You are on page 1of 11

774

JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 20, NO. 3, JUNE 2011

Modeling, Simulation, and Design Guidelines for


Piezoresistive Affinity Cantilevers
Manoj Joshi, Prasanna S. Gandhi, Member, IEEE, Rakesh Lal,
V. Ramgopal Rao, Senior Member, IEEE, and Soumyo Mukherji

AbstractThis paper presents design guidelines for piezoresistive affinity cantilevers for operation in liquid environments. For
the first time, we consider the interdependence of various functional elements (such as biological, mechanical, and electrical) of
the cantilever, their dependence on material choice, microfabrication processes, and geometry, and the resultant effects on the mechanical and electrical sensitivities of the cantilever. The cantilever
design guidelines that include material selection as well as determination of geometrical dimensions are proposed. As an example,
we have designed and simulated a multilayer piezoresistive silicon
nitride affinity cantilever for performance in a liquid environment
under constraints imposed by microfabrication and electrical and
mechanical considerations. Systematic steps toward optimization
of geometrical dimensions include initial analytical estimates of
geometrical dimensions, followed by finite-element modeling and
analysis of such cantilevers under the applied surface stress. Simulation studies brought forth the limitation on maximum obtainable
R/R as well as the nonlinear behavior of the cantilever which
was not observed in analytical estimates.
[2010-0116]
Index TermsAffinity cantilever, nonlinear, piezoresistive.

I. I NTRODUCTION

IEZORESISTIVE affinity cantilevers have great potential


in microsystems used for sensing biomolecules, such as in
point-of-care systems. Such prototype systems using piezoresistive cantilevers in Wheatstone bridge configurations with
integrated microfludics have been demonstrated [1][3]. These
cantilevers have a multilayer structure in which the strainsensitive layer is sandwiched between a structural layer and an
encapsulation layer. Selective immobilization of biomolecules
on either top or bottom surface of the cantilever generates differential surface stresses on the opposite faces of the cantilever
which leads to a bending of the cantilever, thereby inducing a
change in resistance of the strain-sensitive layer incorporated
within it. For maximum sensitivity of cantilever biosensors,
the selective immobilization of biomolecules is a prerequisite,
since immobilization on both faces is expected to elicit a weaker
response [4]. Studies have concentrated on the mechanical and
Manuscript received April 25, 2010; revised February 28, 2011; accepted
March 5, 2011. Date of publication April 25, 2011; date of current version
June 2, 2011. Subject Editor A. Seshia.
M. Joshi is with R&D, Taiwan Semiconductor Manufacturing Company
Limited, Hsinchu 300-77, Taiwan (e-mail: manojjoshi05@gmail.com).
P. S. Gandhi, V. R. Rao, and S. Mukherji are with the Indian Institute
of Technology Bombay, Mumbai 400076, India (e-mail: gandhi@iitb.ac.in;
rrao@ee.iitb.ac.in; mukherji@iitb.ac.in).
R. Lal, retired, was with the Indian Institute of Technology Bombay, Mumbai
400076, India. (e-mail: rlal@ee.iitb.ac.in).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/JMEMS.2011.2140353

electrical design of multilayer piezoresistive cantilevers, such


as those for atomic force microscope cantilevers [5] as well as
affinity cantilevers [6], [7]. For those affinity cantilevers, the
electrical sensitivity, i.e., change in resistance due to applied
surface stress (R/R) is less than one part per million (1
106 1 107 ) which can get suppressed by electrical noise.
Also, previous work does not comprehensively discuss engineering figures of merit such as signal-to-noise ratio (SNR),
power dissipation, and mechanical stability in liquid media,
thus posing practical limitations in the realization of cantilever
sensors.
The interdependence of various functional elements (biological, mechanical, and electrical) of piezoresistive affinity
cantilevers, their dependence on choice of materials, processing, and geometry, and the resultant effects on mechanical
sensitivity (change of deflection with applied stress) and electrical sensitivity (change of resistance with applied stress) make
the design of affinity cantilevers challenging. This is further
complicated by the fact that these cantilevers are expected
to work in a liquid environment. Selective immobilization of
biomolecules, packaging, and testing are other important considerations in the overall design of such systems [8]. In spite
of the complexity in designing such cantilevers, there are no
standard guidelines available in the literature.
In this paper, we report a design of piezoresistive affinity
cantilevers based on a combination of analytic formulas and
numerical simulations. We did not intend to optimize the
sensor for maximum biosensitivity, e.g., detection of lowest
possible analyte concentration. However, for a chosen surface
stress value, our aim was to maximize electrical sensitivity by
obtaining the highest R/R and SNR. Mechanical sensitivity
parameters such as resonant frequency and spring constant were
used as design constraints for stable and noise-free operation
of the cantilever. The design guidelines considering the interdependence of various functional domains of the cantilever, as
well as the criteria for material selection and determination of
geometrical dimensions of the cantilever, are proposed. To illustrate the proposed design, an example of a piezoresistive silicon
nitride cantilever used for the detection of antibodyantigen
binding on the cantilever surface is presented. Surface stress
generated due to the antibodyantigen interaction was modeled
using a MEMS simulator, and the cantilever response was
obtained. In order to optimize the cantilever geometry, simple
analytical equations were used for initial estimates of cantilever dimensions and subsequently fine-tuned using numerical
simulations. Simulation studies demonstrated the limitation on
maximum achievable electrical sensitivity that was many folds

