Access to this full-text is provided by Wiley.
Content available from Advanced Science
This content is subject to copyright. Terms and conditions apply.
RESEARCH ARTICLE
www.advancedscience.com
Ultrasensitive Linear Capacitive Pressure Sensor with
Wrinkled Microstructures for Tactile Perception
Chunyu Lv, Chengcheng Tian, Jiashun Jiang, Yu Dang, Yang Liu, Xuexin Duan,
Quanning Li, Xuejiao Chen, and Mengying Xie*
Ultrasensitive flexible pressure sensors with excellent linearity are essential
for achieving tactile perception. Although microstructured dielectrics have
endowed capacitive sensors with ultrahigh sensitivity, the compromise of
sensitivity with increasing pressure is an issue yet to be resolved. Herein, a
spontaneously wrinkled MWCNT/PDMS dielectric layer is proposed to realize
the excellent sensitivity and linearity of capacitive sensors for tactile
perception. The synergistic effect of a high dielectric constant and wrinkled
microstructures enables the sensor to exhibit linearity up to 21 kPa with a
sensitivity of 1.448 kPa−1and a detection limit of 0.2 Pa. Owing to these
merits, the sensor monitors subtle physiological signals such as various
arterial pulses and respiration. This sensor is further integrated into a fully
multimaterial 3D-printed soft pneumatic finger to realize material hardness
perception. Eight materials with different hardness values are successfully
discriminated, and the capacitance of the sensor varies linearly (R2>0.975)
with increasing hardness. Moreover, the sensitivity to the material hardness
can be tuned by controlling the inflation pressure of the soft finger. As a proof
of concept, the finger is used to discriminate pork fats with different hardness,
paving the way for hardness discrimination in clinical palpation.
1. Introduction
Tactile perception is important in detecting physical interactions,
and has shown great promise in wearable electronics, intelli-
gent robotics and prosthetics.[1–3 ] Ultrasensitive flexible capaci-
tive pressure sensors are essential components of tactile sensing.
Enormous effort has been devoted to improving the performance
C. Lv, C. Tian, J. Jiang, Y. Liu, X. Duan, Q. Li, X. Chen, M. Xie
State Key Laboratory of Precision Measuring Technology and Instrument
School of Precision Instrument and Opto-electronics Engineering
Tianjin 300072, P. R. China
E-mail: mengying_xie@tju.edu.cn
Y. Dang
College of Artificial Intelligence
Nankai University
Tianjin 300350, P. R. China
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/advs.202206807
© 2023 The Authors. Advanced Science published by Wiley-VCH GmbH.
This is an open access article under the terms of the Creative Commons
Attribution License, which permits use, distribution and reproduction in
any medium, provided the original work is properly cited.
DOI: 10.1002/advs.202206807
of pressure sensors.[1] Nevertheless, most
reported sensors suffer from nonlin-
ear sensitivity with increasing pressure,
which leads to compromised resolution
and redundant post-data processing and
conversion.[4] It is highly desirable to de-
velop a pressure sensor with both high
sensitivity and excellent linearity over a
broad pressure range for practical applica-
tions.
Ultrahigh sensitivity and linearity, two of
the most critical prerequisites for accurate
tactile sensing, are still outstanding chal-
lenges for pressure sensors. Currently, em-
bedding microstructured dielectric materi-
als with a low Young’s modulus to obtain
high compressibility has proven to be one
of the most effective methods for improv-
ing sensitivity.[5–8] These microstructures
are often engineered using complex and ex-
pensive technologies such as photolithogra-
phy and etching. Additionally, some studies
have demonstrated that an increase in the
dielectric constant can effectively improve
the sensitivity of the sensors.[9,10 ] These
dielectrics with high permittivity are commonly fabricated by
mixing high-dielectric-constant particles or conductive fillers into
the polymer matrix.[11,12 ] However, the above two methods play a
role mainly in a narrow low-pressure range. When the pressure
increases, the sensitivity drops dramatically, which leads to poor
linearity and limits the practical applications. Therefore, recent
studies have focused not only on ultrahigh sensitivity, but also on
improving the linearity of pressure sensors. Zhou et al. studied
the structural design and proposed a dielectric layer with hierar-
chical microdomes to achieve a stepwise contact between the di-
electric layer and electrode, which led to outstanding linearity (R2
=0.99) and a sensitivity of 0.065 kPa−1in an ultra-broad pressure
range.[13] In addition, a method to improve linearity by tuning the
filler contour and concentration has also been demonstrated. For
example, when common spherical nickel particles were replaced
by spiky nickel particles as filler materials, the hybrid composite
exhibited extraordinary linearity (R2=0.999) and the pressure
range was up to 1.7 MPa.[14] However, the sensitivity was only
0.0046 kPa−1. Therefore, although the above approaches enable
sensors with either ultrahigh sensitivity or excellent linearity, si-
multaneously realizing high sensitivity and linearity for pressure
sensors remains a great challenge.
Adv. Sci. 2023,10, 2206807 © 2023 The Authors. Advanced Science published by Wiley-VCH GmbH
2206807 (1 of 10)
www.advancedsciencenews.com www.advancedscience.com
Ultrasensitive and linear pressure sensors have great poten-
tial for tactile perception such as health monitoring and hard-
ness discrimination. Pressure sensors capable of reliable real-
time monitoring of arterial pulses[15] have attracted attention
because of their importance in cardiovascular disease diagno-
sis and noninvasive usage. To monitor such subtle physiologi-
cal signals, sufficiently high sensitivity and resolution are pre-
requisites for the sensors. Additionally, the tactile feedback of
contact pressure and the information of the object materials[16]
have been promoting safe and dexterous manipulations. Hard-
ness is a material mechanics-related physical property, and its sig-
nificance in guaranteeing the safety of objects has been demon-
strated, especially in agricultural picking and minimally invasive
surgery.[17,18] In addition, embedding the tactile sensors capa-
ble of hardness discrimination into a silicon fingertip can fur-
ther improve the safety of manipulations.[19] Moreover, soft pneu-
matic robots play an important role in decreasing the damage
of substances from traditional rigid grippers, thus improving
the security of manipulations.[20] Therefore, inspired by novel
soft robotics, combining pressure sensors with a soft pneumatic
chamber to realize hardness tactile perception is a practical and
promising method.
