You are on page 1of 41

Micromechanical Actuators

by Gijs Krijnen and Niels Tas


MESA+ Research Institute, Transducer Technology Laboratory
University of Twente, Enschede, The Netherlands

Preface
This report gives a brief overview of micromechanical actuators, a field which has attracted
great activity in the past 10-15 years. Due to the broad nature of the subject it can only
form a first sampling and it is impossible to go into any substantial detail on the many
subjects that could fill chapters or entire books on there own.
The initial goal was to describe the many different actuator principles, ranging from
electrostatic to electrowetting, from shape memory alloys to piezo-electric and so on.
However, it turned out that this was a too ambitious set-up for a report of limited size as
well as for the limited amount of time we could spend. Moreover, others have written entire
books about micro-actuators, showing the difficulty in treating the many published results
in considerable depth. Therefore, we were forced to limit ourselves. We decided to focus on
electrostatic micro-actuators since these have been investigated far more than any of the
other principles. Moreover, due to their relative easy, mostly CMOS-compatible fabrication,
they show a rich variety of implementations which makes them excellent for serving the
purpose of exemplifying the exciting possibilities delivered by micro-electro-mechanical
actuators.
Micromechanical actuators have become possible due to the development of micromachining technologies and many of the concepts and principles on which the actuators
rely can only be used courtesy of this technology. That is why a the first part of this report
is dedicated to this technology. This report forms the counterpart of a report on micromechanical-sensors which was written in the framework of the ASSET programme as well.
To be self-contained the technology decriptions have been included in this report but can be
skipped by those familiar with the previous report.

Introduction
Micromechanical actuators form a class of actuators which we here loosely define as
actuators which are made by micromachining technology (disregarding the actual size of
the actuators). In practice, many of the actuators complying to this definition will indeed
be rather small as compared to their "macro" counterparts although the latter ones can
be quite small as well (especially when fabricated using precision engineering
technology). In this part of the report we will not go in detail into all the specifics of
micromechanical actuators, simply since we neither have room nor time to cover an area
of research being so wide and so multidisciplinary in nature. What we will do, though, is
give a short overview of some important aspects of micromechanical actuators, discuss
some examples of actuators, and compile a list of references which will be useful for
further reading1.

In this respect there is one publication which has been very useful for the technology part of this report:
[Elwenspoek00]

GK+NT

MESA+ Research Institute

ASSET

Micromechanical Actuators

Miniaturisation and scaling laws


One aspect of micromechanical actuators is that they are made by micromachining
technology which basically has been adopted from microelectronics. As such, the field of
micromechanical actuator research and fabrication forms a part of a wider field called
Micro System Technology (MST) (an acronym mostly used in Europe) or Micro ElectroMechanical Systems (MEMS). Although in the field of MEMS there is no driving force to
reduce feature sizes per se, the efforts made in this direction in the field of microelectronics are beneficially. It offers MEMS researchers and engineers a toolbox
increasingly meeting their requirements in terms of accuracy and feature size.
If one had to give an indication of the feature sizes of MEMS devices one could say that
these roughly have dimensions of the order of 0.1 micrometre to a couple of millimetres,
albeit that this is by no means meant as a restricting definition. Although these
dimensions may look modestly small in respect of today's advent of nanotechnology,
there is still a huge transition going from "macro-actuators" to microactuators which is
related to various scaling laws. This may be easily appreciated looking at a unit-cell in
the form of a cube with (linear) unit dimension l. Since the volume, and hence the mass,
are proportional to l3, and the surface is proportional to l2, the ratio of body to surface
forces will be proportional to l. In other words, scaling down a "macro-actuator" from 1
cm to 10 micrometer, means that the surface forces will play a 1000-fold more important
role. This is for example reflected in the fact that gravity normally does not play an
important role in micromechanical actuators2. Looking at diffusion, be it particles or
fluxes (e.g. heat), diffusion times scale with l2. Hence, thermal processes will be 100
times faster when one reduces a system by a factor of 10. More examples of scaling laws
can be found in [Elwenspoek00].
These examples show that we cannot (always, simply) transfer our (intuitive) understanding of mechanical systems from the macro-world to the microdomain. In fact, we
always have to look carefully at the systems, especially at the numbers that describe
specific properties of the physics that we study (e.g. the Reynolds number). By doing so,
we will find limitations as well as new possibilities of microactuators relative to their
macro-world counterparts.

Systems aspects of MEMS


There is an important aspect of MEMS-devices: being small, and generally producing
small forces and signals, they do not trivially interface with the macro-world. Therefore,
micromechanical devices cannot be viewed as separate entities but they should actually
be considered as microsystems involving not only the actuating and/or sensing
microstructure but the interfacing and packaging as well. Hence, the design and
development-process of microactuators should simultaneously include the interfacing
and packaging aspects in order to successfully come to a working microsystem. In some
cases, where one cannot rely on standard technology (i.e. in the form of foundryprocesses3), it may be even necessary to design the microfabrication processes hand in
hand with the microsystems.
2

E.g.: the deflection of a beam under its own weight scales as l2. So a ten times smaller beam, bends 100
times less.
An example of a foundry service is the MUMPS process offered by Microelectronics Center of North
Carolina (MCNC). This process currently contains a surface micromachining technology consisting of 3
polysilicon structural layers in combination with the required sacrificial layers and a siliconnitride layer
(i.e. for electrical isolation).

GK+NT

MESA+ Research Institute

ASSET

Micromechanical Actuators

The foregoing also reveals the multi-disciplinary nature of microsystems technology; not
only are the mechanical properties of structures important, but generally all the physical
domains that take part in the transduction of the signal. This may involve such things
as the actual actuation (e.g. an electrostatic combdrive), mechanics (e.g. a bending
flexture) and possible signals in various domains for sensing to allow for feedback (e.g.
read-out of a displacement) and electronics for further signal processing. Moreover, the
microsystem needs to be packaged. Although packaging of electronic integrated circuits
is not a trivial thing, packaging a microsystem may even be more challenging by the very
sole reason that a microsystem should somehow "be open" to the environment in order
to benefit from the actuation or allow for sensing. Sometimes it may be possible to have
actuation/sensin completely within the package (e.g. a Digital Mirror Device with an
optically transparent window) but normally, apart from an electrical interconnection,
one needs at least one other physical connection. Also from this point of view, the design
and development of microsystems requires multidisciplinary teams and system
approaches, rather than sole monodisciplinary depth.

MEMS technology
In this section some very basic remarks will be made about MEMS technology. Rather
than going into any detail of technological matters, the purpose of this section is to give
some background of the technology that has enabled the development of microactuator
structures, which are the prime subject of this report and which will be discussed in the
next section.

Silicon
Although the term MEMS is not restricted to silicon micromachining, most of today's
MEMS technology is based on silicon. This is, off course, for good reasons. Silicon monocrystalline wafers offer a good combination of qualities, ranging from ideally elastic (no
creep4 or hysteresis) to a good heat conductor, from low to intermediate electrical
conductivity (depending on type and doping), from having a small thermal expansion
coefficient to being stable up to high temperatures [Petersen82]. More importantly,
silicon wafers are produced and used on a large scale for integrated microelectronics
resulting in low prices and availability of all kinds of compatible equipment. Additionally
(with certain restrictions on technology) silicon wafers allow for monolithic integration of
mechanical and electronic functions in one and the same chip offering a high potential
for both cheap and advanced microsystems5. Therefore, the majority of MEMS devices is
made on silicon wafers as the starting substrate.

Photolithography
Probably the most decisive characteristic for calling something "a MEMS device" is the
use of photolithography in its fabrication. Basically, photolithography comes down to
applying a layer of a light-sensitive material (photoresist) on a flat substrate,
4

Creep is the phenomena that a material continues to deform under a constant load. Hysteresis on the
other hand is the effect that the deformation of a material is dependent on its history, being for example
different for in-or decreasing loads.
A good example of a highly developed, advanced as well as economic MEMS device, showing both
mechanical and electronic functionality, is the impact sensor often found in today's cars for triggering of
airbags.

GK+NT

MESA+ Research Institute

ASSET

Micromechanical Actuators

illuminating it by some source using some pattern (a mask) thereby making the
illuminated parts either soluble6 or insoluble and then removing the soluble parts.
Spin photoresist

Silicon

Oxidise

Develop

Etch oxide & remove


photoresist

Illuminate
Mask

Light

Figure 1: Various steps in a photolithography process (after [Elwenspoek00]).

The way the photoresist is illuminated can vary. One way is to have a mask which is
more or less pressed against the photoresist-layer (contact). In another set-up a mask is
close to the substrate (proximity) whereas in the third method a mask is put in a
projection system to have a projection of the mask on the photoresist-layer. The
photoresist-layer is usually the first layer used to further shape the underlying
structure. This is normally done by material-selective etching processes (i.e. in which
the photoresist layer is etched much slower than the underlying material). On their turn,
the underlying layers may be well used as mask layers for further processing7. Since
photolithography only allows to define two-dimensional patterns, the structures that can
be fabricated are only moderately shaped in the direction perpendicular to the plane of
the photoresist layer. However, special etching processes as well as repetitive use of
deposition-photolithography-etching-cycles can reduce this limitation (see e.g. figure 2).

Growth and deposition


To make structures in or on silicon, one needs almost always additional materials.
Either to obtain specific properties related to these materials or because one needs a
layer to shield the underlying material during a specific process (i.e. etching). To this end
various methods exist. E.g. layers can be grown by using part of the underlying layer in
an oxidation process (i.e. the growth of siliconoxide on top of silicon, thereby using the
top part of the silicon wafer). Alternatively all required material for growth is transported
to the substrate. Here again there are choices; either the material gets to the substrate
in the form it needs to be grown in or in the form of intermediate products which react to
the eventual compound at the substrate. In the first case, depending on the material
there may be a choice of deposition methods. For atomic species (metals) evaporation
and/or sputtering may be possible. Evaporation is only possible for materials with a
high vapour pressure and can only be carried out under vacuum conditions. Sputtering
on the other hand relieves the high vapour pressure condition by using charged
particles (i.e. argon ions) to remove material from a target prior to deposition on a
substrate. For compound materials Chemical Vapour Deposition (CVD) may be more
suitable. Important materials that can be deposited by CVD are silicon-nitride
(chemically inert, hard material, electrical insulator with moderate thermal conductivity),
6

Photo resists are said to be of the positive type if they become soluble when exposed to light. Negative
photoresists become insoluble when exposed to light.
This is for example the case when a thin silicon-oxide layer is etched to make openings to the underlying
silicon. The silicon, then, is etched using another etching process which does not etch the silicon-oxide
layer.

GK+NT

MESA+ Research Institute

ASSET

Micromechanical Actuators

poly-silicon (having properties which may resemble those of single crystalline silicon and
often used as structural layer and/or conductor (when doped)), silicon-oxide (often used
as mask material or sacrificial layer, see below) and phosphor silicate glass (PSG, mainly
used as sacrificial layer).

Doping
Doping of silicon, mostly with either boron or phosphor, is not only a means for
changing the electrical properties of silicon. It also changes mechanical properties as
well as the etching behaviour in certain etching solutions. For example etch rates of
anisotropically wet chemical etched silicon drop significantly when silicon is doped with
boron. This fact can be used to have well-defined stopping positions for etching, rather
than timed etching. Doping can be obtained in various ways. One is ion implantation
where high energy ions from an accelerator hit the silicon and penetrate to a certain
depth, precisely controlled by the energy of the ions, into the substrate. Implantation
energies of keV's are required to reach depths of a few 100 nm. Depending on the usage
of the doping an annealing step may follow to spread the doping more homogeneously
over a thicker layer. Another method of doping is by indiffusion at elevated
temperatures. This can for example be accomplished by putting the silicon wafers, side
by side with boron or phosphorous wafers, in a tube at high temperature. Doping can
also be achieved by using dopant sources while depositing poly-silicon layers in CVD
equipment. This process is often used to fabricate electrically conducting or
piezoresistive structures. Moverover, this deposition method is very well suited to
fabricate movable, conducting, mechanical stiff structures as required for
microactuators (e.g. comb-drives, see below).

