Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

A high resolution capacitive MEMS force sensor with novel

bionic swallow comb arrays for multiphysics measurement


This paper was downloaded from TechRxiv (https://www.techrxiv.org).

LICENSE

CC BY-NC-SA 4.0

SUBMISSION DATE / POSTED DATE

19-05-2022 / 24-05-2022

CITATION

Gao, Wendi (2022): A high resolution capacitive MEMS force sensor with novel bionic swallow comb arrays
for multiphysics measurement. TechRxiv. Preprint. https://doi.org/10.36227/techrxiv.19793593.v1

DOI

10.36227/techrxiv.19793593.v1
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 1

A High Resolution MEMS Capacitive Force Sensor


with Bionic Swallow Comb Arrays for Ultralow
Multiphysics Measurement
Wendi Gao, Cunlang Liu, Libo Zhao, Qijing Lin, Zhuangde Jiang, Dong Sun, Fellow, IEEE

 signals do not only suffer from small amplitude but they are also
Abstract—Precise force sensing is essential for the mechanical accompanied by strong background noise from the cellular
characterization and robotic micromanipulation of biological environment [9], [10]; therefore, force sensing techniques with
targets. In this work, a high-resolution MEMS capacitive force high sensitivity and resolution are in urgent demand.
sensor was proposed for measuring ultralow multiphysics. A In practice, atomic force microscopy (AFM) has long been
bionic swallow structure that contained multiple feathered comb adopted for micro-force measurement. However, except for
arrays was designed for reducing chip dimension and eliminating
undesirable mechanical cross-coupling effect. The comb structure
labor-intensive optical alignment and adjustment procedures,
was optimized for maximum sensitivity, linearity, and compact AFM mostly suffers from a short measurement range [11] and
chip size. Utilizing a novel interconnection configuration, limited operation space [12]. A microelectromechanical system
interferences derived from parasitic capacitance and electrostatic (MEMS) sensor, which takes advantage of a task-oriented
forces exerted negligible effects on the sensor output. The design, high sensing performance, easy integration with end
proposed bionic force sensor was fabricated following a simple effectors, and mass production [13], is a competitive alternative
three-mask process and integrated with ASIC readouts. Its for this task. Compared with MEMS force sensors based on the
measuring sensitivity was 7.151 fF/nm, 0.529 aF/nN, and 4.247 principles of piezoresistivity [14], [15], piezoelectricity [16],
pF/g for displacement, force, and inclination measurements, [17], optical detection [18], [19], and field-effect transistors
respectively. The proposed sensor had a large measurement range
of 1000.00 nm and 13.83 µN with a high linearity of 0.9998. The 1-
[20], [21], a capacitive sensor is superior in terms of high
σ resolution was 0.0328 nm and 0.4436 nN, and the noise floor sensitivity [22], direct integration ability with application-
resolution was 0.0044 nm √ and 0.0597 nN/ √ for specific integrated circuit (ASIC) readouts [23], and easy
displacement and force measurements, respectively. The bias fabrication [24]; it has become the mainstream in ultralow
stability of Allan deviance was 0.0050 nm and 0.0678 nN at an multiphysics measurement.
integration time of 0.65 s. The proposed bionic swallow sensor Many capacitive sensors based on area-variant or gap-
exhibited considerable improvement over existing capacitive variant type combs have been developed. An area-variant type
sensors and feasibility for ultralow multiphysics measurement in capacitive sensor with nonlinear buckling anti-springs was
biomedical applications.
developed with a noise floor resolution of 0.054 nm/√Hz [25];
Index Terms—MEMS Capacitive sensor; Bionic design;
Mechanical characterization; Robotic micromanipulation;
however, its deformation sensitivity was as low as 2.55 aF/nm.
Biomedical applications. An asymmetric structure with three anti-springs was proposed
I. INTRODUCTION as an area-variant type capacitive sensor [26]; its sensitivity was
increased to 242.27 aF/nm and its noise floor resolution can
T HE mechanical characterization of micro/nano-scaled
materials is of considerable importance. In particular,
Young’s modulus, viscosity, and other biomechanical
reach 0.0088 nm/ √Hz . Compared with area-variant type
capacitive sensors, a gap-variant type sensor with higher
properties have been intensively measured to investigate intrinsic electrical sensitivity of 1.73 fF/nm and force resolution
biological status and biomedical reactions in living tissues [1], of 680 nN was proposed for characterizing fruit fly flight
embryos [2], cells [3], and organelles [4]; this process requires behavior [27]. A gap-variant type capacitive force sensor was
precise force and deformation sensing information below nano- fabricated via a narrow initial capacitor gap of 0.9 µm; the
Newton and nanometer levels. To increase automation extent sensitivity of this sensor increased to 2.58 fF/nm but at the costs
and reduce operation damage in cell surgery, force and of resolution, measurement range, and linearity [28]. Increasing
inclination sensing provide robotic manipulation tasks with a the number of capacitive combs in a single chip has been widely
real-time feedback via interaction force regulation [5], [6] and utilized in the aforementioned practices to amplify sensor
operation orientation control [7], [8]. However, biological output. However, more combs can result in large chip size, a

This work was supported by grants from the National Key Research and Innovation, Xi’an Jiaotong University (Yantai) Research Institute for
Development Program of China (Ref. no. 2021YFB3201704), National Natural Intelligent Sensing Technology and System, School of Mechanical
Science Foundation of China (Ref. no. 52105589, U1909221), and the China Engineering, Xi’an Jiaotong University, Xi’an, China (e-mail:
Postdoctoral Science Foundation (Ref. no. 2021M692590). (Corresponding wendi_gao@xjtu.edu.cn or libozhao@xjtu.edu.cn).
author: Libo Zhao). Dong Sun is with the Department of Biomedical Engineering, City
Wendi Gao, Cunlang Liu, Libo Zhao, Qijing Lin, Zhuangde Jiang are with University of Hong Kong, Hong Kong, China, and also with International Joint
the State Key Laboratory for Manufacturing Systems Engineering, Laboratory for Micro/Nano Manufacturing and Measurement Technologies,
International Joint Laboratory for Micro/Nano Manufacturing and Xi’an Jiaotong University, Xi’an, China. (e-mail: medsun@ cityu.edu.hk).
Measurement Technologies, Overseas Expertise Introduction Center for
Micro/Nano Manufacturing and Nano Measurement Technologies Discipline
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 2