1057-7157/$26.00 2011 IEEE

JOSHI et al.: MODELING, SIMULATION, AND DESIGN GUIDELINES FOR AFFINITY CANTILEVERS

higher than its counterparts reported in [6] and [7]. It also


demonstrated nonlinear behavior of the cantilever which was
not seen in analytical estimates.
II. D ESIGN G UIDELINES
A piezoresistive affinity cantilever has four functional layers:
1) an immobilization layer; 2) an anchor layer to which the immobilization layer is attached; 3) a piezoresistive layer; and 4) a
structural layer. We assume immobilization of biomolecules on
the top surface of the cantilever. The tip deflection and change
in resistance (R/R) of the piezoresistor are measures of the
mechanical and electrical sensitivities, respectively.
The design process is iterative with material properties and
electromechanical effects impacting each other. However, we
can notionally divide the design process into material selection
and geometry determination.
A. Selection of Materials
1) TargetProbe Biomolecule Pair: Depending on the target biomolecules (which have to be detected), an appropriate
affinity probe needs to be identified. Immobilization of such
biological affinity probes on solid substrates can be achieved
by adsorption, covalent bonds, entrapment in gels, etc. [9]. The
immobilization using covalent bonds can allow a thin layer
of biomolecules [10] with lesser impact on the mechanical
properties of the cantilever. The biomolecule layer immobilized
using this technique should be as less susceptible as possible to
changes in pH, temperature, or ionic strength of the surrounding
medium [9], [11] and should show low desorption of biomolecules from the surface of the cantilever. It does not require
a diffusion barrier like a membrane, gel, or polymer matrix
between the cantilever surface and the immobilized molecule;
the analyte can react with the sensor surface more readily. This
improves the sensor kinetics as well as reduces the amount
of biomolecules required for sensor fabrication. Thus, covalent
attachment is a preferred choice for the immobilization.
2) Surface Layer for Immobilization: Preferably, the cantilever should be incubated in an appropriately designed liquid
cell or flow channel leading to selective immobilization on
either the top or the bottom surface. This can be achieved
with a surface of immobilization layer chemically different
(isoelectric points) than structural layers such as gold against
silicon nitride. Another alternative is chosen: The immobilization technique should allow the selective immobilization (e.g.,
selective functionalization of SiO2 using aminosilanization as
against the Si3 N4 surface) [12], [13]. Ideally, such an anchor
layer film should be infinitesimally thin and have a low Youngs
modulus. Preferably, it should be an insulator for thermal and
electrical isolations between the piezoresistive layer and immobilized biomolecules. However, the selected immobilization
protocol affects the selection of material for the anchor layer.
Gold [14], silicon dioxide [15], and silicon nitride films [16]
have been widely used as an anchor layer. Gold film can be
used as a reflecting surface for laser light and is often used for
immobilization in affinity cantilevers for optical detection of
deflection. However, in the case of piezoresistive cantilevers, if

775

gold film is used for immobilization, an additional insulating


layer (e.g., SiO2 and Si3 N4 ) between the piezoresistive layer
and the gold film is essential. Such an additional insulator layer
has an adverse impact on the sensitivity of the cantilever.
3) Strain-Sensitive Layer: The strain-sensitive layer may be
made of semiconductor or metal with the former providing
higher sensitivity due to high gauge factor. Although doped
single crystal silicon has higher sensitivity to strain than doped
polysilicon, the latter offers the advantage of surface micromachining and is preferred as a strain-sensitive layer. It also
exhibits a lower temperature coefficient of resistance (0.04%
per degree Celsius) than single crystal silicon (0.14% per
degree Celsius), which improves thermal stability in liquid environment [17]. (111)-oriented P-type boron-doped polysilicon
with dopant concentration about 1019 cm3 offers a high gauge
factor and is preferred as the piezoresistive layer [17]. SNR is a
key performance index for minimum detectable stress changes
within the piezoresistive layer and can be optimized based
on piezoresistor geometry, doping concentration, and anneal
temperature [18]. The internal stresses within the polysilicon
film can be minimized by controlling the deposition process
parameters [18].
4) Structural Layer: The material for this layer has to be
chosen such that biomolecules do not get attached to this layer
due to the selected immobilization protocol. For example, one
can use silicon nitride as a structural layer when a gold (chemically different surface) or silicon dioxide (e.g., silanization
protocol for selective surface functionalization) film is used as
an immobilization layer, thus allowing cantilever incubation in
one solution [13], [14]. Cantilever sensors with hydrophobic
surfaces are prone to stiction in liquid environments and can
be susceptible to nonspecific adsorption of biomolecules. A
good inert surface for biosensors can be obtained by selective
surface modification. For example, polyethyleneglycol-covered
surfaces are hydrophilic in nature and known to prevent adsorption of biomolecules.
The electrical sensitivity (R/R) of the piezoresistive affinity cantilever is a function of the Youngs modulus of the
materials used, the distance between the piezoresistive layer,
the neutral axis of the cantilever, as well as the thickness [(13),
Appendix B]. Higher sensitivity can be achieved by increasing
the distance between the piezoresistive layer and the neutral
axis, which can be obtained by selecting a structural layer material having a higher Youngs modulus than the piezoresistive
layer. For example, if p-type polysilicon (E = 150 GPa) is
used as a piezoresistive layer, silicon nitride (E = 280 GPa)
is a better choice than silicon dioxide (E = 75 GPa) as a
structural layer material.
B. Determination of Geometrical Dimensions
1) Mechanical Effects of TargetProbe Interaction:
Antigenantibody interactions on the cantilever surface produce surface stress [19]. In order to determine the geometrical
dimensions of an affinity cantilever, the approximate magnitude
and nature of the stress (compressive or tensile) need to be
known. For certain classes of probetarget interactions, the
approximate values may be obtained from literature. For

776

JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 20, NO. 3, JUNE 2011

Fig. 1. (a) Conversion of surface stress into surface force used for surface
boundary conditions. Surface force = Surface stress (in newtons per meter)
Width (in meter). (b) Change in the direction of surface force after bending of
the cantilever.

