The MEMS Handbook MEMS Applications (2nd Ed) - M. Gad el Hak Episode 1 Part 10 ppsx

7.5.1 Microgrippers and Other MicrotoolsThe first presented microrobotic device was based on in-plane electrostatic actuation [Kim et al., 1992].This microgripper had two relatively thin

has been presented [Ebefors, 2000] Most of these techniques could be arranged in array configurationsfor distributed micromotion systems (DMMS).

For out-of-plane actuators using an external force field it is difficult to control each individual actuator

in a large array of folded structures Therefore, a synchronous jumping mode is used to convey the objects

or move the device itself This jumping mode involves quick actuation of all the actuators simultaneously,which forces the object to jump When the object lands on the actuators (located in their off position),the object has moved a small distance and the actuators can be actuated again to move (walk or convey)further [Liu et al., 1995]

A critical aspect of large distributed micromotion systems based on arrayed actuators and distributed(or collective) actuation is the problem associated with the need for the very high yield of the actuators[Ruffieux and Rooij, 1999] Just one nonworking actuator could destroy the entire locomotion principle.Therefore, special attention must be paid to achieve high redundancy by parallel designs wherever possi-ble These aspects will be further addressed in Section 7.5.3

As pointed out in Section 7.2, a microrobotic device can be either a simple catheter with a steerable joint(Figure 7.2a), or a complex autonomous walking robot equipped with various microtools as in Figure7.2o Between these two extremes are microgrippers and microtools of various kinds, as well as micro-conveyers and walking microrobot platforms Each of these three microrobotic devices will be presentedmore in detail in the following discussion Section 7.6 describes more complex microrobotic systemswhere both microtools and actuators for locomotion are integrated to form so-called microfactories ordesk-to-manipulation stations Also, multirobot systems and communication between microrobots insuch multirobot systems will be discussed

Actuator (a)

A closed nozzle

(c)

FIGURE 7.7 Electrostatic-controlled pneumatic actuators for a one-dimensional contact-free conveyance system (a) Concept for the arrayed pneumatic conveyer (i.e., contact-free operation) (b) and (c) Mechanism for flow control

by electrostatic actuation of nozzle when (b) the electrostatic nozzle is in the normal situation (off) and (c) when voltage

is applied to one electrode (on) (Illustration printed with permission and courtesy of H Fujita, University of Tokyo.)

7.5.1 Microgrippers and Other Microtools

The first presented microrobotic device was based on in-plane electrostatic actuation [Kim et al., 1992].This microgripper had two relatively thin gripping arms (thin-film deposited polysilicon) as shown inFigure 7.8 Other microgrippers based on quasi-three-dimensional structures with high-aspect-ratiosfabrication techniques (beams perpendicular rather than parallel to the surface) have also been presented[Keller and Howe, 1995; 1997] (Figure 7.9) These kinds of grippers, so-called over-hanging tools, areformed by etching away the substrate under the gripper

One critical parameter for the in-plane technique is how to achieve actuators with large displacementand force generation capabilities Thermal actuators are known for their ability to generate high forces Athermal actuator made from a single material would be easy to fabricate, but the displacement due tothermal expansion of a simple beam, for example, is quite small This is a general drawback for in-planeactuators that occurs independently of the fabrication technique used However, by using mechanical lever-age, large displacements can be obtained, as was demonstrated by Keller and Howe (1995) They used areplication and micromolding technique, named HEXSIL, to fabricate thermally actuated microtweezersmade from nickel and later in polysilicon [Keller and Howe, 1997] In the HEXSIL process [Keller, 1998a]

Alignment

window

V-groove

PSG Poly

7 mm 1.5 mm 400 µm

500 µm Support cantilever

the mold is formed by deep trench etching in the silicon substrate A sacrificial layer of oxide is deposited

in the silicon mold which is then filled with deposited polysilicon Then the polysilicon structure is releasedfrom the mold by sacrificial etching of the oxide Afterwards, the mold can be reused by a new oxide andpolysilicon deposition process One advantage of this process is the ability to make thick (100 µm or greater)polysilicon structures (quasi-three-dimensional structures) on which electronics can be integrated Figure7.9b shows a close-up of the leverage design for the HEXSIL microtweezer; a large beam is resistively heated