In this work, we present an ultrasensitive flexible capac-
itive pressure sensor with a broad linearity range based
on a spontaneously wrinkled multiwall carbon nanotube/
polydimethylsiloxane (MWCNT/PDMS) composite dielectric
layer. Compared with complex fabrication technologies, the di-
electric composite film was fabricated by a simple spin-coating
process and wrinkled microstructures were formed owing to the
aggregation of MWCNT. The combination of these wrinkled mi-
crostructures and percolating MWCNT fillers contributes to ul-
trahigh sensitivity and linearity. This was achieved by tuning
the MWCNT concentration. The proposed sensor with 2.6 wt%
MWCNTs can achieve a high sensitivity of 1.448 kPa−1and excel-
lent linearity (R2=0.9982) in the pressure range of 0.005–21 kPa.
The sensor also showed a fast response time, good durability, and
could even detect pressures as low as 0.2 Pa. Furthermore, the
concept of pressure sensors has numerous merits. The feasibil-
ity of tactile perception applications in a different real-life context
is demonstrated, especially for continuous health monitoring by
recording various artery pulsations and respiration. In addition, a
flexible pressure sensor was attached to a 3D printed pneumatic
soft finger to discriminate eight materials with different hardness
values. The results show a linear relationship between the capac-
itance change and material hardness (covering the range of 5.5
HA to 54.5 HA), and the sensitivity to hardness can be enhanced
by increasing the inflation pressure. Moreover, we successfully
discriminated between the hardness of soft fat and stiffened fats
of pork, which paves the way for application in clinical palpation.
2. Results and Discussion
2.1. Sensor Fabrication and Characterization
Figure 1a shows a schematic of the flexible pressure sen-
sor, where the MWCNT/PDMS dielectric layer was sand-
wiched between two indium tin oxide/polyethylene terephthalate
(ITO/PET) flexible electrodes. The key component of this sen-
sor is a microstructured dielectric layer, which is fabricated by
simply spin-coating a mixture of MWCNT and PDMS. The fab-
rication process of the flexible pressure sensor is shown in Fig-
ure 1b. First, the MWCNT powder was uniformly dispersed into
a hexane solution via a 20 min ultrasonic treatment. The PDMS
precursor was then poured and magnetically stirred to ensure a
homogeneous distribution. The as-prepared mixture was spin-
coated onto a clean glass substrate at 250 rpm for 20 s and then
placed in a vacuum desiccator to remove residual hexane. Subse-
quently, the MWCNT/PDMS composite film was cured at 90 °C
for 1 h. Finally, the prepared MWCNT/PDMS dielectric layer was
cut into 1 ×1cm
2pieces and sandwiched between two identi-
cal ITO/PET electrodes to form a capacitive sensor, as shown in
Figure 1c. In this work, to systematically investigate the impact
of MWCNT fillers on the dielectric constant and the sensor per-
formance, four different MWCNT weight ratios, that is, 1 wt%, 2
wt%, 2.6 wt% and 3 wt% in the PDMS matrix were studied. Table
S1, Supporting Information, summarizes the dielectric constants
of the composite dielectrics with various MWCNT weight ratios
at different frequencies (10, 100, and 300 kHz). The characteri-
zation method is described in the Characterization and Measure-
ment section. The dielectric constant increases with the increas-
ing amount of MWCNT and reaches a maximum value of 13.61
at 2.6 wt%. When the MWCNT weight ratio increased further
than 2.6 wt%, the dielectric constant slightly decreased. This can
be attributed to the fact that the collapse of the microcapacitors
formed in the PDMS matrix and the adjacent WMCNTs start to
form ohmic conductive paths, which leads to the transition of the
sensing material from the dielectric to conductivity.[21,22]
Additionally, the impact of MWCNT weight ratios on the mor-
phology of dielectric layers was explored. The surface and cross-
sectional morphologies of the 1 wt% and 2 wt% MWCNT/PDMS
are shown in Figures S1 and S2, Supporting Information, respec-
tively. The surface of the composite film with 1 wt% MWCNT
was completely flat, while the film with 2 wt% MWCNT started
to have small bumps with an average height of ≈10 μm. Scan-
ning electron microscope (SEM) of the 2.6 wt% MWCNT/PDMS
are shown in Figures 1d and 1e, respectively. It is obvious that
the 2.6 wt% MWCNT/PDMS composite is rougher and possesses
spontaneous wrinkles on the surface, which can also be observed
in the optical images in Figure S2a,b, Supporting Information.
According to Figure 1d and 1e, the thickness of the 2.6 wt% com-
posite film is ≈160 μm, and the average height and width of the
surface wrinkles are ≈45 and ≈280 μm, respectively. This also
indicates that the size of the wrinkles increased with increasing
MWCNT concentration. The irregular and randomly distributed
wrinkles are attributed to the agglomeration of CNTs owing to the
strong van der Waals forces, high aspect ratio, and low bending
stiffness of the nanotubes.[23] Furthermore, to illustrate the im-
portance of these wrinkles on the sensor performance, we fabri-
cated a flat film with a 2.6 wt% MWCNT/PDMS film via the blade
coating technique, as shown in Figure S1a, Supporting Informa-
tion.