high possibility of geometric asymmetry, unavoidable parasitic the head region, and the structure is supported by four straight
capacitance, and electrostatic interferences, reducing device beams at the corners of the wing region and two similar beams
dynamics and accuracy. On the basis of a closed-loop position at the tail region. Four internal fixed plate arrays are placed on
feedback measurement principle, a nulling mechanism wherein the two cantilevers. The two resting fixed comb arrays are
the deformed internal structure was driven back by a set of placed on the outside frame, forming six comb arrays (Clo, Cli1,
comb actuators was introduced [29]; the force and displacement Cli2, Cri1, Cri2, and Cro) with corresponding movable plate
resolution were increased to 2 nN and 0.2 nm, respectively. A arrays.
proper driving signal should be computed and applied to the The comb arrays in the swallow body and wings region move
actuators, leading to a slow signal response [30], [31]. One laterally when a lateral force Fy is applied to the probe.
alternative is utilizing an additional actuator to adjust the Consequently, the gap between the movable and fixed comb
working position independently [32], [33]; the force resolution plates changes, which is expressed as
of such devices can reach 0.514 nN. However, this phenomenon
= , (1)
can only occur within a small measurement range of 0.230 µN.
Measuring minor and fragile biological targets with feedback is where Iy = is the moment inertia of the beam vertical
problematic because the actuation movement can probably
produce unnecessary slippery interactions and operation section; and Lb, Wb, and Tb are the length, width, and thickness
damages [34], [35] in biological cells. To meet the requirements of the supporting beam, respectively. E is the Young’s modulus
for biomedical multiphysics measurement, more effective effort of the silicon material. Interferences along the x and z axis may
should be initiated to improve the sensitivity, resolution, and cause deflections and then generate a disturbance in the signal
linearity of capacitive sensors simultaneously. of interest. Thus, they should be eliminated in the mechanical
In this study, a gap-variant type capacitive MEMS force design. A flexible supporting beam with a long length is
sensor with high performance was developed. The major
contributions of this work are as follows. First, inspired by a
gliding swallow, a bionic sensitive structure with multiple
feathered comb arrays was proposed. This structure largely
reduces chip dimension and eliminates undesirable mechanical
cross-coupling effect. Second, an optimization analysis of
mechanical and capacitive sensitivities was performed over the
comb structure wherein sensitivity and measurement ranges
were maximized while nonlinearity was suppressed. Third, a
novel decoupling sensing configuration was proposed.
Interferences derived from parasitic capacitance and
electrostatic forces were largely dismissed, leading to high-
resolution detection with low noise. Fourth, the proposed bionic (a)
force sensor was fabricated via a simple three-mask process and
integrated with ASIC readouts. Fifth, the fabricated sensors
were calibrated for deformation, force, and inclination
characterization, proving their overwhelming performances and
high potential in ultralow biomedical measurement.
The remainder of this paper is organized as follows. Section
II describes the design, optimization, and fabrication of the
proposed bionic sensor. In the Section III, the characterization
experiments of the fabricated sensors are presented, followed
by the performance discussion of the sensor. Conclusions are
provided in the last section.
II. SENSOR DESIGN AND FABRICATION (b)
A. Bionic Swallow Structure Design Fig. 1. Bionic design of the proposed bionic swallow sensor. (a) Schematic of the proposed
bionic swallow structure. (b) Schematic of the proposed decoupling sensing configuration.
Capacitive combs are widely adopted in capacitive sensor
design. However, a rare association has been observed that desirable for high sensitivity, while the structure is inevitably
comb plates exhibit an arrangement similar to swallow wing involved in vertical bending deformation as beam length
feathers. Swallow typically flies with a gliding gesture to increases. A high aspect ratio beam is normally adopted for
maintain body balance with high kinetic stability even under enhancing the moment inertia of the beam horizontal section Iz
thunderstorms. Similarly, interferences from the sensing = , while sensor sensitivity deteriorates. Therefore, a
environment can produce a huge unbalance for the sensing
delicate trade-off should be achieved between stiffness along
structures of capacitive sensors, and thus, should be overcome.
Inspired by this bionic observation, a swallow movable these axes. For the swallow structure, a zero vertical bending
structure was designed in which six feathered movable plate point exists along the body and wing regions, wherein vertical
arrays were placed in the body and wing regions, as shown in loading produces contrary translation and rotation motions. The
Fig. 1(a). The bionic swallow structure has a beak probe tip in comb arrays are placed near this balanced position to obtain a
negligible vertical deformation.
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 3