by the application of current, and subsequently expansion causes other beams in the link system to rotateand open the tweezer tips When cooled, the contraction of the thermal element closes the tweezers.Leverage and linkage systems (sometimes combined with gears for force transfer) are useful techniquesfor obtaining large displacements or forces that can be used for thermal actuation as well as electrostaticcomb-drive actuators [Rodgers et al., 1999] Several publications on design optimization schemes for var-ious leverage techniques applied to thermal actuators (so-called compliant microstructures) have beenpresented [Jonsmann et al., 1999] Another way to achieve a leverage effect is to use clever geometricaldesigns for single material expansion One such method is the polyimide-filled V-groove (PVG) jointtechnology shown in Figure 7.10 The PVG joint technique has also been used for microconveyers andwalking microrobots, as will be described in Sections 7.5.2 and 7.5.3 The purpose of the PVG jointmicrogripper in Figure 7.10 is easy integration with a walking microrobot platform

Several publications on LIGA-based microgrippers have been presented The reason for using LIGA is

to get quasi-three-dimesnsional structures (thick structures) similar to the HEXIL tweezers in Figure 7.9.The LIGA process has also been used to produce single material (unimorph) in-plane thermal actuatorsfor micropositioning applications Guckel et al (1992) presented an asymmetric LIGA structure with one

“cold” and one “hot” side to generate large displacements (tenths of a millimeter) with relatively low powerconsumption, as illustrated in Figure 7.20a More recently, this approach was used by Comotis and Bright(1996) for surface micromachined polysilicon thermal actuators With this in-plane actuator they havesuccessfully fabricated over-hanging microgrippers

As an alternative to single-material expansion actuators, bimorph structures could also be used forout-of-plane acting gripping arms [Greitmann and Buser, 1996] A bimorph microgripper for automatedhandling of microparts is shown in Figure 7.11 This device consists of two gripping arm chips assembledtogether Each gripper arm has integrated heating resistors for actuation of the bimorph and tactilepiezoresistors for force sensing

Several different approaches to obtain three-dimensional microgrippers working out-of-plane like theones in Figures 7.10 and 7.11 exist One commonly used approach is use of the surface-micromachinedpolysilicon microhinge technology shown in Figures 7.3a–e Such microhinges have been used both formicrogrippers and for articulated microrobot components [Pister et al., 1992; Yeh et al., 1996]

FIGURE 7.9 Photograph of fabricated HEXSIL tweezers (a) Overview of the overhanging microtweezers with a compliant linkage system (b) Close-up of the 80-µm-tall HEXSIL tweezers The tip displacement between the closed and open position is typically 40 µm with a time constant 0.5 s for a typical actuation power of 0.5 A at 6 V (Reprinted with permission from MEMS Precision Instruments, Berkeley, CA and courtesy of C Keller.)

One major drawback of these microgrippers (mainly based on thermal or electrostatic actuation) isfound in biological applications Microgrippers based on thermal, magnetic, or high-voltage electric actu-ation could easily kill or destroy biological and living samples The pneumatic microgripper presented by Kim

et al [Ok et al., 1999] avoids such problems An alternative to heating grippers (such as the ones shown inFigures 7.10 and 7.11, which require relatively high heating temperatures) is the use of shape memoryalloy (SMA) actuators Microgrippers based on SMA often require lower temperatures than thermallyactuated bimorph or unimorph grippers SMA-based three-dimensional microgrippers have been used togrip (clip) an insect nerve for recording the nerve activity of various insects [Takeuchi and Shimoyama, 1999]

FIGURE 7.11 Photograph of a microgripper based on bimorph thermal actuation and piezoresistive tactile sensing.

(Reprinted with permission from Greitmann, G., and Buser, R [1996] Sensors and Actuators A 53[1–4], pp 410–415.)

