www.acsnano.org
Filtered Mechanosensing Using Snapping
Composites with Embedded MechanoElectrical Transduction
Hortense Le Ferrand,†,‡,§ André R. Studart,*,† and Andres F. Arrieta*,‡
†
Complex Materials, Department of Materials, ETH Zürich, Vladimir-Prelog-Weg 5, 8093 Zürich, Switzerland
School of Mechanical Engineering, Purdue University, 585 Purdue Mall, West Lafayette, Indiana 47907, United States
Downloaded via PURDUE UNIV on May 5, 2019 at 02:28:30 (UTC).
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
‡
S Supporting Information
*
ABSTRACT: Mechanosensing is ubiquitous in natural systems. From the skin
ridges of our finger tips to the microscopic ion channels in cells, mechanosensors
allow organisms to probe their environment and gather information needed for
processing, decision making, and actuation. Despite technological advances in
synthetic mechanosensing, it remains challenging to achieve this functionality at
the scale of large stiff structures where both the amount of data to sense locally and
the diversity of input stresses that the sensors have to withstand require highly
tunable systems. Filtered sensing using mechanical displacement is an effective strategy developed by organisms to cope
with large sets of stimuli. Inspired by this biological strategy, we fabricate bistable elements that can passively filter
mechanical inputs, translate them into electrical signals, and be reset to their original sensing state using an external
magnetic field. These multiple functionalities are achieved using hierarchically structured composites that can be arranged
in large-area arrays. The filtering capability and fast passive response of our mechanosensors are experimentally
demonstrated using simple electrical circuits and magnets. Thanks to their scalability and applicability to a wide range of
material systems, these low-power sensors are avenues for the fabrication of load-bearing structures that are able to sense,
compute, communicate, and autonomously adapt in response to external magneto-mechanical stimuli.
KEYWORDS: mechanosensing, bioinspired composite, hierarchical microstructure, electrical conductivity, anisotropy
S
To address this challenge, it is crucial to develop passive and
selective mechanosensing strategies to filter the meaningful
part of the mechanical perturbation of interest using fast and
scalable sensing technologies.12 Real-time filtering of background noise is possible by using, for example, dynamic
wireless systems whose resonance frequency depends on the
variation of strain.13,14 The signals obtained are then filtered by
wireless readers through intensive digital data processing.15,16
Despite the effective filtering effect, such systems are power
intensive and thus difficult to implement in large area and in
mobile applications. Analog circuits have been utilized to
decrease power consumption and increase the throughput of
analog to digital interfaces. As an example, distributed selfpowered body sensor networks have been developed to
monitor the health status of a patient with minimum power
and to even harvest energy from daily movements.17,18
In contrast to synthetic approaches, biological systems have
evolved strategies to achieve low-power mechanosensing.
Instead of electrical filtering, living organisms in nature often
employ distributed mechanical filters intimately integrated
ensing of external mechanical stimuli is a key feature
used by living organisms and synthetic devices to probe,
respond, and adapt to their environment.1,2 The demand
for advanced sensing systems in structural applications has
increased significantly in recent years with the development of,
for example, soft robots,3−6 morphing wings for airplanes,7−9
and reconfigurable buildings.10 During their lifetime, these
structures sustain mechanical loads that vary in time and
location at their interacting surface. Monitoring of these local
mechanical inputs is important to determine the status of the
structure, detect flaws or damage, provide information about
the source of the mechanical stimuli, and permit these
structures to respond in dynamic environments. Such
interactivity and dynamic functionalities will become increasingly important for future digital technologies involving realtime monitoring of physical interactions of adaptive structures
with the surrounding environment and with living systems.