Bulk micromachining
Roughly, bulk micromachining is the technology by which structures are made in the
silicon substrate rather than on top of it by selectively removing material. The purpose of
this technology is to make structures that are released or undercut such that elements
come into existence that can somehow move with respect to the frame of the wafer
[Kovacs98]. The technology has many aspects [Elwenspoek99] but probably one of the
most important is the etching of silicon, be it a small layer, or entirely through the
silicon wafer. There exist various methods for etching. The most important flavours are
wet-chemical etching and dry etching (e.g. Reactive Ion Etching, RIE). Etching methods
can also be distinguished on the basis of the etch-profiles they produce: isotropic (etchspeed equal in all directions leading to rounded structures), anisotropic (etching speed
highly dependent on the crystallographic directions leading to sharp edges and corners)
and directional (mostly due to geometrical effects of dry etching. Examples of anisotropic
wet-chemical etching are etching in KOH (potassium hydroxide), EDP (ethylenediamine
pyrocatechol) or TMAH (tetramethyl ammonium hydroxide) solutions (see figure 2 left).
Isotropic wet-chemical etching is obtained using solutions containing e.g. fluoride and
can be tailored using doping or voltage or light assisted etching [Tjerkstra99]. Dry
etching has a great variety of implementations which share some similarities: they all
take place in a vessel which can be brought at "certain vacuum" (10-5-0.5 Torr) and they
all use accelerated (reactive) particles (ions, atoms) for etching. Additionally chemical
species may be added to assist in etching (speed and selectivity) or passivation. Etching
mechanisms can vary from physical removal due to bombardment (directional) to
chemical reactions followed by removal of (volatile) reaction products. By combination of

GK+NT

MESA+ Research Institute

ASSET

Micromechanical Actuators

both effects, tailoring their mutual influence, it is possible to obtain etching processes
that vary between isotropic and almost perfectly directional.

Figure 2: Left: example of a structure fabricated in silicon by anisotropic etching in a KOH


solution [Oosterbroek99]. Right: structures etched out of silicon by means of powderblasting
[Wensink00].

The etching methods mentioned above share the feature that they generally result in
moderate etching speeds (microns/minute). Sometimes one may want to etch over large
depths and large areas in the shortest times possible with limited requirements on
accuracy. In this case a technique well-known from precision engineering may come in
handy: powder-blasting. The technology is relative new in micromaching but seems to be
able to "fill the gap" in etching needs between micromachining and precision engineering
(10-100 micron, see figure 2 right).

Surface micromachining
Although bulk-micromachining offers many possibilities to fabricate interesting
structures, entirely new structures have become possible due to surface micromachining
[Howe83, Howe86, Bustillo98]. One of the goals of this technology is to make small
features just above the (silicon) substrate which can move (partly) free from the
substrate. This is accomplished by using one layer as a sacrificial layer. Etching away
this layer, a structural layer becomes free from the substrate. The technology is drawn
schematically in figure 3. In a first step (A) a sacrificial layer is deposited on the silicon
substrate. Subsequently this layer is shaped using photolithographic and etching
techniques (B). Next a layer of the structural material is deposited over the sacrificial
layer (C). Again using photolithography and etching the structural layer is shaped and
holes are made to allow etching fluid to reach the sacrificial layer in selected areas (D).
Finally the sacrificial layer is completely removed by selective wet-chemical etching
thereby partly releasing the structural layer. Various combinations of structural and
sacrificial materials exist. E.g. silicon nitride and poly-silicon, gold and titanium, nickel
and titanium, polyimide and aluminium, tungsten and silicon-dioxide and aluminum
and polymer. However, the most widely used surface micromachining technology uses
poly-silicon as the structural material and either silicon-dioxide or phosphor silicate
glass (PSG) as the sacrificial layer [Elwenspoek99].

GK+NT

MESA+ Research Institute

ASSET

Micromechanical Actuators

Figure 3: Surface micromachining. Left: processing sequence for micromachining (after


[Elwenspoek99]). Right: examples of structures made by surface micromachining: a ball
bearing like groove for a wobble motor [Legtenberg96a] and a shuffle-motor [Tas00]

Bonding and Chemical Mechanical Polishing


To even further enhance the possibilities and complexity of micromachined devices
(partly structured) wafers (of various materials) can be bonded together. Again various
methods exist. Anodic bonding can be applied to a pair of a good conducting layer
(metal, silicon) and a slightly conducting layer (glass). A voltage applied over the
waferpair induces a small depletion layer in the glass wafer resulting in a large electrical
field over a short distance near the interface of the wafers. The resulting electrostatic
pressure brings the wafers in such close contact that they may join together. Executing
the process at high temperatures (400-500 C) will allow the formation of covalent bonds
at the glass-conductor interface. Anodic bonding is possible for wafers with a roughness
of less than 1 micrometer. Another method of bonding is silicon fusion bonding. This
type of bonding can occur between layers that have a roughness of 1 nanometer or less.
For wafers having such smooth surfaces, they do not contact at 3 points but at large
areas. The bonding of wafers covered with native oxide is thought to originate from
hydrogen bridging [Stengl89] which can be promoted by annealing in a wet-oxidation
oven at temperatures of about 700 C.
Since the requirements for silicon fusion bonding, especially the extreme smoothness of
the wafers, are generally hard to meet, it is a difficult process. However, there is a
technology which can help to make the process more applicable: Chemical Mechanical
Polishing (CMP).

GK+NT

MESA+ Research Institute

ASSET

Micromechanical Actuators

Figure 4: Left: Schematic view of a CMP machine. Right: LPCVD Si3N4 surface before (top)
and after (bottom) CMP (RMS roughness values are reduced from 3.6. nm to 0.4 nm [Gui98])

In CMP the wafer is placed on a rotating polishing head and moved under a certain
pressure over a polishing pad. A slightly alkalic slurry containing nanometer sized silica
particles is added. The combined action of wear and etching yields the smooth surfaces.
This process can also be applied to other materials among others silicon nitride, silicon
oxide and poly silicon [Gui98]. It appears that all materials that can be polished down to
a roughness better than 1 nm are bondable [Berenschot94].

Some actuator backgrounds


Brief transducer theory
Energy and force
Actuators are structures that allow to input energy in one domain to get a certain type of
motion (translation, rotation) in the mechanical domain. Hence they are transducers
with at least two ports in two physical domains. Although there exist a number of types
of transducers we will restrict our discussion to the buffering generator-type transducers
(i.e. non-dissipating, non-modulated transducers). These type of transducers can be
largely described and understood from a thermo-dynamical analysis.

Az

+
x
y

u
Ax

Ay
Figure 5: Schematic of an electrostatic actuator consisting of two charged plates which can
move relative to each other.
GK+NT

MESA+ Research Institute

ASSET

Micromechanical Actuators

Central in the description is (are) always a (various) buffer(s) of energy. Next, to describe
the transducer it is necessary to determine the energy input into the structure and to
derive an expression for the entire energy contained within the structure. By carefully
looking at the total differential of these energy functions it is possible to relate
input/output of energy from one port to another.
As an example we will look at an electrostatic actuator given by two charged plates from
which the upper plate can move with respect to the lower plate (see Figure 5). Neglecting
stray-fields we can describe the energy contained in the volume V between the plates by
the electrostatic energy contained in the electrical field.
W = wV =

1
D2
EDV =
V
2
2 0

(1)

where W is the energy, w the energy density, E the electrical field, D the electrical
displacement and 0 the permittivity of vacuum. D can be easily related to the charge
density on the capacitor plates using Maxwells equations:
W=

2
V
2 0

(2)

Now writing the volume between the overlapping area of the plates as xyz, where z is the
distance between the plates, W can be written as:
W=

q 2z 2
= W (q , V )
2 0 V

(3)

In order to input energy into the system we can do one of two things: we can displace
the movable plate, inputting mechanical work into the system, or we can charge the
capacitor, inputting electrical work into the capacitor. Both contributions lead to:
dW = pdV + udq

(4)

here p is the external pressure which we need to apply to counteract the electrostatic
pressure encountered when we try to change the volume containing the electrostatic
energy. Taking the derivatives of W with respect to V and q, while keeping the other
variable constant, we find:
q 2z 2
W
p=
=
2 0 V 2
V q

(5)

W
qz 2
=
u =
q V 0 V

(6)

Note that p=negative, i.e. the electrostatic pressure tries to decrease the volume between
the plates and we need a negative pressure to counteract this8. It is interesting to look at
the external force that we need to apply to move the plate in a certain direction. This
force can be found from:
W i
Fi = A i p = A i

i V q

(7)

Although a negative external pressure may seem rather odd, by multiplying the pressure with an area we
simply find a force pointing outward the volume i.e. to increase the volume work needs to be done by the
external force.

GK+NT

MESA+ Research Institute

ASSET

Micromechanical Actuators

where i is x, y or z and Ai is the applicable area perpendicular to the direction of


movement (i.e. Ax=yz). Using the above formula we find for movement of the plate in the
direction parallel to the plates (i.e. in x or y direction):
2
q 2z 2 1
y
W x
= q z = u2 0
Fx = A x
= zy
x 2 0 xyz yz
x 2 0 A z
2z
x V q

(8)

where use has been made of q=uC=uAz0/z. Note that Fx points in the negative xdirection (see Figure 5).
For the force in the direction perpendicular to the plates we find:
2
q 2z 2 1
0Az
W z
= q
Fz = A z
= u2
= xy
z 2 0 xyz xy
2 0 A z
2z 2
z V q

(9)

In this case the force points in the positive z-direction.


Other types of actuators can be described in an analogues way the prerequisite being
that one must be able to derive an energy function like (3) in terms of extensive
parameters like q and V. In order to get transduction it is always necessary that (at least
one of) the buffer function(s) depends on some variable(s) (e.g. in the above case the
capacitance is the buffering element in the system and it depends on x, y and z).

Stability
The stability of transducers can be investigated by looking at effective spring-constants.
These are found by taking the derivatives:
K eff ,i = i Fi

(10)

where i is again x, y or z. For any transducer, Keff,i should be positive in order to have a
stable transducer [Tas00]. Most often actuators are stabilised by means of an additional
spring (with spring constant K). The resulting effective spring constants are given by:
K eff ,x (q ) =

q 2z
+K
x 2 2 0 A z

(11)

K eff ,x (u ) = K

for the lateral movement and


K eff ,z (q ) = K
K eff ,z (u ) = u 2

0Az
z

(12)

+K

for the normal movement. From the above formulas it follows that lateral operated
electrostatic transducers cannot become unstable whereas normal operate electrostatic
transducers can get unstable when they are controlled by the voltage on the plates.
When no external force is applied to stabilise the transducer the effective springconstant decreases with increasing voltage, becoming zero at the pull-in voltage:
u pi =

8 Kz 30
27 0 A z

(13)

where z0 is the distance between the plates when no voltage is applied, i.e. when the
spring is relaxed.

GK+NT

MESA+ Research Institute

10

ASSET

Micromechanical Actuators

Some actuator scaling laws


Electrostatic forces9
The electrostatic force between capacitor plates is given by (7) with W obtained from (3).
Assuming that we are able to keep w constant (this is for E or D () is constant) as we
scale the actuator, we find that
Electrostatic energy:
Electrostratic force:

Wel~l3.
Fel~l2.

where l is a unit dimension. Consequently, electrostatic forces increase relative to the


volume when the systems shrink.
The maximum field in air is not independent of the distance of capacitor plates if they
approach closely. This has been observed first by Paschen. Figure 6 shows the
breakdown voltage in air and Ar as a function of the product of plate separation and
pressure [Thornton00]. If the separation between the plates is roughly of the order of the
mean free path of molecules in air, the maximum field increases.

Figure 6: The DC breakdown voltage as a function of the gas pressure p and the electrode
spacing d for plane parallel electrodes in air and Argon. Such curves are determined
experimentally and are known as Paschen Curves [Thornton00]

If one is able to use materials with a high dielectric constant (e.g. PZT - lead zirconate
titanate has r=1300) the energy density and the total output-force can be increased over
the actuator with air gap by the amount of the dielectric constant. In theory this
increases the maximum force of electrostatic micromotors to about 1 N.

Piezoelectric forces
Piezoelectric forces obtained by capacitor like structures filled with piezoelectric
materials can be described analogues with electrostatic actuators. Under the assumption of constant attainable energy-density one finds:
Piezoelectric energy:
Piezoelectric force:

Wel~l3.
Fel~l2.