example, the approximate surface stress generated on a


cantilever due to antibodyantigen binding is 5 mN/m [4]
and that due to the surface stress induced by the adsorption
of thiolated DNA molecules on gold-coated microcantilever
sensors is on the order of a few newtons per meter [20].
The deflection of affinity cantilevers is normally much
smaller than the length of the cantilever [1]. For such a low
range of deflections, nonlinearity arrived due to cantilever
geometry and material is neglected. However, consideration
of nonlinearity arrived due to changing simulation boundary
conditions [changing surface force direction in Fig. 1(b)] leads
to higher simulation accuracy [21]. Thus, nonlinear simulations
based on finite-element analysis predict the mechanical and
electrical responses of the cantilever well.
2) Cantilever Shape: Piezoresistive affinity cantilever designs of T-shape, inverse T-shape, narrow as well as broad
rectangular shapes, V-shape, and long- as well as short-based
U-shapes have been reported in [22]. A systematic performance
comparison of these cantilevers using constant mass, thickness,
and surface area is discussed. Among those cantilevers, inverse
T-shape and long-based U-shape cantilevers show the highest
surface stress and deflection sensitivity.
Final release of cantilevers often employs wet silicon etching
at elevated temperatures. For U-shape cantilevers, faster release
is possible due to simultaneous etching from the outer and inner
edges of the cantilever. The reduced time of exposure to a wet
etchant results in a lower probability of damage to the ultrathin
cantilevers, thereby increasing the process yield. Therefore, a
long-based U-shape cantilever [Fig. 1(a)] is a preferred choice.
3) Cantilever Thickness: From the analytical formulas
[(13), Appendix B], thickness is the only geometrical parameter
which decides the electrical sensitivity of piezoresistive affinity cantilevers. The minimum thickness of the immobilization
layer may be limited by considerations of chemical, physical,
and biological stability and functionality requirements. For
maximum electrical sensitivity, the piezoresistive layer must be
infinitesimally thin and smooth and must be located close to the
surface where maximum surface stress is developed. However,
to get the required piezoresistive properties, the minimum
thickness is limited by the locally optimized deposition process
parameters and noise considerations. Based on such limitations
described and taking into account the desired mechanical and
electrical sensitivities, analytical calculations yield initial estimates of the desired thickness of the structural layer.

4) Cantilever Width: Determination of the width of affinity


cantilevers is governed by conflicting requirements in terms
of uniformity/reproducibility of immobilization versus electrical power dissipation and mechanical stability. Cantilevers
of smaller widths will have lower planar areas for immobilization with edge effects significantly affecting the uniformity of immobilization and quantitative reproducibility of the
targetprobe interaction. However, during the release of cantilevers using wet silicon etching, the cantilevers with larger
widths need higher etching times, thereby impacting on the
process yield. Furthermore, for a fixed measurement bias, as
the width of the cantilever increases, SNR increases along with
electrical power dissipation. Hence, the width of the cantilever
has to be determined based on biological, microfabrication, and
electrical performances.
5) Cantilever Length: In the microfluidic environment, the
mechanical stability of longer cantilevers gets affected due
to turbulence caused by circular velocity vectors near the
cantilever tip [23]. However, in longer cantilevers, one may
notice lower electrical power dissipation and higher SNR in the
piezoresistive layer. Thus, the length of affinity cantilevers has
to be optimized under considerations of mechanical stability
and performance parameters such as power dissipation and
SNR.
6) Mechanical and Electrical Constraints in Determining
Geometric Dimensions: The performance of piezoresistive
affinity cantilevers is governed by both mechanical and electrical sensitivities. Mechanical parameters like spring constant
and resonant frequency need to be used to design cantilevers of
desired mechanical sensitivity and stability. The spring constant
of affinity cantilevers has been reported to be in the range of
1 mN/m [24]. In this exercise, it was chosen to be approximately 1 mN/m. A cantilever with such low spring constant is
highly susceptible to external mechanical noise. Therefore, it is
possible to minimize the noise by mechanical shielding and/or
designing the cantilever with adequate resonant frequency. In
this exercise, it was chosen to be 5 kHz [24].
Electrical power dissipation and SNR are two important
constraints under which the cantilever design should be optimized. Increase in temperature due to the electrical power
dissipation can destroy the cantilever and/or may impact the
binding constant of the molecular interaction with target molecules; however, such studies are yet not reported. Assuming an
excitation voltage of 5 V, in liquid medium, such cantilevers can
safely handle power up to 2 mW [24]. Hence, power dissipation
was limited to this value. Higher SNRs can be achieved with
reduction in internal noise by controlling design and process
parameters as well [25]. In this paper, the desired figure of merit
of SNR was chosen as 3 and above.
The cantilever under design uses selective immobilization [13] on the top surface. Thus, surface stress due to
antibodyantigen binding is expected only on the top surface
of the cantilever. The thickness of the biolayer (including
surface modifier, linker, and probe molecule) is on the order of
1020 nm with a Youngs modulus of 1 MPa [26], which
are much lower than the cantilever thickness and the Youngs
modulus of the cantilever material. Furthermore, it is unlikely
that the biolayer contributes in any way toward increasing

JOSHI et al.: MODELING, SIMULATION, AND DESIGN GUIDELINES FOR AFFINITY CANTILEVERS

777

TABLE I
M ICROSCALE M ATERIAL P ROPERTIES U SED IN S IMULATION [17], [18], [27], [28]

the stiffness in the direction of bending. Hence, the stiffness


contributed by the biolayer can be neglected and only its effect
as a surface force that occurs due to the surface stress was
considered. In this paper, the surface stress of 5 mN/m [4] was
used for design purposes.