The enticing perspectives for the forthcoming digital era are
however accompanied by major technical challenges. Given the
large dimensions, the amount of data generated by state-of-theart sensors positioned at the surface of such systems is
enormous, posing a large scale big data challenge along with a
high energy cost.11
© 2019 American Chemical Society
Received: February 8, 2019
Accepted: March 29, 2019
Published: March 29, 2019
4752
DOI: 10.1021/acsnano.9b01095
ACS Nano 2019, 13, 4752−4760
Article
Cite This: ACS Nano 2019, 13, 4752−4760
Article
ACS Nano
Figure 1. Schematics describing the sensing and actuating principles of a bird wing (A) and the bioinspired concept of bistable
mechanosensing developed in this study (B). (A) Under airflow, the deflection of covert feathers of birds above a threshold of ∼30° triggers
a neuronal electrical discharge, which is further processed by the neuronal network to actuate the shape of the wing. The signal is reset after
the input air flow ceases. (B) In the envisioned bioinspired analogue, a bistable mechanosensor could be used to monitor the airflow across
the wing of an aircraft. When the air flow is sufficient to snap the bistable sensor, a change in electrical conductivity is detected. This
electrical signal can be further processed through a central unit to provide the relevant actuation to the wing of an airplane. Resetting to the
initial condition is possible by applying a magnetic field.
to which they can be exposed to. Typically, the maximum flow
that soft hydrogel hair-like sensors can withstand is in the
range of 100 mm/s,29 which is about 3 orders of magnitude
lower than the airflow around an aircraft. To enable passive
and selective sensing for airflow monitoring and haptic
interactions over large areas, such microstructured mechanosensors should be easily scalable and exhibit sufficient strength
to sustain the high mechanical forces it may be subjected to
during operation.
In this paper, we propose a general method for fabricating
stiff composites with mechanosensing capabilities that are
programmed within the hierarchical architecture of the
material. To achieve fast and selective mechanosensing, the
hierarchical material is designed to showcase three main stressdependent functionalities. First, the composite exhibits a
bilayer design that gives rise to mechanical bistability. By
setting the minimum load required to snap the composite to
the “on-state”, this bistable behavior is used to program the
sensitivity threshold to an external mechanical trigger. Second,
one of the layers contains metallic particles that form an
electrically conductive path only when the composite snaps
into the “on-state”. Analogous to the synaptic connections
between neurons, the reduction of the distance between
metallic particles within the composite enables the transmission of electrical signals to the main processing unit for
further computation and decision making. Third, ferromagnetic particles are also added to the composite to enable
resetting of the bilayer to the “off-state” using an external
magnetic field. The analogies of this composite architecture
with the mechanosensorial system in the wing of birds are
schematically shown in Figure 1. In the following, we describe
the manufacturing and hierarchical structure of the proposed
composite and discuss the multifunctional properties and
macroscopic behavior of the material. A proof-of-concept
example of successive actuation and resetting operations is
then shown to illustrate the three main functionalities
programmed within the material’s architecture. Finally, we
characterize the dynamical response and demonstrate the
within their materials’ architecture. Programming the sensing
functionality at the material level reduces significantly the
amount of information that needs to be handled by a
centralized signal processing unit. The resulting neuromechanical system is able to simultaneously sense and actuate,
therefore minimizing the time and energy required to perform
dynamic biological functions.19−21 In a typical example of
biological sensing, geometrical features in the form of hairs or
cilia are positioned on the legs of spiders, on the wings of bats,
and on the skin of fish, to mechanically sense the flow
conditions of the surrounding environment.22−25 Another
illustrative example are the covert receptors associated with
contour feathers in the dorsal wings of birds.26 When the
airflow increases the elevation angle of the feathers above a
threshold of 30°, the mechanoreceptor neural cells located at
the base of the feathers are mechanically actuated and emit a
burst of electrical signal at a frequency that depends on the
elevation angle. After this electrical discharge, the neuron
regains its initial state, until another change in elevation angle
above the threshold is recorded. Acting as a high-pass filter,
this sensory mechanism presumably enables the bird to
measure the airspeed and to adapt its wing shape and direction
to maximize flight efficiency.26
Many attempts have been made to fabricate bioinspired
passive mechanosensors that rely on the filtering effect of hairlike structures to obtain flow data relevant for flight.27−33 Flow
sensitivity comparable to that of the mechanosensory biological systems (i.e., fish) combined with minimum detectable
velocities as low as 0.008 mm/s has been achieved.29 Using
photolithography, these micro- and nanoelectromechanical
systems (MEMS and NEMS) can be organized into arrays of a
few millimeters to a couple of centimeters to provide accurate
sensing and high signal-to-noise ratios.29 However, the small
and delicate sizes of these MEMS and NEMS pose significant
challenges to apply such mechanosensors on the highly
populated sensing networks envisioned, for example, in largearea airplane wings. The presence of hydrogels or soft
polymers in these sensors also restricts the mechanical loads
4753
DOI: 10.1021/acsnano.9b01095
ACS Nano 2019, 13, 4752−4760
Article
ACS Nano
Figure 2. (A) Hierarchical architecture of a bistable mechanosensor illustrated by electron and optical micrographs (top) and explanatory
schematics (bottom). (B) Relationships between the composition and the properties of a single layer of composite material: (i) Coefficient
of thermal expansion α, parallel (α∥) or perpendicular (α⊥) to the alignment direction, as a function of the volume fraction fAl2O3 of stiff
alumina platelets. The lines correspond to theoretical calculations (see eqs S1 and S2 in Supporting Information). (ii) Electrical conductivity
σ of nonaligned composites as a function of the volume fraction f Ni in nickel flakes. The blue line corresponds to a fit using a percolation
model (eq S3) with a percolation threshold ( f Ni,c) of 4. (iii) Magnetic remanence B as a function of f Ni fitted by a linear regression.
