Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Characterization and operation of a mechanically actuated silicon

microgripper
Marius M. Blideran,a兲 Monika Fleischer, Wolfgang Henschel, and Dieter P. Kern
Institute of Applied Physics, University of Tübingen, Auf der Morgenstelle 10, 72076 Tübingen, Germany
Jochen Sterr, Klaus Schock, and Stephan Kleindiek
Kleindiek Nanotechnik GmbH, Aspenhaustrasse 25, 72770 Reutlingen, Germany
Matthias G. Langer, Karsten Löffler, and Francois Grauvogel
Sensory Biophysics Group, Department of Applied Physiology, University of Ulm, Albert-Einstein-Allee 11,
89081 Ulm, Germany
共Received 6 February 2006; accepted 28 August 2006; published 4 December 2006兲
Since the manipulation of biological objects is usually performed in a life-sustaining environment,
electrical fields or thermal gradients through the liquid may cause perturbations. The authors present
a microgripper fabricated in silicon by a combination of bulk and surface micromachining processes
that exhibits several advantages compared to previous reports. In order to avoid any possible
perturbation caused by electrical fields, their microgripper is mechanically actuated. The complete
system including the microgripper, a piezoactuator, and a nanomanipulator is described in detail
together with manipulation of micrometer sized glass spheres. © 2006 American Vacuum
Society. 关DOI: 10.1116/1.2357961兴

I. INTRODUCTION objects to be manipulated. The stability during the gripper’s


Understanding biological matter to its deepest levels re- operation is strongly increased by the three different sections
quires a better understanding of the matter at the micron and of thicknesses: 452 ␮m thickness of the support base, 52 ␮m
submicron levels. Studying the properties of small particles thicknesses of “the support arms” and “contact area,” and
or biological structures, which is our final target, requires 11 ␮m for “the actuated segment.” The components of the
their individual handling and manipulation. actuated segment are shown in Fig. 1共b兲: “thin arms,”
Early manipulations were carried out, e.g., using an “hinges,” and parallel “tips.” Figure 1共c兲 illustrates the con-
atomic force microscope,1 ranging from moving 30 nm tact area to which a copper wire was glued, the other end of
GaAs particles on a surface with the tip of the cantilever2 to the wire being attached to a piezomotor. The motion pro-
surgery on living cells, using a sharpened atomic force mi- vided by the piezomotor is transferred along the copper wire
croscopy tip.3
to the contact area and further along the long “pullers” to the
The previously reported manipulators can be grouped ac-
actuated segment. The flexure and with it the closing of the
cording to their gripping mechanism as follows: thermal
actuation—by passing an electrical current through a gripper tips is realized through the hinges. The four support bridges
with different thermal expansion coefficients along its connecting the contact area to the thick support arms reduce
structure,4–8 electrical actuation—applying a potential differ- significantly the out-of-plane motion. The tips, shown in Fig.
ence between the tips,9–11 and mechanical actuation— 1共d兲, are 1 ␮m wide and 3 ␮m long.
applying a mechanical force on a segment of the gripper, a
force that leads to the movement of the tips.12,13 Beside these
gripping mechanisms, optically driven micromanipulators
with a focused laser beam14 and magnetically driven three-
finger grippers15 were also reported. III. SIMULATION
We present a mechanically actuated microgripper that A. Theoretical considerations
eliminates the disadvantage of generating electrical fields or
thermal gradients and which promises better mechanical sta- For obtaining optimum geometrical parameters of the
bility during its operation. gripper we performed simulations using the finite element
method 共FEM兲 simulation program FEMLAB. This program
II. DESCRIPTION OF THE MICROGRIPPER provides direct output of the maximum surface stress called
CONCEPT “von Mises stress.” The “von Mises criterion” known as
In Fig. 1, scanning electron microscope images of one “maximum distortion energy criterion” or Maxwell-Huber-
structure are shown. Through its narrow angle of only 25° Hecky-von Mises theory16 states that a material will yield if
between its arms the gripper ensures an easy approach to the the distortion 共strain兲 energy reaches a critical value.17 The
strain energy is proportional to the von Mises stress, which is
a兲
Electronic mail: marius-marcel.blideran@uni-tuebingen.de defined by the equation

