurmg
circmt The nmnedate apphcatlon lies m the blomedlcal
field, where (especmlly for telemetry purposes) reduction of power consumption 1s the maJor design cnterlon The first devtce was developed for use m
cardiology [21], and almed at low power consumption
In a second version, an oscillator clrcult was added m
bipolar technology to yield an overall current dram of
20 pA at a supply from 2 5 to 10 V Figure 6 illustrates
a cross section of this ploneenng structure
It has been steadily nnproved to incorporate more
advanced electronics (for temperature compensation)
and finally to include a dummy reference capacitor
[22,23] Ko and his co-workers mtroduced the idea of
mcorporatmg a dummy capacitor in the same cavity to
reduce the effects of parasitic capacitance They also
compared square with circular shaped electrodes and
illustrated that the highest sensitivity can be obtamed
for a circular electrode (of about 36% of the central
area of the diaphragm [24,25] Pressure sensors for
blomedlcal apphcatlons were the very first technology
drivers to simulate research m this direction [26-281
Numerous devices followed the first ones, ullth ever-decreasmg sensor sizes Rest capacltlve values of a mere
0 5 pF [29,30] and, m a later development stage, even
going down to 0 3 pF [31], have been reported by the
Mlchlgan group of Wise Evidently, such low mtnnslc
capacitive values cannot be handled wlthout the matchmg arcmtry close to them In the last two papers, the
capacitor and the processmg network were separate
chips, avoldmg compatlblhty problems when trymg to
merge the two devices on one die An interesting approach to further munaturlzmg the size was proposed
in ref 30, where first anodlc bonding 1s performed, after
which almost the full frame of the membrane die 1s
etched away (see Fig 7) This 1s an excellent example of
how standard techniques can still lead to novel and
successful designs The arcult chip can be processed by
any foundry The only drawback of this approach 1s the
protection of the clrcult chip when applied ~1 uvo
The group of Esaslu m Japan has proposed a senes
of mterestmg combmatlons of sensors wth a matched
measuring circuit m one package [32-341 All of these
devices are based on the approach of first bonding the
wafers and then doing a consecutive backside etch to
yield a high degree of mmlatunzatlon
Furthermore, a
special technique 1s adopted to connect the device all
electromcs are encapsulated, and only the backside of
the capacltlve membrane 1s exposed to the outside
world To contact the electronics, a glass feedthrough 1s
realized by ultrasomc drdlmg, sputtermg coating of the
hole and finally connecting the lead wire by conductive
epoxy (see Fig 8) Although the seahng 1s more advanced in this device, more caution 1s reqmred for the
realization of the CMOS/mlcromachmed device chip
In the search to improve sensor speaficatlons many
more solutions have been proposed Special concern
Year
1980
1982
1982
1983
1984
1985
1990
1986
1987
1987
1987
1989
1989
1990
1990
Authors
Sander, Knuttl
and Memdl
Ko, Bao and Hong
Lee and Wise
Ko, Shao, Fung,
Shen and Yeh
Snuth, Pnsbe, Shott
and Memdl
Hanneborg and Ohlkers
Srmth, Bowman
and Mandl
Chau and Wise
Mlyoshl, Akxyama,
Shmtaku, Inann
and HtJklgawa
ShoJl, Nwse, Esasht
and Matsuo