Cured polyimide

Heaters for thermal actuation

∆ x

Polyimide joint

actuator

Piezoresistors for force sensing

Silicon arm for gripping

Silicon arms on-position (heated)

Silicon arms off-position (cold)

4 piezoresistors in a full Wheatstone bridge configuration for force sensing

FIGURE 7.10 Concept of a microgripper fabricated by polyimide V-groove joints (Ebefors, T et al [2000] “A

Robust Micro Conveyer Realized by Arrayed Polyimide Joint Actuators,” IOP Journal of Micromechanical

Microengineering 10[3], pp 77–349.) The polyimide in the V-grooves expands due to heating and the gripping arms

are opened Self-assembling out-of-plane rotation of the arms as well as the leverage effect for single material sion are accomplished by the well-controlled geometrical shape of the V-groove Polysilicon resistors are used both as resistive heaters and as strain gauges for force sensing.

expan-In the biotechnology field — for example, the growing area of genomics and proteomics — microtools formanipulation of single cells are of major importance In particular, massive parallel single-cell manipulationand characterization by the use of microrobotic tools are very attractive In this type of application, themicrogrippers usually must operate in aqueous media Most of the microgrippers presented so far in this

a SiO2 Ti Au BCB PPy Cross section at a-a

FIGURE 7.12 Schematic drawing of the process steps for fabricating microrobotic arms (in this case, an arm with three fingers arranged in a 120° configuration) based on hinges (micromuscle joints) consisting of PPy(DBS)/Au bimorph structures (A) Deposition and patterning of a sacrificial Ti layer (B) Deposition of a structural Au layer and etching of the isolating slits (C) Patterning of BCB rigid elements (D) Electrodeposition of PPy (conductive poly- mer) (E) Etching of the final robot and electrode structure and removal of the sacrificial layer Each microactuator is

100 µm  50 µm The total length of the robot is 670 µm, and the width at the base is either 170 or 240 µm ing on the wire width) (Reprinted with permission and courtesy of E Jager, LiTH-IFM, Sweden.)

(depend-review cannot operate in water because of electrical short-circuiting, etc One possible solution is to useconductive polymers Such conductive polymers, which undergo volume changes during oxidation and reduc-tion, are often referred to as electroactive polymers (EAPs) or micromuscles These kinds of micromuscleshave been used as joint material for microrobotic arms for single-cell manipulation devices [Smela et al.,1995; Jager, 2000a,b] Figure 7.12 describes the fabrication of a microrobot arm based on a polypyrole(PPy) conductive polymer During electrochemical doping of PPy, volume changes take place which can

be used to achieve movement of micrometer-size actuators The actuator joints consist of a PPy and goldbimorph structure, and the rigid parts between the joints consist of benzocyclobutene (BCB) The conjugatedpolymer is grown electrochemically on the gold electrode, and the electrochemical doping reactions take place

in a water solution of a suitable salt The voltages required to drive the motion are in the range of a few volts.One of the many experiments conducted with the various robot arms fabricated with the PPy micro-muscles is shown in Figure 7.13 The drawback of microrobotic devices based on the conductive polymerhinge (or “micromuscles”) is that they cannot operate in dry media

7.5.2 Microconveyers

Recently, a variety of MEMS concepts for realization of locomotive microrobotic systems in the form ofmicroconveyers have been presented [Riethmüller and Benecke, 1989; Kim et al., 1990; Pister et al., 1990;Ataka et al., 1993a,b; Konishi and Fujita, 1993; 1994; Goosen and Wolffenbuttel, 1995; Liu et al., 1995;Böhringer et al., 1996; 1997; Nakazawa et al., 1997; 1999; Suh et al., 1997; 1999; Hirata et al., 1998; Kladitis

et al., 1999; Ruffieux and Rooij, 1999; 2000; Smela et al., 1999; Ebefors et al., 2000] The characteristics forsome of these devices are summarized in Table 7.3, where the microconveyers are classified in two groups:contact-free or contact systems, depending on whether the conveyer is in contact with the moving object

or not, and synchronous or asynchronous, depending on how the actuators are driven Examples of bothcontact and contact-free microconveyance systems were shown in Figures 7.6 and 7.7, respectively

TABLE 7.3 Overview of Some Microconveyance Systems

Moved Length per

CF: pneumatic air bearing Slow Flat Si pieces/ 100–500 µm/ Not specified [Pister (for low-friction levitation)  1.8 mg max at 1–2 Hz et al., 1990] electrostatic force for driving

CF: magnetic levitation 7.1 mm/s b Nd–Fe–B Not specified Not specified [Kim et al.,