The electromechanical performance of flexible capacitive
sensors with various MWCNTs weight ratios was studied using
the experimental setup shown in Figure 2a. Figure 2b shows the
relative capacitance change (C−C0/C0) in response to an applied
pressure Pfor pressure sensors with different MWCNT ratios,
where Cand C0are the capacitances in the loading and initiation
states, respectively. The sensitivity of the capacitive pressure
Adv. Sci. 2023,10, 2206807 © 2023 The Authors. Advanced Science published by Wiley-VCH GmbH
2206807 (2 of 10)
www.advancedsciencenews.com www.advancedscience.com
Figure 1. Proposed capacitive sensor. a) Schematic of the capacitive sensor and photograph of the sensor. b) Schematic illustration of the fabrication
process. i) Mixing the MWCNT into PDMS matrix, ii) spin-coating the MWCNT/PDMS mixture, iii) curing the MWCNT/PDMS membrane in an oven,
and iv) releasing the membrane from the substrate. c) Photograph of our capacitive sensor. d) SEM photograph of the surface of the MWCNT/PDMS
dielectric layer. e) Cross-section SEM image of the dielectric layer.
sensor is defined as S=𝛿(C−C0/C0)/𝛿P. According to the
sensitivity summary of the counterpart with 0, 1.0 wt%, 2.0 wt%
and flat-2.6 wt% MWCNTs (Table S2, Supporting Information),
it is clear that the sensitivity increases with the increasing ratio
of MWCNT owing to the increase in the dielectric constant and
gradual formation of numerous wrinkles on the surface. The
finite element analysis (Figure S3, Supporting Information)
shows the intense stress concentration on the wrinkles, which
plays a significant role in sensitivity improvement. However, the
sensitivity of these sensors gradually decreased as the pressure
increased, leading to poor linearity in the sensing range. This
is unfavorable for practical applications because it requires an
additional complicated readout circuit for data processing and
conversion. Therefore, in this study, with the same MWCNT ra-
tio of 2.6 wt%, the capacitive sensor based on a microstructured
dielectric layer with randomly distributed wrinkles exhibited an
ultrahigh sensitivity of 1.448 kPa−1and excellent linearity (R2=
99.82%) over a broad pressure range of 0.005–21 kPa, as shown
in Figure 2c. This sensitivity is 3.5 times higher than that of flat
2.6 wt% film and 145 times higher than that of pure PDMS in
the low- to medium-pressure range (2–21 kPa).
Based on the spontaneously wrinkled dielectric layer, excel-
lent sensitivity, and linearity of the pressure sensor are achieved
owing to the synergistic effect of spontaneous microstructures
and the composite material. At the structural level, randomly dis-
tributed microstructures cause stress concentrations. At the ma-
terial level, MWCNTs act as a high-dielectric-constant dopant,
and the composite follows the percolation threshold theory. Gen-
erally, a capacitive pressure sensor is defined as the change in
capacitance of a typical parallel plate capacitor with respect to the
pressure change, as shown in Equation (1), where 𝜖0,𝜖r,Aand
dare the permittivity of vacuum, relative dielectric constant of
the dielectric material, overlapping area of the two electrodes and
distance between the top and bottom electrodes, respectively. In
our sensors, the overlapping area of electrode A is fixed and can
be considered constant. Therefore, the capacitance is determined
by 𝜖rand d. To clearly explain the change in 𝜖rand dunder pres-
sure, the dielectric layer was divided into a wrinkled layer and a
membrane layer, as shown in Figure 1d. To comprehensively il-
lustrate the sensing mechanism of this linear pressure sensor, a
simplified working flow chart and equivalent circuits are shown
in Figure 2d.
C=𝜀0𝜀r/Ad(1)
The wrinkled layer consists of two parallel-connected capac-
itors, C1and C2.C1is a capacitor with a vacuum air dielectric
layer and C2is a capacitor with a wrinkled MWCNT/PDMS di-
electric layer. The membrane layer was a capacitor with a flat
MWCNT/PDMS film as the dielectric layer, represented by C3,
which was connected to C1and C2. In the initial state, no pres-
sure is applied to the sensor, and three capacitors are formed and
remain at the initial values, as shown in Figure 2d-i. Once the
sensor is subjected to external pressure, its capacitance increases
Adv. Sci. 2023,10, 2206807 © 2023 The Authors. Advanced Science published by Wiley-VCH GmbH
2206807 (3 of 10)
www.advancedsciencenews.com www.advancedscience.com
Figure 2. Sensing performances of the sensor. a) The experimental setup. b) Sensitivities of capacitive sensor with different MWCNT/PDMS ratios
and surface morphology. c) Sensitivity of the sensor with 2.6 wt% MWCNT/PDMS dielectric layer. d) Sensing mechanism of the capacitive sensor with
natural wrinkles. Equivalent circuit of the sensor i) without pressure, ii) when the sensor was subjected to low pressure, and iii) when the sensor was
subjected to high pressure. e) Response and relaxation time of the capacitive sensor with 2.6 wt% MWCNT/PDMS when applied pressure was 0.9 kPa.
f) Real-time capacitive change rate of the sensor subjected to step pressure. g) Responses to ultralight objects (with the weight of 23 and 2 mg) and
the limit of the pressure sensor, around 0.2 Pa. h) Hysteresis performance of the flexible pressure sensor. i) Stability of the sensor under the pressure
of 17 kPa for 5000 cycles. j) Comparative study of the pressure-sensing performance with respect to sensitivity and detection range between our sensor
and recently reported ones.
