MEMS and NEMS

Christopher Hierold
ETH Zurich, Micro and Nanosystems
Tannenstrasse 3, CH-8092 Zurich, Switzerland,
This paper is about the impact of scaling on the system performance of mechanical
inertia sensors and first activities towards nano mechanical sensors. Permanent cost
pressure will result in continuous efforts to integrate more functions into further miniaturized systems. As a consequence microsystems (MEMS) may also incorporate
functional nano devices such as carbon nanotubes in the future. Therefore an overview of recent activities for the application of carbon nanotubes with a focus on mechanical sensors is provided.
Keywords: microsystem, MEMS, nanosystem, carbon nano tube, sensor

Introduction
Requests for further miniaturization of microsystems may certainly result in serious
efforts to integrate nano scale functional devices in microsystems. New materials with
new properties on the nano scale will emerge. In this paper, limitations for microsystems scaling and current research activities in the field of nanostructures are discussed
in general and first proposals to realize mechanical sensors based on carbon nanotubes (CNT) are introduced in particular. This paper is mostly about mechanical sensors and systems on the micro and nano scale.
After more than 20 years of microsystems research and development it seems inappropriate to include a definition of microsystems once again. But due to some evolution
of these definitions over the years it should be mentioned that microsystems or MEMS
(both phrases are used as synonyms) are considered in this paper as systems that
involve electronic and non-electronic elements and functions on the micro scale. This
may also include sensing functions, signal acquisition and processing, control, actuation, display functions and means for performing chemical and biochemical interactions. The system aspects comprise also system partitioning, (V)LSI of non-electronic
elements, calibration, signal-to-noise-ratio, stability, reliability and first of all assembly,
packaging and test. This definition follows in most points the one given by Senturia [2].
Prominent examples of such microsystems and products are TIs DMDTM device [3],
ADIs inertia sensors [4] or Infineons surface micromachined integrated pressure sensors [5], [6].
Consequently, a definition for nanosystems follows: Nanosystems involve electronic
and non-electronic elements and functions on the nano scale for sensing, actuation,
signal processing, display, control and / or interface functions.
Today there are yet not many examples available that might fulfill this definition of
nanosystems. Two examples might be IBMs Millipede [1] and Samsungs field emission display based on carbon nanotubes [7] the function of which is defined by electron
emitting CNTs sealed in a flat vacuum package that also incorporates the picture generating phosphorus layers at a distance of 200 m from the CNTs. Besides Samsung
other companies like Motorola and Sony have been engaged in the development of
field emission displays (FED). An excellent overview and summary of the state-of-theart is provided by Itoh [8].

Proceedings of the GMe Forum 2005

C. Hierold

Scaling of Capacitive Force Sensors

Cost issues are in the foreground of the discussion about miniaturization. Researchers
and product developers are spending tremendous efforts to further miniaturize integrated systems. They even accept the fate that in some cases, miniaturization does not
support the physical system function per se. This is one of the most important differences between microelectronics and microsystems: While in microelectronics miniaturization and further integration, following Moores Law, have succeeded in better performing transistors and systems until recently (smaller, faster, cheaper), inertia sensors
do not benefit from scaling in general.
Table 1 summarizes the analysis of the system performance of three generalized and
simplified examples of MEMS, i.e. pressure, acceleration and yaw rate sensors [9]. The
performance is discussed with respect to scaling.
Pressure p

Sensing force Fs = 2A0 p

SNRopt
~ 10log (const. 2)
x/d0=const.
SNRopt
~ 10log (const. -2)
0,mech=const.

Acceleration a

Yaw rate

Fs = 3m0 a

Fs = 24m0v0

~ 10log (const. 2)

~ 10log (const. 2)

~ 10log (const. 1)

~ 10log (const. 2)