Lorentz force for driving 8–17 mg

C: array of thermobimorph 0.027– Flat Si piece/ ∆x  80 µm 8  2  16 legs/ [Ataka polyimide legs c (cantilevers); 0.5 mm/s 2.4 mg (f  f c; 33 mW)/ 500 µm/total et al., 1993] electrical heating (asyn) f c 10 Hz area: 5  5 mm 2

CF: array of pneumatic valves; Not Flat Si piece/ Not specified/ 9  7 valves/ [Konishi electrostatically actuated specified 0.7 mg f  1 Hz 100  200 µm 2 / and Fujita,

(pressure) total area: 2  3 mm 2 1994] C: array of magnetic inplane 2.6 mm/s d Flat Si pieces/ ∆x  500 µm/ 4  7  8 flaps/ [Liu et al., flap actuators; external 222 mg f c 40 Hz 1400 µm/total 1995]

C: array of torsional 5 µm Slow Flat glass piece/ ∆x  5 mm/fc 15,000 tips 180  [Böhringer high; Si-tips; electrostatic ⬇1 mg high kHz-range 240 mm 2 /total area: et al., 1996;

(for flat Si/ 60 mg 50 µm/total area: et al., 1998]

(thermal)/f chigh a hexagonal chip kHz-range approx 18 mm 2

(piezoelectric) C: array of erected c Si-legs; 12 mm/s h Flat Si pieces  ∆x  170 µmh 2  6 legs/500 µm/ [Ebefors thermal actuation of external load/ ( f  f c; 175 mW)/ total area: et al., 2000] polyimide joints (asyn) 3500 mg f c 3 Hz i 15  5 mm 2

a C  contact; CF  contact-free; asyn  asynchronous; syn  synchronous.

b The superconductor requires low temperature (77 K).

c Self-assembled erection of the legs.

d Estimated cycletime ⬇25 ms (faster excitation results in uncontrolled jumping motion) and 0.5 mm movements on

8 cycles [Liu, 1995].

e For flat sliders The velocity depends on the surface of the moving slider (critical tolerances of the slider dimensions).

f Manual assembly of the erected leg.

g Depends on the magnet and surface treatment.

h Possible to improve with longer legs and more V-grooves in the joInternational.

iPossible to improve Thinner legs with smaller polyimide mass to heat would increase the cut-off frequency, f c larger displacements at higher frequencies.

Contact-free systems have been realized using pneumatic, electrostatic, or electromagnetic forces tocreate a cushion on which the mover levitates Magnetic levitation can be achieved by using permanent mag-nets, electromagnets, or diamagnetic bodies (a superconductor) The main advantage of the contact-freesystems is low friction The drawback of these systems is their high sensitivity to the cushion thickness(load dependent), while the cushion thickness can also be quite difficult to control Also, this kind of con-veyance system often has low load capacity.

Systems where the actuators are in contact with the moving object have been realized based on arrays

of moveable legs erected from the silicon wafer surface The legs have been actuated by using differentprinciples such as thermal, electrostatic, and magnetic actuation Both synchronous driving [Liu et al.,1995] and the more complex, but also more effective, asynchronous driving modes have been used.The magnetic [Nakazawa et al., 1997; 1999] and pneumatic [Hirata et al., 1998] actuation principlesfor contact-free conveyer systems have a disadvantage that they require a specially designed magnetmover or slider which limits the usefulness With a contact system based on thermal actuators it is possi-ble to move objects of various kinds (nonmagnetic, nonconducting, unpatterned, unstructured, etc.);however, the increased temperature of the leg in contact with the conveyed object may be a limitation insome applications The contact-free techniques have been developed mainly to meet the necessary crite-ria for a cleanroom environment, where a contact between the conveyer and the object may generate par-ticles that would then serve to restrict its applicability for conveyance in clean rooms

A microconveyer structure based on very robust polyimide V-groove actuators has been realized [Ebefors

et al., 2000] This conveyer is shown in Figure 7.14 In contrast to most of the previously presented

Θ

K T H

- S 3

Cured polyimide

Metal

∆x

∆z

Leg position when heater "off"

Leg position when heater "on"