Adv. Sci. 2023,10, 2206807 © 2023 The Authors. Advanced Science published by Wiley-VCH GmbH
2206807 (4 of 10)
www.advancedsciencenews.com www.advancedscience.com
instantly. In Figure 2d-ii, when a small pressure is applied to the
sensor, the irregular microstructures of the dielectric layer are
compressed gradually, which induces a decrease ind. Because the
microstructured layer consists of air and wrinkles, the Young’s
modulus is lower than that of the membrane layer. Therefore,
a rapid increase Δdwas observed in the microstructured layer,
whereas Δdof the membrane layer was negligible. According
to the general Lichtenecker mixing rule,[24] the relative dielec-
tric constant 𝜖rof the dielectric layer is defined in Equation (2),
where 𝜖Air,𝜖Com posit e, VAir and VCompos ite represent the permittivity
of vacuum air, dielectric constant of the sensing material, volume
percentage of air and MWCNT/PDMS composite, respectively.
Here, 𝜖Air =1and𝜖Compos ite for the 2.6 wt% MWCNT/PDMS com-
posite is 13.61, as shown in Table S1, Supporting Information.
When the sensor is compressed, air is gradually substituted for
the MWCNT/PDMS, leading to an increase in 𝜖r. Therefore, at
stage ii, ΔC1and ΔC2induced by the deformation of the wrinkled
dielectric layer determine the sensor response to pressure. C3can
be regarded as a constant and has a minor effect on the equivalent
circuit. When the pressure increased further, as shown in Fig-
ure 2d-iii, the irregular wrinkles collapsed and released the air.
The microstructured layer becomes a flat membrane layer, and
𝜖requals that of 𝜖Compos ite . Therefore, the capacitance of capacitor
ΔC3determines the response of the entire sensor. The capaci-
tance change completely depends on the deformation of the di-
electric layer and the change in the dielectric constant,[14] where a
higher dielectric constant compensates for Δdowing to the high
Young’s modulus of the flat MWCNT/PDMS composite, result-
ing in a large capacitance change. Therefore, at stages ii and iii,
the combination of wrinkled microstructures and a high dielec-
tric constant enables the sensor with 2.6 wt% MWCNT/PDMS
dielectric layer with high sensitivity and linearity. However, for
other MWCNT concentrations, the mismatch between the mi-
crostructured morphology and dielectric constant compels them
to exhibit a segmented response to pressure.
𝜀r=𝜀Air VAir +𝜀Composi te VCompos ite (2)
In addition to sensitivity and linearity, response time is another
critical factor for various tactile applications. Figure 2e shows a
fast response time of 123 ms and recovery time of 86 ms under
a pressure of 0.9 kPa, which is slightly higher than the response
time of human skin perception.[25] Figure 2f shows the stepwise
response of the sensor when a step pressure of 0–20 kPa was
applied to the sensor. The inset shows the sensor response to
a small pressure ranging from 0.005 to 1 kPa. Furthermore, it
was found that the sensor response was not dependent on the
temperature, as shown in Figure S4, Supporting Information. To
comprehensively demonstrate the real-time response of the sen-
sor, it was attached to the index finger of a glove to distinguish
the pressure when the human hand grasped and released a plas-
tic cup, as shown in Figure S5, Supporting Information, where
the capacitance increased when the amount of water increased.
To further investigate the capability of perceiving subtle pressure
beyond the normal detection range, our sensor was used to sense
airflow using an air blower placed 10 mm above the sensor, sim-
ply indicating the ability to detect human respiration, as shown
in Figure S6, Supporting Information. Concurrently, Figure 2g
shows the sensor response to a domestic rice of 23 mg and an
ultralight esculent millet of 2 mg (representing the pressure of
≈2.3 and ≈0.2 Pa, respectively), which indicates that compared
with a 0.5 Pa detect limit of the commercial force gauge, the de-
tection limit of our sensor can be as low as 0.2 Pa. These excellent
performances demonstrate that our sensor possesses both high
resolution and precision. In addition, the hysteresis test was car-
ried out by loading and unloading pressure on the sensor, and
a maximum hysteresis of 6.78% was achieved, as shown in Fig-
ure 2h. Moreover, the proposed sensor exhibited excellent robust-
ness and durability, as shown in Figure 2i. No obvious degrada-
tion was observed under a pressure of 17 kPa for 5000 cycles, in-
dicating the excellent long-term stability of this pressure sensor.
In Figure 2j, in terms of sensitivity and linearity, the compari-
son between our work and other recently reported sensors based
on microstructured dielectric layers or high-permittivity fillers
was studied, and the details are listed in Table S3, Supporting
Information.[26–30 ] Most of the reported sensors consist of two or
more linear ranges within the dynamic range. For instance, the
porous Ecoflex-MWCNT dielectric presented the highest sensi-
tivity of 6.42 kPa−1but a small sensing range of 2 kPa.[9] Another
copper calcium titanate-wrapped hybrid sponge dielectric exhib-
ited a wide sensing range of up to 125 kPa, but a low sensitivity of
0.026 kPa−1.[11] Notably, our sensor exhibits a better trade-off be-
tween sensitivity and linearity over a wide sensing range, which is
realized using only a simple and low-cost fabrication technique.
Additionally, based on the characterization of the performance,
our pressure sensors can be further integrated into a sensor ar-
ray with a PDMS layer encapsulated on the top for pressure dis-
tribution perception. Each pixel size was 5 mm ×5 mm, which
can be further miniaturized in the future. Figure 3aandFigure
S7, Supporting Information, show the sequential data from nine
pixels collected using a microprocessor (Arduino Uno) and mul-
tichannel collector (MUX). When the 3D printed “T”, “J” and “U”-
shaped letter blocks were positioned on the top of the sensor ar-
ray (Figure 3b), the mappings of the capacitance change are pre-
sented in Figure 3c–e and each inset exhibits the vertical view of
the corresponding histogram. It is obvious that the histograms
agree with the profiles of the letter blocks, which further demon-
strates the outstanding response to the pressure and provides a
potential for various practical applications of tactile perception,
such as physiological signal monitoring and the discrimination
of material hardness.