Furuta, Esashl,
Shojl and Matsumoto
Puers, Peeters and Sansen
Backlund, Rosengren,
Hok and Svedbergh
Kandler, Elchholz,
Manoh and Mokwa
2x6
4x4
pressure
pressure
10
2x3
array 81 x
loo P0
3x3
pressure
pressure
2x35
2
25
10
35
8
35x07
05
1X5
sacrificial
layer,
polysilicon
KOH bulk
bulk, KOH
bulk, KOH
bulk, EDP
surface etch
EDP, bulk
bulk
bulk
7
22
bulk
bulk,
hydrazme
bulk, KOH
22
6
12
bulk,
7
hydrazme
bulk
Etch
22
G (PF)
7
pressure
pressure
pressure
pressure
pressure
2x6
2x4
pressure
pressure
4x4
2x4
3x3
Dunenslons
(mm’>
pressure
pressure
pressure
Type
TABLE 1 Detads of capacltlve-type sensors fabncated m s~bcon technology
na
fusion
bonding
anodx
anodx
&on
Sl-SI
na
anodlc
anodlc
sputtered
glass
anodlc
ano&
anodx
anodlc
ano&c
Seal
SC CMOS
LC clrcmt
only
no
separate,
CMOS
CMOS
il0
separate
mtegrated
oscliIator,
bipolar 10pm
separate ctip
Integrated
os&Iator,
bipolar 10 pm
Integrated
CMOS
separate
integrated,
bxpoIar
general
eye-pressure
general
car&o
blomedlcal
general
cardlo
cardlo
general
cardlo
blomed~cal
general
general
cardlo
schmltt
oscdlator,
hpolar
Apphcatlon
Circutt
prehmmary
results
LCtuned by
pressuretransponder
system
FEM analysts
& hneanzation
small assembly,
complete
backslde etch
direct f&on
bondmg
NI dtaphragm,
sacnticxil layer,
cheap large
batch prod
mnuatmxzed
companscm with
plez,oresistive
devices
low TCO and
dnft
temperature
compensation
ref capacitor
Integrated wzth
sensmg capacitor
effect sealmg on
TCO and TCS
membrane, drxft
aspects
ring versus square
early device
Dtscusslon
47
69
39
33
27
49
30
23
36,
37
22
25
26
24
21
Ref
3
99
bulk, EDP
bulk
surface etch
bulk. KOH
bulk,
hydrazme
10
16
10
9
9
12
15
X2
28x36
7x1
35x35
3x4s
3x45
*5x5
36x36
5x6
accelerometer
accelerometer
accelerometer
accelerometer
accelerometer
accelerometer
accelerometer
accelerometer
1988
1990
1990
1990
1991
1991
1991
1991
Schlaak, Arndt, Steckenbom,
Gevatter, Qesewetter and
Grethen
Muck,
Olney
%del, Rledel, Kolbeck,
Kupke and Komger
Suzuki, Tuclutam, Kate, Ueno,
Yokota, Sato and Esash~
Kloeck, SW&I, TucIntam,
Mlkl, Matsumoto,
Sato,
Kolde and Sugsawa
Payne and Dmswood,
and Goodenough
Peeters, Vergote, Puers
and Sansen
Ura and Esashl
bulk
bulk
bulk
KOH bulk
20
8x6
accelerometer
Rudolf, Jomod, Bergqvlst,
Leuthold and Bencze
1987,
1990
Etch
G (PF)
Year
Authors
Dunenslons
(mm’)
1 (continued)
Type
TABLE
anodlc
fusion
bond
+ anodlc
na
anodlc
anodlc
anodlc
anodlc
ano&c
Seal
59
65
66
61
62
transparent
electrodes on glass
allow observation
fully integrated
device, no bulk
nucromachmmg
highly symmetrical,,“›
tunable damping
s&on-oxnutnde
suspension
general
automotlve
general
general
Integrated
CMOS
separate
separate
57
servo actlon
general
separate
58
64
55,
56
60
device
pg
Ref
good hneanty
pg resolution,
symmetnc
suspension
commercial
force balancmg
resolution
Discussion
general
positlon
sensmg
space, fight
control
spacecraft
Apphcatlon
separate
separate
separate
separate
Circuit
z
101
bond.lng
pads
\
bipolar
Fig 6 The Stanford
1980
capacltlve
pressure sensor, measunng
orlglnal
3 mm x 3 mm and mcorporatmg
size
transducer
a bipolar convertmg
cmxnt, as proposed
m
chip
glass carrier
Rg 7 In order to obtam mnumal size, the sensor chip IS firstly anodlcally bonded to the glass carrier, and then etched down usmg EDP,
which stops on the boron-doped ddked