flakes within the individual composite layers. By tuning the
concentration of nickel flakes to be close to the percolation
threshold, the compressive strain induced within the composite
layer in one of the two stable states enables physical contact
between adjacent flakes, thus forming an electrically conductive network. Finally, the magnetic properties of the
composite are directly enabled by the presence of the
ferromagnetic nickel flakes within the layers. We found that
a nickel flake concentration of 4 vol % is enough to make
composites with a magnetization level of ∼0.03 emu·mg−1,
which is sufficient for magnetic actuation using a hand-held
external magnet (see Supporting Information Figure S1).
Composites with this hierarchical design were manufactured
using a magnetic alignment technology that provides control
over the orientation of platelets and flakes in a fluid, which is
later converted into a polymer matrix by simple curing.35−38
To demonstrate that our strategy can be applied for actual
applications where structural demands can be high, we use stiff
alumina platelets that are easily available in large quantities and
are already used in many structural applications. Such
microplatelets are on average 10 μm in diameter and 180
nm in thickness. In contrast to the long carbon fibers used in
conventional composite technology,39,40 these platelets allow
for controlled microstructuring of the composite at finer length
scales below 10 μm. Magnetic alignment of alumina platelets in
an epoxy matrix has already been shown to be suitable for the
fabrication of mechanically bistable shells.36 To make them
responsive to magnetic fields, the alumina microplatelets were
decorated with iron oxide nanoparticles of 12 nm diameter by
a simple van der Waals attraction in deionized water. Along
with the alumina platelets, nickel flakes of an average diameter
of 20 μm and thickness of 500 nm were also incorporated into
suitability of these sensors for large-area and fast sensing with
stress selectivity and low power consumption.
RESULTS AND DISCUSSION
Hierarchical Design and Fabrication Process. To
achieve the multiple coupled functionalities proposed (Figure
1), our mechanosensing composite is organized in a
hierarchical structure (Figure 2) comprising: micron-sized
platelets (level 1), aligned platelets in a polymer matrix (level
2), bilayer architecture (level 3), and macroscopic geometry
(level 4). Such a hierarchy from microelements to macroscopic
form is reminiscent of natural materials.34 Below, we describe
how this hierarchical architecture is designed and manufactured to achieve the three stress-dependent functionalities: (i)
mechanical bistability, (ii) filtered sensing via snappinginduced electrical percolation, (iii) and magnetic restoring of
the initial configuration. The structure−property relationships
resulting from this hierarchical architecture are then discussed
and quantified (Figure 2A).
The mechanical bistability of our composites arises from a
bilayer design in which two orthotropic plies are orthogonally
combined. In such orthogonal arrangement, the direction of
highest stiffness and coefficients of thermal expansion of each
individual layer lies perpendicular to one another. This
structure recalls the layered and anisotropic microstructures
found in morphing natural composites such as the bilayers in
Bauhinia seedpods or in pine cones.32 Orthotropy is achieved
here by aligning microplatelets in deliberate directions within
the polymer matrix of each composite layer. In addition to
mechanical bistability, our composite is designed to be
electrically conductive in one of the bistable configurations.