2.2. Physiological Signals Monitoring
Cardiovascular disease is the leading cause of death and dis-
ability worldwide. The tactile perception of physiological signals
is critical for cardiovascular disease prevention because it indi-
cates the physiological functions of the circulatory and respira-
tory systems.[31] As shown in Figure 4, our pressure sensor can
acquire various subtle artery pulsation signals from the radial
artery of the wrist, brachial artery, carotid artery, and frontal tem-
poral artery. The corresponding locations of these arteries (Fig-
ure 4a) can be found in several published studies.[15,32 ] To a c -
quire the pulsation information, the flexible pressure sensor was
attached to human skin with a preload with the aid of a com-
mercial medical adhesive bandage.[33] Figure 4b shows the real-
time output signal when the pressure sensor was placed over
Adv. Sci. 2023,10, 2206807 © 2023 The Authors. Advanced Science published by Wiley-VCH GmbH
2206807 (5 of 10)
www.advancedsciencenews.com www.advancedscience.com
Figure 3. 3×3 Sensor array for pressure distribution perception. a) Experimental setup. b) Schematic diagram of the sensor array with T, J, U-shaped
letter blocks placed on the top for pressure distribution perception. c–e) Histogram of the pressure distribution for three letters.
the radial artery of a 24-year-old healthy volunteer, and the fast
Fourier transformation of this signal is shown in Figure 4d. Ac-
cording to the peak of 1.157 Hz, the pulse frequency was evalu-
ated to be ≈70 beats/min, which falls within the normal range of
a healthy person. Figure 4c presents the period of the waveform
(the fourth radial arterial signal in Figure 4b) with a regression
fitting that clearly distinguishes three typical peaks, including the
percussion wave (P-wave), tidal wave (T-wave) and diastolic peak
(D-wave), which are consistent with the clinical monitoring.[34]
Similarly, Figures 4e and 4f show the results of the sensor tactile
perception when our pressure sensor was placed on the volun-
teer’s brachial artery and neck over the carotid arterial vessels.
A waveform and pulse frequency similar to those of the radial
artery signals was monitored. It should be noted that the low-
frequency fluctuation superimposed on the carotid arterial signal
was due to the respiration of the volunteer. Simultaneously, the
sensor was attached to the temporal artery, as shown in Figure 4g,
and the real-time output exhibited periodic temporal artery pulsa-
tion. Several high-amplitude spikes are attributed to eye blinks,
which induce a strain on the skin that is greater than the pres-
sure of arterial pulsation. In addition to various pulsations, hu-
man respiration is also important for evaluating human health
conditions.[35] Similar to the detection of the breeze of air men-
tioned in the sensor characterization, the respiration of a human
was also monitored. The sensor was placed 10 mm away from the
human nose and the real-time respiration signal was recorded, as
shown in Figure 4h. With the above arterial pressure and respira-
tion perception, our flexible sensor shows excellent potential for
physiological signal monitoring, from detecting normal arterial
pulsations to even subtle pulsations.
2.3. Hardness Tactile Perception Based on Soft Pneumatic Finger
Hardness, as a physical property reflecting material mechanics,
has a critical impact on safe and dexterous manipulation, espe-
cially in the era of intelligent devices with humanoid perceptions.
Therefore, hardness discrimination has become significant tac-
tile feedback in manipulation and has attracted considerable in-
terest. In this study, we developed a soft pneumatic finger using
multimaterial 3D printing technologies and integrated our pres-
sure sensor to realize the tactile perception of hardness. Figure
5a shows a schematic of the 3D printed finger and our pressure
sensor (5 ×5mm
2) was attached to the soft pneumatic cham-
Adv. Sci. 2023,10, 2206807 © 2023 The Authors. Advanced Science published by Wiley-VCH GmbH
2206807 (6 of 10)
www.advancedsciencenews.com www.advancedscience.com
Figure 4. Flexible pressure sensor for continuous physiological signals monitoring. a) Schematic illustration of various possible monitoring artery
positions. b) Real-time radial arterial pulsation signal when the sensor is bonded on the wrist surface. c) A period of radial arterial signal extracted from
the fourth waveform of the figure (b). d) Fast Fourier transformation for radial arterial pulsation signal to evaluate the frequency of pulsating. e) Real-time
brachial arterial pulsation when a young volunteer wears the sensor on the inside of the elbow. f) Real-time carotid arterial pulsation signal and the shift
are probably due to the respiration leading to the skin stretching. g) Real-time temporal arterial pulsation signal, the higher peaks are attributed to the
regular blinking. h) Real-time respiration signal from the pressure sensor when the sensor is positioned ≈10 mm below the nostril. The inset of each
subfigure shows the position of the sensor that is attached on the skin.
ber. The detailed dimensions of the fingers can be found in Fig-
ure S8, Supporting Information. The finger had three inflation
chambers, each connected to an airway. Two materials with dif-
ferent hardness were used to fabricate the finger: the inflation
chamber was composed of soft material, and the remaining parts
were made of reinforced hard material to restrict their expansion.
Figure 5b presents the working cycle of a soft pneumatic finger
with a regulated inflation pressure. First, the finger is in an initial
state without inflation pressure. The object mounted on a linear
motor gradually approached the finger, and the custom-made ex-
perimental platform is shown in Figure S9, Supporting Informa-
tion. Once the sensor attached to the finger contacts the object,
inflation pressure is applied to the pneumatic chamber, thereby
expanding and pushing the sensor toward the object. The pres-
sure on the sensor results in a change in the capacitance. Here, a
soft conductive thread (resistance <1Ω) was selected to transmit
the output of the pressure sensor because it hardly affected the
dynamic movement of the pneumatic chamber. Finally, when the
inflation pressure was released, the finger and sensor recovered
to their initial state and accomplished a hardness discrimination
cycle.