3239 J. Vac. Sci. Technol. B 24„6…, Nov/Dec 2006 1071-1023/2006/24„6…/3239/5/$23.00 ©2006 American Vacuum Society 3239
3240 Blideran et al.: Characterization and operation of a mechanically actuated silicon microgripper 3240

FIG. 1. Description of the microgrip-


per’s concept and components: 共a兲 the
complete gripper, 共b兲 the actuated seg-
ment, 共c兲 the contact area, and 共d兲 the
tips.

␴0 = 冑 共 ␴ 1 − ␴ 2兲 2 + 共 ␴ 2 − ␴ 3兲 2 + 共 ␴ 3 − ␴ 1兲 2
2
,
important objective was to keep the maximum von Mises
stress below the critical value. Different designs were studied
starting with already presented models8 that were scaled to
where ␴1, ␴2, and ␴3 are the principal stresses. The “von fulfill our purposes and cantilever based structures. After
Mises” criterion states that a structure will behave elastically most of the initial models failed because they exhibited mo-
when tion out of the gripper’s plane we decided that one with a
␴0 ⬍ ␴ y , robust “thin arm” of a least 5 ⫻ 5 ␮m2 cross section will be
more suitable for our applications. Only a small region on
where ␴y is the yield strength of the material and defines the the thin arm was narrowed constituting the hinges of the
limit for its elastic behavior. Since silicon is a brittle mate- gripper. For simplicity and symmetry reasons, first only one
rial, it exhibits no plastic behavior and therefore fractures thin arm structure was simulated, one result being shown in
will appear if ␴y is exceeded. For silicon the fracture strength Fig. 2共a兲 for an arm with curved hinges. All the geometrical
is 7 GPa.18 parameters were studied: the position of the hinges along the
arms, their length, possible shape and minimum width, the
B. Simulation results
cross section and length of the thin arm, and the angle made
Our first concern during design simulations was the me- with the vertical. Besides the maximum value of the von
chanical stability of the microgrippers. Therefore the most Mises stress, other important aspects were considered: the

FIG. 2. Simulation results for 共a兲 one


thin arm with curved hinges and 共b兲
the functional part.

J. Vac. Sci. Technol. B, Vol. 24, No. 6, Nov/Dec 2006


3241 Blideran et al.: Characterization and operation of a mechanically actuated silicon microgripper 3241

FIG. 3. Dependency of 共a兲 maximum “von Mises stress” and displacement at the puller on minimum width w for a gripper with curved hinges. 共b兲 Maximum
“von Mises stress” and displacement at the puller on angle ␣.