This section on scaling has largely been adapted from [Elwenspoek00]

GK+NT

MESA+ Research Institute

11

ASSET

Micromechanical Actuators

One difference with electrostatic transducers is that the occurrence of breakdown will
strongly be determined by the quality of the materials used. This on its turn may largely
depend on the deposition methods and conditions under which thin films are fabricated.

Magnetic forces
Magnetic forces can be calculated in an analogous way. The magnetic field energy
density is given by (with B, H and the magnetic induction, the magnetic field and the
permeability, respectively)
w mag =

1
B2
BH =
2
2

(14)

so one is tempted to assume that the energy in a gap of a magnetic circuit scales with
the volume and the forces with the surface, but this conclusion is too quick. For
magnetic fields obtained by current carrying wires, the magnetic induction in the gap is
given by
B=

1 n
I
0 L

(15)

with n the number of turns of wire, L the total length of the coil and I the current
through the coil. When shrinking the coil and the magnet, it may be possible to keep n
constant, but the cross-section S of the wires must shrink, so that it will be impossible
to keep I constant. It is more reasonable to keep the current density constant, but this
means that B scales like l! The energy density then scales like l2. We have:
Magnetic energy: Wmag~ l5.
Magnetic force:
Fmag~ l4.
Some care is necessary here: The maximum current density is limited by the dissipation,
accordingly by the maximum temperature the system can stand. Small systems loose
their heat faster than larger systems: heat flow is proportional to the temperature
gradient, which is proportional to l-1. Taking this into account, one would end up with
Fmag~ l2. However, it depends very much on the details of the design if one really can use
the large temperature gradients. Furthermore, the assumption that the number of wires
is independent on the system size becomes doubtful if the coils shrink below 1 mm3. In
practice, magnetic actuators smaller than 1 mm3 are hardly feasible. Any comparison
between magnetic motors and electrostatic motors will end up that at the small end the
electrostatic motors are stronger, while at the large end the magnetic motors are
stronger.

Electrostatic microactuators and micromotors


As well as there is an abundance of macro-scale actuators, there is an abundance of
micromechanical actuators. There is both a large variety of actuation means as well as
possible implementations. Moreover, the obtained movements can also show variety:
linear motion, stepping motion, rotary motion, and combinations of movements with
multiple degrees of freedom (either wanted or unwanted). It will be clear that such a
variety cannot be covered in a report of 30-40 pages or so. Even in books explicitly
targeting microactuators only a sampling and brief description of the many published
actuators can be found. A good example of such a book is [Tabib-Azar97] where a large
collection of microactuars can be found. Therefore in this report a small number of
GK+NT

MESA+ Research Institute

12

ASSET

Micromechanical Actuators

actuators will be described, just the ones that the authors came across and found
interesting to mention here. As such this section only will give a flavour which hopefully
serves as an invitation for further reading.
For various reason we have limited this report to electrostatic actuators. For one
electrostatic micro-actuators have been investigated far more than any of the other
principles. Likely one of the reasons for this is their relative easy, mostly CMOScompatible fabrication. After all, all one needs for an electrostatic actuator is conductive
insulating and structural materials which all happen to be available in CMOS
compatible technology. Many other actuator principles lack this property since they
often require specific materials (magnetic, piezo-electric transducers). Another reason to
choose electrostatic microactutors is that they show a rich variety of implementations
which makes them excellent as examples of the exciting possibilities delivered by microelectro-mechanical actuators.
Following Mehregany [Mehregany91] we define motors as actuators capable of
unrestrained motion in at least one degree of freedom. In case of a linear motor, the
motion will practically always be restrained by the size of the guidance. It is called a
motor, if the generated stroke is created by an addition of small motions, which in
principle can be repeated infinitely10.
The history of electrostatic motors goes back to the 18th century. Andrew Gordon
constructed the motor known as the electric bells. Around the same time, Benjamin
Franklin built his spark motors, which were able to generate a considerable power of ca.
0.1 W [Moore73]. The size of these motors was huge: the rotor was a disc with a diameter
up to around one meter. The power density of these motors was therefore small, caused
by the fact that in large air gaps the break-down electric field and therefore the energy
density is small. It is because of the low power density that electrostatic motors have not
found wide-spread application. An improvement was made by Poggendorf [Poggendorff70], who constructed a so called corona motor. In the Franklins motor only the
electrodes at the edge of the rotating disc were charged and contributed to the torque
generation. In the corona motor sharp needles spray charge on a glass disc. Therefore,
the whole disc surface became active in the torque generation. Variable capacitance
motors were introduced by Zipernowski in 1889 [Zipernowski89]. In this type of motor,
forces are generated by electrostatic attraction between oppositely charged (conductive)
capacitor plates. For more details and references, we refer to Jefimenko [Moore73],
which gives an extensive overview of realized electrostatic motors.
Bolle and coworkers have studied electrostatic motors in the 1960s [Bolle69]. They
constructed rotary electrostatic motors with a size in the order of 1-10 cm. Both
synchronous variable capacitance and asynchronous induction motors were studied.
Motors have been constructed using precision fabrication techniques. Bolle reports an
output power of 600 W at 1 rev/s, driven by 200 V for a synchronous motor. Starting in
1988 several electrostatic micromotors have been developed within MEMS research
groups. These include side drive motors [Fan88, Tavrow90], top drive motors
[Mehregany90a], outer-stator wobble motors [Jacobson89, Trimmer 89], outer-rotor
wobble motors [Mehregany91, Furuhata93] and lower-stator wobble motors
[Legtenberg95]. In literature also a number of linear electrostatic motors has been found
[Fujita88, Akiyama93a, Akiyama93b, Yeh95, Yeh96, Egawa90, Tas97, Tas98].

10

Parts of this section have been published in [Tas00].

GK+NT

MESA+ Research Institute

13

ASSET

Micromechanical Actuators

Linear motion
Combdrive
A large number of capacitors in parallel, arranged like two combs of interdigitated
fingers, one of which is able to move in a certain direction, is usually called a combdrive.
The electrostatic working principle of such devices is easily understood from equation
(9). In order to obtain stability and co-linear movement in two directions flextures are
required since the electrostatic force always enforces the combs to move inwards, see
Figure 7.

Figure 7: Left: Close-up of a large comb-drive actuator with flexures in the foreground.
Right: schematic of the structure of a single unit of a combdrive.
(Courtesy of R. Legtenberg [Legtenberg96b]).

In a way comb-drives are synonymous with the success of MEMS in general and surface
micromachining in particular since the structures have been used in countless applications both as actuators and sensors (i.e. in gyroscopes and accelerometer sensors).
Since comb-drives are most often fabricated in surface micromachining technology the
surface of the combs is generally small leading to either small forces or small
capacitances hampering capacitive measurement schemes. However, as the number of
fingers per comb can be varied (between certain limits), the structures show a nice
scaling behaviour (as far as output force is concerned, of course at the expense of lower
resonance frequencies at increasing number of fingers).
Comb-drive structures exhibit various advantages. They have no sliding surfaces making
them virtually showing no wear. The force of the drives is independent of the position of
the combs when one uses (the preferred) voltage control as can be seen from equation
(9). Moreover the control of the drives is relatively simple since only one voltage is
required per drive (as compared to the multiphase control signals required for stepping
motors). Last but not least the structures are unconditionally stable as far as the motion
parallel to the fingers is concerned. However, sideways pull-in of the fingers may occur
at higher voltages since the perpendicular forces acting from both sides of each finger
only cancel each other in the case of perfect symmetry [Legtenberg96a]. By symmetry
breaking (i.e. due to thermal motion or non-ideal processing) one of the perpendicular
force-terms may easily overrule the other since these forces increase like x-2. The
resulting instability poses an upper limit to the allowable voltages, next to breakdown
effects. Which of the two prevails depends largely on the precise design of the combdrives. If the side instability is the dominant effect, it can be derived that the maximum
deflection obtained by comb-drives is limited to [Legtenberg96b]:
GK+NT

MESA+ Research Institute

14

ASSET

Micromechanical Actuators

y max = d

kx
y
0
2k y
2

(16)

where y0 is the overlap of the fingers in the relaxed state, kx and ky are the lateral
stiffness of the fingers of the comb and the stiffness of the compliant structure
respectively and d is the distance between the fingers.
For further reading, studies on shaped finger comb-drives can be found in [Ye98],
application of comb-drives in laser beam scanning in [Kiang98], polyimide comb-drives
for lateral as well as perpendicular motion are described in [Kamiya99].

Gap closing actuators


Gap closing actuators basically consist of (one or more) parallel plate capacitors, in
parallel like in a combdrive geometry, moving in the direction normal to the fingers of
the comb (z-direction in figure 5). In general the movement per step is limited to a couple
of microns at most but using such motors in a stepping motion may result in a much
larger travel range [Tas97]. Control of the actuators is relatively straightforward: since
the pull-in effect is used to obtain a well defined step only the load influences the
required minimum excitation voltage.

Figure 8: Schematic of an array of two parallel plate gap-closing actuators [Tas97].

Figure 8 shows a schematic of a gap-closing actuator array [Tas97]. The bumper protects
the actuator against short-circuit (otherwise a insulator should be applied). The drive
voltage is applied between the stator plates (grey) and the moveable plates (black). The
moveable part is kept on the same potential as the substrate preventing it to be pulledin onto the substrate.

Walking or crawling motion


Walking motion was introduced to linear actuators by Burleigh [May75] in a patent for a
piezoelectric inch-worm motor. The motor consists of three basic elements: A front clamp,
a back clamp and a body which can contract (figure 9). The walking cycle starts with
only the front clamp activated (a). In the next stage the body contracts (b) and the back
clamp shifts forward. Next the back clamp is activated (c) and the front clamp is released
(d). The body relaxes and extends (e), shifting the front clamp forward. Finally the front
clamp is activated (f) and the back clamp released (a).

GK+NT

MESA+ Research Institute

15

ASSET

Micromechanical Actuators
front clamp

body

back clamp

(a)

(d)
stator shaft

(b)

(e)

(c)

(f)

Figure 9: Walking cycle of the Burleighs inch-worm motor.

Burleigh [Inchworm] has implemented the principle in several piezo-electrically driven


motors. These have a resolution of 4 nm and up to 200 mm of travel. The actuators have
a circular cross-section with a diameter in the order of 1 cm and a length of several cm,
depending on the maximum travel distance. The max. load is 10 N and the max. speed is
1.5 mm/s. Koster has implemented the walking concept in combination with a
mechanical amplification mechanism to increase the walking velocity [Koster94]. The
motor has a peak velocity of 30 mm/s, because of heating by dissipation in the PZTactuators, the continuous max. velocity is 3 mm/s. The motor includes a special
clamping design, which is less sensitive to variations of the diameter of the shaft, along
which the mechanism walks.

Walking motor
In the motors presented so far, the moving part contains the clamp- and displacement
actuators. Walking motion can also be implemented with a passive moving part, which is
driven by a pair of drive-units (legs), each containing a clamp and a pull actuator. The
drive-units are fixed to the substrate. This concept has been implemented by Yeh et al.
[Yeh95, Yeh96] for electrostatic actuation of microrobots. The implementation is realised
using the commercially available MUMPs surface micromachining process [Koester99].
clamp pull actuator anchor
c1

(a)

p1

(e)

shuttle
p2
c2

(b)

guidance

(f)

(c)
(g)

(d)

(h)

Figure 10: Cycle of the linear motor with passive moving part and drive
units fixed to the surroundings. See text.
GK+NT

MESA+ Research Institute

16

ASSET

Micromechanical Actuators

Figure 10 illustrates the walking cycle. The walking cycle starts with both pull actuators
relaxed, and clamp c1 activated to hold the shuttle (a). The shuttle is displaced by pull
actuator p1 (b) and clamp c2 is activated (c) to take over the clamping from c1. Next, c1
is released (d), p1 is relaxed (e) and p2 contracts to displace the shuttle (e). Then c1 is
activated (f), c2 is released (g), and p2 is relaxed (a). Clamping is generated by
electrostatic attachment of the clamp shoes to the shuttle. The step motion is generated
by gap-closing actuator arrays, each consisting of 10 rotor-stator beam pairs with a 60
m overlap, working in parallel. The actuator arrays are made of 3.5 m thick
polysilicon, which is obtained by stacking of two layers of the MUMPs process in which
the motor is realized. The produced force was estimated at 6 N with 35 V driving
voltage. The maximum deflection measured was 40 m in 20 steps of 2 m [Yeh96].