structure, forming the mmlature frame and membrane (dotted hne shows the ongmal chip size)
The processmg chip dself IS embedded m a cavity m the glass carrier
Al
wire
glass counter
Part
f
\electronlc
Fig 8 Fully sealed device mcorporatmg
an electronic
detection of the capacmve changes A glass feedthrough
the connechon to the outslde
contact
epoxyglue
clrcult
nrcult for
provides
has been gven by the Norwegan group [35-371 to
lower stress effects m the membrane (which cause temperature effects m the sensltlvlty curve) fis can be
achieved by retammg the idea of bonding the s&con
chip to a glass carrier, but usmg thm sputtered glass
layers on slhcon substrates instead In conJunction,
many efforts have been spent m better understandmg
capacitive transducmg phenomena, and more precisely
many proposals have been made m order to reduce
parasmc effects, either by special arcmtry or by specific
electrode configurattons, and-to look for a better lmear
response of the device [38-431
Also, more recently, more and more efforts are being
made to merge CMOS processes vvlth capacltlve pressure
sensors [44-481 The mam drive behind this approach 1s
to mmmnze parasitic effects, and to look for acceptable
process flows for the slbcon foundries This has partially
Rg 9 Prmctple of a surface nucromachmed
pressure sensor by
etchmg the oxide from underneath the polysdlcon, a cavity can be
created The doped p+ regton IS used as the movable electrode for
the sensor
become possible by the use of sacrficlal layer technology,
first introduced by Guckel and Burns [49] The method
consists m etchmg away the silicon dloxlde from between
the upper LPCVD polyslhcon and the n-type slhcon, as
shown m Fig 9 A cavity 1s reahzed, which 1s equal to
the orrgmal oxide tluckness, e g , 1 m The formation
of stress-free polysticon IS a possible tic&y,
but the
advantage 1sa higher yield per wafer (since no space-consummg backside etchmg 1s reqmred), and an automatic
over-pressure stop for pressure sensors [50] Also, the
double-sided alignment that ISreqmred m bulk nucromachmmg can be avoided Moreover, this process 1s fully
compatible Hrlth the standard CMOS process
Other approaches were to use an alternative to alkalme-based etchants, such as TMAH, to perform the
bulk etch of the &con, thus avoldmg possible contamination of the CMOS process by KOH [51,52]
102
Accelerometers
Accelerometers form the second type of devices for
which the capantlve measuring prmaple has proved
successful Although the first devices, developed only
ten years ago, were rather prlmltlve [53,54], the first
one showed a degree of mlnlatu~zatlon never obtamed
m any of the later devices It conslsted of a cantllevered
beam, made out of slhcon oxide, the end of which was
provided with a deposlted dot of gold, which acts as a
selsmlc mass The capacitance value was a mere 4 fF,
such that a buffer stage had to be Implemented on the
same dre, m l~edlate
proxmuty of the cantilever The
second device was nothing more than a suspended plate
that served as the selsmlc mass The device was soon
developed into a more sensitive accelerometer, using a
mesa structure to detect the acceleration [5.5,56] The
capacitive prmclple also allowed It to be used recrprotally, 1e , by applymg an electrostatic voltage, the