This is implemented by the incorporation of aligned nickel
4754
DOI: 10.1021/acsnano.9b01095
ACS Nano 2019, 13, 4752−4760
Article
ACS Nano
Figure 3. Sensing−restoring cycle of a bistable mechanosensor consisting of a 10 cm square composite shell: (A) Schematics of the sensor in
the off-state displaying a curvature κ0 and high electrical resistance parallel (R∥) and perpendicular (R⊥) to the direction of alignment of
flakes in the top layer. (B) Sensing of an external force F applied as depicted in the picture (top): curvature κ (black dots), force applied
(white dots), and resistances R∥ (bold) and R⊥ (thin) as a function of the displacement ε of the point at the center of the shell. The snapthrough transition at the displacement threshold εth is highlighted in red. (C) Sensor in the on-state featuring a curvature κ1 and a low
electrical resistance. (D) Restoring of the shell’s initial curvature by applying a magnetic field H as shown in the picture (bottom): curvature
κ (black dots), reaction force (white dots), and resistances R∥ (bold) and R⊥ (thin) as a function of the magnetic field strength H. The snapthrough transition at the magnetic field threshold Hth is highlighted in red.
and restoring functionalities within the composite bilayer
requires tuning of three specific properties of individual layers:
thermal expansion coefficient, electrical conductivity, and
magnetic susceptibility. These properties can be tuned by
modifying the chemical composition and structural features at
each length scale of the composite (Figure 2B).
For a given composite geometry and processing conditions,
the coefficients of thermal expansion (CTE) of the individual
layers determine the shape change that the orthogonal bilayer
undergoes during cooling and thus the mechanical bistability
of the system. The CTE of individual layers can be controlled
by the concentration and orientation of the reinforcing alumina
particles used in the composite. Earlier work has shown that
the CTE parallel to the platelet orientation (α∥) decreases with
the concentration of reinforcing particles (Figure 2B(i)). For
the perpendicular direction, the CTE value (α⊥) first increases
with the platelet content before it steadily drops for particle
concentrations higher than approximately 5 vol %. These
trends follow the behavior expected from a simple rule of
mixtures (see Supporting Information).36 Overall, the difference between the CTE values in the two orthogonal directions
(α⊥ − α∥) remains constant at about 4 × 10−5 K−1 for platelet
concentrations above 5 vol % (see Supporting Information).
Similarly, each layer presents stiffness anisotropy that increases
with the concentration in platelets. At 10 vol % of alumina
platelets, the Young’s modulus parallel to the direction of
alignment reaches 6 GPa, whereas it is only of 1.2 GPa in the
perpendicular direction.36
the composite to provide the magnetic and electrical properties
required in our hierarchical design. These two anisotropic
microparticles exhibit the ultrahigh magnetic response
necessary for biaxial alignment throughout the composite
using low rotating magnetic fields.37 Indeed, low magnetic field
strengths ranging from 10 to 100 mT were sufficient to align
the microparticles during composite fabrication.36 Composite
bilayers are obtained by sequentially casting, aligning, and
curing the mixture of particles in the epoxy matrix and turning
the direction of the alignment by 90° between the two layers.
We have demonstrated previously that the sequential casting
with a final curing of the bilayer at 100 °C leads to an intimate
bonding between the two layers.36 Therefore, the second layer
was directly casted and cured on top of the first and already
consolidated layer, enabling the partial infiltration of the liquid
uncured material at its surface and the formation of strong
chemical bonds between the layers. In the final material, the
interface between the epoxy matrix of the two layers could
barely be distinguished (see Figure S2). Bistability in the sheetlike bilayer composite arises during cooling by setting the
temperature of the matrix to room temperature. Since the
composite is constrained and unable to deform during the
setting processes, internal stresses are accumulated and
released only upon unmolding.36 The release of internal
stresses eventually leads to a macroscopic deformation of the
bilayer into one of the two possible stable states.
Properties of Individual Composite Layers. The
implementation of mechanical bistability, filtered sensing,
4755
DOI: 10.1021/acsnano.9b01095
ACS Nano 2019, 13, 4752−4760
Article
ACS Nano
Figure 4. (A) Electrical circuit with integrated bistable mechanosensor used to demonstrate the filtered sensing capabilities of the
hierarchical composite shell (left). (B) Displacement ε of the center of the shell (black) and its electrical resistance (blue) as a function of
time t, showing the filtering of noise signal below the displacement threshold εth (highlighted in red) and the restoring of the initial state
using an external magnetic field H. (C) Dynamics of the snapping event characterized by DIC and electrical resistance measurements. The
displacement ε (black, left) and the electrical resistance ΔR (blue, right) are shown as a function of time t during the snap-through transition
(gray area highlighted in (B)). (D) Pictures and corresponding schematics of a large-area sensing array comprising four bistable
mechanosensing elements integrated onto an airplane wing.