For tactile hardness perception, we fabricated seven PDMS
blocks with different hardness values by controlling the mixing
ratio of the PDMS base and curing agent, including 1:5, 1:10,
1:12.5, 1:15, 1:17.5, 1:20 and 1:22.5. Additionally, we used Ecoflex
00–30 to fabricate a sample with lower hardness, as shown in Fig-
ure 5c. The hardness of the eight materials was measured using
a commercial hardness meter, and the results are shown in Fig-
Adv. Sci. 2023,10, 2206807 © 2023 The Authors. Advanced Science published by Wiley-VCH GmbH
2206807 (7 of 10)
www.advancedsciencenews.com www.advancedscience.com
Figure 5. Soft robotic finger embedded with tactile sensor for hardness discrimination. a) Schematic of the soft finger with flexible pressure sensor
attached to the inflation chamber. b) The schematic of a working cycle of the soft finger for the hardness discrimination. c) The photograph of 8 polymer
blocks with different hardness. d) Measured hardness of polymer blocks by using a commercial hardness meter. e) Photographs of the chamber expansion
under the different inflation pressures. f) Sensor response versus material hardness under different inflation pressures. g) Photographs of a block of
pork fat without and with beans inside. h) Real-time signals of the pressure sensors when the soft finger discriminates the raw fat with different hardness.
ure 5d. It is clear that with an increase in mixing ratios, the PDMS
hardness linearly raises from 19.5 HA to 54.5 HA, as shown
in Figure S10, Supporting Information. Compared to PDMS,
Ecoflex 00–30 exhibits the smallest hardness of 5.5 HA. To ex-
plore the reliability of hardness discrimination based on pneu-
matic fingers, we investigated the sensor output five times by ap-
plying four different inflation pressures to drive the pneumatic
chamber: i) 10, ii) 16, iii) 26, and iv) 28 kPa. Images of cham-
ber expansion under different inflation pressures are shown in
Figure 5e. The chamber gradually expanded when the finger was
driven by an increase in inflation pressure. Before performing
the hardness discrimination, as shown in Figure S11, Support-
ing Information, no interface was observed by running a working
cycle without contacting objects and eliminating the influence of
Adv. Sci. 2023,10, 2206807 © 2023 The Authors. Advanced Science published by Wiley-VCH GmbH
2206807 (8 of 10)
www.advancedsciencenews.com www.advancedscience.com
chamber expansion on the sensor response. Figure 5f presents
the capacitance of the sensor as a function of hardness under four
different inflation pressures, and a video was recorded (Movie S1,
Supporting Information). For a specific inflation pressure, when
the material hardness increased, the capacitance of the sensor
increased linearly. This finding indicates that the combination of
the pressure sensor and soft pneumatic finger can effectively dis-
tinguish material hardness. Meanwhile, the linear relationship
(R2>0.975) between the sensor capacitance and material hard-
ness remained despite the change in inflation pressure. More im-
portantly, the gradient of the fitting curves improved from 0.026
to 0.096 pF/HA with an increase in the inflation pressure from
10 to 28 kPa, which can be explained in Figure S12, Supporting
Information. As shown in Table S4, Supporting Information, the
hardness of the chamber increases when the inflation pressure
increases. Thus, a harder chamber leads to an increase in the rel-
ative pressure between the sensor and objects, which induces a
larger capacitance change and thereby achieves a higher sensi-
tivity. However, considering the contradiction between the soft
finger durability and hardness discrimination sensitivity, an in-
flation pressure of 26 kPa was selected as the optimal working
condition, and repeatability was tested with 20 cyclic measure-
ments, as shown in Figure S13, Supporting Information.
To further demonstrate the versatility of hardness perception
based on the soft pneumatic finger and pressure sensor, we at-
tempted to provide a preliminary evaluation of the difference be-
tween normal tissue and abnormal rectal tumors following digi-
tal rectal examination (DRE). Currently, DRE is one of the most
direct diagnostic approaches to rectal cancer. It can be used to
evaluate rectal canal health in terms of tumor hardness and rec-
tal conditions (i.e., narrowed, swollen, eroded). Therefore, based
on this simple and effective diagnosis, we employed pork fat with
different hardness values to imitate the rectum in healthy and
harder tumors. The hardness of the fat was tuned by embed-
ding the beads, as shown in Figure 5g. Figure 5h shows the sen-
sor response to soft fat (section I) and stiffened fat (section III)
when the finger was driven by an inflation pressure of 26 kPa.
Evidently, stiffened fat exhibits a larger capacitance change than
soft fat, which preliminarily proves the feasibility of discriminat-
ing fats with different hardness values based on our finger and
pressure sensor. Furthermore, the sensor response was studied
when the object approached and then pressed the finger without
inflation pressure in two ways: i) step-by-step with 0.5 mm incre-
ments; and ii) continuous motion at a speed of 1.5 mm s−1.The
two methods employed the same compression displacement of
2 mm. The recorded capacitance changes in two different com-
pressed ways are shown in sections II and IV of Figure 5h. It can
be observed that the capacitance change for an object with a spe-
cific hardness is consistent despite the different pressing meth-
ods, which indicates that the discrimination is only determined
by the compression displacement rather than the pressing pro-
cess. These results provide a foundation for assessing rectal nar-
rowing. For stepwise compression, the corresponding stepped ca-
pacitance change shows the ability to evaluate compression dis-
placement. Thus, by combining the evaluation results under in-
flation pressures of 0 and 26 kPa, our pneumatic soft finger with
a pressure sensor paves the way for clinical palpation and demon-
strates the potential to screen for abnormal conditions of the rec-
tum.