Figure 11: Microscope picture of a walking motor (1x1 mm2 ). The motor consists of a
shuttle and two drive units at both sides of the shuttle. Each drive unit consists of a clamp
actuator, a pull actuator, a clamp shoe and elastic beams that connect the actuators and the
shoe [Tas97].

In [Tas97] the design and fabrication of a linear electrostatic stepper motor is described.
Fabrication is simple using only a single mask surface micromachining process. The
motor consists of two drive units which alternately generate a step to move a shuttle.
Stepping cycles showed a deflection of 15 m, with steps in the order of 2 m, the
deflection being limited by the suspension. Generated force was 3 N at 40 V driving
voltage. The friction and adhesion in the clamps was measured and a friction coefficient
of 0.8 0.3 was found. The adhesion in the clamp was just low enough to release the
clamping shoe.

Inertial drive
The inertial drive was presented in MEMS by Higuchi [Higuchi90]. Figure 1.2 shows the
principle: A load with mass mload is connected to the help mass mhelp, by an actuator
element. The load is pressed on to the substrate with a force Fclamp, which induces
friction between them. The motor is actuated by rapid extension of the actuator (b). Both
mload and mhelp are accelerated with al and ah respectively. If the inertial force of mhelp
acting on mload is larger than the friction force, than the load will shift leftward. In the
next phase (c) the actuator generates a contraction force, and both masses are
decelerated slowly. If this is done subtle enough the inertial forces will be small
compared to the friction force, and mload will rest. Only mhelp is decelerated with ahelp
pointed leftward. At the end of the contraction, the actuator suddenly generates an
GK+NT

MESA+ Research Institute

17

ASSET

Micromechanical Actuators

extension force, in order to turn the direction of movement of mhelp. Due the mutual
inertial forces, both mload and mhelp are accelerated outward, and we have returned to
phase (b).
Fclamp

mload

Fclamp

mhelp

mload

mhelp

actuator

(a)

ah

(c)

Fclamp

Fclamp

al

mload

mhelp

al

ah

mhelp

mload

ah

(d)

(b)

Figure 12:Operation principle of an inertial drive

Higuchi [Higuchi90] has successfully used this concept to move robot arms. The typical
mass mhelp is in the range of 1-10 g. Stacked piezos with a size of 2x3x10 mm3 and a
driving voltage < 150 V were used. Step-sizes generated were between several nanometers up to ten m. The inertial drive has been improved by Ikuta et al. [Ikuta91,
Ikuta92] by including clamp actuators in order to control the friction between the
moving load and the substrate or the guiding rail.

Elastic Inertial drive


This propulsion principle is based on the conversion of an oscillation, into directional
motion. The conversion is accomplished by mechanical rectification using an elastic fin
connected to the translator [Uchiki91].
Fclamp

m1

vx1(t)
ay2(t)

Fclamp

k1

m2

m1
k1
1/2 k2

1/2 k2

Ff(t)
F21(t)

(a)

(b)

Figure 13: Operation principle of an elastic inertial drive.

Figure 1.3 illustrates the principle. The configuration consists of a translator with mass
m1 which is in contact with the oscillating mass m2 by means of the fin which is

GK+NT

MESA+ Research Institute

18

ASSET

Micromechanical Actuators

connected to m1 by a hinge joint. The elastic deformation of the fin is represented by the
its rotation and by the spring k1. The forces working on the translator m1 are shown in
figure 1.3b. The translator is clamped onto the oscillating mass with a constant force
Fclamp. Mass m1 moves in the x-direction with a speed vx1(t). Mass m2 is connected to the
substrate by the spring k2. The oscillation of m2 is represented by its acceleration ay2(t).
There are two forces working on the tip of the fin: The contact force F21(t) and the friction
force Ff (t), which we assume to be proportional to F21(t) with the friction coefficient .
The mechanical rectification is based on variation of F12 and therefore of the friction
force Ff , due to the oscillation of m2. Assume that the resonance frequency (k2 / m2) is
larger than the resonance frequency (k1 / m1). If m2 > m1 this means that k2 >> k1. It
means that the mass m1 is not able to follow the oscillation of m2. As a result the spring
k1 is deflected. In the positive flank of a2y the spring will be compressed, and F21 will be
large. Because k1 is placed obliquely, the compression of k1 also induces acceleration of
m1 in x-direction. This acceleration is limited by the friction force, and is therefore
proportional to F21. During the negative flank of a2y the spring k1 extends so F21 will be
smaller than in the positive flank. During the extension of k1 the friction force will tend
to decelerate m1. Because F12 is smaller in the extension phase, the decelerating friction
force will be smaller than in the compression phase. Therefore, in the complete cycle
there is a net acceleration of m1 to the right.
The concept has been implemented by Racine et al. [Racine93]. They build a hybrid
rotation motor with a silicon micromachined oscillator, activated by a ZnO thin-film
layer. The rotor was made by a laser cut metal film, with tilted flexures made by pressing
in a mould. The size of the assembled motor was 6x6x1.5 mm3. The motor driven by 10
VRMS delivered a free rotational speed up to 200 rpm, and a torque of 50 nNm. A similar
motor with a PZT thin film actuation has been realized by Muralt et al. [Kohli95]. The
use of PZT in stead of ZnO results in a three times higher speed per Volt, and a factor
two larger torque per Volt [Kohli95].

Impact Drive
Impact-driven microactuators were first presented by Lee et al. [Lee93]. In their design
the slider was driven by impact of a flexure connected to a resonating mass-spring
system. This system is almost similar to the inertial elastic drive presented in the
previous section. The difference is that the flexure is connected to the resonating mass.
Daneman [Daneman95] created microvibromotors driven through stiff, oblique impact by
resonating masses (Figure 1.4). The resonators are actuated by electrostatic comb-drives
[Tang89a, Tang89b, Tang90]. In resonance, the maximum produced electrostatic force is
amplified, roughly by the resonator quality factor. For laterally driven microstructures
the quality factor can be as high as 100 in air [Daneman95, Cho94]. In order to create
impact, at least 4 driving cycles at the resonance frequency were needed. To obtain
repeatable steps, 6 driving cycles were applied for each step. The typical measured stepsize was in the order of 0.3 - 0.4 m, at a driving voltage of 50 Vdc and 12.5 Vac (p-p)
superimposed [Daneman95]. Slider velocities over 1 mm/s, and displacements of the
slider up to 100 m have been reached. The value of the generated forces have not been
reported, but they were large enough to overcome the (adhesive) static friction of the
slider with the substrate and the flange. The linear microvibromotors have been
successfully employed to position a micromachined mirror in a laser-to-fiber coupler
system [Daneman96].

GK+NT

MESA+ Research Institute

19

ASSET

Micromechanical Actuators

impact arm

elastic
suspension
comb-drive
actuator

slider

Figure 14: Top view of a impact driven linear motor, by Daneman et al. [Kohli95]. This so
called microvibromotor consists of a slider which is driven by oblique impact of resonating
impact arms. The resonators are driven by comb-drive actuators. The motor contains four
drive units, two for each driving direction.

Zigzag Drive
The zigzag concept has been proposed by Anderson et al. [Anderson91] in 1991. Koga et
al. [Koga96] have implemented the concept, using (bulk) micromachining of Pyrex and
silicon wafers, and anodic bonding to join the glass and silicon parts.
stator
electrode

+V
+V

translator

(a)

0
0
(c)

0
0

+V
+V

+V
0

0
+V

(b)

(d)

+V
0

0
0
+V

Figure 15: Principle of operation of the zigzag drive.

The principle of operation is illustrated in figure 1.5. The motor consists of a translator
with a length l zigzagging in a channel, between four electrodes. Electrostatic forces are
generated by applying a voltage difference between the grounded slider and the stator
electrodes. The cycle is as follows: (a) The translator is attracted by both upper
electrodes. (b) One of the lower electrode attracts the back of the translator. The
translator rotates around the front contact line, and the back slightly moves forward. (c)

GK+NT

MESA+ Research Institute

20

ASSET

Micromechanical Actuators

The front is attracted by the lower front electrode. Rotation is around the back contact
line, so the front moves forward (d) The back moves to the upper back electrode and
moves forward. Finally the front moves upward (a) and the cycle is completed.
By making the zigzagging motion illustrated in figure 1.5, small steps are made. For
small angles g / l, the leftward movement x of the translator as defined in figure 1.5a, is
given by gy / l - y2 / 2l, with y the upward movement, l the translator length, and g the
gap. Due to the large transformation ratio (leftward movement: upward movement), the
generated lateral force can be much larger than the normal attractive forces between the
stator electrodes and the translator.
The dimensions of the fabricated motors are 6 x 7 x 1.2 mm [Koga96]. Motors have been
tested successfully. They were operated at 80-250 V, and could travel 1 mm at a speed
of 0.3 mm/s. At 250 V the produced force is in the order of 0.1 mN. This is smaller than
what can be calculated from the large transformation ratio and the (normal) electrostatic
forces. Probably, the lateral force is limited by the friction force in the stator-translator
contact, which will be in the order of the normally acting electrostatic forces.

Travelling Field Surface Drive


Egawa and Higuchi [Egawa91] introduced their multi-layered electrostatic film actuator
in 1990. This first version was an electrostatic pulse driven induction motor. It is
propelled by electrostatic forces between charges on the stator electrodes, and induced
charges on the highly resistive slider surface. Therefore, it has been called the image
charge stepping actuator. By moving the voltage pattern on the stator electrodes, the
slider is dragged and pushed by the electrostatic forces.
resistive layer

slider
electrode
stator

insulator

+++

---

+++

---

+++

--+V
-V
0

(a)

---

+++

---

+++

+++

---

+++

---

+++

--+V
-V
0

(b)

---

---

+++

+++

---

+++

+++

---

+++

--0
+V
-V

(c)

---

+++

---

+++

+++

---

+++

---

+++

--0
+V
-V

(d)

Figure 16: Schematic of a travelling field inductive surface drive motor.

GK+NT

MESA+ Research Institute

21

ASSET

Micromechanical Actuators

The motor consists of a stator with electrodes, and a slider with two layers: an insulating
layer and a highly resistive top layer, in which a charge distribution can be induced by
the electric field generated by applying a voltage pattern to the electrodes. Figure 1.6
shows the principle of operation. Initially, a charge distribution is induced on the
resistive layer by applying a voltage pattern on the stator electrodes (figure 1.6b). Next,
the voltage pattern at the stator electrodes is shifted to the right (figure 1.6c). The
attractive electrostatic forces force the slider to shift to the right, to the next equilibrium
position (figure 1.6d). Stage (c) and (d) can indefinitely be repeated. After shifting the
voltage pattern three times, the initial voltage pattern returns. These actuators have
some attractive features: Because the slider does not contain electrodes which have to
be contacted, there are no wires or (conductive) sliding contacts needed between the
slider and the outside world. Furthermore, due to the repulsive forces at the start of the
propulsive phase no bearings are needed, as the propulsive forces reduce the friction
significantly. In the realised motors, the stator electrodes and the translator surface
coating were applied on PET films with a thickness of about 100 m. Copper electrodes
were plated on the stator to create the electrodes. The translator was coated with highly
resistive layer [Egawa90, Egawa91] either an anti-static agent [Egawa90], a resin
[Egawa91] or polyurethane mixed with carbon particles [Egawa91] were used. For proper
operation the resistivity of the translator surface was chosen around 1014 [/square]
[Egawa90, Egawa91]. In order to increase the break-down field strength, the actuator
was mounted in a package filled with dielectric fluid (Fluorinert, by 3M). Several design
optimisations have been published, for example the optimal stator-translator electrode
spacing compared to the electrode pitch, in order to obtain maximum force generation
[Egawa91], the introduction of 18 m glass balls between the stator and the slider to
reduce friction [Niino94], and special electrode design which reduces the force ripple
[Yamamoto98]. With the pulse driven induction motor a maximum force of 8 N and a
power to weight ration of 5W/kg have been reached, using a driving voltage of 800 V.
The motor weighs 110 g and consists of 40 stacked layers. The outer dimensions are 20
x 14 x 50 mm3 [Niino94].
Niino [Niino94, Niino92] introduces three other driving principles for the film actuators:
ac-driven induction, pulse-drive variable capacitance, and ac-driven variable
capacitance motors [Niino94]. The variable capacitance motors have the advantage of a
larger power generation, and the disadvantage of larger complexity due to the necessary
of electrodes on and electrical connections to the slider [Niino94].