ratio between the motion along x and z directions at the tip, maximum surface stress moves to the bottom of the arm and
xt / zt, and the ratio between the motion along x at the tip and increases with increasing angle. To minimize the von Mises
the motion along the z axis of the puller, xt / z p. By providing stress a design with 12.5° was chosen.
a large ratio xt / zt we actually ensure that the object to be The next step was to complete the microgripper, namely,
grabbed will still be located between the tips at the moment to extend the actuated segment until it could connect with the
of their closing. The motion of the puller can be translated macroworld. At this point one aspect was very important: the
directly into the motion provided by the piezoactuator; there- thickness of “the support bridges” which basically deter-
fore increasing xt / z p means that a bigger stroke of the tips mines the force needed to operate the gripper. The maximum
along the x axis will be achieved for a smaller effort from the von Mises stress ranges from 321 to 473 MPa when the
piezobars. On the other hand, increasing those ratios results thickness T 关Fig. 2共b兲兴 is increased from 5 to 10 ␮m and is
in an increase of the structure’s stiffness and therewith an thus uncritical during the design process. The force though
increase of the maximum von Mises stress. Therefore a com- increases substantially from 8 up to 33 mN. Under these
promise between these aspects had to be found. forces the maximum expansion of the approximately 3 mm
One thin arm structure 5 ⫻ 5 ␮m2 in cross section was long⫻ 100 ␮m thick copper wire is about 100 nm, represent-
chosen and sets of simulations were performed, in turn, for ing 10% of the motion required to close the gap. We used the
each parameter, keeping all the other fixed. As expected, in- commercial glue UHU 300 Endfest with a mechanical
creasing the length of the hinges from 10 to 30 ␮m led to a strength of 15⫻ 106 N / m2. This value was compared with
decrease in the gripper’s stiffness of 13%, and of the maxi- the load necessary to close the gap between the tips of grip-
mum von Mises stress from 650 to 496 MPa, and required a pers with 5 – 10 ␮m thicknesses of the support bridges. Since
two and a half times smaller force to close the gap between the calculated maximum required load for the 10 ␮m struc-
the tips. At the same time the displacement at the puller tures is 4 N / m2, corresponding to 33 mN acting on a
increased by 31%. Out of the three different shapes consid- 100 ␮m diameter wire, and is thus of the same order as the
ered for the hinges: parallel 关Fig. 1共b兲兴, curved 关Fig. 2共a兲兴, maximum load supported by the glue, we considered only
and triangular, the parallel exhibited the greatest von Mises structures with 5, 6, 7, and 8 ␮m for fabrication.
stress of 803 MPa but for a smaller force. The changes in As a result of the simulations, we have chosen for fabri-
behavior of the designs in terms of motion were practically cation microgrippers with the following sizes and features:
nonexistent. In Fig. 3共a兲, the dependency of the von Mises hinges in parallel, curved, and triangular shapes with 20 ␮m
stress and motion of the pullers on the minimum width w of length and 1.5 ␮m minimum width, on arms making an
the hinges is plotted. Varying w from 1 to 2 ␮m showed angle of 12.5° and widths of the thin arms of 5, 6, 7, and
only a 3% decrease in the puller’s motion. As a compromise, 8 ␮m.
we have chosen grippers of all three shapes with 20 ␮m
length and 1.5 ␮m minimum width for fabrication.
IV. SYSTEM DESCRIPTION
Another important aspect to analyze is the angle between
the arm and the vertical. In Fig. 3共b兲 the results when the The microgripper is made from silicon 共Si兲 and the fabri-
angle ␣ is increased from 10° to 40° are presented. The dis- cation process, as described elsewhere,19 consists of four op-
placement of the puller along z triples for ␣ = 40°. For angles tical lithography steps and five deep reactive ion etching
up to 25°, the maximum von Mises stress appears on the steps. The starting Si wafer has two buried silicon oxide
hinges and decreases with increasing angle. From 25°, the layers 共SiO2兲 that act as etch stop during the processing. The

JVST B - Microelectronics and Nanometer Structures


3242 Blideran et al.: Characterization and operation of a mechanically actuated silicon microgripper 3242

FIG. 4. Schematics of the piezomotor including the microgripper and the


piezoactuator.