3. Conclusion
In this study, we propose a flexible capacitive pressure sensor
with a spontaneous wrinkled MWCNT/PDMS dielectric layer
fabricated via a simple spin-coating process. The sensor exhib-
ited an ultrahigh sensitivity of 1.448 kPa−1and excellent linearity
(R2=0.9982), which was attributed to the synergistic effect of the
wrinkled microstructures and high dielectric constant. Further-
more, owing to the stress concentration resulting from the spon-
taneously wrinkled microstructures, this sensor offers a num-
ber of high-performance characteristics, including a low limit of
detection, fast response and release, high durability, and robust-
ness. These exceptional performances endowed this sensor with
potential tactile perception for physiological health monitoring
and pressure distribution. Additionally, to improve the safety and
dexterity of various manipulations, such as robotic grasping and
minimally invasive surgery, we developed a material hardness
perception method based on a soft pneumatic finger and the pro-
posed pressure sensor. Driven by a pneumatic finger, our sensor
exhibited a linear response to material hardness, and the sensi-
tivity was tunable via the inflation pressure of the finger. Further-
more, inspired by DRE for rectal cancer, the finger was used to
successfully discriminate pork fat with different hardness. In the
future, integrating a sensor with a compliant encapsulating struc-
ture into a pneumatic finger will be taken into consideration. This
method can provide a foundation for replacing the human finger
to realize more stable and accurate rectal health screening in the
clinical setting.
4. Experimental Section
Fabrication of Pneumatic Finger:The pneumatic finger was printed us-
ing a commercial multimaterial 3D printer (Objet500 Connex3, Strata-
sys). The capacitive pressure sensor was connected to a conductive silver
thread using silver paste, and the sensor was attached to a plane wall con-
nected to a soft pneumatic chamber using silicon paste. The geometric
details of the soft fingers are shown in Figure S8, Supporting Informa-
tion. The finger has three inflation chambers, and each connects to an
airway to inflate these chambers. The finger body printed with hard rubber
(Shore A 60) as a reinforced structure increased the pressure durability of
the finger body and prevented the finger body from expanding under in-
flation. The soft pneumatic chamber was printed with soft rubber (Shore
A 30), leading to faster expansion during inflation. Therefore, the cooper-
ation of materials with different stiffnesses achieved partial expansion of
the soft finger, which efficiently enhanced the pressure when the sensor
contacted the blocks. It is noted that the silicon paste was used to seal the
connection between the finger air passageways and flexible gas tubes. In
addition, conductive silver fibers were bonded to the grooves of the finger
body using silicon paste. Finally, a glossy finish was selected to ensure the
smoothness of the prepared finger.
Characterizations and Measurements:Optical and SEM images were
captured using an Olympus optical microscope and a scanning electron
microscope (ZEISS SIGMA). The force load was applied with a linear mo-
tor (PZG650-R05AG-E, Japan) that offers a gentle pressure gradient as a
consequence of a minimum displacement change of 2 μmandaforce
gauge with a resolution of 0.0005 N (Mark-10, USA.) The pressure was cal-
culated using the relationship between the force and electrode area. The
capacitance of the sensor output was measured using a high-precision
(around 10−5pF) LCR meter (Keysight E4980AL) with a supply voltage
of 1 V and a typical measurement frequency of 100 kHz. For the capac-
itance value acquisition of the sensor array, the MUX (CD74HC4067) by
Arduino control was used for selecting the measurement channel, with
MATLAB software assistant and the array data were acquired into a PC,
Adv. Sci. 2023,10, 2206807 © 2023 The Authors. Advanced Science published by Wiley-VCH GmbH
2206807 (9 of 10)
www.advancedsciencenews.com www.advancedscience.com
facilitating the result process. For the hardness tactile perception experi-
ment, the inflation pressure of the pneumatic chamber of the soft finger
was controlled using a proportional valve (VEAB-L26-D7-Q4-V1-1R1, Ger-
many) with STM32. The feedback inflation pressure was monitored using
a pressure meter (TM510; Tecman, China). A shore hardness meter (Type-
A) was used to measure the hardness of PDMS and Ecoflex.
The Dielectric Constant Test of Dielectric Layer:The intrinsic dielectric
constants of PDMS and MWCNT/PDMS composites with different doping
concentrations were measured using an LCR meter. The MWCNT/PDMS
samples were fabricated into flat membranes to prevent interference from
surface microstructures. The effective dielectric constant of the hybrid
composite was calculated using Equation (3), where Cis the initial ca-
pacitance of a capacitor, dis the thickness of the composite, which is fixed
at 1.15 mm. Ais the area of the dielectric layer (9.44 ×9.48 =89.5 mm2),
which is identical to the electrode area.
𝜀r=Cd∕A𝜀0(3)
Supporting Information
Supporting Information is available from the Wiley Online Library or from
the author.
Acknowledgements
This work was supported by the National Natural Science Foundation of
China (grant number 62001325 and 52075384). The authors appreciate
the support from China Association for Science and Technology and China
Instrument and Control Society. The authors would like to thank Wenlan
Guo, Chen Sun, Bohua Liu, and Chongling Sun for microscope and per-
formance testing.
Conflict of Interest
The authors declare no conflict of interest.
Data Availability Statement
The data that support the findings of this study are available from the cor-
responding author upon reasonable request.
Keywords
hardness discrimination, physiological signal monitoring, pressure sen-
sors, soft pneumatic finger, spontaneous microstructures, tactile percep-
tion
Received: November 20, 2022
Revised: February 8, 2023
Published online: March 15, 2023
[1] S.R.A.Ruth,V.R.Feig,H.Tran,Z.Bao,Adv. Funct. Mater. 2020,30,
2003491.
[2] Y. Zang, F. Zhang, C.-a. Di, D. Zhu, Mater. Horiz. 2015,2, 140.
[3] K. Keum, J. Eom, J. H. Lee, J. S. Heo, S. K. Park, Y.-H. Kim, Nano Energy
2021,79, 105479.