Scratch Drive
This propulsion principle is invented by Akiyama [Akiyama93a, Akiyama93b], and has
been implemented as an electrostatic, surface micromachined motor. The operation
principle of the scratch drive actuator (SDA) is illustrated in figure 17.
The SDA consists of a thin (1 - 2 m) polysilicon plate, with a so called bushing at one
end of the plate (figure17a,d). Figure 17a-c illustrate the stepping cycle. In the activated
state (figure 17a) the plate is attracted to the lower electrode (substrate). As the
actuation voltage switches from positive in negative, the plate relaxes shortly (fig 17b).
This causes the bushing to rotate. Because the friction between the bushing and the
insulator is higher than between the plate and the substrate the bushing does not slide,
and the centre of rotation is the line where the bushing touches the insulator. Therefore,
the result of this movement is that the plate shifts slightly forward (figure 17b). The plate
is pulled down again, and the bushing slides forward. Next, the state of figure 17a is
reached again and the cycle can be repeated.

GK+NT

MESA+ Research Institute

22

ASSET

Micromechanical Actuators

Figure 17: Operation principle of the Scratch Drive Actuator


(from [Akiyama93a, Akiyama93b]

In tested SDAs the length of the plate is between 50 and 80 m, and bushing height is
around 1.5 m [Akiyama93b, Akiyama95]. The step-size increases with increasing
driving voltage, and lies between 10 and 100 nm, for driving voltages between 40 V and
200 V. Measured speeds range between 5 and 80 m/s, at driving frequencies between
100 and 1000 Hz [Akiyama93a, Akiyama95]. The maximum measured force is 63 N at
112 V pulse peak voltage [Akiyama95]
A problem with the SDA is that the operation principle is not fully understood. In
particular, it is not clear why during the relaxation of the plate, the friction in the
bushing-insulator contact is larger than in the plate-insulator contact. Consequently, it
is not clear too if there are critical parts in the structure. For example, the shape of the
bushing edge touching the insulator might be important to obtain the desired frictional
behaviour. In addition, it is difficult to predict whether wear in the bushing-insulator
contact can cause malfunction. Despite these problems, the SDA has become popular in
the MEMS society, because of its simple geometry and the compatibility with
standardised surface micromachining processes, such as the Multi-User MEMS
processes (MUMPs) offered by MCNC [Koester99]. Nice examples of application of the
SDA are the lifting of a polysilicon plate to create suspended inductors and variable
capacitors [Fan98], and the integrated assembly of XYZ-stages for optical scanning and
alignment [Fan97].

Shuffle motor
An electrostatic stepping motor with large range, large force (1 mN), large speed (1
mm/s) and high accuracy (10-100 nm) was designed and implemented by N. Tas [Tas98,
Tas00]. The motor was coined Shuffle-motor and does function much the same way as
an inchworm [Inchworm] (see also figure 9). A schematic of the operation is shown in
figure 18.

GK+NT

MESA+ Research Institute

23

ASSET

Micromechanical Actuators

100 m

200 m

Figure 18: Left: artist impression of a Shuffle motor. Centre: schematic operation of a shuffle
motor. Right: SEM micrograph of a realised motor [Tas98].

The shuffle motor obtains its motion from a 6 phase operation applied by three stator
electrodes to actuate a front and back clamp and an actuator plate. In sequence: a) the
front foot gets clamped, b) the central actuator plate is pulled in leading to a slight
shortening and shifting the back foot, c) the back foot is clamped, d) the front foot is
released, e) the actuator plate is released leading to a small shift of the front foot and f)
the back foot is released. The moving poly-silicon structure is suspended by two
flextures which simultaneously serve as conductors to get the moving structures on
ground level. In between the lower poly-silicon conduction lines and the moving parts a
Si3N4 layer serves as an insulator. Apart from the flextures, in principle the motor can
have unrestrained motion in one direction.

Figure: 19: Left: the frame of the shuffle motor consists of an elevated part (medium gray)
and a lower part (black). The lower part composes the four clamp feet, resting on the
electrodes. The elevated part consists of the folded support springs and the large stiff parts
that connect the clamp feet and the actuator plate (light gray). The support springs are
anchored at the black squares. The stretch springs push the two frame parts apart when the
plate is released. Right: close up around one of the two inner stretch springs, showing the
different levels in the motor [Tas98].

One of the main advantages of the design is the lever-action of the actuator plate. The
pull-down distance of the plate ( 2 m) is resolved in a shortening of the distance by a
reduction factor of about 20 leading to small steps of the motor displacement. At the
same time the force is increased by the same amount. Another advantage is the relative
GK+NT

MESA+ Research Institute

24

ASSET

Micromechanical Actuators

large actuator plate area. Calculated output forces are as high as 1 mN. Forces,
experimentally obtained from the deflection of the support flextures, are about one order
smaller than the design values. Since pull-in voltages were pretty much as expected
(about 13 V), the reduced output forces were attributed to considerable slip of the
clamps.

Discretized motion: the MEMS-DAC


Recently a very interesting paper was published on a microelectromechanicaly digital to
analog converter by Toshiyoshi et al. [Toshiyoshi00]. Quoting from their abstract:
The mechanism is an N-stage network of connected suspensions, in which an electro-static
actuator is attached to the longer suspensions of compliance 2C, and N of such unit structures
are connected side by side with the shorter suspensions of compliance. Each actuator is an
electrostatic shuttle moving back and forth between the driving electrodes, and is operated by
the corresponding digit of the input code. The N-bits of local displacement accumulate in the
suspension network to synthesize an analog output, which is proportional to the analog value
coded with the N-bit input. The output displacement is independent of the fluctuation of the
driving voltage since the traveling distance of the shuttle is clipped by mechanical stoppers. We
call the mechanism a microelectromechanical digital-to-analog converter (MEMDAC) since the
function is equivalent to the electrical digital-to-analog converter known as the R2R resistor
net-work.

Figure 20: Left: Schematic (only upper part of symmetrical structure) (top) and realised
MEMSDAC (bottom). Right: output displacement vs. digital input code (top) and nonlinearity
(bottom) (from [Toshiyoshi00])

GK+NT

MESA+ Research Institute

25

ASSET

Micromechanical Actuators

One of the interesting things of the MEMDAC is that it can be operated in an open loop
configuration while still not being dependent on a critically controlled operating voltage.
Moreover, it does not exhibit the hysteresis behaviour so often found in other mechanical constructions (e.g. rack an pinion).
The results obtained by Toshiyoshi are impressive: for a 4 bit MEMDAC with a stroke of
5.8 m with a step of 0.38 m they measure an integral nonlineariy of only 0.25 times
the least significant bit (LSB) (see figure 20 right). Depending on input code fundamental
mechanical resonances between 2-10 kHz are found and operating voltages are at a
(rather high) 150 V. The authors expect that LSB distances down to 20 nm are
attainable for an 8 bit MEMDAC at equal stroke of 5.8 m.

Restrained rotational actuation


Torsional microactuators
Simple restrained rotational movement is found for example in MEMS scanning mirrors
applications like mirror displays. When tilting angles are relatively small, rotation can be
obtained using planar structures connected to hinges made of thin beams acting as
torsional spring-elements. Examples of these type of structures can be found in
[Bhler97].

Figure 21: Schematic of a torsional micromirror. Depending on which address


electrode is used the mirror is either tilted in positive direction (a) or negative
direction (b). Electron micrograph of realised torsional micromiror (c) (Photograhs from
[Bhler97]) Electron micrograph of a asymmetric comb-drive torsional scanning mirror (d)
(picture from [Yeh00])

A schematic of a restricted rotational movement is shown in figure 21. In general forces


need not be very high and control of the structures is obtained using moderate voltages
(tens of volts). Due to the small thickness of the mirror-plates the weight is low resulting
in reasonable bandwidths of tens of kHz. Tilting angles are roughly in the range of -3 to
3 degrees at control voltages of about 15 V [Bhler97].
Torsional actuation by means of asymmetric comb-drives, fabricated by a combination of
surface and bulk micromachining on SOI (Silicon On Insulator) is described in [Yeh99,
Yeh00]. In this case the tilting angle is in the range of -2..2 degrees at control voltages of
about 20 V. Natural frequencies are a couple of tens of kHz.
Somewhat larger tilting angles, ranging between -6 and 6 degrees at 12 V control
voltage, are obtained by Degani [Degani98]. Here the structure is very much comparable
to the structure of Buhler shown in 21.

GK+NT

MESA+ Research Institute

26

ASSET

Micromechanical Actuators

Figure 22: Left: part of an array of tilting mirrors in the TI DMD device with the centre mirror
removed. Middle: schematic of the tilting mirrors with drive electrodes, landing sites and
torsion hinges. Right: complete packaged device (bottom) ([TIDMD00])

Without doubt the most well-known tilting mirror MEMS device is the Digital Mirror
Device (DMD) fabricated and commercialised by Texas Instruments. This device has so
much matured that entire chips with integrated electronics and pulse-width modulators
can now be purchased commercially. The systems are coined the Digital Light
Processors and are available up to sizes of 1920*1080 pixels for HDTV projectors.

Curved electrode actuators and active joints


Rather simple actuators can be obtained by using conducting levers (with or without
internal stress) and opposing electrodes. One such device has been proposed in
[Elwenspoek92] where a bent bimorph with conducting upper layer is electrostatically
pulled in towards the electrode on the substrate (see Figure 23 left). For rectangular
levers, analysis of the systems predict a bistable behaviour in which the bimorph is
either on top of the substrate electrode or in its relaxed state. Critical voltage for a
structure with a thickness of about 1 m, a radius of curvature of 1 mm and a Youngs
modulus of 1011 Nm-2 is predicted to be 30 V. The bistable nature of the active joint can
be removed from the device by giving the electrodes peculiar shapes (e.g. triangular).
Using induced stress of Au/Cr coated poly-silicon layers Chen et al. [Chen99 ] fabricated
bent actuators for out-of plane micromirror displacement (see Figure 23 middle).
Actuators were fabricated by means of the Multi-User MUMPs MEMS processes.
Displacements of as much as 300 m were obtained using operating voltages of 20 V.
Switching speeds were sub-millisecond (<600 sec) and operation cycles of more than 14
million times could be obtained.

GK+NT

MESA+ Research Institute

27

ASSET

Micromechanical Actuators

Figure 23: Left: Active joint consisting of a conducting bimorph layer with stress gradient in
relaxed (top) and activated (bottom) states [Elwenspoek92]. Middle: polysilicon beam bent by
a stress-gradient electrostatically pulled in to move a micromirror (from [Chen99 ]). Right:
Electrostatic curved electrode actuators as proposed in [Legtenberg97a]. Top: schematic of
the structure (one half). Bottom: realised structure.

In plane electrostatic curved electrode actuators have been described in [Legtenberg97a].


Here a conducting lever can move parallel to the plane of the substrate. The lever as well
as the substrate are connected to ground whereas the curved electrodes are connected
to a positive voltage (one at the time). Using a curvatures of varing order n, i.e.
(x)=max(x/L)n, with L the length of the electrodes, it was found that pull-in occurs at
orders n<2 whereas for orders n>2 the levers get restrained by the electrode structures
resulting in the absence of pull-in. I.e. for these curvatures the lever becomes very well
controllable by the voltage. At the same time output force of the structures are
calculated to be roughly between 2 and 10 N at a control voltage of 100 V. Since the
lever may stick to the electrodes when pulled-in, either a dielectric is required (see
Figure 23 right, top) or stand-off bumpers have to be incorporated into the electrodes
(see Figure 23 right, bottom). Output swings of as much as 25 m have been obtained
for structures with a length of 300 m. Partial pull-in could experimentally be observed
between stand-off bumpers and with dielectric (Si3N4) spacers with large surface
roughness.

Rotation motors
Rotation motors have been fabricated in many forms with a variety of drive geometries.
Quoting form Fujitas excellent overview paper [Fujita98]:
The first electrostatic micromotors with diameters of 60-l 20 mm were developed by Fan et
al. These are called side-drive motors since they utilize the electrostatic force that acts between
the edges of a rotor and a stator, both of which were made with polysilicon. Rotational speed
was on the order of 500 rpm. The speed was relatively low because the sliding surface
generated substantial friction force even when silicon nitride film was deposited on the surface
to reduce the friction. Later improvements by Mehregany et al. enabled rotational speeds of up
to 15000 rpm and continuous operation for more than a week. They reduced the clearance

GK+NT

MESA+ Research Institute

28

ASSET

Micromechanical Actuators

between the rotor and the shaft and formed three dimples under the rotor for both support and
electrical contact.