placed away from the middle of the fixed combs, leaving an


asymmetric air spacing d1, d2 between two neighboring comb
plates, as shown in Fig. 3. When the structure moves a distance
of Δd, the capacitance of the differential comb pair C1, C2 can
be expressed as
= + , (2)
= + , (3)
where ε is the permittivity of air, and A is the overlapped area
(a) (b)
of the comb plates. Therefore, capacitance change ΔC can be
described as
=2 − . (4)
For a relatively small Δd, ΔC equals 2 − and
exhibits full linearity. A maximum sensitivity of 2 can
be obtained for a large gap d2 >> d1. However, a large air spacing
increases chip size, resulting in low output sensitivity.
Therefore, a comb optimization process should be performed to
(c) achieve balance among sensitivity, linearity, and sensor size. In
Fig. 2. Mechanical simulation of the proposed bionic swallow structure. (a) Simulated
displacement along y axis. (b) Simulated displacement along x axis. (c) Simulated the current work, electromechanical simulations were
displacement along z axis. conducted over comb sizes by using COMSOL
To investigate the deformation behavior of the proposed MULTIPHYSICS. In the simulation, the movable swallow
swallow structure, finite element method (FEM) simulations structure was deformed at 1 µm with a step size of 0.2 µm. To
were conducted using COMSOL MULTIPHYSICS with the simplify computational complexity, the swallow sensor was
parameters listed in Table I. In the simulations, forces along the modeled with one pair of differential comb plates. Considering
x/y/z axis were applied to the probe tip. Consequently, the fabrication difficulties and accuracy in our laboratory, a
deformation along the y axis was 1.063 µm for a 10 µN loading small but achievable gap spacing d1 of 3 µm was adopted for
along this axis, while forces along the x/z axis produced only a maximizing sensitivity. Then, gap d2 was scanned for high
small deformation of 0.018 nm and 0.045 nm, respectively, as linearity and compact size. The overlapped length and thickness
of the comb plates were 375 µm and 50 µm, respectively. The
shown in Fig. 2(a). For the deformation along the x/z axis, six
positive combs C1 exhibited an increasing potential
comb arrays exhibited similar deflection behavior but with
concentration within its narrow air spacing while the potential
different values. The average deformation of comb array Clo in distribution in the negative combs C2 became sparse during
the wing region is illustrated in Figs. 2(b) and (c). The largest structure movement, as shown in Fig. 4(a).
deformation along the x axis was 0.382 nm because the adopted
straight beams exhibited considerable resistance to stretching
and compression. The largest vertical deformation was 0.040
µm for a 10 µN force, and this value was only 3.7% of the
deformation along the y axis. However, vertical interferences
can hardly reach such level because vertical loadings only
account for a small proportion in lateral measurement. By
employing this bionic structure design, the proposed swallow
structure is highly sensitive to loadings along the operation
direction but insusceptible to interferences from other
directions. Fig. 3. Tri-plate differential comb configuration.
TABLE I The relationship between capacitive change and comb
GEOMETRY PARAMETERS OF THE PROPOSED SWALLOW SENSOR
Parameters
displacement is illustrated in Fig. 4(b). Sensing performance
Value
Width of supporting beams Wb 5 µm under different spacing gaps is summarized in Table II. For a
length of supporting beams Lb 800 µm, 600 µm* small gap d2 of 5 µm, the differential comb pair achieved the
Structure Thickness Tb 50 µm
Length of center mass 2000 µm lowest sensitivity of 26.04 fF/µm and linearity of 0.9957.
Width of center mass 5000 µm Similar to the theoretical analysis, capacitance sensitivity and
Number of plates in each array 110
linearity improved with an increment in gap d2. Capacitance
* 800 µm was for the wing beams, 600 µm was for the tail beams.
sensitivity reached 37.52 fF/µm and linearity approached
B. Comb Configuration and Optimization 0.9975 when d2 increased to 10 µm. Notably, linearity remained
Gap-variant type combs exhibit high sensitivity but suffer constant while sensitivity rarely changed once d2 exceeded over
from a low linear measurement range because of their narrow 15 µm, indicating that the interferences from the second term in
gap. In the current work, a tri-plate differential comb Equations (2) and (3) were nearly eliminated. Therefore, the
configuration was adopted to improve linearity and initial space gaps d1 and d2 were set as 3 µm and 15 µm,
measurement range, wherein the movable comb plates were respectively, in this design to obtain high linearity and maintain
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 4

compact chip size simultaneously, wherein the sensitivity of layer surface. Third, a thick photoresist layer was spun onto the
one differential comb pair reached 37.52 fF/µm and linearity handle layer and exposed with the third mask, followed by deep
was 0.9975. Considering the number of comb plates in three RIE for exposing the movable swallow structure from the
differential arrays, theoretical sensitivity can reach 12.451 backside, which was then released via isotropic etching of a 2
fF/nm. µm sacrificial silicon oxide layer.

(a) (b)
Fig. 4. Electrical simulation of the proposed bionic swallow sensor. (a) Simulated potential
distribution of one differential comb pair. (b) Effect of asymmetric spacing gaps on comb
capacitance for 1µm comb movement.
TABLE II
SENSING PERFORMANCE OF ONE TRI-PLATE COMB PAIR UNDER DIFFERENT SPACING GAP
d1 d2 Sensitivity Linearity Comb Size
Case (µm)
(µm) (µm) (fF/ µm)
1 3 5 26.04 0.9957 8
2 3 10 35.97 0.9971 13
3 3 15 37.76 0.9974 18
4 3 20 38.37 0.9974 23

A decoupling sensing configuration was proposed on the


basis of this tri-plate differential comb design. Comb arrays Clo, Fig. 6. Microscopic image of the fabricated sensor.
Cri1, and Cri2, which exhibited the same spacing layout, shared Microscopic observation of the fabricated sensor is
the same electrical potential and were connected to PAD2, illustrated in Fig. 6. The comb plate deformed from the original
serving as differential comb pairs with arrays Cli1, Cli2, and Cro, position because of the structural inner stress generated
as shown in Fig. 1(b). The vertical deformation resulting from throughout the whole fabrication process, and the measured
z axis disturbances was identical for each differential comb spacing gaps d1 and d2 were 3.45 µm and 16.97 µm,
array pair, and it can be eliminated in this differential output. respectively. Consequently, the sensitivity of the electrical
Moreover, the electrostatic forces of the movable swallow comb should deteriorate and be verified in the next section. The
structure from arrays Clo, Cri2, and Cri2 can be counteracted by width of the supporting tail and wing beams Wb was measured
those from arrays Cli1, Cli2, and Cro. Meanwhile, electrostatic in- as 5.80 µm. On the basis of Equation (1), stiffness along the y
plane torsion can be eliminated by the electrostatic forces. axis of the fabricated sensor should be 1.14 times that of the
Therefore, electrical interferences can also be largely decoupled designed one. To limit overloaded movement, 3 µm trenches
and high resolution is guaranteed. existed between the outside frame and the movable swallow
structure at the wing and tail regions. The fabricated sensor also
had a step profile tip with a measured width of 2.95 µm and
length of 29.87 µm. This tip was capable of the
micromanipulation and biophysical measurement of small-
scaled mammalian cells at the micrometer level.
D. Chip Packaging and Readouts
The fabricated sensor was packaged on a customized print
Fig. 5. Three-mask fabrication process of the proposed bionic swallow sensor. circuit board (PCB). A transparent poly(methyl methacrylate)
. (PMMA) shelter was glued on top of the sensor to shield it from
C. Device Fabrication falling dust and airflow disturbances, as shown in Fig. 7. A 24
The designed swallow sensor was fabricated using a 4-inch bit ASIC chip (AD7746, Analog Devices) with a resolution of
silicon on insulator (SOI) wafer with a simple three-mask 4 aF and a measurement range of ±21 pF was employed for
process, as shown in Fig. 5. First, a gold layer with a thickness converting capacitance change into digital output. Notably, the
of 50 µm and a low resistivity of 0.02 Ω∙cm was deposited onto total capacitance of the connected comb arrays initially
the device layer. A photoresist layer was deposited and exposed approached the upper limit, and capacitance enlargement
with the first mask. The deposited gold layer was etched to form during the swallow structure’s movement may exceed the
interconnection pads. Second, another photoresist layer was measurement range. Therefore, a range extension circuit was
deposited and patterned with the second mask. The device layer connected to the corresponding comb pads before conversion,
of the SOI wafer was etched with reactive ion etching (RIE), wherein an operational amplifier (AD8515, Analog Devices)
serving as the movable structure. To protect the patterned operated as a low-impedance source for fully charging the
swallow structure, a silicon wafer was bonded to the device sensing capacitance of two comb array series, as shown in Fig.
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 5