[4] X. Cui, F. Huang, X. Zhang, P. Song, H. Zheng, V. Chevali, H. Wang,
Z. Xu, iScience 2022,25, 104148.
[5] S. Li, K. Dong, R. Li, X. Huang, T. Chen, X. Xiao, Sens. Actuators, A
2020,312, 112106.
[6] J.C.Yang,J.-O.Kim,J.Oh,S.Y.Kwon,J.Y.Sim,D.W.Kim,H.B.Choi,
S. Park, ACS Appl. Mater. Interfaces 2019,11, 19472.
[7] Y. Luo, J. Shao, S. Chen, X. Chen, H. Tian, X. Li, L. Wang, D. Wang, B.
Lu, ACS Appl. Mater. Interfaces 2019,11, 17796.
[8] J. Hwang, Y. Kim, H. Yang, J. H. Oh, Composites, Part B 2021,211,
108607.
[9] J. Choi, D. Kwon, K. Kim, J. Park, D. D. Orbe, J. Gu, J. Ahn, I. Cho, Y.
Jeong, Y. Oh, I. Park, ACS Appl. Mater. Interfaces 2020,12, 1698.
[10] D. M. Grannan, J. C. Garland, D. B. Tanner, Phys. Rev. Lett. 1981,46,
375.
[11] A. Chhetry, S. Sharma, H. Yoon, S. Ko, J. Y. Park, Adv. Funct. Mater.
2020,30, 1910020.
[12] L. Zhang, Z.-Y. Cheng, J. Adv. Dielectr. 2011,01, 389.
[13] B. Ji, Q. Zhou, M. Lei, S. Ding, Q. Song, Y. Gao, S. Li, Y. Xu, Y. Zhou,
B. Zhou, Small 2021,17, 2103312.
[14] J. Wu, Y. Yao, Y. Zhang, T. Shao, H. Wu, S. Liu, Z. Li, L. Wu, Nanoscale
2020,12, 21198.
[15] K.-H. Ha, W. Zhang, H. Jang, S. Kang, L. Wang, P. Tan, H. Hwang, N.
Lu, Adv. Mater. 2021,33, 2103320.
[16] W.Yang,M.Xie,X.Zhang,X.Sun,C.Zhou,Y.Chang,H.Zhang,X.
Duan, ACS Appl. Mater. Interfaces 2021,13, 55756.
[17] Z. Zhang, J. Zhou, Z. Yan, K. Wang, J. Mao, Z. Jiang, Comput. Electron.
Agric. 2021,181, 105959.
[18] L. Zhang, F. Ju, Y. Cao, Y. Wang, B. Chen, Sens. Actuators, A 2017,266,
197.
[19] S. Takamuku, G. Gómez, K. Hosoda, R. Pfeifer, 2007 IEEE 6th Int.
Conf. Dev. Learn. 2007,1.
[20] M. Xie, M. Zhu, Z. Yang, S. Okada, S. Kawamura, Nano Energy 2021,
79, 105438.
[21] P. Wei, X. Guo, X. Qiu, D. Yu, Nanotechnology 2019,30, 455501.
[22] G. Liu, Y. Chen, M. Gong, X. Liu, Z.-K. Cui, Q. Pei, J. Gu, C. Huang,
Q. Zhuang, J. Mater. Chem. C 2018,6, 10829.
[23] T. C. M, J. Pitchaimani, Appl. Math. Modell. 2022,103, 68.
[24] H. Niu, S. Gao, W. Yue, Y. Li, W. Zhou, H. Liu, Small 2020,16, 1904774.
[25] J. Yang, S. Luo, X. Zhou, J. Li, J. Fu, W. Yang, D. Wei, ACS Appl. Mater.
Interfaces 2019,11, 14997.
[26] Q. Su, Q. Zou, Y. Li, Y.Chen, S.-Y. Teng, J. T. Kelleher, R. Nith, P. Cheng,
N. Li, W. Liu, S. Dai, Y. Liu, A. Mazursky, J. Xu, L. Jin, P. Lopes, S. Wang,
Sci. Adv. 7, eabi4563.
[27] C. Mu, J. Li, Y. Song, W. Huang, A. Ran, K. Deng, J. Huang, W. Xie, R.
Sun, H. Zhang, ACS Appl. Nano Mater. 2018,1, 274.
[28] Y. Wan, Z. Qiu, Y. Hong, Y. Wang, J. Zhang, Q. Liu, Z. Wu, C. F. Guo,
Adv. Electron. Mater. 2018,4, 1700586.
[29] S. Kang, J. Lee, S. Lee, S. Kim, J.-K. Kim, H. Algadi, S. Al-Sayari, D.-E.
Kim, D. Kim, T. Lee, Adv. Electron. Mater. 2016,2, 1600356.
[30] Y. Kim, H. Yang, J. H. Oh, Mater. Des. 2021,197, 109203.
[31] Z.Shen,F.Liu,S.Huang,H.Wang,C.Yang,T.Hang,J.Tao,W.Xia,
X. Xie, Biosens. Bioelectron. 2022,211, 114298.
[32] H. Ouyang, J. Tian, G. Sun, Y. Zou, Z. Liu, H. Li, L. Zhao, B. Shi, Yu.
Fan,Yi.Fan,Z.L.Wang,Z.Li,Adv. Mater. 2017,29, 1703456.
[33] P. Rwei, C. Qian, A. Abiri, Y. Zhou, E.-F. Chou, W. C. Tang, M. Khine,
Adv. Mater. Interfaces 2022,9, 2200294.
[34] W. W. Nichols, Am. J. Hypertens. 2005,18,3S.
[35] X. Wang, Z. Liu, T. Zhang, Small 2017,13, 1602790.
Adv. Sci. 2023,10, 2206807 © 2023 The Authors. Advanced Science published by Wiley-VCH GmbH
2206807 (10 of 10)
Available via license: CC BY
Content may be subject to copyright.