electrode can be reposItioned urlth respect to the or?glnal drlvmg force By domg so, the voltage reqmred to
restore the equlhbnum posltlon becomes a measure of
p7740 Pymx
the acceleration Thrs principle IS better known under
the term ‘force balancmg’ and became popular m accelerometer development because It allows the range
of the device to be expanded by several orders of
magnitude [ 57-591
Other designs do not focus on fork-clang
systems, but concentrate on the geometry of the structure
as a means to Improve the dynamic behavlour One
posslhhty 1s to look for highly s~met~cal structures
The lmmedlate merrt of this approach IS to obtain
extremely high cross-sensltmlty reduction [ 60,611 Both
the symmetry of the structure and the capacltlve detecbon prmclple guarantee a speaficatlon of this parameter which 15 some orders of magnrtude better than that
of the best commercial devices m other technologies
Fjgure 10 gves a thr~-dImensIona
nnpresslon of this
accelerometer
In some designs, the sprmg elements of the accelerometer are no longer s&con and other layers can be used
This IS espectally the case for designs where rmmaturlzatlon becomes Important the size of the selsmlc
UCASO
Rg 10 Three-drmenaonal wew of a fully symmetric accelerometer, reahzed as a four-layer structure, mcludmg both s&on fusron and
anodlc bonding [61]
103
mass becomes too small to be susceptible to the desired
acceleration Thinner suspension systems can be a solution In two cases, slhcon mtnde membrane 1s used
[62, 631 The first uses a set of 64 slhcon mtrlde beams
to suspend the selsmlc mass, whereas the second device
uses a composite membrane of nitride and oxlmtnde m
order to cancel stress effects This device can therefore
be made as small as 1 mm x 1 mm x 1 mm
It is intriguing to encounter fully commercial capacitive acceleration sensors, a fact which does not appear
wth pressure sensors The reason has to be sought m
the fact that the capacltlve principle allows for a much
better performance than any other detectlon prmclple
This IS particularly the case m space and military apphcatlons, where there 1s a strong need for precise accelerometers The Endevco device 1s a good example
[64] of such a commercial product It contains a wellcalculated damping system, and it IS shown that the
device 1s much more sensitive, stable and has a higher
resonance frequency than Its plezoreslstlve counterpart
Another commercial product 1s the Analog Device
acceleration sensor, developed for the automotlve sector (alrbag system) [65,66] It 1s a very interesting
device, built around an mterdlgltlzed structure, realized
by reactwe Ion etching of a free-standing polyslhcon
layer (surface mlcromachmed) The axls of sensltlvlty IS
thus m the plane of the sensor Itself What IS surprlsmg
is to see this company, with so far no known experience
m mlcromachmmg, capable of putting a high state of
the art device into the market (surface nucromachmed,
chip surface = 10% sensor, 90% slgnal processmg, meludes force balancing, sensltlve m the chip surface),
obviously dwarfing all the research efforts of so many
other centres
TONOMETER
MODULE
PMMA LENS
L--I--I----
Rg
rmmvw!%oIIp
I1 Schematic drawmg of the postenor chamber lens,
telemetnc tonometer
equpped with the
Example of blomedtcal deuce development