three Si layers separated by the SiO2 define the 11, 52, and
452 ␮m cross sections along the microgripper. Following the
simulations, grippers with different shapes for the hinges: FIG. 6. Sequences of manipulating a 5 ␮m glass sphere: 共a兲 gripping the
trapezoidal, curved, and parallel, were fabricated. The total sphere, 共b兲 lifting and moving the sphere, 共c兲 setting the sphere on the
length of the grippers is 4.25 mm and the width at the end of surface again, and 共d兲 releasing the sphere.
the base is 2 mm. The gaps between the tips were 6.5 and
8.5 ␮m.
positioning resolutions are 0.25 nm for the linear range and 5
Once the microgrippers were fabricated, they were at-
and 3.5 nm for the horizontal and vertical ranges,
tached to the piezomotor as described in Fig. 4. The system
respectively.
consists of two piezobars. Each piezobar has depicted in
black and yellow the polarity of the faces. At the bottom side
of the lower bar a silicon plate was glued and the microgrip- V. APPLICATIONS
per was attached on top of it. On the bottom side of the upper First, the function of the microgripper was studied and
bar one end of a 100 ␮m copper wire was soldered while its structures with trapezoidal, curved, and parallel “hinges”
other end was glued on top of the microgripper’s “contact were tested. The gaps of 6.5 and 8.5 ␮m between “the tips”
area.” The inner sides of the piezobars were grounded, while were closed before the 80 V voltage limit was reached, the
on the outer sides a positive potential was applied. By having voltage that corresponds to the maximum expansion range of
the bars positioned such that their inner and outer sides are of the piezobars. The closing procedure was repeated for more
opposite polarity, the upper piezobar contracts and the lower than ten times for several of the structures as an endurance
piezobar extends when applying a positive voltage. Thus the test. One structure had 5 ␮m thickness T of the support
action of the two piezos is essentially added. The design also bridges, 5 ␮m width of the thin arms, and 1.5 ␮m wide
ensures that differences in the thermal expansion of the pi- hinges. A second structure featured the following sizes:
ezobars and the silicon structure do not translate into a force 8 ␮m thick support bridges, 7 ␮m wide thin arms, and 2 ␮m
on the contact area, which would move the tips. wide hinges. Differences between the two operating proce-
The piezomotor, including the microgripper and the pi- dures were practically unidentifiable under the optical micro-
ezoactuator, was attached to a micromanipulator from scope and more importantly, both structures were intact after
Kleindiek Nanotechnik20 共see Fig. 5兲. This is a high precision the ten times closing operation.
positioning tool with a linear operating range of 12 mm and As a proof for the grippers’ function, in Fig. 6 we show a
240° rotation range in the horizontal and vertical plans. The sequence of images from the manipulation of 5 ␮m glass
spheres. At the first attempt the glass spheres with diameters
of 3 – 5 ␮m were spread on a glass substrate. Due to the
electrostatic charges accumulated on the spheres and on the
substrate a precise control was practically impossible. The
spheres jumped as soon as they were released from the pres-
sure of the tips and left the visual field of the optical micro-
scope. By choosing a gold coated substrate the gripping op-
eration presented no more difficulties and was successful
from the first attempt. In Fig. 6共a兲, the microgripper is al-
ready positioned around the sphere. Since the sphere we
wanted to handle was one of the biggest we considered its
size as being 5 ␮m. The scratch that crosses the optical mi-
crographs on the diagonal was our reference during manipu-
FIG. 5. Complete system: microgripper, piezoactuator, and Kleindiek lation. In the next image 关Fig. 6共b兲兴 the sphere is lifted and
Micromanipulator. the microgripper moved to the upper side of the scratch. At

J. Vac. Sci. Technol. B, Vol. 24, No. 6, Nov/Dec 2006


3243 Blideran et al.: Characterization and operation of a mechanically actuated silicon microgripper 3243

this point the gripper is out of focus. Next, we brought the ACKNOWLEDGMENTS
gripper back in contact with the surface 关Fig. 6共c兲兴. In Fig. Special thanks go to Peter Nommensen and Johannes
6共d兲 the sphere is settled on the surface and released from the Auber from the “Institute for Micro and Information Tech-
tips. nology” for fruitful discussions during the fabrication pro-
cess. Financial support from “The German Federal Ministry
of Education and Research” under Contract No. 0312017C is
gratefully acknowledged.