8. An electrolytic capacitor Cd1 and a ceramic capacitor Cd2 implemented for manipulating the two motorized stages, and a
were connected in parallel for frequency decoupling. The graphical user interface (GUI) was integrated into the on-site
resistors R1 and R2 can scale down the excitation voltage by a computer for visualized operations.
range magnification factor of 4.5, covering capacitance change B. Displacement Measurement
for a maximum measurement range of 94.5 pF. The initial A rigid glass cube mounted onto the left specimen stage was
capacitance difference of the two differential arrays C1 and C2 employed for displacement experiments. The packaged sensor
is 0.21 pF, indicating high consistency of the fabricated comb was screwed onto a 3D-printed PMMA plate through bolt holes
arrays. on the PCB board and then screwed onto the right manipulation
stage. In the beginning, the height of the specimen and the
manipulation stage was adjusted so that the sensor probe and
cube can be fully focused for clear observation. The
manipulation stage then moved the sensor probe to approach
the cube, and a reflection of the probe tip appeared inside the
cube once it got close to the cube side surface, as shown in Fig.
10. The sensor probe moved with a smaller step and lower
velocity until the recorded capacitance signal experienced a
jump, indicating that the probe was in full contact with the cube
side surface. Then, the manipulation stage initiated step
Fig. 7. Packaged sensor on the customized PCB board.
movements with a time interval of 200 ms. After each step
movement, the stage remained in the same position for the
subsequent nine intervals. The capacitance signal was recorded
with a sampling rate of 9.1 Hz, which was about two times the
stage motion rate. Considering the considerable stiffness
difference between the sensor and the tested glass cube, the
cube was rarely deformed but sensor deformation was
equivalent to stage step size in each movement.
The sensor was first tested with a step size of 100 nm for a
total deformation of 1 µm. Fig. 11(a) illustrates the capacitance
Fig. 8. Sensor readout circuit with an ASIC chip AD7746 and range extension circuit.

Fig. 10. Micromanipulation process for the displacement measurement.

change that exhibited a stepwise increase during stage


movement. However, contact slip and environmental
disturbances could result in a ±20 nm deviation of the stage
motion, as reflected through fluctuations of the recorder
Fig. 9. Robotic micromanipulation system. manipulation stage position signal shown in Fig. 11(b).
Deformation sensitivity was calculated using the slope factor
III. SENSOR CHARACTERIZATION AND MEASUREMENT
between the average capacitance change and stage position. The
A. Setup of the Robotic Micromanipulation System derived sensitivity was 7.185 fF/nm, and linearity was 0.9996.
A robotic microscopy system was built and adopted for Then, the sensor was tested with a larger step size of 200 nm,
sensor characterization. It comprised three modules, namely, wherein stage motion deviation accounts for a smaller
imaging, manipulation, and control, as shown in Fig. 9. A proportion, as shown in Fig. 11(d). Consequently, capacitance
laboratory-made orthographic microscope integrated with change presented fewer fluctuations than in the 100 nm tests, as
multiple objective lenses and a CCD camera (CS895CU, shown in Fig. 11(c). The calibration results demonstrated good
Thorlabs) was used for specimen observation. The agreement with the results of the 100 nm experiment, which had
manipulation module consisted of two motorized stages. (1) a deformation sensitivity of 7.115 fF/nm and a linearity of
The tested object is mounted on the left specimen stage 0.9995, as shown in Fig. 11(e). To verify measurement
(P611.ZS, PI), which exhibited a high vertical accuracy of 0.2 accuracy and repeatability, calibration attempts were repeated
nm and a large movement range of 100 µm. (2) two more times for each experiment. The average and standard
Micromanipulation tools can be assembled on the right stage, deviation of the displacement sensitivity were 7.151 fF/nm and
which had a planar positioner (P611.2S, PI) with a high 0.035 fF/nm, respectively, which were smaller than the
resolution of 0.2 nm and a movement range of 100 µm and a designed sensitivity of 12.461 fF/nm at a spacing gap d1 of 3
vertical positioner (XP-611.Z100S, XMT) with a positioning µm but were consistent with the measurement spacing gap d1 of
resolution of 1 nm and a movement range of 100 µm. In the 3.45 µm.
control module, a closed-loop position controller was
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 6

and environmental disturbances exerted negligible effects on


capacitance change. Fig. 13(b) illustrates the average
capacitance change under the corresponding position, which
was 21.676 fF for each step movement ΔD. Considering the
measured deformation sensitivity and AFM cantilever stiffness,
the average sensor deformation Δd was 3.031 nm and sensor
stiffness k was 13.485 N/m. The calibration experiments were
(a) (b) repeated five times to eliminate system error. The calibration
results are summarized in Table IV. The average and standard
deviation of sensor stiffness were 13.524 N/m and 0.026 N/m,
respectively. When the measured deformation sensitivity was
divided by the calibrated sensor stiffness, the average force
sensitivity was 0.529 fF/nN and linearity was better than
0.9998.
(c) (d)

Fig. 12. Micromanipulation process for the force measurement.