In general, rotation derived from rotational actuators can only be obtained from a multistator excitation. This leads to the disadvantage that control circuitry has to include the
effects of inertia and load in order to have the stator-excitation in phase with the actual
movement of the rotating body. This disadvantage can be overcome using linear to
rotational movement converters as e.g. used in [Sandia00].

Variable capacitance motor


Earliest attempts to obtain MEMS electrostatic rotation motors were based on variable
capacitances ordered in a cylindrical symmetry. Having a multi-phase segmented stator
in combination with a rotor made out of a certain number of salient poles forming
capacitances with the stator blades, a multiphased torque production is obtained. See
figure 24. In order to minimise out of plane forces and possible clamping an additional
conducting (poly-silicon) layer can be used underneath the (grounded) rotor [TabibAzar97].

Figure 24: Left schematic of an electrostatic variable capacitance multi-phase


rotation motor [Tabib-Azar97]. Right: side-drive salient pole electrostatic
rotation motor made in SiC [Yasseen99].

The electrostatic torque produced by the motor can be derived from:


u 2 0 Rh
q 2g
W
W
T = F R = A pR = A
=
R = hg
R=
2g
2 0 2 Rh
V q
V q

(17)

where h and g are the height of and the gap between the stator and the rotor
respectively and R is the radius of the rotor.
Since these type of motors are barely conceivable without surface micromachined rotors,
and since the simplest technological implementation requires the stators and rotor to be
in one (poly-silicon) layer, most earlier variable capacitance motors are side-driven.
However, since the height of the rotor and stators is rather low the torque production is
limited as can be understood from equation equation (17). Moreover the rotor is
surrounded by stators making mechanical connections troublesome. Solutions to these
problems have been obtained using surface micromachining processes with mulptiple
conductive (poly-silicon) layers. These led, amongst other things, to top driven motors.
GK+NT

MESA+ Research Institute

29

ASSET

Micromechanical Actuators

These were not very successful since this type of motor suffered from out of plane forces
leading to tilting of the rotor and associated wear and tear [Tabib-Azar97].
A side drive made by Mller et al. [Mller97] utilises the variable capacitances of the
micromotor in an oscillation circuit driven at about 10-100 kHz. This way they are able
to increase the effective voltage by which the motor is controlled. The result is an
increase of a factor of 15 in efficiency delivering output at 5 V control signals comparable
to 75 V for the situation without oscillation ciruits.

Wobble motors
In order to reduce friction and wear, researchers have searched for modes of operation
that show less sensitivity to contact forces. One of the improvements was obtained in the
form of wobble motors as shown schematically in Figure 25.

Figure 25: Left schematic of a wobble motor with radial gap. Right: flange bearing (top) and
center pin bearing with needle shaped bushing (bottom) for a wobble motor
(figures from [Rymuza99]).

Again quoting Fujita [Fujita98] the advantages of the wobble motor can be understood:
Even for improved micromotors, friction has been a major problem. One solution is to replace
the sliding contact at the center axis of the device with a rolling contact. A type of micromotor,
called a wobble motor, was designed on this principle and was realized using surface
micromachining. Its rotor is a smooth ring, and, using electrostatic attraction, it rotates
eccentrically without slipping at the contact with the shaft. Since the circumferential distance of
the rotor hole is slightly longer than that of the shaft, the rotor really revolves a fraction of a
circle after one eccentric rotation. This results in two advantages for the motor, i.e., reduction of
friction and higher torque at low speed. An outer rotor design that permits easier extraction of
motion is also possible. The use of rolling motion in many other geometries has been
demonstrated, although the fabrication processes for these actuators were not fully IC
compatible.

The gear-ratio of wobble motors is defined by [Samper98]:


Gear ratio =

Rr
Rr Rb

(18)

where Rr is the rotor radius and Rb the bearing radius. Basically the gear-ratio is the
ratio between the number of revolutions of the driving voltage along the stator pattern
and the number of revolutions of the rotating disc. The larger this number is the higher
GK+NT

MESA+ Research Institute

30

ASSET

Micromechanical Actuators

the torque can be at given motor design. Gear-ratios roughly ranging between 10 and
1000 have been published. The rotational movement of wobble motors is far from
smooth [Legtenberg98]. The smoothness of the rotation, the angle dependence of the
torque and the effective (i.e. average) torque, depend, amongst other things, on the
number of stator poles, the control of the stator voltage (i.e. single or multiphase11), the
inertia of the disc and the load on the motor.
In [Legtenberg98] a wobble motor was proposed in which the disc and the shaft are
formed out of one layer of polysilicon. The bearing is formed by underetching of silicon
followed by additional polysilicon overgrowth (see Figure 26).

Figure 26. Left: Schematic of an axial gap wobble motor (top and middle) and a realized
shaft bearing. Right: Processing sequence of the wobble motor presented in [Legtenberg98].
(a) LPCVD of SixNy, LPCVD, and etching of polySi. (b) LPCVD of SixNy and contact window
etching. (c) PECVD of SiO2. (d) RIE of SiO2 and SixNy and plasma etching of silicon wafer. (e)
LPCVD of SiO2. (f) LPCVD of polySi and sputtering of Si and anisotropic etching. (g)
Sacrificial layer etching and backside metallization.

The stator electrodes are below the rotor instead of on the side. This enables larger
actuation area as well as an improved accessibility for e.g. driving of gear-trains. Having
rotor dimensions of 50 and 100 m radius torques in the nanoNewtonmeter range were
achieved. Amongst others this is made possible by the a large gear-ratio inherent in this
11

In this context multi-phase means that two or more neighbouring stator electrodes are exited in
overlapping time intervals to prevent the rotor to stick to one stator electrode.

GK+NT

MESA+ Research Institute

31

ASSET

Micromechanical Actuators

type of wobble motor. Gear-ratios as large as 1100 are obtained leading to increased
torque at reduced angular velocity (eliminating the use of additional gear-trains). Single
phase as well as multi-phase control were applied in order to obtain smooth operation
and small angle-dependence on the torque.
In [Yasseen99] wobble motors made in a silicon-carbide surface micromachining process
are described. Evidently, the advantages of SiC are mainly in improved friction and wear
characteristics due to its hardness and chemical inertness, the downside, difficult
processing conditions, stemming from the same properties. Actuator operation is
obtained up to temperatures of about 500 0C.

Figure 27. Schematic of the construction of a twin-stator wobble motor design fabricated by
LIGA [Samper98, Samper99].

Wobble motors produced in a LIGA12 technology for use in catheters are described in
[Samper98, Samper99]. In the design the authors investigate the optimisation of torque
production by using both the outer and inner rings of the rotor for electrostatic driving.
This requires additional insulation on the shaft. Furthermore the motor was driven
using closed loop control, employing position detection by a segmented bearing, allowing
for adaptation to loads and inertial effects. Motors with minimum rotor-stator gaps of 5
and 10 m and bearing clearances of 5 and 10 m (producing devices with gear ratios of
225 and 112.5) have been designed. The devices were analyzed for excitation potentials
of 100 V at various frequencies. Expected torque amounted to values as large as 0.6
Nm. Measured torques were somewhat smaller at 0.1 Nm at 120 V excitation.

Mechanical Transformations
Transformation of Rotary Motion to Linearly Motion
Transformation of rotary into linearly motion is a concept which is often employed in
macro-actuators. An example is the car with a combustion motor: Linear motion of the
cylinders is transformed into rotary motion by the crankshaft, and the rotary motion is
transformed into linearly motion by the wheels. This principle has successfully been
tested in micromechanics by Legtenberg et al. [Legtenberg96c, Legtenberg97b] who
coupled a rotary wobble motor and a gear rack (figure 28). The advantage of this concept
12

LIGA: LithographieGalvanoformungAbformung, a German acronym for (x-ray) Lithography, Electroplating and Moulding. Quoting [Thorndell98]: The technique employs synchrotron radiation for deep
exposure, typically several hundred micrometres. Because of the low absorption and diffraction of x-rays,
the resolution is submicrometre and the aspect ratio around 1000. Very soon the developed structures were
found suitable for metal filling by electrodeposition .... Later, another step was added, where the metal
structure is used as a mould insert for production of plastic templates.

GK+NT

MESA+ Research Institute

32

ASSET

Micromechanical Actuators

is that infinite motion can easily be generated by a rotary motor. The disadvantage is
that if a gearwheel and rack are used for the transformation, there will be back lash.
This might be avoided if frictional contact is used for the transformation.

(a)

(b)

Figure 28: Overview (a) and close up (b) of a wobble motor driving a gear rack
[Legtenberg96c, Legtenberg96c]. Note the anchoring of the gear rack in slots which were
etched in the wafer prior to the structural polysilicon deposition.

Transformation of rotational to rotational movement: gear trains


Would it not be great to use gear boxes in micromachines the way they are used in the
macro-world? It turns out that this is possible using a multi-layer surface-micromachining process having multiple poly-silicon layers. In this process, it is possible to
have tooth-wheels with different numbers of teeth in different poly-layers anchored to
each other. Each of the tooth-wheels on their turn can be in mechanical contact with
neighbouring tooth-wheels, thus forming a micro-world implementation of a gearbox.

Figure 29: Left: impression of a multi- sacrificial layer etching surface micromachining
process (the steps are additional to the ones shown in figure 3). Rigth: a micromachined
gearbox fabricated by researchers at Sandia [Sandia00].

Transformation of linear to rotation to linear motion


Not only rotational to linear motion conversion is possible but also the other way
around, linear to rotational conversion can be obtained. This kind of conversion is
implemented, amongst others, by researchers at Sandia [Sandia00] (see figure 30).

GK+NT

MESA+ Research Institute

33

ASSET

Micromechanical Actuators

Figure 30: Overview (left) and close up (right) of a pair of comb-drives, driving a tooth-wheel,
driving a linear rack [Sandia00].

Their approach is to have two comb-drives working in perpendicular directions, being


operated 90 degrees out of phase, like a sine and cosine. The combined motion forms a
circular motion which drives a toothwheel. The toothwheel on its turn drives a linear
rack. Using this system researchers are able to drive a 1 mm long linear rack into
position in 48 ms, i.e. a velocity of 2 cm/s!

Microtribology
Due to the large surface to volume ratio in micromechanics, surface forces become
relative large compared to body forces. To give an indication: the gravitation force acting
on surface micromachined elements is typically in the range of 0.1 - 10 nN. The
adhesion force between surface micromachined elements can easily become in the order
of 1 N and larger. This can cause friction much larger than the friction caused by the
gravitational load. [Mehregany90b, Yeh95, Yeh96, Lim90, Kaneko91]. For walking
motion it is necessary to create controllable friction in the clamps. Friction in the clamps
should be low (say < 1 N) if the clamps are not activated (zero load), and should be large
(> 100 N) if the clamps are activated (large load). In particular the first demand can be
troublesome to achieve, because due to the adhesion between the two contacting
surfaces in the clamp, significant friction can remain if the clamp is not activated.
Generally two friction mechanisms are distinguished: ploughing and shearing friction.
Ploughing friction exists if asperities of the harder contact material create scratches in
the softer material [Bowden50]. Shearing friction (also called adhesive friction) exists if
asperities slide across the counter surface, continiously creating and breaking bonds. In
MEMS contacting surfaces often consist of rather hard materials ((poly)silicon, silicondioxide, silicon nitride) and are rather smooth. It is therefore expected that adhesive
friction is the main friction mechanism.
From macro-tribology it is known that the friction force is almost proportional to the load
force, and is almost independent of the apparent contact area (Amontons Law).
According to Bowden and Tabor [Bowden50] the shearing friction equals the product of
the real contact area and a characteristic maximum shear stress in the interface. If the
contacting asperities deform plastically, it follows from this model that the real contact
area is proportional to the load and independent of the apparent contact area. Therefore
the friction will be proportional to the load. In 1966 Greenwood and Williamson showed
GK+NT

MESA+ Research Institute

34

ASSET

Micromechanical Actuators

that Amontons Law also applies to rough surfaces of which the contacting asperities
deform elastically [Greenwood66]. This result is important for MEMS contacts: in MEMS
devices materials are often hard and smooth. This implies that elastic deformation will
frequently be encountered in MEMS contacts.
In the 1970s Johnson, Kendall and Roberts [Johnson71] and Derjaguin, Muller and
Toporov [Derjaguin75] derived expressions for the contact area and pull-off force for a
single asperity elastic contact, including adhesive load. Fuller and Tabor [Fuller75]
implemented the JKR single asperity model in the statistical model of Greenwood and
Williamson, in order to calculate the separation force of rough, elastic, adhesive
contacts. A similar approach has been followed by Maugis [Maugis96] to calculate the
real contact area in an elastic adhesive contact, as a function of externally applied load,
and elasticity, roughness, adhesion parameters. A combination of this latter model and
the Bowden and Tabor model provides us with a friction model for rough, elastic,
adhesive contacts. From Maugis [Maugis96] it follows that in adhesive elastic contacts
the real contact area and thus the friction force strongly depends on the surface
topography. This is confirmed by experimental results [Chang88].