A good example for the Justified application of capacitive sensing prmclples IS the blomedlcal field, especlally when mformatlon has to be transmltted from the
mslde of the (human) body to the outslde world For
such telemetry systems, low power consumption, high
sensltmlty, low temperature drift and a good stablhty of
the sensors are a prerequlslte Capacltlve sensors can
cope with these demands As an example, the project of
eye momtormg 1s gven In ophthalmology, the eye
pressure 1s an important parameter However, it IS a
very slowly varying slgnal and 1s moreover very small m
comparison v&h the atmospheric pressure which on
average lies 2000 Pa above it, mth a dally cychc vanatlon m the range of 500 Pa In order to be able to
monitor this signal on a contmuous basis, implantable
eye-pressure momtonng devices have been proposed
They all rely on capactttve sensors [5,67,68] Whereas
the first two rely on a purely passlve system (one C, one
L whtch act as a resonant arc&),
the last approach 1s
a fully active device, which picks up Its power from a
tuned co11 system It was developed m our laboratory
by Van Schuylenbergh and Peeters, and IS dlustrated m
Fig 11 [ 51 A commercially available PMMA lens IS
machined down to leave only the central optic part A
cyhndncal space of 3 5 mm ID and 8 mm OD by
0 5 mm is available Total power delivery by the tuned
cod system IS hrmted to 100 VW The Figure illustrates
the cn-cular tamer m glass, which contams the receiving
co11 on Its bottom part, and the mterconnectlon dlagram for the electromc circmts and the (fixed) sensor
electrode at the top side A dedicated switch capacitor
clrcmt 1s used to make a dflerentlal measurement between the reference capacitor and the pressure sensor
The interface chip, shown m Fig 12, IS an adapted
version of a former development [40], and runs at a
mere 35 pA
Fig
12 MIcrophotograph
of the custom-designed
CMOS
switched capacitor circuit It measures 19 mm x 2 4 mm and accepts unregulated supply voltages between 4 5 and I5 V
References
1 U Zelbstem, Capteurs de pression de fluIdes, m G Asch (ed ),
L.es Capfeurs en Ins~rumerziatzonIndustrzelle, Bordas, Pans,
1987, pp 595-599
2 S Timoshenko and S Womowsky-Kneger,
Theory of Plates
and Shells, McGraw I&II, New York, 1959
3 E Komgsberg, Capablhtles of implatable transducers, m L
Harnnson (ed ), Res Anzmals zn Med, DHEW (NIH) 72-333,
1973, pp 1145-1150
4 H Sandler, T Fryer and B Datnow, Smgle channel pressure
telemetry umt, J Appl Physrol, 26 (I 969) 235 -238
5 K Van Schuylenbergh, E Peeters, B Puers, W Sansen and A
Neetens, An unplantable telemetric tonometer for dnect mtraocular pressure measurements, Absrr 1st Eur Conf Bzomed
Eng, Nzce, France, 1991, pp 194- 195
6 J Atkmson, D Shurtleff and E Foltz, Radio telemetry for the
measurement of mtracramal pressure, J Neurosurg , 27 (1967)
428-432
7 T Fryer, Telemetry of intracranial pressure, BIorelemetry Patzent Monztorzng, S(2) (1978) 88-122
8 PhilIps Industrial AutomatIon, Dam sheet of PSC-Transmztter
PD3 for DzflerentzalPressure and FIow Measurements, 1987
9 T Fryer, A pressure telemetry system utibzmg a capaatancetype transducer, Bzorelemetry III, Academic Press, New York,
1976, pp 279-282
IO A Moore and K Anderson, An Implantable chrome pressure