III. I NITIAL D ESIGN OF A P IEZORESISTIVE


A FFINITY C ANTILEVER
The considerations of design presented in the previous section are being illustrated with an example of a multilayer silicon nitride piezoresistive affinity cantilever. In this cantilever,
silicon nitride is the structural layer and p-type polysilicon is
the piezoresistive layer. A thin layer of silicon nitride is used
as surface layer for immobilization on top of the polysilicon
layer. Such a silicon nitride layer can undergo through oxygen
plasma treatment followed by aminosilanization and antibody
immobilization [10]. Gold contact pads are used for the electrical probing of the piezoresistive layer. According to the Ashby
approach for material selection, material properties of macro(bulk) structures can be different than the microstructures [27].
Such properties are functions of length scale and details of
processing techniques employed. The correlation of macroand microstructure properties in [27] was used to choose the
microscale properties of the cantilever material. Important material properties of silicon nitride, polysilicon, and gold films
used for analytical calculations and for simulation studies are
as shown in Table I.
The analytical calculations using formulas listed in Appendixes A and B were performed using Microsoft Excel. The
estimation process comprised of an initial selection of geometrical dimensions followed by iterative adjustments under
the constraints mentioned earlier. The final estimates at this
stage for the length and width of the cantilever were 200 and
40 m, respectively. The thicknesses of silicon nitride, polysilicon, and oxynitride layers were estimated at 100, 45, and 5 nm,
respectively.

IV. S IMULATION AND D ESIGN M ODIFICATIONS


The aim of the simulation study was to model the effects of
stress generated due to probetarget biomolecular interaction
on the cantilever surface and to determine cantilever response.
Such simulations aided in deciding on the exact geometrical
dimensions (thickness, width, and length) of the cantilever.

Fig. 2. Schematic of piezoresistive silicon nitride cantilever used in


simulation.

A. Modeling of the Cantilever Structure


and Boundary Conditions
Using the parameters estimated in the previous section, the
cantilever structure modeled within the CoventorWare MEMS
simulator is shown in Fig. 2.
The length of the piezoresistor in this U-shaped cantilever is
2 Lleg W/2, where Lleg = L W/2. The cantilever structure was meshed using Manhattan bricks with parabolic-order
mesh elements by choosing the aspect ratio of cantilever mesh
elements the same as cantilever geometry.
The compressive surface stress of 5 mN/m (simulating an
increase in repulsive forces on the surface due to the interaction
of the biomolecules) was converted into surface force (Fig. 1)
and applied on the top surface of the cantilever. The cantilever
meshed model was simulated in the nonlinear mechanical and
piezoresistive domains in order to obtain endpoint deflection
and R/R, respectively.
B. Bending of Cantilever and Strain Distribution
Within the Cantilever
It can be seen from Fig. 3 that the bending of the cantilever
near the fixed end is almost negligible and is more noticeable
from the middle of the length of the cantilever. The deflected
profile of the bent cantilever is different for a point force applied
to the free end compared with having surface stress distributed
over the length of the cantilever. For the cantilever under
study, the bending curvature (Fig. 4) is maximum between
the fixed end and middle of the cantilever. The bent profile
of the cantilever bears a good qualitative similarity with the
experimental results reported in [29], thereby validating the

778

JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 20, NO. 3, JUNE 2011

Fig. 3. End-point deflection of cantilever showing predominant bending begins close to the middle of its length.
Fig. 5. Thickness optimization of silicon nitride cantilever for maximum
R/R.

Fig. 4. Stress distribution within the cantilever showing unequal and oppositein-nature stress near the top and bottom surfaces of the cantilever. (Photograph
is magnified by 100 times in Z direction).

modeling of the affinity cantilever. It can also be seen that, in


affinity cantilevers, the stress developed near the top surface,
on which surface stress was applied, is much higher than the
stress developed at the bottom surface. This is in concurrence
with reported findings [2], which mentions that, although the
simple BernoulliEuler model of cantilever bending (small
deflections) would predict that the surface strains generated
at the opposite faces of a cantilever would have the same
magnitude, they are, in fact, different. Hence, to get maximum
stress sensitivity, the piezoresistive layer must be located at a
level within the cantilever where maximum stress is expected.
C. Thickness Optimization of Affinity Cantilever
The constraints in processing stages limit our ability to
actually get required functionality from piezoresistive layers
which are less than 45 nm thick. As a result, the optimization
of the thickness of the total cantilever structure can only be done
by varying the thickness of the structural layer.
As shown in Fig. 5, R/R increases with the reduction of
total thickness and it attains a maximum at a certain value of
thickness below which it falls. The difference between the maximum sensitivities obtained from numerical simulations and analytical calculations is attributed to the nonlinear behavior due
to changing surface force direction, which was not considered
for the first-order estimations in analytical calculations. The
decrease in R/R after the maximum with reducing thickness is
mainly due to the shifting of the neutral axis toward the piezoresistive layer. Accordingly, a cantilever thickness of 100 nm
may be selected for maximum realistic R/R. However, at this

thickness, the values of the mechanical performance parameters


such as spring constant [(3), Appendix A], resonance frequency
[(4), Appendix A], and electrical performance parameters such
as SNR [(14) and (20), Appendix B] need to be estimated. At a
thickness of 100 nm, the spring constant and resonant frequency
of the cantilever are 0.23 mN/m and 3.26 kHz, respectively,
which are lower than the design constraints discussed earlier
(k 5 mN/m, f 5 kHz, SNR > 3, and PD < 2 mW). It
may be possible to achieve the desired performance parameters
by retaining the thickness of 100 nm and reducing the length
or increasing the width of the cantilever. However, such a
strategy may cause degradation in SNR and higher power
dissipation [(16), Appendix B] within the cantilever. Therefore,
the thickness of the structural layer has to be increased at the
cost of a slight decrease in R/R. It was observed that, at a
total thickness of 150 nm and at assumed length and width
of 200 and 40 m, respectively, the value of spring constant
was 0.83 mN/m, resonant frequency was 4.89 kHz, and SNR
was 4.03. Since the SNR is not particularly low, one can settle
with 150 nm as the total thickness of the cantilever. The spring
constant and resonant frequency can be improved further while
analyzing the issues related to the width and the length of the
cantilever.
D. Width Determination of Affinity Cantilever
The reproducibility of the immobilization process of biomolecules on a cantilever surface and, thereby, the reproducibility of its response increase with the increase in its width.
Experimental results (e.g., Fig. 6) show that the immobilization
density is high and irregular near the edges of the cantilever.
Our further investigations suggest that it is mainly due to the
rounding of cantilever edges during the wet release of the cantilever and is independent of its width. The biomolecules on the
rounded edges can have undesirable orientations and may not
contribute in a proportionate fashion toward the development
of surface stresses along the cantilever top surface. Thus, for
immobilization on cantilevers of lower width, the ratio of planar
surface area to the area of rounded edges decreases, thereby
having an adverse effect on its reproducibility. It is therefore

JOSHI et al.: MODELING, SIMULATION, AND DESIGN GUIDELINES FOR AFFINITY CANTILEVERS

779

Fig. 6. Micrograph of FITC-tagged goat antihuman IgG immobilized silicon cantilever observed under an (a) optical microscope and (b) fluorescent microscope,
showing edge effect.