− Ground

l

Right backward, x−r

FIGURE 7.14 (a) Principle for the rotational movements on a test conveyer using robust PVG joints The left and right side can be actuated separately like a caterpillar Each leg has a size of 500  600  30 µm (b) Photograph show- ing different (undiced) structures used to demonstrate the function of the polyimide-joint-based microconveyer One

conveyer consists of two rows of legs (12 silicon legs in total) Two sets of legs (six each of xand x
) are indicated in the photograph (compare Figure 7.6c ) For the conveyer with five bonding pads, the right and left rows of legs can be controlled separately for possible rotational conveyance (c) SEM photographs showing Si legs with a length of

500 µm (d) The microconveyer during a load test The 2-g weight shown in the photograph is equivalent to 350 mg

on each leg or 16,000 times the weight of the legs (Note: Videos of various experiments involving this microconveyer are available at http://www.s3.kth.se/mst/research/gallery/conveyer.html/ or http://www.iop.org/Journals/jm )

microconveyance systems, the PVG joint approach has the advantage of producing robust actuators withhigh load capacity Another attractive feature of this approach is the built-in self-assembly by which oneavoids time-consuming manual erection of the conveyer legs out of plane Some of the conveyers listed

inTable 7.3require special movers (e.g., magnets or sliders with accurate dimensions) The PVG joint veyer solution is more flexible because one can move flat objects of almost any material and shape Thelarge actuator displacement results in a fast system that is less sensitive to the surface roughness of themoving object By using individually controlled heaters in each actuator, an efficient asynchronous driv-ing mode has been realized, which also allows a parallel design giving relatively high redundancy for actu-ator failure The first experiments with the conveyer showed good performance, and one of the highestreported load capacities for MEMS-based microconveyers was obtained The maximum load successfullyconveyed on the structure had a weight of 3500 mg and was placed on a 115-mg silicon mover, as shown

con-in Figure 7.14d Conveyance velocities up to 12 mm/s have been measured Both forward–backward andsimple rotational conveyance movements have been demonstrated The principle for rotating an object

by a two-row conveyer is shown in Figure 7.14a The lifetime of the PVG joints actuator exceeds 2  108load cycles and so far no device has broken due to fatigue

The most sophisticated microrobotic device fabricated to date is the two-dimensional microconveyersystem with integrated CMOS electronics for control which has been fabricated by Suh et al (1999).The theories on programmable vector fields for advanced control of microconveyance systems presented

by Böhringer et al (1997) were tested on this conveyer Several different versions of these conveyers havebeen fabricated throughout the years [Suh et al., 1997; 1999] All versions are based on polyimide ther-mal bimorph ciliary microactuator arrays, as shown in Figure 7.15

TiW resistor

High CTE polyimide

Wet Etch access vias

tion

Up (off) Down (on)

7.5.3 Walking MEMS Microrobots

In principle, most of the microconveyer structures described in the previous section could be turnedupside down to realize locomotive microrobot platforms For contact systems, that means that the devicewill have legs for walking or jumping The contact-free systems relying on levitation forces will “float”over the surface rather than walk Such systems seem more difficult to realize than the contact-operatingrobots The focus for the rest of this section is therefore on contact microrobot systems for walking.Although it seems straightforward to turn a microconveyer upside down, most of the existing convey-ers do not have enough load capacity to carry their own weight Further, there are problems on how tosupply the robot with the required power As illustrated in Figure 7.16, power supply through wires mayinfluence the robot operation range, and the stiffness of the wires may degrade the controllability toomuch On the other hand, telemetric or other means of wireless power transmission require complex elec-tronics on the robot Because many actuators proposed for microrobotics require high power consumption,the limited amount of power that can be transmitted through wireless transmission is a big limitation for potential autonomous robot applications To avoid the need for interconnecting wires, designs based

on solar cells have been proposed, as have low-power-consuming piezoelectric actuators [Ruffieux andRooij, 1999; Ruffieux, 2000], electrostatic comb-drives [Yeh et al., 1996], or inch-worm actuators [Yehand Pister, 2000] suited for such wireless robots For wire-powered robots, a limited amount of wires ispreferable, which implies that simple leg actuation schemes (on–off actuation as used for the walkingrobot platform in Figure 7.6c) are required if complex onboard steering electronics are to be avoided.Several proposals for making totally MEMS-based microrobots include the possibility of locomotion(e.g., walking) powered with or without wires

7.5.3.1 Examples of Walking Microrobots Designs

Several different principles used to actuate the different legs on a walking microrobot have been posed, most of them mimicking principles used in nature Some of the most feasible principles for walk-ing MEMS microrobot platforms are listed below

pro-FIGURE 7.16 Influence of energy cable on microrobots (Illustration printed with permission and courtesy of

T Fukuda, Nagoya University, Japan.)