sensor, Proc 4th Int Conf Sohd-State Sensors and Acruarors
(Transducers ‘85). Phriadelphza, PA, USA, June 7-11, 1985,
pp 189-192
II R Wolffenbuttel and R Van Kampen, An Integrable capacltlve angular dtsplacement sensor with Improved hneanty,
Sensors and Actuators A, 25-27 (1991) 835-843
12 B Puers, S Pasczcynskl and W Sansen, Assessment of thick
tilm fabncatlon methods for force (pressure) sensors, Sensors
and Actuators, 12 (1986) 57-76
13 M Neuman, A Beret and E O’Connor, Capacitive sensors
for measurmg finger and thumb tip forces, IEEE Fronlzers
Eng Comp zn Health Care, 1984, pp 436-439
I4 K Petersen, S&on as a mechanical matenal, Proc IEEE, 70
(1982) 420-457
I5 K Bean, Amsotroplc etchmg of s&on, IEEE Tram Electron
Dewes, ED-25 (1978) 1185- 1193
16 A Bohg, Ethylene dmmme-pyrocatechol-water
rnlxture
shows etchmg anomaly m boron doped silicon, J Electrochem Sot, 118(1971) 401
I7 T Jackson, M Tischler and K Wise, An electrochenncal
P-N Junctlon etchstop for the formation of sdlcon nucrostructures, IEEE Electron Deuzces Letr , EDL-2 (1981) 4445
I8 G Walhs and D Pomerantz, Field assisted glass-metal sealing, J Appl Phys , 40 (1969) 3946-3950
I9 A Brooks and R Donovan, Low temperature electrostatic
sihcon to slhcon seals using sputtered borosrhcate glass, J
EIec@ochem Sot , 119 (1972) 545-546
20 S Clark and K Wise, Pressure sensmvlty m amsotropically
etched thin-diaphragm pressure sensors, IEEE Tram Electron
Deuzces, ED-26 (1979) 1887-1896
21 C Sander, J Knuttl and J Memdl, A monohthlc capacitive
pressure sensor with pulse period output, IEEE Tram Elecbon Deuzces, ED- 17 (1980) 927-930
22 M Smith, M Pnsbe, J Shott and J Memdl, Integrated
cm&s for a capacltwe pressure sensor, Proc IEEE Fronrrers
Engzn Comp nz Health Care, 1984, pp 440-443
23 M Smith, L Bowman and J Memdl, Analysis, design and
performance of capacltlve pressure sensor IC, IEEE Tram
Bzomed Eng , BME-33 ( 1986) 163- I74
24 W Ko, M Bao and Y Hong, A high sensltwity Integrated
Cmzuit capacltlve pressure transducer, IEEE Trans Electron
Deuzces, ED-29 (1982) 48-56
25 W Ko, B Shao, C Fung, W Shen and G Yeh, Capacitive
pressure transducers Hrlth Integrated arcmts, Sensors and Actuators, 4 (1983) 403-411
26 Y Lee and K Wise., A batch-fabncated
&con capacltlve
pressure transducer vvlth low temperature sensltlvlty, IEEE
Trans Elecfron Deorces, ED -29 ( 1982) 42 -48
27 S Shojl, T Nlsase, M Esashl and T Matsuo, Fabncatlon of
an Implantable capacltlve type pressure sensor, Proc 5th Int
Conf Solzd-State Sensors and Actuators (Transducers ‘87),
Tokyo, Japan, June 2-5, 1987, pp 305-308
28 J Sunnnto, G Yeh, T Spear and W Ko, Sdlcon diaphragm
capacitive sensor for pressure, flow, acceleration and altitude
measurements, Proc 4th Int Conf Solzd-Slate Sensors and
Actuators (Transducers ‘87), Tokyo, Japan, June 2-5, 1987,
pp 336-339
29 H Chau and K Wise, Scaling hnuts m batch fabncated
slhcon pressure sensors, Proc 3rd Inf Conf Sohd-Stare Sensors and Actuators (Transducers ‘85), Phzladelphia,PA, USA,
June 7-11, 1985, pp 174-177
30 H Chau and K Wise, An ultra-rnmlature solrd-state pressure
sensor for a cardiovascular
catheter, Proc 4th Int Conf
Solzd-State Sensors and Actuators (Transducers ‘87), Tokyo,
Japan, June 2-5, 1987, pp 344-347