Fig. 7.

Width versus (a) deflection and spring constant and (b) R/R of silicon nitride cantilever.

logical to conclude that the influence of this edge effect in


wider cantilevers will be lower than the narrower ones.
The width of the cantilever impacts on power dissipation
and SNR. From the analytical formulas, it may be presumed
that the deflection of the cantilever is independent of the width.
Simulations were performed to determine the best width under
the discussed conditions of operation and constraints. The
thickness (150 nm) determined in the earlier section was used
along with a length of 200 m.
The cantilever width was varied from 20 to 60 m. Simulations show a small decrease in the deflection with the increase
in the width of the cantilever [Fig. 7(a)]. This departure from
the earlier assumption is due to the fact that the cantilever not
only bends downward but wide and thin cantilevers also curve
downward along their widths near the free end [30]. This causes
an increase in stiffness of the cantilever and, in turn, reduce the
expected deflection and R/R.
Fig. 7(b) shows that, as the cantilever width increases, the
electrical power dissipation and SNR within the cantilever
increase. Higher power dissipation is due to reducing the electrical resistance of the piezoresistor. Since the volume of the
piezoresistor is proportional to the cantilever width, a larger
width causes an increase in the number of free carriers within
the piezoresistor and, hence, an increase in SNR [(17)(20),
Appendix B]. In this paper, the width determined earlier
(40 m) was found to be optimal under the conditions specified.

Fig. 8. Length versus R/R and deflection of silicon nitride cantilever.

E. Length Determination of Affinity Cantilever


The analytical formulas predict that R/R is independent of
the length of the piezoresistive cantilever although this may affect other electrical and mechanical characteristics. Cantilevers
of varying length (100 to 200 m) were simulated, keeping the
thickness and width constant at values decided in earlier sections. As shown in Fig. 8, deflection of the cantilever increases
with an increase in cantilever length. As in the case of the
width, the cantilever shows a slight decrease in R/R as well.

780

JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 20, NO. 3, JUNE 2011

TABLE II
O PTIMIZED G EOMETRICAL D IMENSIONS OF
S ILICON N ITRIDE C ANTILEVER

studies show that they do not increase linearly with the applied surface stress. One possible reason is the aforementioned
change in the surface force directions, which is addressed in the
finite-element analysis performed during simulation but not in
the analytical solutions.
V. C ONCLUSION

TABLE III
P ERFORMANCE PARAMETERS OF O PTIMIZED
S ILICON N ITRIDE C ANTILEVERS

The deviation is consistent with the results published in [30]


and is due to the fact that, for longer cantilevers, change in the
direction of the surface force, as the cantilever bends, is more
pronounced.
This causes reduction in expected deflection, thereby reducing R/R. The mechanical (spring constant and resonant
frequency) and electrical performance (power dissipation and
SNR) parameters for various lengths were estimated. Those
parameters at the length of 180 m closely fit (Table III) to
the desired performance parameters and was the best choice for
the length of the cantilever.
The geometrical dimensions of the optimized silicon nitride
cantilever are summarized in Table II and their performance
parameters are summarized in Table III.
F. Performance of Optimized Piezoresistive Cantilever
The performance of the designed cantilever under different
surface stress conditions (approximating possible variations
due to the nature of immobilization and biomolecular interaction) was investigated using simulation tools. Fig. 9 shows
that the deflection and R/R of the cantilever increase with the
applied surface stress. The electrical sensitivity of the cantilever
was many folds higher than those of the simulation studies
reported earlier [6], [7]. This demonstrates that the design
approach considering the interdependence of various functional elements (e.g., biological, mechanical, and electrical) of
the piezoresistive cantilever and their dependence on material
choice, microfabrication processes, and geometry improved the
sensitivity of the cantilever. From the analytical equations of
affinity cantilevers, these parameters are linear functions of the
surface stress applied on the cantilever. However, simulation

The design guidelines of piezoresistive affinity cantilevers


supported by analytical calculations and numerical simulations
for performance in a liquid environment under the constraints
imposed by microfabrication and electrical and mechanical
considerations have been reported. The effect of stress generated on a cantilever surface due to antigenantibody interaction was modeled using a MEMS simulator. The simulation
results obtained were in concurrence with the experimental and
postulated characteristics of stress-sensitive affinity cantilevers
published by other investigators. It may be concluded that the
thickness of the piezoresistive affinity cantilever is the principal
geometrical dimension which decides the electrical sensitivity of the cantilever. Contrary to expectations from analytical
formulas, the width and length of the cantilever also affect
electrical sensitivity, albeit to a lesser degree. This obviously
improved the accuracy of the analyses although it brought in
(expected) mismatches with the analytically derived parameters
and solutions (Table IV).
A PPENDIX A
M ECHANICAL P ERFORMANCE PARAMETERS
The position of the neutral axis in the cantilever is given
by [2]

Ei Zi hi
(1)
ZN = 
Ei h i
where Ei is the Youngs modulus of the ith layer, Zi is the
position of the ith layer from the neutral axis, and hi is
the thickness of the ith layer. The effective stiffness of the
composite cantilever is given by [6]

   h3
i
2
+ hi (Zi ZN )
EI = W
Ei
(2)
12
i
where (Zi ZN ) is the distance between the center of the ith
layer and the neutral axis.
The effective spring constant ke of the multilayer cantilever
is given by [6]
ke =

3W (EI)
L3

(3)

where the value of (EI) can be calculated from (2).