Ciliary motion was used by Ebefors et al (1999) for eight-legged robots, as illustrated in Figures 7.6and 7.17.

Elliptical leg movements adopted from the animal kingdom have been proposed by Ruffieux and Rooij(1999) and Ruffieux (2000) and are illustrated in Figure 7.18 Similar to this concept is the principle used

by Simu and Johansson (2001) Their microrobot concept consists of both a walking and a microtool unitfor highly flexible micromanipulation Instead of using thin-film piezolayers and silicon legs as didRuffieux, Johansson and Simu have made robot legs in a solid, multilayer, piezoceramic material mounted

on a glass body [Simu and Johansson, 1999] (see Figure 7.22)

FIGURE 7.17 Photograph of the microrobot platform, used for walking, during a load test The load of 2500 mg is equivalent to maximum 625 mg/leg (or more than 30 times the weight of the robot itself) The power supply

is maintained through three 30-µm-thin and 5- to 10-cm-long bonding wires of gold The robot walks using the asynchronous ciliary motion principle described in Figure 7.6(c) The legs are actuated using the polyimide V-groove joint technology described in Figure 7.10 (Photograph by P Westergård and published with permis- sion.) (Note: Videos of various experiments performed using the microrobot shown here are available at

FIGURE 7.18 The rotational leg walking principle A few hundred cells (shown in the right picture) are arranged into a hexagonal array, inside a triangular grid frame that provides stiffness and room for interconnection The sim- plest form of gait requires two phases so that half the actuators are in contact with the ground, where friction trans- mits their motion to the device, while the other half is preparing for the next step (compare to the CMS technique depicted in Figure 7.6) The bottom left figure illustrates the serial interconnections of the actuator resulting in six

independent groups of beams (From Ruffieux, D et al [2000] in Proceedings IEEE 13th International Conference on

Micro Electromechanical Systems [MEMS 2000], pp 662–667 With permission.)

Gait mimicry of six-legged insects (similar to that of a crab [Zill and Seyfarth, 1996]) has been posed by several research groups — for example, the six-legged microrobot by Yeh et al (1996) and Yehand Pister (2000) and the multilegged microrobot prototype by Kladitis et al (1999) These concepts areillustrated in Figures 7.19 and 7.20.

pro-Inch-worm robots [Thornell, 1998] or slip-and-stick robots [Breguet and Renaud, 1996] that mimick aninch-worm or caterpillar are attractive microrobot walking principles, as these techniques take advantage

of the frictional forces rather than trying to avoid them The scratch-drive actuator principle [Mita et al.,1999] works according to this principle by firmly attaching to the surface only half of the robot or actu-ator body, then extending the spine (or middle part) of the body before anchoring the other half, releas-ing the first grip, shortening the spine, swapping the grip, and so on in a repeatable forward motion.Vibration fields and resonating thin-film silicon legs with polyimide joints have been used by Shimoyama

et al [Yasuda et al., 1993], as illustrated in Figure 7.21

Because friction in many cases poses a restriction on microrobots due to their small size, solutions thattake advantage of this effect (e.g., inch-worm robots) rather than trying to avoid it are the best suited formicrogait This trend can also be seen in the evolution of micromotors for microrobotic and other appli-cations Currently, many researchers try to avoid bearings or sliding contacts in their motors and instead

Bandgap ref.

ring oscillator

Finite state machine Low-power CMOS controller Solar cell

Charge pump

Si die

Solar array chip

Motors CMOS

hv

FIGURE 7.19 First version of the microrobot prototype based on surface micromachined microhinges for joints; each leg has three degrees of freedom and is comprised of two 1.2-mm-long, rigid polysilicon links and electrostatic step motors for movement (not included in the SEM-photo) (Bottom) Concept for a new design based on a solar- powered silicon microrobot Various components can be made separately and then assembled Surface micromachined hinges are used for folding the legs (see Figures 7.3d and e) Each leg has two links, and each link will be actuated by an inchworm motor [Yeh and Pister, 2000] (Printed with permission and courtesy of K Pister, BSAC-Berkeley, CA.)

make use of friction in wobble and inch-worm motors [Yeh and Pister, 2000] or actuators with flexiblejoints [Suzuki et al., 1992; Ebefors et al., 1999] without the drawback of wearing out.