Finally, the addition of ferromagnetic nickel flakes also
equips the composite with the magnetic properties needed to
restore the sensor to its active state after snapping events
triggered by the environment. To quantify the magnetic
properties of the composites, we measured the remanent
magnetic field B of cured composite plies as a function of the
concentration of nickel flakes (f Ni). Our results show that the
remanent magnetization of the composites varies linearly with
the volume fraction f Ni (Figure 2B(iii)), as expected from
theory (see Supporting Information). Based on these
experimental and theoretical analyses, a volume fraction of 6
vol % of alumina platelets and 4 vol % of nickel flakes was
selected for the fabrication of bistable composites featuring
mechanosensing capabilities. Indeed, hierarchical composite
bilayers with this composition were found to develop a natural
curvature (κ0) of 4 ± 1 m−1 after fabrication, if square-shaped
specimens with individual layer thicknesses and widths of,
respectively, 200 μm and 10 cm are used (Figure 2A).
Filtered Sensing and Restoring. To demonstrate the
principle of our bistable mechanosensor in action, we
performed cycles of sensing and restoring using a square
composite shell manufactured with optimized composition and
geometry (Figure 3). The sensing and restoring functionalities
are illustrated by tracking the actual shell curvature, the force
exerted on (or by) the shell, and the electrical resistance of one
of the composite layers as a function of the displacement (ε)
imposed during sensing and the magnetic field (H) applied
during restoring.
In its initial convex state (Figure 3A), the top layer of the
sensor is under tensile strain, which keeps a large distance
between the nickel flakes and ensures a high electrical
resistance (R). Applying a mechanical displacement through
a force localized at the center of the shell, the initial curvature
To ensure that switching between the stable states results in
an electrical signal that can be used for passive sensing, the
composite layers are designed to significantly change their
electrical conductivity depending on the curvature of the
bilayer. This is possible by incorporating electrically conductive
nickel flakes at concentrations close to the threshold needed to
form a percolating network of contacting flakes. We explored
this idea by measuring the effect of the concentration of nickel
flakes on the electrical conductivity of the epoxy matrix (Figure
2B(ii)). As predicted by percolation theory, the electrical
conductivity σ in one ply changes with the concentration of
electrically conductive elements (f Ni) according to the power
law: σ = C × (f Ni − f Ni,c)t, where C is a constant, f Ni,c is the
volume fraction of flakes at the percolation threshold, and t is a
system-dependent exponent. Because of this power law
dependence, small variations in flake concentration around
the percolation threshold lead to a dramatic increase in the
global conductivity of the composite.41,42 In our composites,
the nickel flakes were found to percolate above 4 vol % with a
critical exponent t of 4 (Figure S3). This a very high exponent,
twice higher than a typically observed t of 2, probably resulting
from the anisotropy of the flakes.42 In addition to the flake
concentration, the global conductivity of the composite close
to the percolation threshold also depends on the state of local
internal stresses and strains within the material. Compressive
stresses decrease the distance between conducting particles
and thus promote the formation of an electrically conductive
percolating network. Conversely, tensile stresses have the
opposite effect. Since switching between two mechanical stable
states changes the internal stresses within the composite layer
from tensile to compressive and vice versa, the binary
mechanical states of our composite bilayers can be directly
translated into binary electrical states.
4756
DOI: 10.1021/acsnano.9b01095
ACS Nano 2019, 13, 4752−4760
Article
ACS Nano
κ0 decreases continuously (Figure 3B). When the displacement
of the central point of the shell reaches a threshold
displacement (εth) of 5.5 mm, the sensor snaps into its
concave geometry (state 1). For a given loading distribution,
this threshold displacement can be tuned by varying the
geometry and materials properties of the composite shell. We
note that the applied force (Fapp) reaches a maximum before
the snapping event. This effect is commonly observed in
bistable shells and is usually associated with the development
of frictional forces between the shell and the rod used to
impose the mechanical displacement.43 After the transition, the
top ply of the shell, initially under tension, becomes
compressed. This brings the nickel flakes into contact, thereby
decreasing the electrical resistance of the top composite layer.