The total mass mTotal of the cantilever can be estimated as

(di hi ).
(4)
mTotal = LW
i

The resonant frequency of the multilayer cantilever is given


by [22]

ke
fres = 0.32
.
(5)
mTotal

JOSHI et al.: MODELING, SIMULATION, AND DESIGN GUIDELINES FOR AFFINITY CANTILEVERS

Fig. 9.

781

Applied surface stress versus (a) deflection and (b) R/R of optimized silicon nitride cantilever.

The approximate end-point deflection of the cantilever is


given by [2]
=

1 2
L
2

(6)

and is calculated by [2]


=


i


Ei h i

s ZT
i
ZT j=0 hj +

hi
2

2
+

1
3

hi
2

(7)

Rpiezo =

where ZT is the position of the top surface layer with respect to


the neutral axis and is given by [2]



i
hi
Ei h i
j=0 hi 2

.
(8)
ZT = i
Ei h i
i

The minimum detectable end-point deflection of the cantilever in liquid medium is given by [24]

4KB f
min =
.
(9)
2Qkfres
The approximate minimum detectable surface stress can be
estimated by [24]

4 KB T f (1 )1/2 (E1 )1/2 L
smin =
(10)
3
QW
where 1 and E1 are the material density and Youngs modulus
of the structural layer (assuming infinitely thin piezoresistive
layer).
A PPENDIX B
E LECTRICAL P ERFORMANCE PARAMETERS
The resistivity of the P-type piezoresistive polysilicon layer
can be estimated as
R =

1
.
p qP

The mobility of holes in polysilicon material is highly


process parameter dependent. At the doping concentration
near 1019 cm3 , the mobility of holes in polysilicon is approximately one-third of the mobility of holes in silicon
[31], [32].
The electrical resistance of the piezoresistive layer can be
estimated using (11)

(11)


2 Lleg +
W
2

W
2

hpoly

(12)

The change in the piezoresistance due to the surface stress


developed on top of the cantilever is given by [3]



R
hR
ZT ZT j=0 hj + 2

R


= K

s
2



R
i
hi
1 hi 2
Z
E
h

h
+
+
T
i i i
j=0 j
2
3
2
K

1
s .
i Ei h i

(13)

The affinity cantilever under study is expected to be of use


in on-chip Wheatstone bridge configuration. In that case, the
change in output voltage of the Wheatstone bridge can be
estimated as
Vout =

1 R
VB .
4 R

(14)

However, for the design and simulation study demonstrated


in this paper, the bias voltage is applied between the two pads
of the cantilever whose prototype is shown in Fig. 1. Hence,
all the electrical performance parameters demonstrated in this
paper are only for a single piezoresistive cantilever, and some
of them will improve greatly when this cantilever is employed
in the Wheatstone bridge. Current through the piezoresistor can
be estimated as
Ipiezo =

VB
.
Rpiezo

(15)

782

JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 20, NO. 3, JUNE 2011

TABLE IV
TABLE OF VARIABLES

SNR is one of the important performance parameters of the


piezoresistive cantilever. The external noise generated from the
surrounding can be minimized by performing the experiments
in the controlled environment. However, the internal noise
within the cantilever needs to be controlled using the process
parameters [25]. The internal noise sources considered for the
design purpose are Johnson noise and Hooge noise, which
are generated within the piezoresistive layer of the cantilever.
Johnson or white noise power vJ2  due to the thermal fluctuation of charge carriers within the piezoresistor is given
by [2]



vJ2 = 4KB T f Rpiezo .

(17)

The Hooge noise voltage power in the range of frequency


from fmin to fmax is given by [2]


 2  VB2
fmax
vH =
ln
(18)
N
fmin
where is a material constant of piezoresistive polysilicon
and has strong dependence on the annealing temperature and
time [25]. The total number of carriers in the piezoresistor can
be estimated as
N = P Lleg W hpiezo .

(19)

2
 in the piezoresisThe total internal noise voltage power vN
tive cantilever is the sum of the Hooge noise power and Johnson
noise power.
SNR is the figure of merit for the piezoresistive cantilevers.
Neglecting the external noise and the vibration noise, SNR can
be estimated as

2
Vout
S
=
.
(20)
2
N
vN

R EFERENCES

Power dissipation in the piezoresistor can be estimated as


PD = (Ipiezo )2 Rpiezo .

(16)

[1] K. W. Wee, G. Y. Kang, J. Park, J. Y. Kang, D. S. Yoon, J. H. Park,


and T. S. Kim, Novel electrical detection of label-free disease marker
proteins using piezoresistive self-sensing micro-cantilevers, Biosens.
Bioelectron., vol. 20, no. 10, pp. 19321938, Apr. 2005.
[2] P. A. Rasmussen, J. Thysen, O. Hansen, S. C. Eriksen, and A. Boisen,
Optimized cantilever biosensor with piezoresistive read-out, Ultramicroscopy, vol. 97, no. 14, pp. 371376, Oct./Nov. 2003.
[3] J. Thaysen, A. D. Yalinkaya, P. Vettiger, and A. Menon, Polymer based
stress sensor with integrated readout, J. Phys. D, Appl. Phys., vol. 35,
no. 21, pp. 26982703, Nov. 2002.
[4] Y. Arntz, J. D. Seelig, H. P. Lang, J. Zhang, P. Hunziker, J. P. Ramseyer,
E. Meyer, M. Hegner, and C. Gerber, Label-free protein assay based on a
nanomechanical cantilever array, Nanotechnology, vol. 14, no. 1, pp. 86
90, Jan. 2003.
[5] R. P. Ried, H. J. Mamin, B. D. Trrris, L. S. Fan, and D. Rugar,
6-MHz 2-N/m Piezoresistive atomic-force-microscope cantilevers with
INCISIVE tips, J. Microelectromech. Syst., vol. 6, no. 4, pp. 294302,
Dec. 1997.
[6] N. S. Kale and V. R. Rao, Design and fabrication issues in affinity cantilevers for bioMEMS applications, J. Microelectromech. Syst., vol. 15,
no. 6, pp. 17891794, Dec. 2006.
[7] M. Yang, X. Zhang, K. Vafai, and C. S. Ozkan, High sensitivity
piezoresistive cantilever design and optimization for analyte-receptor
binding, J. Micromech. Microeng., vol. 13, no. 6, pp. 864872,
Nov. 2003.