The main problem associated with the fabrication of silicon robots is to achieve enough strength in themoveable legs and in the rotating joints Most efforts to realize micromachined robots utilize surfacemicromachining techniques which results in relatively thin and fragile legs [Yeh et al., 1996] have pro-posed the use of surface micromachined microhinges for joints, poly-Si beams for linkage to the trian-gular polysilicon legs, and linear electrostatic stepper motor actuation for the realization of a microrobot,

as illustrated in Figure 7.19

A similar approach for walking microrobots was used by Kladitis et al (1999) They also used the hinge technique to fold the leg out of plane, but instead of area-consuming comb drives for actuation, theyused the thermal “heatuator” principle integrated in the leg, as illustrated in Figure 7.20 To further improvethe robustness and load capacity of their robot, they used 96 legs arranged in six groups instead of using a

by Joule heating 1 cm

Line A

Line B

Square wave input

Gold bondwires

Walking surface Motion of legs

270 µm Leg motion

Leg motion

Spring wire

FIGURE 7.20 (a) Schematic illustration of a single-material asymmetric thermal actuator for in-plane actuation (also called “heatuator”) The actuator structure was originally made of nickel in a LIGA process, although surface- micromachined polysilicon structures have also been used (b) An out-of-plane folded heatuator for microrobotic applications (c) Photograph of the microrobotic device in a conveyance mode (belly up) (d) Close-up of one of the

heatuator legs erected out of plane (Part (a) printed with permission from Guckel, H et al [1992] in Technical Digest,

Solid-State Sensor and Actuator Workshop, pp 73–75.) (Part (b) printed with permission from Kladitis, P et al [1999]

in Proceedings IEEE 12th International Conference on Micro Electro Mechanical Systems (MEMS ’99), pp 570–575.)

(Photographs courtesy of P Kladitis and V Bright, MEMS Group at the University of Colorado at Boulder.)

six-legged robot However, that robot structure could only withstand a load four times the dead weight of therobot That was not enough to obtain locomotive walking but was high enough for conveyance applications.For both microrobots described above, which were based on surface micromachined polysilicon micro-hinges, the polysilicon legs were manually erected out of plane The use of the microhinge technique maycause problems because of wear after long-term actuation Shimoyama et al [Suzuki et al., 1992; Miura et al.,1995] introduced the concept of creating insect-like microrobots with exoskeletons made from surface-micromachined polysilicon plates and rigid polyimide joints that have low friction and therefore lower therisk for wearing out These microrobots were powered externally by a vibrating field (no cables were needed)

as illustrated in Figure 7.21 By cleverly designed robot legs having different masses and spring constants andthus different mechanical resonance frequencies, the leg to be actuated can be selected by applying a certain

Adder Piezoelectricvibrator

Left turn Right turn

Straight advance Full stop

(b)

(c)

Driving force

Vibration

Polyimide springs Supporting legs

Body

Kicking legs Masses

(a)

Base Activated by ƒ 2

FIGURE 7.21 (a) The principle for a selective power supply through a vibrating energy field [Yasuda et al., 1993] The microrobot has several resonant actuators with mutually exclusive resonance frequencies The power and control signals to the robot are obtained via the vibrating table (b) The four possible kinds of actions (c) Photograph of the 1.5  0.7-mm 2 surface micromachined microrobot The legs have different spring constants and masses, resulting in different resonance frequencies Polyimide is used for the soft springs and the polyimide joints are used for the erected

leg (Printed with permission from Yasuda et al [1993] Technical Digest, 7th International Conference on Solid-State

Sensors and Actuators (Transducers ’93), pp 42–45, June 7–10, Yokohama, Japan.)