Table 1: Influence of miniaturization and scaling on system performance for pressure,

acceleration and yaw rate sensors [9]. < 1 is the scaling factor for the dimensions of these simplified sensor models.
The signal-to-noise-ratio (SNR) at the output of a virtual ground amplifier is taken as a
measure of the performance of all three systems. For comparison reasons all sensors,
even the pressure sensor, though it would not be straight forward to realize, are assumed to be differential capacitive sensors, forming a capacitive bridge at the input of
the amplifier. The following evaluation also applies if one of the sensor capacitors is
fixed and considered as a reference sensor (reduced sensitivity).
All three sensors measure a sensing force Fs as a result of the physical unit applied
that displaces the sensing electrode against a spring force. The change of the distance
to a counter electrode is measured by the change of the respective capacitance Cs.
Equation (1) gives the SNR of the sensor systems in dependence of the capacitance
change and a number of constants that depend on technology, only. Furthermore, the
sensor capacitance Cs is assumed to be smaller than the parasitic capacitances Cpar of
the setup, even in case of monolithic integration of MEMS and the circuit.
Numerous constraints exist when scaling a particular device, including design, process
and electro-mechanical considerations. Most of them are design specific and therefore
cannot be considered from a general viewpoint, but two constraints are common to all
measurement systems, namely the minimum required measurement range and the
minimum required measurement bandwidth. While the scaling of the mass is predefined by the scaling of the geometrical dimensions, the spring constant is considered as
a free design parameter. Its choice has to satisfy these two constraints.

MEMS and NEMS

SNRopt

Cs

= 10 log consttech

C par

(1)

9 Vin Vd , sat
2

Cs << C par ,
Bel:
Vin:
kB:
T:

consttech =

4k BTBel L2

Electrical bandwith
Input voltage
Boltzman constant
Absolute temperature

:
Carrier mobility
L:
Gate length
Vd,sat: Saturation voltage

In the case of an open loop system (case x/d0 = const. in Table 1) the minimum required measurement range, which is related to the maximum relative displacement,
dominates the choice of the spring constant.
Force feedback systems (case mech = const. in Table 1) can overcome this limitation, if
the feedback force can handle the measurement range. Then the minimum required
mechanical bandwidth, which determines the noise bandwidth, determines the choice
of the spring constant.
The results of scaling on the pressure and inertia sensors are achieved by substituting
for Cs in Eqn. (1) with expressions achieved by scaling down the sensors geometry
(Table 1) It is obvious that in general scaling of inertia sensors will reduce the system
performance of miniaturized systems ( < 1 is the scaling factor for the dimensions of
these simplified sensor models). Only in the case of force feedback systems with reduced spring constants compensating for the reduction of size and mass accelerometers are scaling invariant. Pressure sensors are less sensitive to scaling, or may even
improve, if the thickness of the sensors membrane can be easily scaled down (reducing spring constant).
In concluding this section about scaling, it is shown that at least in the case of yaw rate
sensors miniaturization will not improve the sensors performance per se. There are
possible solutions available that will help escape this scaling trap. These solutions will
be implemented by complex system designs.

Nano Mechanical Sensors

CNT integration into MEMS for actuation and electrical interfacing has been proposed
to characterize the electro mechanical properties of nanotubes [11], [12]. Figure 1
shows the realization of that kind of a test stand [13]. A single walled carbon nanotube
(SWCNT) is assembled between two fixed Ti/Au electrodes (50 nm thick). AFM cantilevers can be used to apply forces and bending moments on the suspended SWCNT by
deflecting the released cantilever structure. It is obvious that this kind of MEMS structures can also be used as electromechanical interfaces for CNT based mechanical
sensors. Figure 2 shows clearly the embedded carbon nanotube (CNT).
Current research on CNT based nano transducers could be classified roughly in two
categories: (i) Sensors that take advantage of the small dimensions of the tube to interact with structures and surfaces on atomic and molecular level and (ii) sensors that
take advantage of unique (e.g. electronic) properties of nanotubes to interact with the
macroscopic environment.

C. Hierold

Fig. 1: SEM image of a suspended single walled carbon nano-tube, contacted by two
fixed electrodes and a released cantilever (50 nm Ti/Au). The dashed lines indicate anchor areas [13].