JOSHI et al.: MODELING, SIMULATION, AND DESIGN GUIDELINES FOR AFFINITY CANTILEVERS

[8] Y. C. Lee, B. A. Parviz, A. Chiou, and S. Chen, Packaging for microelectromechanical and nanoelectromechanical systems, IEEE Trans.
Adv. Packag., vol. 26, no. 3, pp. 217226, Aug. 2003.
[9] P. Peluso, D. S. Wilson, and D. D. Huu-Tran, Optimizing antibody immobilization strategies for the construction of protein microarrays, Anal.
Biochem., vol. 312, no. 2, pp. 113124, Jan. 2003.
[10] M. Joshi, S. Singh, B. Swain, S. Patil, R. Dusane, V. R. Rao,
and S. Mukherji, Anhydrous silanization and antibody immobilization
on hotwire CVD deposited silicon oxynitride films, in Proc. IEEE
INDICON, 2004, pp. 538541.
[11] R. A. Williams and H. W. Blanch, Covalent immobilization of protein
monolayers for biosensor application, Biosens. Bioelectron., vol. 9, no. 2,
pp. 159167, 1994.
[12] Y. Wayne and P. C. Robert, Multilayer alkoxysilanes silylation of oxide
surface, Langmuir, vol. 17, no. 19, pp. 58825888, 2001.
[13] M. Joshi, V. R. Rao, and S. Mukherji, AFM characterization and selectivity of immobilization of antibodies in bio-MEMS, in Proc. IBEC,
Singapore, Sep. 2004.
[14] S. Nakata, N. Kido, M. Hayashi, M. Hara, H. Sasabe, T. Sugawara, and
T. Matsuda, Chemisorptions of proteins and their thiol derivatives onto
gold surfaces: Characterization based on electrochemical nonlinearity,
Biophys. Chem., vol. 62, no. 13, pp. 6372, Nov. 1996.
[15] J. N. Lin, J. Herron, J. D. Andrade, and M. Brizgys, Characterization
of immobilized antibodies on silica surface, IEEE Trans. Biomed. Eng.,
vol. 35, no. 6, pp. 466471, Jun. 1998.
[16] A. Tlili, M. A. Jarboui, A. Abdelghani, D. M. Fathallah, and
M. A. Maaref, A novel silicon nitride biosensor for specific antibodyantigen interaction, Mater. Sci. Eng., vol. 25, no. 4, pp. 490495,
Jun. 2005.
[17] P. J. French, Polysilicon: A versatile material for microsystems, Sens.
Actuators, vol. 99, no. 1/2, pp. 312, Apr. 2002.
[18] T. I. Kamins, Design properties of polycrystalline silicon, Sens.
Actuators, vol. 23, no. 13, pp. 817824, Apr. 1990.
[19] A. M. Moulin, S. J. OShea, and M. E. Welland, Microcantilever-based
biosensors, Ultramicroscopy, vol. 82, no. 14, pp. 2331, Feb. 2000.
[20] R. Marie, H. Jensenius, and J. Thaysen, Adsorption kinetics and mechanical properties of thiol-modified DNA-oligos on gold investigated by
microcantilever sensors, Ultramicroscopy, vol. 91, no. 14, pp. 2936,
May 2002.
[21] CoventorWare Analyzer, Reference Guide, MEMS and Microsystems
Design, 2004.
[22] S. Sukuabol, D. K. Sood, and G. Rosengarten, Geometric optimization
of SU-8 piezoresistive cantilever sensors for biochemical application, in
IEEE Int. Conf. ISSNIP, Melbourne, Australia, 2005, pp. 247252.
[23] M. T. Clark and M. R. Paul, The stochastic dynamics of rectangular and
V-shaped atomic force microscope cantilevers in a viscous fluid and near
a solid boundary, J. Appl. Phys., vol. 103, no. 9, pp. 094910-1094910-9,
May 2008.
[24] J. Thaysen, Cantilever for bio-chemical sensing integrated in micro
liquid handling system, Ph.D. dissertation, Mikroelectronik Centre,
Technical Univ. Denmark, Lyngby, Denmark, Jun. 2001.
[25] J. A. Harley and T. W. Kenny, 1/F noise considerations for the design and
process optimization of piezoresistive cantilevers, J. Microelectromech.
Syst., vol. 9, no. 2, pp. 226235, Jun. 2000.
[26] J. W. Kim, Y. Yamagata, B. J. Kim, S. Takeuchi, and T. Higuchi, Biosensor based on SU-8 cantilever by using the electro spray deposition of
proteins, in Proc. 7th Int. Conf. Miniaturized Chem. Biochem. Anal. Syst.,
Squaw Valley, CA, Oct. 2003, pp. 399402.
[27] V. T. Srikar and S. Mark Spearing, Material selection in micromechanical
design: An application of Ashby approach, J. Microelectromech. Syst.,
vol. 12, no. 1, pp. 310, Feb. 2003.
[28] N. Maluf, An Introduction to Microelectromechanical Systems
Engineering. Boston, MA: Artech House, 2004, ser. MEMS series.
978-1-58053-590-8.
[29] S. Jeon, N. Jung, and T. Thundat, Nanomechanics of a self-assembled
monolayer on microcantilever sensor measured by a multiple-point deflection technique, Sens. Actuators B, Chem., vol. 122, no. 2, pp. 365368,
Mar. 2007.
[30] F. T. Goericke and W. P. King, Modeling piezoresistive microcantilever
sensor response to surface stress for biochemical sensors, IEEE Sensors
J., vol. 8, no. 8, pp. 14041410, Aug. 2008.
[31] P. Nishida and C. Sah, A physically based mobility model for MOSFET
numerical simulation, IEEE Trans. Electron Devices, vol. ED-34, no. 2,
pp. 310320, Feb. 1987.
[32] K. Suzuki, Unified minority carrier transport equation for polysilicon
or hetromaterial emitter contact bipolar transistor, IEEE Trans. Electron
Devices, vol. 38, no. 8, pp. 18681877, Aug. 1991.