GK+NT

MESA+ Research Institute

35

ASSET

Micromechanical Actuators

References
[Akiyama93a]

T. Akiyama, K. Shono, A New Step Motion of Polysilicon Microstructures, Proc.


IEEE Micro Electro Mechanical Systems Workshop 1993, Fort Lauderdale,
Florida, USA, Febr. 7-10, 1993, pp. 272-277.

[Akiyama93b]

T. Akiyama, K. Shono, Controlled Stepwise Motion in Polysilicon


Microstructures, J. Microelectromechanical Systems, vol. 2, no. 3, 1993, pp.
106-110.

[Akiyama95]

T. Akiyama, H. Fujita, A Quantitative Analysis of Scratch Drive Actuator Using


Buckling Motion, Proc. IEEE Micro Electro Mechanical Systems 1995,
Amsterdam, The Netherlands, Jan. 29 - Febr. 2, 1995, pp. 310-315.

[Anderson91]

K. Anderson, J. Colgate, A Model of the Attachment/Detachment cycle of


Electrostatic Micro Actuators, DSC-Vol. 32 (MEMS), ASME, 1991, pp. 255-268.

[Berenschot94]

J. Berenschot, J. Gardeniers, T. Lammerink and M. Elwenspoek, "New


applications of r.f.-sputtered glass films as protection and bonding layers in
silicon micromachining", Sensors & Actuators A 41-42, 338 (1994).

[Bolle69]

Bolle B., Electrostatic Motors , Philips technical review, vol. 30, no. 6/7, 1969,
pp. 178-194.

[Bowden50]

Bowden F.P., Tabor D., The friction and lubrication of solids, Oxford Clarendon
Press, 1950.

[Bhler97]

J. Bhler, J Funk, J. Korvink, F. Steiner, P. Sarro and H. Baltes, Electrostatic


Aluminum Micromirrors Using Double-Pass Metallization, J. Microelectromechanical Systems, vol. 6, June 1997, pg 126 - 135.

[Bustillo98]

J. Bustillo, R. Howe, R. Muller, "Surface micromaching for


microelectromechanical systems", Proc. IEEE, Vol-86, 1998, p 1552-1574.

[Chang88]

W. Chang, I. Etsion, D. Bogy, Static friction coefficient model for metallic rough
surfaces, J. Tribology, Vol. 110, Jan. 1988, pp. 57-63.

[Chen99]

R. Chen , H. Nguyen, M. Wu A low voltage micromachined optical switch by


stress-induced bending, Proceedings MEMS conference, Heidelberg, Germany,
1999.

[Cho94]

Y. Cho, A. Pisano, and R. Howe, Viscous Damping Model for Laterally


Oscillating Microstructures, J. Microelectromechanical Systems, Vol. 3, no.2,
June 1994, pp. 81-87.

[Daneman95]

M. Daneman, N. Tien, O. Solgaard, A. Pisano, K. Lau, R. Muller, Linear


Microvibromotor for Positioning Optical Components, Proc. IEEE Micro Electro
Mechanical Systems Workshop 1995, Amsterdam, Jan. 29- Febr. 2, 1995, pp.
55-60.

[Daneman96]

M. Daneman, N. Tien, O. Solgaard, K. Lau, R. Muller, Linear VibromotorActuated Micromachined Microreflector for Integrated Optical Systems, SolidState Sensors and Actuators Workshop Hilton Head, South Carolina, June 2-6,
1996, pp. 109-112.

[Degani98]

O. Degani, E. Socher, A. Lipson, T. Leitner, D. Setter, S. Kaldor, and Y.


Nemirovsky, Pull-In Study of an Electrostatic Torsion Microactuator, J.
Microelectromechanical Systems, Vol. 7, 1998, pp. 373-376.

GK+NT

MESA+ Research Institute

36

ASSET

Micromechanical Actuators

[Degani98]

O. Degani, E. Socher, A. Lipson, T. Leitner, D. Setter, S. Kaldor and Y.


Nemirosky, Pull-in study of an electrostatic torsion microactuator, J. MEMS,
Vol 7, 1998, pg. 373-379.

[Derjaguin75]

B. Derjaguin, V. Muller, Y. Toporov, Effect of contact deformations on the


adhesion of particles, J. Colloid Interface Sci., Vol. 53, no.2, Nov. 1975, pp. 314326.

[Egawa90]

S. Egawa, T. Higuchi, Multi-Layered Electrostatic Film Actuator, Proc. IEEE


Micro Electro Mechanical Systems Workshop 1990, Napa Valley, CA, USA, Febr.
11-14, 1990, pp. 166-171.

[Egawa91]

S. Egawa, T. Niino, T. Higuchi, Film Actuators: Planar, Electrostatic SurfaceDrive Actuators, Proc. IEEE Micro Electro Mechanical Systems Workshop 1991,
Nara, Japan, Jan. 30 - Febr. 2, 1991, pp. 9-14.

[Elwenspoek00] M. Elwenspoek and R. Wiegerink, "Mechanical microsensors", Springer Verlag,


Heidelberg, to be published in 2000.
[Elwenspoek92] M. Elwenspoek, L. Smith and B Hk, Active joints for microrobot limbs, J.
Micromech. Microeng., Vol. 2, 1992, pg. 221-223.
[Elwenspoek99] M. Elwenspoek and H. Jansen, "Silicon Micromachining", Cambridge University
Press, Cambridge, New York, 1999.
[Fan88]

L. Fan, Y. Tai, R. Muller, Integrated Movable Micromechanical Structures for


Sensors and Actuators, IEEE Trans. On Electron Devices, Vol. 35, no.6, June
1988, pp. 724-730.

[Fan97]

L. Fan, M. Mu, K.D. Choquette, M. Crawford, Self-Assembled Microactuated


XYZ stages for Optical Scanning and Alignment, Proc. 1997 Int. Conf. on SolidState Sensors and Actuators (Transducers 97), Chicago, USA, June 16-19, 1997,
pp.319-322.

[Fan98]

L. Fan, R. Chen, A. Nespola, M. Wu, Universal MEMS Platforms for Passive RF


Components: Suspended Inductors and Variable Capacitors, Proc. IEEE Micro
Electro Mechanical Systems Workshop 98, Heidelberg, Germany, Jan. 25-29,
1998, pp. 29-33.

[Fujita88]

H. Fujita, A. Omodaka, The Fabrication of an Electrostatic Linear Actuator by


Silicon Micromachining, IEEE Trans. on Elec. Dev., vol. 35, no. 6, June 1988.

[Fujita98]

H. Fujita, Microactuators and micromachines Proc. IEEE, vol. 86, August


1998, 1721-1732.

[Fuller75]

K. Fuller, D. Tabor, The effect of surface roughness on the adhesion of elastic


solids, Proc. R. Soc. Lond. A., vol. 345, 1975, pp.327-342.

[Furuhata93]

T. Furuhata, T. Hirano, L. Lane, R. Fontana, L. Fan, H. Fujita, Outer Rotor


Surface-Micromachined Wobble Micromotor, Proc. IEEE Micro Electro
Mechanical Systems Workshop 1993, Fort Lauderdale, Florida, USA, Febr. 7-10,
1993, pp. 161-166.

[Greenwood66]

J. Greenwood, J. Williamson, Contact of nominally flat surfaces, Proc. R. Soc.


Lond. A, Vol. 295, 1966, pp. 300-319.

[Gui98]

C. Gui, Direct wafer bonding with chemical mechanical polishing, Ph.D.


Dissertation, University of Twente, November 1998, ISBN 90-36512328.

[Higuchi90]

T. Higuchi, Y. Yamagata, K. Furutani, K. Kudoh, Precise Positioning Mechanism


Utilizing Rapid Deformations of Piezoelectric Elements, Proc. IEEE Micro Electro
Mechanical Systems Workshop 1990, Napa Valley, CA, USA, Febr. 11-14, 1990,
pp. 222-226.

GK+NT

MESA+ Research Institute

37

ASSET

Micromechanical Actuators

[Howe83]

R. Howe and R. Muller, "Polycrystalline silicon micromechanical beams", J.


Electrochem. Soc., Vol-130, 1983, p 1420-1423.

[Howe86]

R. Howe and R. Muller, "Resonant microbridge vapour sensor", IEEE Trans.


Elect. Dev., Vol-33, 1986, 499.

[Ikuta91]

K. Ikuta, A. Kawahara, S. Yamazumi, Miniature Cybernetic Actuators Using


Piezoelectric Device, Proc. IEEE Micro Electro Mechanical Systems Workshop,
Nara, Japan, 1991, pp. 131-136.

[Ikuta92]

K. Ikuta, S. Aritomi, T. Kabashima, Tiny Silent Linear Cybernetic Actuator


Driven by Piezoelectric Device with Electromagnetic Clamp, Proc. IEEE Micro
Electro Mechanical Sytems 92, Travemunde, Germany, Febr. 4-7, 1992, pp.
232-237.

[Inchworm]

Inchworm Motors, see www.burleigh.com.

[Jacobson89]

S. Jacobson, R. Price, J. Wood, T. Rytting and M. Rafaelof, A design Overview of


an Eccentric-motion Electrostatic Microactuator (the Wobble Motor), Sensors
and Actuators, vol. 20, 1989, pp. 1-16.

[Johnson71]

K. Johnson, K. Kendall, A. Roberts, Surface energy and the contact of elastic


solids, Proc. R. Soc. Lond. A, Vol. 324, 1971, pp. 301-313.

[Kamiya99]

D. Kamiya, T. Hayama and M. Horie, Electrostatic comb-drive actuators made of


polyimide for actuating micromotion convert mechanisms, Microsystem Techn.,
Vol 5, 1999, pg. 161-165.

[Kaneko91]

R. Kaneko, Microtribology related to MEMS, Proc. IEEE Micro Electro


Mechanical Systems Workshop, Nara, Japan, 1991, pp. 1-8.

[Kiang98]

M. Kiang, O. Solgaard, K. Lau and R. Muller, Electrostatic comdrive-actuated


micromirrors for laser-beam scanning and positioning, J. MEMS, Vol 7, 1998,
pg. 27-37.

[Koester99]

Koester et al., MUMPs Design Handbook 4.0 (may 1999), Cronos Integrated
Microsystems, see mems.mcnc.org/mumps.html.

[Koga96]

A. Koga, K. Suzumori, T. Miyagawa, M. Sekimura, Attachment/Detachment


Electrostatic Micro Actuators for Pan-tilt Drive of a Micro CCD Camera, Proc.
IEEE Micro Electro Mechanical Systems Workshop 1996, San Diego, California,
USA, Febr. 11-15, 1996, pp. 509-514.

[Kohli95]

P. Muralt, M. Kohli, T. Maeder, A. Kholkin, K. Brooks, N. Setter, R. Luthier,


Fabrication and Characterization of PZT thin-film vibrators for micromotors,
Sensors and Actuators A, vol. 48, 1995, pp. 157-165.

[Koster94]

M. Koster, A walking piezo motor, Conf. Proc. Actuator 94, Bremen, Germany,
June 15-17, 1994, pp. 144-148.

[Kovacs98]

G. Kovacs, N. Maluf and K. Petersen, "Bulk micromachining of silicon", Proc.


IEEE, Vol-86, 1998, p 1536-1551.