31 J Ji, S Cho, Y Zhang, K NaJafi and K Wise, An ultramuuature CMOS pressure sensor for a multlplexed carchovascular
catheter, Proc 6th Int Conf Sohd-State Sensors and Actuators (Transducers ‘91), San Franczsco, CA, USA, June 24-28,
1991, pp 1018-1020
32 K Furuta, M Esash~, S ShoJl and Y Matsumoto, Cathetertip capa&ve pressure sensor, Tech Dzgest, 8th Sensor Symp ,
Japan, 1989, pp 25-28
33 Y Matsumoto,
S Shall and M Esashl, A numature mtegrated capawbve pressure sensor, Tech Dzgest, 9th Sensor
Symp , Japan, 1990, pp 43-46
34 T Kudoh, S ShoJl and M Esashl, An integrated munature
capacltlve pressure sensor, Sensors and Actuators A, 29 (1991)
185-193
105
35 A Hanneborg, T Hansen, P Ohlkers, E Calson, B Dahl and
0 Holwech, A new Integrated capantlve pressure sensor with
frequency modulated output, Proc 3rd Int Conf Solui-State
Sensors and Actuators (Transducers ‘85), PhdaaZphra, PA,
USA, June 7-11, 1985, pp 186-188
36 A Hanneborg and P Ohlkers, A capaclhve &con pressure
sensor v&h low TCO and iugh long-term stablhty, Sensors and
Acfuators, A21-A23 (1990) 151-154
37 A Hanneborg, M Nese, H Jacobsen and R Hobn, Sihcon to
thin film anodlc bondmg, MME’92 Workshop Dzgest, Leuuen,
Be$rum, 1992, pp 9-19
38 B Puers, E Peeters and W Sansen, CAD tools m mechamcal
sensors design, Sensors and Actuators, 17 (1989) 423-429
39 B Puers, E Peeters, A Vanden Bossche and W Sansen, A
capacltwe pressure sensor with low impedance output and
active suppression of parasltx effects, Sensors and Actuators,
AZI-A23 (1990) 108-114
40 P Pons, G Blasquez and N Ratrer, Harmomc response of
silicon capacitive pressure sensor, Sensors and Actuators A,
25-27 (1991) 301-305
41 V Artyomov,
E Kudryashov,
V Shelenshkevlch
and A
Shulga, Sihcon capaative pressure transducer with mcreased
modulation depth, Sensors and Actuators A, 28 (1991) 223230
42 L Rosengren, J Soderkvlst and L Snuth, Mlcromachmed
sensor structures mth linear capacltlve response, Sensors and
Actuators A, 31 (1992) ZOO-205
43 P Pons and G Blasquez, Transient response of capacitive
pressure sensors, Sensorr and Actuators A, 32 (1992) 616-621
44 B Puers, A Vanden Bossche, E Peeters and W Sansen, An
Implantable pressure sensor for use m cardiology, Sensors and
Actuaiors, A21 -A23 (1990) 944-947
45 J Kung and H Lee, An Integrated ax-gap capacitor process
for sensor apphcations,
Proc 6th Int Conf Sold-State Sensors and Actuators (Transducers PI), San Francwco, CA, USA,
June 24-28, 1991, pp 1010-1013
46 M Kandler, J Etchholz, Y Manoh and W Mokwa, CMOS
compatible capacitive pressure sensor with read-out electronics, in H Reich1 (ed ), Mxrosystems, Sprmger, New York,
1990, pp 574-580
47 F Schnatz, U Schoneberg, W Brockherde, P Kopystynslu,
T Mehlhom, E Obermeler and H Benzel, Smart CMOS
capacitive pressure transducer with on-chip calibration capab&y, Sensors and Actuators A, 34 (1992) 77-83
48 S Mlyoslu, H Akiyama, H Shmtaku, Y Inaml and M
Hgdcigawa, A new fabmation
process for capacltlve pressures sensors, Proc 4th Int Conf Sohd-State Sensors and
Actuators (Transducers ‘87), Tokyo, Japan, June 2-5, 1987,
pp 309-311
49 H Guckel and D Bums, Planar processed polyslhcon sealed
cavities for pressure transducer arrays, IEDM Tech Digest,
1984, pp 223-225
50 H Guckel, D Bums, H Busta and J Detry, Laser recrystalhzed plezo-resistive micro-diaphragm
sensor, Proc 3rd Int