783

Manoj Joshi received the M.Tech. degree in electrical engineering and the Ph.D. degree in bio-MEMS
from the Indian Institute of Technology Bombay,
Mumbai, India, in 2001 and 2007, respectively.
Since 2007, he has been a Principal Engineer
and Member of Technical Staff with R&D, Taiwan
Semiconductor Manufacturing Company Limited,
Hsinchu, Taiwan. He is currently working on 28-nm
node high- and metal-gate CMOS device development. He has authored 24 peer-reviewed journal and
conference papers, along with U.S., Chinese, and
Indian patents. His research interest includes device physics, layout-dependent
effects in advanced CMOS technologies, and bio-MEMS.

Prasanna S. Gandhi (M99) received the B.Eng. degree in mechanical engineering from the University
of Bombay, Mumbai, India, in 1994, the M.Tech.
degree in mechanical engineering from the Indian
Institute of Technology Bombay, Mumbai, in 1996,
and the Ph.D. degree in mechanical engineering from
Rice University, Houston, TX, in 2001.
Since 2001, he has been a Faculty Member, currently an Associate Professor, in the Department of
Mechanical Engineering, Indian Institute of Technology Bombay. He has coordinated and set up a new
laboratory, Suman Mashruwala Microengineering Laboratory, for research in
microdomain and has successfully completed several research projects sponsored by the government and private sector. He has been a qualified teacher of
stress-relieving and life-enhancing techniques of the Art of Living Foundation
by Sri Sri Ravishankar. He has authored over 50 peer-reviewed conference
and journal papers, along with one U.S. patent and three Indian patents
(pending). His research interests are in the areas of MEMS and microsystems,
mechatronics, and nonlinear dynamical systems and control.
Dr. Gandhi was a recipient of the 2006 BOYSCAST fellowship from the
Government of India and the Prof. J. R. Issac Fellowship.

Rakesh Lal received the B.Tech. degree in electronics and communication engineering from the Indian
Institute of Technology (IIT) Kharagpur, Kharagpur,
India, the M.D. degree in electronics from NUFFIC,
The Hague, The Netherlands, and the Ph.D. degree
in electrical engineering from IIT Kanpur, Kanpur,
India.
He was a faculty member at IIT Bombay, Mumbai,
India, and retired as a Professor from the Electrical Engineering Department. He has also been a
consultant to industry for designing computer-aided
measurement systems for motors and transmission components and to the Government of India on electronics and computer policies. His research interests
include physics and modeling of semiconductor devices, radiation and highfield effects in MOS and bipolar devices, and instrumentation for device characterization. He has also worked extensively on a variety of chemical, radiation,
and biosensors, many as part of an interdisciplinary group encompassing the
Departments of Chemistry, Chemical Engineering, and Materials Science.

784

JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 20, NO. 3, JUNE 2011

V. Ramgopal Rao (M98SM02) received the


M.Tech. degree from the Indian Institute of Technology (IIT) Bombay, Mumbai, India, in 1991, and the
Dr.Ing. degree from the Universitaet der Bundeswehr
Munich, Munich, Germany, in 1997.
From 1997 to 1998 and again in 2001, he was
a Visiting Scholar in the Department of Electrical
Engineering, University of California, Los Angeles.
He is currently a Professor in the Department of
Electrical Engineering and the Chief Investigator for
the Centre of Excellence in Nanoelectronics at IIT
Bombay. He has over 280 publications in the area of electron devices and
nanoelectronics in refereed international journals and conference proceedings
and is the holder of 15 patents issued or pending. He serves on the editorial
boards of various other international journals.
Dr. Rao is a Fellow of the Indian National Academy of Engineering, the
Indian Academy of Sciences, and the National Academy of Sciences in India.
He is an Editor for the IEEE T RANSACTIONS ON E LECTRON D EVICES in
the CMOS Devices and Technology area. He is a Distinguished Lecturer of
the IEEE Electron Devices Society and has served on the program/organizing
committees of a large number of international conferences in the area of
electron devices. He was Chairman of the IEEE AP/ED Bombay Chapter
during 20022003 and currently serves on the executive committee of the
IEEE Bombay Section besides being the Vice-Chair of the IEEE Asia-Pacific
Regions/Chapters Subcommittee. He was the recipient of the coveted Shanti
Swarup Bhatnagar Prize in Engineering Sciences awarded by the Honorable
Prime Minister of the Government of India in 2005 for his work on electron
devices. He is also a recipient of the 2004 Swarnajayanti Fellowship from the
Department of Science and Technology, the 2007 IBM Faculty Award, the 2008
Materials Research Society of India Annual Prize, and the 2009 TechnoMentor
award from the Indian Semiconductor Association.

Soumyo Mukherji received the B.Tech. degree in


instrumentation engineering from the Indian Institute
of Technology (IIT) Kharagpur, Kharagpur, India,
in 1989, the M.S. degree in mechanical engineering from Colorado State University, Fort Collins, in
1992, and the Ph.D. degree in biomedical engineering from the University of North Carolina, Chapel
Hill, in 1997.
He is currently a Professor in the Department
of Biosciences and Bioengineering, IIT Bombay,
Mumbai, India. His research interests include physical, chemical, and biological sensing systems (macro and micro) for medical/
biological applications and telemedicine systems.

You might also like