Fig. 2: AFM image of a contacted single walled carbon nanotube (diameter approx.
1.2 nm) on 1.5 m SiO2 before HF release. Gold electrodes on top are 30 nm
thick [13].
Important representatives of the first category (i) are CNTs that are used as AFM probe
tips [14]. CNTs are placed on the tips of conventional AFM probes either by manipulation or direct catalytic growth. The effective radii of those CNT tips are reported to be in
the range of 3 nm (SWCNT) 6 nm (MWCNT) [15] and at least of half size of etched
silicon tips. An increase in lateral resolution of up to 70% was achieved by imaging gold
nanostructures compared to the resolution achieved with silicon tips. Additional advantages are the typical high aspect ratio of the cylindrical CNT that is advantageous for
imaging narrow and steep features and the elastic buckling of the tube above a low
critical force that avoids the damage of the device under test (e.g. organic and biological samples). Furthermore it was shown [16] that the CNT tips end can be modified to
create probes that can sense matter on the molecular level by very distinct chemical
functionality.
Mechanical sensors of category (ii) are reported in the area of strain sensors. A substrate (matrix) material (e.g. polymer) is filled with CNTs and investigated by Raman
spectroscopy [17]. In SWCNT the position of the D* Raman band strongly depends on
the strain transferred from the substrate to the nanotubes. The spatial resolution of the

MEMS and NEMS

measurement is around 1 m and is limited by the spot of the Raman laser. This
method is used to measure stress fields around defects in polymers and tensile strain
in materials.
A new approach to apply SWCNTs as stress sensors on the macro scale is reported in
[18]. SWCNTs mixed with DMF (N,N-dimethylformamide), filtered and dried, resulting
in a 10 m thick film of randomly orientated SWCNT bundles (buckypaper) is fabricated
and attached to a device under test. The resistance of the CNT layer is measured by a
four point probe technique. The measured voltage shift is proportional to the applied
stress or strain. Unfortunately there is no discussion about the performance of those
CNT strain sensors compared to conventional strain gauges. However, the authors
claim that the described approach allows incorporating CNTs directly into various materials (e.g. composites) to realize integrated strain sensors.
Very recently a PMMA-based micro pressure sensor was reported [19] using bulk multi
walled CNTs as piezoresistive sensing elements. The MWCNTs are attached to gold
electrodes on top of the membrane by AC electrophoretic manipulation of the nanotubes. By applying pressure to the membrane the tubes are strained and the resistance
of the tubes is increasing. The gauge factor of the tubes is estimated and reported to
be 235.
The direct integration of CNTs into MEMS devices could result in the next generation of
nanotransducers for the evaluation of mechanical loads. To realize these nanotransducers, it is mandatory to control and reproduce the assembly, or better the growth, of
nanostructures from one point of catalyst to another. Self assembly of nanostructures
instead of structuring by photolithographic means will be the preferred process technology approach for manufacturing of these structures. Recently [20], [21], field assisted growth of CNTs was demonstrated to control the alignment of CNTs between
two separate areas of Fe catalysts on silicon or molybdenum electrodes, respectively.
Fabrication of nanostructures by means of self-assembly instead of planar photolithography and etch steps will reduce fabrication complexity and costs significantly. In concluding this section about nanosystems one should emphasize that nanosystems are
not just miniaturized microsystems. New process technologies towards self-assembly
and the utilization of new sensing principles based on quantum effects will help to avoid
the scaling trap of microsystems. Basic research is needed to integrate nanostructures
into MEMS on wafer level to characterize performance based on statistical data and to
provide defined electrical and mechanical interfaces of nanostructures to the micro and
macro world.

Acknowledgments
The author thanks Christoph Stampfer and the members of the Micro and Nanosystems Group of ETH Zurich. Special thanks goes to Martin Handtmann, Infineon Technologies AG, Munich, for many discussions and performing calculations on scaling
laws of inertia sensors. The support of the nanotransducers research program by ETH
Zurich (TH 18/03-1) is gratefully acknowledged.

References
[1]

P. Vettiger, et. al.: The Millipede - Nanotechnology Entering Data Storage,

IEEE Transactions on Nanotechnology, Vol. 1, No. 1 , March 2002, pp. 39 55.

[2]

S.D. Senturia Microsystem Design. Kluwer, Boston 20024.

[3]

P.F. van Kessel, L.J. Hornbeck, R.E. Meier, M.R. Douglass: A MEMS-Based
Projection Display, Proceedings of the IEEE , Vol. 86, No. 8, Aug. 1998, pp. 1687
1704.

10

C. Hierold

[4]

J.A. Geen, et.al.S.J. Sherman, J.F. Chang, S.R. Lewis: Single-Chip Surface
Micromachined Integrated Gyroscope With 50 /h Allan Deviation. IEEE Journal of
Solid State Circuits, Vol. 37, No. 12, December 2002, pp. 1860 1866.