[Lee93]

A. Lee, D. Nikkel Jr., and A. Pisano, Polysilicon linear microvibromotors, 7th


Int. Conf. on Solid-State Sensors and Actuators (Transducers 93), Yokohama,
Japan, June 7-10 1993, pp. 46-49.

[Legtenberg95]

R. Legtenberg, E. Berenschot, J. van Baar, Th. Lammerink and M. Elwenspoek,


An electrostatic lower stator axial gap wobble motor:design and fabrication, 8th
Int. Conf. on Solid-State Sensors and Actuators (Transducers 95), and
Eurosensors IX, Stockholm, Sweden, June 25-29, 1995, pp. 404-407.

[Legtenberg96a] R. Legtenberg, "Electrostatic actuators fabricated by surface micromachining


techniques", Ph.D. Dissertation, University of Twente, 1996, ISBN 90-36507960.
GK+NT

MESA+ Research Institute

38

ASSET

Micromechanical Actuators

[Legtenberg96b] R. Legtenberg, A.W. Groeneveld and M. Elwenspoek, Comb-drive actuators for


large displacements, J. Micromech. Microeng., Vol. 6, 1996, pp. 320-329.
[Legtenberg96c] R. Legtenberg, E. Berenschot, M. Elwenspoek, and J.H.J. Fluitman,
Electrostatic Microactuators with Integrated Gear Linkages for Mechanical
Power Transmission, Proc. IEEE Micro Electro Mechanical Systems Workshop
1996, San Diego, CA, USA, Febr. 11-15, 1996, pp. 204-209.
[Legtenberg97a] R. Legtenberg, J. Gilbert, S. Senturia and M. Elwenspoek, Electrostatic curved
electrode actuators, J. MEMS, Vol. 6, 1997, pg. 257-265.
[Legtenberg97b] R. Legtenberg, E. Berenschot, M. Elwenpoek, J.H. Fluitman, A Fabrication
Process for Electrostatic Microactuators with Integrated Gear Linkages, J.
Microelectromechanical Systems, Vol. 6, no. 3, 1997, pp. 234-241.
[Legtenberg98]

R. Legtenberg, E. Berenschot, J.van Baar, and M. Elwenspoek, An Electrostatic


Lower Stator Axial-Gap Polysilicon Wobble Motor Part I: Design and Modeling
and An Electrostatic Lower Stator Axial-Gap Polysilicon Wobble Motor Part II:
Fabrication and Performance, J. MEMS, vol. 7, no. 1, march 1998, pg 79-93.

[Lim90]

M. Lim, J. Chang, D. Schultz, R. Howe, R. White, Polysilicon microstructures to


characterize static friction, Proc. IEEE Micro Electro Mechanical Systems
Workshop, Napa Valley, CA, USA, 11-14 Febr. 1990, pp. 82-88.

[Maugis96]

D. Maugis, On the contact and adhesion of rough surfaces, J. Adhesion Sci.


Technol., vol. 10, no.2, 1996, pp. 161-175.

[May75]

W. May, Jr., Piezoelectric electromechanical translation apparatus, United


States Patent no. 3902084, Aug. 26, 1975.

[Mehregany90a] M. Mehregany, S. Bart, L. Tavrow, J. Lang, S. Senturia and M. Schlecht, A


study of Three Microfabricated Variable-capacitance Motors, Sensors and
Actuators, A21-23, 1990, pp. 173-179.
[Mehregany90b] M. Mehregany, P. Nagarkar, S.D. Senturia, and Jeffrey H. Lang, Operation of
Microfabricated Harmonic and Ordinary Side-drive Motors, Proc. IEEE
Workshop on Microelectromechanical Systems 1990, Napa Valley, CA, USA, 1114 Febr. 1990, pp. 1-8.
[Mehregany91]

M. Mehregany, S. Phillips, E. Hsu, and J. Lang, Operation of Harmonic SideDrive Micromotors Studied through Gear Ratio Measurement, Proc. 6th Int.
Conf. on Solid-State Sensors and Actuators (Transducers 91), San Francisco,
CA, USA, 1991.

[Moore73]

A. Moore (ed.), Electrostatics and its applications, John-Wiley & Sons, New
York, 1973.

[Mller97]

T. Mller, J. Gimsa, B. Wagner, G. Fuhr, A resonant, dielectric micromotor


driven by low ac-voltages (<6V), Microsystem Technologies, Vol 3, 1997 pp. 168170

[Niino92]

T. Niino, S. Egawa, N. Nishiguchi, T. Higuchi, Development of an Electrostatic


Actuator Exceeding 10N Propulsive Force, Proc. IEEE Micro Electro Mechanical
Systems Workshop 1992, Travemunde, Germany, Febr. 4 - Febr. 7, 1992, pp.
122-127.

[Niino94]

T. Niino, S. Egawa, H. Kimura and T. Higuchi, Electrostatic Artificial muscle:


Compact, High-Power Linear Actuators with Multiple-Layer Structures, Proc.
IEEE Micro Electro Mechanical Systems Workshop 1994, Oiso, Japan, Jan. 25 28, 1994, pp. 130-135.

[Oosterbroek99] E. Oosterbroek, "Modeling, design and realization of microfluidic components",


Ph.D. thesis University of Twente, November 1999, ISBN 90-36513464.

GK+NT

MESA+ Research Institute

39

ASSET

Micromechanical Actuators

[Petersen82]

K. Petersen, "Silicon as a mechanical material", Proc. IEEE, Vol-70, 1982, p 420457.

[Poggendorff70]

J. Poggendorff, Ann. Phys., Ser. 2, vol. 139, 1870, pp. 513-546.

[Racine93]

G. Racine, R. Luthier, and N. de Rooij, Hybrid Ultrasonic Micromachined


Motors, Proc. IEEE Micro Electro Mechanical Systems 1993, Fort Lauderdale,
Florida, USA, Febr. 7-10, 1993, pp. 53-59.

[Rymuza99]

Z. Rymuza Control tribological and mechanical properties of MEMS surfaces


Part 1: critical review, Microsystem Technologies 5 (1999), pg 173-180

[Samper98]

V. Samper, A. Sangster, R. Reuben and U. Wallrabe, Multistator LIGAfabricated Multistator LIGA-Fabricated Electrostatic Wobble Motors with
Integrated Synchronous Control, IEEE J. of Microelectromechanical Systems,
Vol 7, 1998, p214-223.

[Samper99]

V. Samper, A. Sangster, R. Reuben, and U. Wallrabe Torque Evaluation of a


LIGA Fabricated Electrostatic Micromotor, IEEE J. of Microelectromechanical
Systems, Vol 8, 1999, p115-123.

[Sandia00]

See e.g. http://www.mdl.sandia.gov/micromachine/gallery

[Sandia00]

See: http://www.mdl.sandia.gov/Micromachine/ (2000).

[Stengl89]

R. Stengl, T. Tan and U. Gsele, "A model for the silicon wafer bonding process",
Jpn. J. Appl. Phys., Vol-28, 1989, p 1735.

[Tabib-Azar97]

M. Tabib-Azar, Microactuators, Kluwer Academic Publishers, 1998, ISBN 07923-8089-4.

[Tang89a]

W. Tang, T. Nguyen, and R. Howe, Laterally Driven Polysilicon Resonant


Microstructures, Proc. IEEE Micro Electro Mechanical Systems 1989, Salt Lake
City, Utah, Febr. 20-22, 1989, pp. 53-59.

[Tang89b]

W. Tang, T. Nguyen, and R. Howe, Laterally Driven Polysilicon Resonant


Microstructures, Sensors and Actuators, Vol. 20, 1989, pp. 25-32.

[Tang90]

W. Tang, T. Nguyen, M. Judy, R. Howe, Electrostatic-comb drive of lateral


polysilicon resonators, Sensors and Actuators, Vol. A21-23, 1990, pp. 328-331.

[Tas00]

N. Tas, Electrostatic Micro Walkers, Ph.D. thesis University of Twente, April


2000, ISBN 90-36514355.

[Tas97]

N. Tas, A. Sonnenberg, A. Sander and M.C. Elwenspoek, Surface


Micromachined Linear Electrostatic Stepper Motor Proc. IEEE Micro Electro
Mechanical Systems, Nagoya, Japan, Jan. 26-30, 1997

[Tas98]

N. Tas, J. Wissink, L. Sander, T. Lammerink and M. Elwenspoek, "Modeling,


Design and Testing of the Electrostatic Shuffle Motor", Sens. Act. A, Vol-70,
1998, pp. 171-178.

[Tavrow90]

L.S. Tavrow, S.F. Bart, J.H. Lang, M.F. Schlecht, A LOCOS Process for an
Electrostatic Microfabricated Motor, Sensors and Actuators, A21-A23, 1990, pp.
893-898.

[Thorndell98]

G. Thornell and S. Johansson, Microprocessing at the fingertips, J. Micromech.


Microeng., Vol 8, 1998, pg 251262. Printed in the UK

[Thornton00]

J. Thornton, in "Deposition Technologies for Coatings", ed. R. Bunshah, Noyes


Publications, New Jersey, Chapter 5

[TIDMD00]

More information on TI DMD device can be found at http://www.ti.com/dlp.

[Tjerkstra99]

R.W. Tjerkstra, (1999, September 23), "Isotropic etching of silicon in fluoride


containing solutions as a tool for micromachining", Ph.D. thesis, University of
Twente, ISBN 90-36513286.

GK+NT

MESA+ Research Institute

40

ASSET

Micromechanical Actuators

[Toshiyoshi00]

H. Toshiyoshi, D. Kobayashi, M. Mita, G. Hashiguchi, H. Fujita, J. Endo and Y.


Wada, Microelectromechanical Digital-to-Analog Converters of Displacement for
Step Motion Actuators, J. Microelectromechanical Systems, vol. 9, no. 2, June
2000, pp. 218-224.

[Trimmer89]

W. Trimmer and R. Jebens, Harmonic Electrostatic Motors, Sensors and


Actuators, vol. 20, 1989, pp. 17-24.

[Uchiki91]

T. Uchiki, T. Nakazawa, K. Nakamura, M. Kurosawa, and S. Ueha, Ultrasonic


Motor Utilizing Elastic Fin Rotor, Jap. J. Appl. Phys., Vol. 30, no. 9b, Sept.
1991, pp. 2289-2291.

[Wensink00]

H. Wensink, H. Jansen, J. Berenschot and M. Elwenspoek Mask materials for


powder blasting J. Micromech. Microeng. 10, 2000 175180.

[Yamamoto98]

A. Yamamoto, T. Niino, T. Higuchi, High Precision Electrostatic Actuator with


Novel Electrode Design, Proc. IEEE Micro Electro Mechanical Systems
Workshop 1998, Heidelberg, Germany, Jan 25-29, 1998, pp. 122-127.

[Yasseen99]

A. Yasseen, C Wu, C Zorman and M. Mehregany, Fabrication and testing of


surface micromachined silicon carbide micromotors, Proceedings MEMS
conference, Heidelberg, Germany, 1999.

[Ye98]

W. Ye, S. Mukherje and N. Macdonald, Optimal shape design of an electrostatic


comb drive in microelectromechanical systems, J. MEMS, Vol 7, 1998, pg. 1626.

[Yeh00]

J. Yeh, C. Hui, N. Tien, Electrostatic model for an asymmetric combdrive, J.


MEMS, Vol 9, 2000, pg. 126-135.

[Yeh95]

R. Yeh, E. Kruglick, K. Pister, Micro Electromechanical Components for


Articulated Microrobots, The 8th Int. Conf. on Solid-State Sensors and
Actuators, (Transducers 95) and Eurosensors IX, Stockholm Sweden, June 2529, 1995, pp. 346-349.

[Yeh96]

R. Yeh, E. Kruglick, and K. Pister, Surface-Micromachined Components for


Articulated Microrobots, J. Microelectromechanical Systems, vol. 5, no. 1,
March 1996, pp. 10-17.

[Yeh99]

J. Yeh, H. Jiang and N. Tien, Integrated polysilicon and DRIE bulk silicon
micromachining for an electrostatic torsional actuator, J. MEMS, Vol 8, 1999,
pg. 456-465.

[Zipernowski89] Zipernowski, Zipernowski electrostatic motors, Elect. World, vol. 14, 1889, pp.
260-.

GK+NT

MESA+ Research Institute

41

You might also like