(e)
Fig. 11. Displacement experiment results of the proposed sensor (a) Capacitance change for
the 100 nm step experiment. (b) Manipulation stage position for the 100 nm step experiment.
(c) Capacitance change for the 200 nm step experiment. (d) Manipulation stage position for
the 200 nm step experiment. (e) Relation between the average capacitance change and sensor
displacement.
TABLE III (a) (b)
EXPERIMENTAL RESULTS OF DISPLACEMENT MEASUREMENTS Fig. 13. Force experiment results of the proposed sensor. (a) Capacitance change for 10
Step Deformation Linearity Average Standard µm manipulation movement with a step size of 1 µm. (b) Relation between the average
No. Movement Sensitivity Sensitivity Deviation capacitance change and stage movement.
(nm) (fF/ nm) (fF/ nm) (fF/ nm) TABLE IV
1 100 7.185 0.9996 EXPERIMENTAL RESULTS OF FORCE MEASUREMENTS
2 100 7.178 0.9997 Capacitive Sensor Sensor Average Standard Force
Linearity
3 100 7.186 0.9998 No. Change Deformation Stiffness Stiffness Deviation Sensitivity
4 200 7.115 0.9995 7.151 0.035 (fF/step) Δd (nm/step) k (N/ m) (N/ m) (N/ m) (fF/ nN)
5 200 7.125 0.9996 1 21.676 0.9998 3.031 13.485
6 200 7.117 0.9996 2 21.574 0.9999 3.017 13.549
3 21.601 0.9997 3.021 13.532 13.524 0.02621 0.529
C. Force Measurement 4 21.633 0.9999 3.025 13.512
5 21.582 0.9998 3.018 13.544
The force-sensing performance of the proposed sensor was
calibrated using the same robotic microscopy system wherein D. Inclination Measurement
an AFM cantilever (Bruker, SNL-10-D) was glued onto the An inappropriate orientation angle between
specimen stage. The sensor probe and AFM cantilever were micromanipulation tools and biological cells can produce
optically focused in the beginning of the experiments. The overloaded interaction force, resulting in operation damage.
manipulation stage moved the sensor probe along the y-axis to The incorporation of displacement, force, and inclination
align the probe tip with the AFM cantilever tip. Then, the sensor sensing into one chip is highly preferred. In this work, an
probe approached the AFM cantilever until a signal jump inclinator system (Y200RA100, JYGD) was utilized for
occurred. Similar to the preceding displacement measurements, inclination characterization, as shown in Fig. 14(a). The
the sensor probe was manipulated to compress the cantilever tip packaged sensor was screwed onto the turnplate that was
with a stepwise movement, as shown in Fig. 12. connected to a controller and manipulated through the GUI of
In accordance with Hooker’s law, the interaction force the on-site computer. The packaged sensor initially headed
between the sensor probe ksΔd should be equal to that of the downward and rotated clockwise from −90° to 90° with a step
cantilever tip kaΔda; thus, sensor displacement Δd exhibits a size of 10°, as shown in Fig. 14(b).
proportional relation = with the cantilever tip Δda. The capacitance change signal was recorded at 9.1 Hz and
presented a trigonometric profile during the 180° inclination
For each step movement ΔD, sensor displacement is equal to
process, as shown in Fig. 15(a). In the beginning, gravity force
. By utilizing the standard procedure, the stiffness of was acting fully on the sensor such that the swallow structure
the AFM cantilever ka was calibrated as 0.041 N/m before the moved forward at the largest extent. Capacitance change was
experiment. Fig. 13(a) presents the capacitance change for a −4.243 pF. Effective loads from gravity force decreased as
total stage movement of 10 µm. The interaction between both inclination angle increased to 0°. Capacitance change returned
tips was solid because of the large step movement ΔD of 1 µm, to 0 pF at the horizontal position. The swallow structure moved
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 7

back once the inclination angle increased above 0°, and and 0.4436 nN for the displacement resolution. Fig. 16(c)
capacitance change achieved its largest value of 4.226 pF at illustrates the power spectral density (PSD) of the noise signal.
90°. Fig. 15(b) illustrates the linear relation between The tested background noise in our laboratory was 0.0316
capacitance change and effective gravity force. Inclination fF/√ at 2.0 Hz, showing that noise floor resolution was better
sensitivity was calibrated as 4.235 pF/g with a high linearity of than 0.0044 nm/√ for deformation and 0.0597 nN/√ for
0.9995. Inclination cycle was also repeated four more times and force measurement. Allan deviation of recorded signals was
presented in Table V. The measured capacitance change also calculated to determine the intrinsic noise in a
exhibited a small increment in the succeeding measurement measurement system as a function of the averaging time, as
attempts. In particular, capacitance change reached −4.272 pF shown in Fig. 16(d). Bias instability appears on the flat region
at −90° and 4.251 pF at 90° in the last measurement. This result at the minimum of the Allan deviation curve and represents the
was primarily due to the accumulated positioning errors and bias of the sensor output changes over a specified period of
installation slippage resulting from the movement of the time. The minimum of Allan deviation of the proposed sensor
connection wire during inclination cycles. However, inclination reached up to 0.0359 fF at 0.65 s for the current laboratory
sensitivity had a small standard deviation of 0.0099 pF/g, environment. The efficient deformation and force resolution
indicating a negligible effect on measurement accuracy. To were 0.0050 nm and 0.0678 nN, respectively, indicating high
eliminate such system errors, the inclination sensitivity of the stability of the fabricated sensor in the laboratory environment.
proposed sensor was calculated by its average value of 4.247
pF/g.

(a) (b)

(a) (b)
Fig. 14 Inclination experiment results of the proposed sensor. (a) Inclinator system with
the assembled sensor. (b) Inclination process of the assembled sensor from -90° to 90°.