[5]

C. Hierold: Intelligent CMOS Sensors. IEEE MEMS 2000, Myazaki, Proceedings

IEEE Thirteenth Annual International Conference on Micro Electro Mechanical
Systems, IEEE 2000, pp. 1-6.

[6]

T. Scheiter, H. Kapels, K.G. Oppermann, C. Hierold, W.M. Werner, H.J. Timme,:

Full integration of a pressure-sensor system into a standard BiCMOS process.
Sensors and Actuators A Physical, 15 May 1998; A67(1-3): pp. 211-214.

[7]

W.B. Choi, et. al.: Carbon-Nanotube Based Field-Emission Displays for Large
Area and Color Applications, J. of Information Display, Vol. 1, No. 1, (Dec 2000),
pp. 59-62,

[8]

S. Itoh, M. Tanaka, Current Status of Field-Emission Displays, Proc. of the IEEE,

Vol. 90, No. 4 (April 2002), pp. 514-520.

[9]

C. Hierold, From micro to nano systems: mechanical sensors go nano, J.

Micromech. Microeng., 14, pp. 1-11, September 2004. www.iop.org/journals/jmm

[10] B.E. Boser, R.T. Howe: Surface Micromachined Accelerometers. IEEE Journal
of Solid-State Circuits, 31(3), March 1996, pp. 366-375,.
[11] R.S. Ruoff, D. Qian, and W.K. Liu, Mechanical properties of carbon nanotubes:
theoretical predictions and experimental measurements C.R. Physique 4 (2003),
pp. 993-1008.
[12] P. A. Williams et. al., Controlled placement of an individual carbon nanotube onto
a microelectromechanical structure, Applied Physics Letters, Vol. 80, No. 14,
2002, pp. 2574-2576.
[13] C. Stampfer, A. Jungen, C. Hierold, Fabrication of discrete carbon nanotube
based nanoscaled force sensors, Proc. IEEE Sensors 2004, Vienna, Oct. 24-27,
2004.
[14] C.L. Cheung, J.H. Hafner, C.M. Lieber: Carbon nanotube atomic force
microscopy tips: Direct growth by chemical vapor deposition and application to
high-resolution imaging, Proc. of the National Academy of Science of the USA,
Vol. 97, No. 8 (Apr. 11, 2000), pp. 3809 3813.
[15] S.S. Wong, et. al., Single-walled carbon nanotube probes for high resolution
nanostructure imaging, Appl. Physics Letters, Vol. 73, No. 23 (7 Dec 1998), pp.
3465 3467.
[16] S.S. Wong, E. Joselevich, A.T. Wooley, C.L. Cheung, C.M. Lieber, Covalently
functionalized Nanotubes as nanometer-sized probes in chemistry and biology,
Nature (letters), Vol. 394, 2 July 1998, pp.52 55.
[17] Q. Zhao, M.D. Frogley, H.D. Wagner, Direction-sensitive strain-mapping with
carbon nanotube sensors, Composites Science and Technology 62 (2002), pp.
147 150.
[18] P. Dharap, Z. Li, S. Nagarajaiah, E.V. Barrera, Nanotube film based on singlewall carbon nanotubes for strain sensing, IoP Nanotechnology Vol. 15 (2004), pp.
379 382.
[19] Carmen K. M. Fung, Maggie Q.H. Zhang, Rosa H. M. Chan, Wen J. Li, A PMMABased Micro Pressure Sensor Chip Using Carbon Nanotubes as Sensing
Elements, Technical Digest, 18th IEEE Int. Conf. on Micro Electro Mechanical
Systems, MEMS 2005, pp. 251 254, 2005

MEMS and NEMS

11

[20] H. Miyashita, T. Ono, M. Esashi, Nanomechanical Structures With An Integrated

Carbon Nanotube. The 12 Int. Conf. on Solid-State Sensors, Actuators and
Microsystems, Transducers03, Boston, Digest of Technical Papers Vol. 1, IEEE
2003 pp. 182
[21] A.Ural, Y. Li, H. Dai, Electric-field-aligned growth of single-walled carbon
nanotubes on surfaces. Appl. Phys. Letters Vol. 81, No. 18, 28. October 2002,
pp. 3464 3466