VI. CONCLUSIONS 1
G. Binnig, C. F. Quate, and C. Gerber, Phys. Rev. Lett. 56, 930 共1986兲.
2
I. Obataya, C. Nakamura, S. W. Han, N. Nakamura, and J. Miyake, Nano
We designed, fabricated, and tested silicon microgrippers Lett. 5, 27 共2005兲.
3
whose gripping mechanism is solely mechanically based. M. Ueda, Y. Baba, H. Iwasaki, O. Kurosawa, and M. Washizu, Jpn. J.
Appl. Phys., Part 1 38, 6568 共1999兲.
This way we eliminated the disadvantages of having electri- 4
G. Hashiguchi and H. Fujita, Proceedings of IEEE Sensors 共IEEE, New
cal fields, potential differences, or thermal gradients in the York, 2002兲, p. 922.
5
medium in which the manipulation is performed. The small N. Chronis and L. P. Lee, J. Microelectromech. Syst. 14, 857 共2005兲.
6
N.-T. Nguyen, S.-S. Ho, and C. L.-N. Low, J. Micromech. Microeng. 14,
angle between “the thin arms” of only 25° facilitates the
969 共2004兲.
approach to the targeted objects especially when they are 7
M. C. Carrozza, A. Menciassi, G. Tiezzi, and P. Dario, J. Micromech.
surrounded by other objects. Those two important advan- Microeng. 8, 141 共1998兲.
8
tages qualify our gripper to be a viable tool for biological Zyvex NanoEffector Microgripper, http://www.zyvex.com
9
P. Boggild, T. M. Hansen, C. Tanasa, and F. Grey, Nanotechnology 12,
applications with cells or other biological structures in a life- 331 共2001兲.
sustaining environment. 10
S. Akita, Y. Nakayama, S. Mizooka, Y. Takano, T. Okawa, Y. Miyatake,
After first tests performed in air and vacuum, the experi- S. Yamanaka, M. Tsuji, and T. Nosaka, Appl. Phys. Lett. 79, 1691
ments confirmed the simulation results in terms of maximum 共2001兲.
11
P. Kim and C. M. Lieber, Science 286, 2148 共1999兲.
surface stress. There are several reports on silicon’s failure 12
D. Petrovic, G. Popovic, T. Petrovic, A. Schneider, E. Huq, H. Detter, and
stress21–23 presenting values from 300 MPa up to 7 GPa. Fol- F. Franek, Proc. SPIE 4928, 63 共2002兲.
13
lowing our endurance test performed on the microgrippers S. K. Jericho, M. H. Jericho, T. Hubbard, and M. Kujath, Rev. Sci.
and considering that the simulations’ results in terms of the Instrum. 75, 1280 共2004兲.
14
S. Maruo, K. Ikuta, and H. Korogi, Appl. Phys. Lett. 82, 133 共2003兲.
maximum surface stress varied from 300 MPa to 1.3 GPa 15
S. Caraffini and J. G. Boyd, Proc. SPIE 3519, 6 共1998兲.
for the stiffer designs we consider that 7 GPa was a reason- 16
R. von Mises, Z. Angew. Math. Mech. 8, 161 共1928兲.
17
able value to refer to during the simulations. The three sec- M. Batista, Arch. Mech. 37, 581 共1985兲.
18
tions of different thicknesses along the gripper increase the K. E. Peterson, Proc. IEEE 70, 420 共1982兲.
19
M. M. Blideran, G. Bertsche, W. Henschel, and D. P. Kern, Microelec-
mechanical stability of the gripper by drastically reducing tron. Eng. 83, 1382 共2006兲.
20
bending and torsional effects. Kleindiek Micromanipulator, http://www.nanotechnik.com
21
Further work will be focused on applications in liquid and C. Wilson and P. Beck, J. Microelectromech. Syst. 5, 142 共1996兲.
22
W. Suwito, M. L. Dunn, and S. J. Cunningham, Proceedings of the Inter-
on biological structures as well as on inorganic objects with
national Conference on Solid-State Sensors and Actuators 共1997兲, p. 611.
sizes under 1 ␮m, and on developing a real-time detection 23
F. Pourahmadi, D. Gee, and K. E. Petersen, Proceedings of IEEE Trans-
method of the gripping moment. ducers, San Francisco 共IEEE, New York, 1991兲, p. 197.

JVST B - Microelectronics and Nanometer Structures

You might also like