(c) (d)
Fig. 16. Noise and stability Measurement of the proposed sensor. (a) Experimental setup of the
noise measurement. (b) Static capacitance noise signal for more than 90 minutes. (c) PSD of the
(a) (b) noise signal. (d) Allan deviation of the noise signal.
Fig. 15 Inclination experiment results of the proposed sensor. (a) Capacitance change during the
180° inclination process. (b) Relation between the capacitance change and effective gravity. IV. PERFORMANCE EVALUATION
TABLE V
EXPERIMENTAL RESULTS OF INCLINATION MEASUREMENTS The performance parameters of the proposed sensor are listed
Inclination Angle Inclination Average Standard in Table VI, and its performance was compared with those of
No Linearity Sensitivity Sensitivity Deviation existing capacitive sensors. On the basis of this bionic swallow
-90° -60° -30° 30° 60° 90°
(pF/g) (pF/g) (pF/g)
1 -4.243 -3.748 -2.224 2.287 3.768 4.226 0.9995 4.235
design, the proposed swallow sensor not only maintained a
2 -4.241 -3.745 -2.218 2.303 3.785 4.249 0.9995 4.245 compact chip size but also possessed an overwhelming sensing
3 -4.256 -3.759 -2.227 2.283 3.763 4.228 0.9995 4.242 4.247 0.0099 performance. Displacement sensitivity was calibrated as 7.15
4 -4.268 -3.775 -2.237 2.293 3.775 4.236 0.9995 4.252
5 -4.272 -3.775 -2.235 2.299 3.784 4.251 0.9995 4.261
fF/nm, which was 29.5 times for the reported area-variant type
capacitive sensor with a value of 242.27 aF/nm [26] and 2.8
E. Noise and Stability Measurement
times for the gap-variant type capacitive sensor with a narrow
The packaged sensor was horizontally mounted on an optical initial gap [28] with a value of 2.58 fF/nm. Benefitting from the
platform (ZDT-P-MOT, Liansheng Tech) under atmospheric proposed three-mask fabrication process, the fabricated
pressure at room temperature (25 °C) in our laboratory, as swallow sensor had low-stiffness supporting beams with fewer
shown in Fig. 16(a). The platform had negligible magnetic fabrication difficulties. Its force sensitivity reached 0.53 fF/nN,
interferences, and its bottom air valve and top cover plates can which was 19.4 times for an area-variant type capacitive sensor
isolate most ground vibrations and airflows. The static with a value of 27.29 aF/nN [26], 9.0 times for the area-variant
capacitance noise signal was sampled with a portable computer type capacitive sensor in [28] with a value of 58.85 aF/nN, and
at 9.1 Hz for more than 90 minutes, as shown in Fig. 16(b). The 1.9 times for the sensor that utilized a nulling mechanism [29].
1σ-resolution of the sensor capacitance signal was 0.2347 fF, Sensor resolution is generally measured using different
which was equal to 0.0328 nm for the displacement resolution methods, such as 1-σ resolution, noise floor resolution, and
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 8