Conf Sold-State Sensors and Actuators (Transducers ‘85),
Phdadeiphla, PA, USA, June 7-11, 1985, pp 182-185
51 H Terabe, Y Fukaya, S Salrural, 0 Tabata, S Sugiyama
and M Esaslu, Capacitive pressure sensor for low pressure
measurements with high overpressure tolerance Tech Digest,
10th Sensor Symp, Japan, 1991, pp 133-136
52 T Nagata, H Terabe, S Sakurai, 0 Tahata, S Suglyama and
M Esasiu, D@al compensated
capacitwe pressure sensor
using CMOS technology for low pressure measurements, Proc
6th Inr Conf Sohd-State Sensors and Actuators (Transducers
Pl), San Francrrco, CA, (ISA, June 24-28, 1991, pp 308-311
53 K Petersen, A Shartel and N Raley, MIcromechanical
accelerometer integrated wth MOS detection cmxntry, IEEE
Trans Electron Devices. ED-29 (1982) 23-27
54 F Rudolf, A nucromechamcal
&pa&ve accelerometer with a
two-point mertlal mass suspens,on, Sensors and Actuators, 4
(1983) 191-198
55 F Rudolf, A Jomod and P Bencze, Slhcon mlcroaccelerometer, Proc 4th Int Conf Sohd-State Sensors and Actuators
(Transducers ‘87), Tokyo, Japan, June 2-5, 1987, pp 395-398
56 F Rudolf, A Jomod, J Bergqvist and H Leuthold, Preclslon
accelerometers wth pg resolution, Sensors and Actuators, A21 A23 (1990) 297-302
57 S Suzuki, S Tuchltarn, K Sato, S Uneo, Y Yokota, M Sato
and M E&u, Semiconductor capacitance-type
accelerometer
with PWM electrostatic servo techmque, Sensors and Actuators
A21-A23 (1990) 316-319
58 H Schlaak, F Amdt, A Steckenbom,
H Gevatter,
L
Kiesewetter and H Grethen, Mlcromechamcal
capacitive acceleratlon sensor with force compensation, m H Relchl (ed ),
Mxrosystems, Spnnger, New York, 1990, pp 617-622
59 B Kloeck, S Suzuki, S Tuclutam, M Mike, M Matsumoto,
K Sate, A Kolde and Y Sugsawa, Motion mvesttgatlon of
electrostatic servo-accelerometers by means of transparent IT0
fixed electrodes, Proc 6th Int Conf Sohd-State Sensors and
Actuators (Transducers ‘91), San Frannsco, CA, USA, June
24-28, 1991, pp 108-111
60 H Sadel, H Riedel, R Kolbeck, G Muck, W KuDke and
M Komger, Capacmve s&on accelerometer wth highly symmetrical design, Sensors and Actuators, A21 -A23 (1990) 312315
61 E Peeters, S Vergote, B Puers and W Sansen, A lughly
symmetrical capaabve micro-accelerometer
arlth single degree
of freedom response, Proc 6fh Int Conf Sohd-State Sensors
and Actuators (Transducers ‘91), San Franczrco CA, USA, June
24-28, 1991, pp 97-100
62 N Ura and M Esaslu, Differential capacitwe accelerometer,
Tech Dtg iOth Sensor Symp, Japan, 1991, pp 41-44
63 B Puers and S Vergote, A subnurnature capacltlve movement
detector using a composite membrane suspension, Sensors and
Actuators A, 31 (1992) 90-96
64 D Olney, Acceleration measurement usmg vanable capacitance, Proc Sensors Nuremberg ‘88, Germany, 1988, pp 149160
Surface nucromacluned
ac65 R Payne and K Dmswood,
celerometer a technology update, SAE Int Automotwe Eng
Congr , Detrort, MI, USA, 1991, pp 127-135
66 F Goodenough,
Axbags boom when IC accelerometer sees
5Og, Electronrc Destgn, (8 Aug ) (1991) 45-56
67 C Collms, Mmlature passive pressure transensor for lmplantmg m the eye, IEEE Trans Womed Eng , BME-14 (1967)
74-83
68 Y Backlund, L Rosengren, B Hok and B Svedbergh, Passive silicon transensor intended for biomechcal, remote pressure morntormg,
Sensors and Actuators, A21LA23 (1990)
58-61