TABLE VI
PERFORMANCE COMPARISON OF THE PROPOSED SENSOR WITH EXISTING MEMS CAPACITIVE SENSORS
Ref. Displacement Force Deformation Force Linearity Measurement Range
Sensitivity Sensitivity Resolution* Resolution*
Non-linear buckling [25] 2.55 aF/nm Na 0.0540 nm/√Hzb Na Na Na
Asymmetric buckling [26] 242.27 aF/nm 27.29 aF/nN 0.0088 nm/√Hzb 0.0759 nN/√Hzb 0.9997 109.38 nm; 0.97 µN
0.2450 nmc 2.1732 nNc
Gap variant [27] 1.73 fF/nm 1.35 aF/nN 0.5313nma 680.0000 nNa 0.9600 781.25nm; 1000 µN
Low initial gap [28] 2.58 fF/nm 58.82 aF/nN Na Na 0.9800 300.00 nm; 13.16 µN
a a
Nulling mechanism [29] 0.20 fF/nm 0.28 fF/nN 1.0000 nm 2.0000 nN Na 555.56 nm; 0.40 µN
a
Double actuators [33] 2.34 V/µN Na Na 0.5140 nN 0.9900 0.23 µN
a a
This work 7.15 fF/nm 0.53 fF/nN 0.0328 nm 0.4436 nN 0.9998 1000.00 nm; 13.83 µN
0.0044 nm/√ b 0.0597 nN/√ b
0.0050 nmc 0.0678 nNc
*
“a” represents the 1-σ resolution; “b” represents the noise floor resolution; “c” represents the Allan deviation resolution
M.Hopyan, S.McNeill, H.Sun, “NANOROBOTS: Intracellular manipulation and
Allan deviation resolution. To the best of our knowledge, the measurement with multipole magnetic tweezers,” Sci. Robot., vol. 4, no. 28, 2019.
measured resolution of the proposed sensor is the highest [5] S. Guo, J. Guo, and W. H. Ko, “Monolithically integrated surface micromachined
touch mode capacitive pressure sensor,” Sensors Actuators, A Phys., vol. 80, no. 3,
among existing capacitive sensors in terms of the pp. 224–232, 2000.
aforementioned resolution metrics. In particular, the 1-σ [6] C. Dai et al., “Robotic Manipulation of Deformable Cells for Orientation Control,”
IEEE Trans. Robot., vol. 36, no. 1, pp. 271–283, Feb. 2020.
resolution of the proposed sensor was better than 0.0328 nm and [7] M. Xie, A. Shakoor, Y. Shen, J. K. Mills, and D. Sun, “Out-of-Plane Rotation Control
0.4539 nN, which were even higher than the sensors based on of Biological Cells with a Robot-Tweezers Manipulation System for Orientation-
Based Cell Surgery,” IEEE Trans. Biomed. Eng., vol. 66, no. 1, pp. 199–207, 2019.
the closed-loop position feedback measurement principle with [8] M. Xie, J. K. Mills, Y. Wang, M. Mahmoodi, and D. Sun, “Automated Translational
a value of 1.0000 nm [29] and 0.5140 nN [33]. To determine and Rotational Control of Biological Cells with a Robot-Aided Optical Tweezers
Manipulation System,” IEEE Trans. Autom. Sci. Eng., vol. 13, no. 2, pp. 543–551,
portability and versatility, the resolution performance of the 2016.
proposed sensor was measured using a portable ASICS chip in [9] P. Yu, W. Liu, C. Gu, X. Cheng, and X. Fu, “Flexible piezoelectric tactile sensor
array for dynamic three-axis force measurement,” Sensors (Switzerland), vol. 16, no.
a test room on the fifth floor. Its performance can be further 6, 2016.
improved when utilizing high-precision instruments on the [10] W. Gao, L. Zhao, Z. Jiang, and D. Sun, “Advanced Biological Imaging for
Intracellular Micromanipulation: Methods and Applications,” Appl. Sci., vol. 10, no.
ground. A high linearity of 0.9998 was achieved for the 20, p. 7308, Oct. 2020.
[11] F. Falleroni, V. Torre, and D. Cojoc, “Cell mechanotransduction with piconewton
proposed sensor under a full measurement range of 1000.00 nm forces applied by optical tweezers,” Front. Cell. Neurosci., vol. 12, no. May, pp. 1–
and 13.83 µN. The proposed sensor was also capable of 11, 2018.
[12] Z. Li, S. Gao, U. Brand, K. Hiller, and H. Wolff, “A MEMS nanoindenter with an
inclination measurement with a high sensitivity of 4.25 pF/g. integrated AFM cantilever gripper for nanomechanical characterization of compliant
Therefore, the proposed bionic swallow sensor exhibits huge materials,” Nanotechnology, vol. 31, no. 30, 2020.
[13] B. Koo and P. M. Ferreira, “An active MEMS probe for fine position and force
potential for ultralow multiphysics measurement, particularly measurements,” Precis. Eng., vol. 38, no. 4, pp. 738–748, 2014.
in biomedical applications. [14] D. Grech et al., “A Quasi-Concertina force-displacement MEMS probe for measuring
biomechanical properties,” Sensors Actuators, A Phys., vol. 275, pp. 67–74, 2018.
V. CONCLUSION [15] M. Maroufi, H. Alemansour, M. Bulut Coskun, and S. O. Reza Moheimani, “An
adjustable-stiffness MEMS force sensor: Design, characterization, and control,”
In this study, a bio-inspired MEMS capacitive force sensor Mechatronics, vol. 56, no. May, pp. 198–210, 2018.
was designed, fabricated, and comprehensively characterized. [16] Y. Xie, Y. Zhou, Y. Lin, L. Wang, and W. Xi, “Development of a microforce sensor
and its array platform for robotic cell microinjection force measurement,” Sensors
The proposed sensor had a bionic swallow structure integrated (Switzerland), vol. 16, no. 4, pp. 1–13, 2016.
with multiple feather shape comb arrays. The mechanical cross- [17] Y. Wei and Q. Xu, “Design of a PVDF-MFC Force Sensor for Robot-Assisted Single
Cell Microinjection,” IEEE Sens. J., vol. 17, no. 13, pp. 3975–3982, Jul. 2017.
coupling effect was largely eliminated. Comb optimization was [18] J. Rajagopalan, A. Tofangchi, and M. T. A. Saif, “Linear high-resolution BioMEMS
force sensors with large measurement range,” J. Microelectromechanical Syst., vol.
performed for high sensitivity and compact chip size. By 19, no. 6, pp. 1380–1389, 2010.
utilizing a novel interconnection configuration, interferences [19] X. Liu, K. Kim, Y. Zhang, and Y. Sun, “Nanonewton force sensing and control in
microrobotic cell manipulation,” Int. J. Rob. Res., vol. 28, no. 8, pp. 1065–1076,
derived from parasitic capacitance and electrostatic forces 2009.
exerted negligible effects on sensor output. The proposed bionic [20] W. Gao, C. Jia, Z. Jiang, X. Zhou, L. Zhao, and D. Sun, “The Design and Analysis of
a Novel Micro Force Sensor Based on Depletion Type Movable Gate Field Effect
force sensor was fabricated using a simple three-mask process Transistor,” J. Microelectromechanical Syst., vol. 28, no. 2, pp. 298–310, Apr. 2019.
and integrated with ASIC readouts. Measuring sensitivity was [21] W. Gao et al., “A novel MEMS force sensor based on Laterally Movable Gate Array
Field Effect Transistor(LMGAFET),” in 2017 IEEE 12th International Conference
7.151 fF/nm, 0.529 aF/nN, and 4.247 pF/g for displacement, on Nano/Micro Engineered and Molecular Systems, NEMS 2017, 2017, pp. 723–727.
force, and inclination measurements, respectively. The [22] Y. Wei and Q. Xu, “A Survey of Force-Assisted Robotic Cell Microinjection
Technologies,” IEEE Trans. Autom. Sci. Eng., vol. 16, no. 2, pp. 931–945, 2019.
proposed sensor had a wide measurement range of 1000.00 nm [23] Y. R. Zhang et al., “Realization of Multistage Detection Sensitivity and Dynamic
and 13.83 µN, with a high linearity of 0.9998. The 1-σ Range in Capacitive Tactile Sensors,” IEEE Sens. J., vol. 20, no. 17, pp. 9724–9732,
2020.
resolution was 0.0328 nm and 0. 4436 nN, and the noise floor [24] Y. Sun and B. J. Nelson, “MEMS capacitive force sensors for cellular and flight
biomechanics,” Biomed. Mater., vol. 2, no. 1, pp. 15–22, 2007.
resolution was 0.0044 nm/ √ and 0.0597 nN/ √ for [25] B. El Mansouri et al., “High-resolution MEMS inertial sensor combining large-
displacement and force measurement, respectively. The bias displacement buckling behaviour with integrated capacitive readout,” Microsystems
Nanoeng., vol. 5, no. 1, 2019.
stability of Allan deviance reached 0.0050 nm and 0.0678 nN [26] H. Zhang, X. Wei, Y. Ding, Z. Jiang, and J. Ren, “A low noise capacitive MEMS
at an integration time of 0.65 s. accelerometer with anti-spring structure,” Sensors Actuators, A Phys., vol. 296, pp.
79–86, 2019.
REFERENCES [27] Y. Sun, D. P. Potasek, S. N. Fry, and B. J. Nelson, “Characterizing fruit fly flight
behavior using a micro force sensor with a new comb drive configuration,” Proc.
[1] M. Dostanić et al., “A Miniaturized EHT Platform for Accurate Measurements of IEEE Int. Conf. Micro Electro Mech. Syst., vol. 14, no. 1, pp. 837–840, 2004.
Tissue Contractile Properties,” J. Microelectromechanical Syst., vol. 29, no. 5, pp. [28] R. Li, Z. Mohammed, M. Rasras, I. (Abe) M. Elfadel, and D. Choi, “Design,
881–887, 2020. modelling and characterization of comb drive MEMS gap-changeable differential
[2] R. Krenger, J. T. Burri, T. Lehnert, B. J. Nelson, and M. A. M. Gijs, “Force capacitive accelerometer,” Meas. J. Int. Meas. Confed., vol. 169, no. August 2020,
microscopy of the Caenorhabditis elegans embryonic eggshell,” Microsystems pp. 1–6, 2021.
Nanoeng., vol. 6, no. 1, 2020. [29] T. Tsuchiya, Y. Ura, K. Sugano, and O. Tabata, “Electrostatic tensile testing device
[3] N. F. Läubli et al., “3D mechanical characterization of single cells and small with nanonewton and nanometer resolution and its application to C 60 nanowire
organisms using acoustic manipulation and force microscopy,” Nat. Commun., vol. testing,” J. Microelectromechanical Syst., vol. 21, no. 3, pp. 523–529, 2012.
12, no. 1, pp. 1–11, 2021. [30] C. Li, G. Cheng, H. Wang, and Y. Zhu, “Microelectromechanical Systems for
[4] Y. S. Wang, X.Ho, C.Tsatskis, Y.Law, J.Zhang, Z.Zhu, M.Dai, C.Wang, F.Tan, Nanomechanical Testing: Displacement- and Force-Controlled Tensile Testing with
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 9

Feedback Control,” Exp. Mech., vol. 60, no. 7, pp. 1005–1015, 2020.
[31] C. Li, D. Zhang, G. Cheng, and Y. Zhu, “Microelectromechanical Systems for
Nanomechanical Testing: Electrostatic Actuation and Capacitive Sensing for High-
Strain-Rate Testing,” Exp. Mech., vol. 60, no. 3, pp. 329–343, 2020.
[32] A. Nastro, M. Ferrari, and V. Ferrari, “MEMS Inclinometer with Tunable-Sensitivity
and Segmented Overlapping Allan Variance Analysis,” 12th AEIT Int. Annu. Conf.
AEIT 2020, pp. 1–6, 2020.
[33] A. Nastro, M. Ferrari, and V. Ferrari, “Double-actuator position-feedback mechanism
for adjustable sensitivity in electrostatic-capacitive MEMS force sensors,” Sensors
Actuators, A Phys., vol. 312, p. 112127, 2020.
[34] W. Gao, A. Shakoor, L. Zhao, Z. Jiang, and D. Sun, “3-D Image Reconstruction of
Biological Organelles With a Robot-Aided Microscopy System for Intracellular
Surgery,” IEEE Robot. Autom. Lett., vol. 4, no. 2, pp. 231–238, Apr. 2019.
[35] W. Gao et al., “Precise Automated Intracellular Delivery Using a Robotic Cell
Microscope System With Three-Dimensional Image Reconstruction Information,”
IEEE/ASME Trans. Mechatronics, vol. 25, no. 6, pp. 2870–2881, Dec. 2020.
Wendi Gao is currently a Assistant Professor in Xi’an
Jiaotong University. He received the joint Ph.D. degrees
in Mechanical Engineering from Xi’an Jiaotong
University (with honors) and in Biomedical Engineering
from City University of Hong Kong in 2020.
His main research interests include MEMS devices,
biomedical sensing and image processing,
micromanipulation and micro-robotics based cell surgery
technology.

Cunlang Liu received his bachelor degree in Mechanical


Engineering from Xi’an Jiaotong University in 2021. He
is currently pursuing the master degree in mechanical
engineering at Xi’an Jiaotong University, Xi’an, China.
His major interest contain MEMS inertial sensors and
computation.

Libo Zhao received M.S. degree in Instrument Science


and Technology in 2003 and Ph.D. degree in Instrument
Science and Technology in 2007, both from Xi’an
Jiaotong University, Xi’an, China. He was a visiting
scholar from Sep. 2014 to Sept. 2015 at the Department
of Mechanical Engineering, University of California at
Berkeley, USA. He is currently working as a professor in
Xi’an Jiaotong University.
His research interests include micro and
nanofabrication technology, MEMS sensor technology, and precision
machining technology.
Qijing Lin received Ph.D. degree in Mechanical
Engineering from Xi’an Jiaotong University Xi’an, China
in 2015. He is currently a Research Assistant with Xi’an
Jiaotong University.
His current research interests include
micromanufacturing/ nanomanufacturing and intelligent
sensing technology.

Zhuangde Jiang received the Ph.D. degree from the


School of Mechanical Engineering, University of
Birmingham, Birmingham, U.K., in 2011. He is currently
a Professor with Xi’an Jiaotong University, Xi’an, China,
and an Academician with the Chinese Academy of
Engineering, China. He is currently the Director of the
Institute of Precision Engineering (IPE), Xi’an Jiaotong
University. He involves in research areas including
MEMS, ultra-precision machining technology and
equipment, and photoelectric testing technology and
instrumentation. Dr. Jiang is the Vice Director of the Chinese Society of Micro-
Nano Technology (CSMNT) and the Executive Director of the Chinese
Mechanical Engineering Society (CMES).
Dong Sun is currently a Chair Professor and Head of the
Department of Biomedical Engineering, and Director of
the Center for Robotics and Automation at City
University of Hong Kong. He received the Bachelor’s
and Master’s degrees from Tsinghua University, Beijing,
China in 1990 and 1994 respectively, and the Ph.D.
degree from the Chinese University of Hong Kong,
Hong Kong in 1997. After performing his Postdoctoral
research at the University of Toronto, Canada, he joined
the City University of Hong Kong in 2000.
His research interests include robotics and the related biomedical engineering
particularly for cellular engineering applications. He is a fellow of Canadian
Academy of Engineering, IEEE and HKIE.

You might also like