Because the conductive flakes are biaxially aligned in a specific
orientation, the drop in electrical resistance depends on the
direction along which the shell is electrically probed. The
larger contact areas between flakes established along the
thickness of the particles (smaller dimension) result in a 3
orders of magnitude drop in resistance when the measurement
is conducted perpendicular to the area of the flakes. This
contrasts with the 10-fold reduction if the measurement is
performed in the other orthogonal direction. The lower
electrical resistance of this concave state indicates that the
filtering sensor was subjected to a displacement higher than its
threshold value εth (Figure 3C).
Once the sensor has transmitted the electrical signal to a
peripheral or main processing unit, the initial convex state of
the composite shell can be restored by applying a similar
deformation in the opposite direction. Instead of a mechanical
input, the threshold displacement is imposed in this case by
remotely applying an external magnetic field. In our experiment, the field was generated by simply approaching a strong
rare-earth magnet close to the shell, while one edge was
maintained clamped (Figure 3D). The attracting force
developed between the permanent magnet and the sensor
deforms the shell toward its original convex configuration
(state 0) if the applied magnetic field strength exceeds a
threshold value Hth of 280 mT. By snapping back to the initial
position 0, the sensor recovers the initial electrical resistance.
After restoring, the bistable mechanosensor generates a
reaction force Freact of 0.6 N, which is in the same order of
magnitude of the maximum force Fapp required to trigger the
transition from state 0 to state 1. The bistable shells can be
operated through several cycles, as long as the internal stresses
responsible for the snapping events are present within the
structure. Although stress relaxation has been observed in the
epoxy matrix utilized in our composites (see Supporting
Information), this can be inhibited by using liquid crystalline
epoxy resins as the matrix44 or through the addition of small
quantities of carbon nanotubes.45
Dynamic Response and Integration. The ability to snap
only when subjected to a threshold displacement allows the
proposed bistable composites to filter mechanical input data
from a noisy dynamic environment, thus saving the power
needed for sensing. We illustrate this filtering effect by
connecting the bistable composite to an external electrical
circuit that visually indicates the binary state of the shell using
a commercial LED (Figure 4). The mechanosensing composite
was integrated into the electrical circuit via a 1 mm-wide silver
electrode deposited on the surface of the shell and
perpendicular to the long axis of the platelets. The sensor
was connected in series with the LED using a simple electrical
setup (Figure 4A). In this configuration, the LED lights up if
the shell is in the concave state and turns off if it snaps back
into the convex geometry. Upon the application of arbitrary
mechanical noise below the critical displacement of the shell
(εth), no snap-through event was triggered, keeping the
electrical resistance low and the LED turned on (Figure 4B).
Once the snap-through was triggered above the threshold value
(εth = 2.5 mm in this case), the electrical resistance increased,
switching off the LED. For the shell used in this specific
example, the percolating network of nickel flakes is enough to
turn the LED off but an insulating network (R = infinite) is
only formed for a short period of time. This transient behavior
resembles the spike of the action potential developed between
neurons in biological systems and is a mechanism to further
reduce the energy required for sensing. Approaching the other
side of the shell with a high strength magnet triggers the snapthrough of the shell to its initial convex configuration, restoring
the sensing capability of the system.
In addition to energy consumption, the time scale required
for sensing is another important parameter for the performance
of the mechanosensor in dynamic environments and in
potential computing applications. To assess this time scale,
we tracked the motion of the snapping shell using a digital
image correlation (DIC) set up (Figure 4C). Measurements of
the shell displacement over time show the oscillatory motion
typically observed after a snapping event. The amplitude of the
oscillation decays over a time period of approximately 200 ms,
which is in agreement with previous experiments with
composite shells of comparable dimensions and composition.36
Because of the fast nature of the snapping event, the period of
the oscillatory motion is as small as 10 ms. This is in line with
the duration of the spike in electrical resistance measured on
the same shell during snapping. If compared with biological
systems, the operating time scale of our mechanosensor lies
between those observed for the action potential in cardiac (100
ms) and nerve cells (1 ms).46 The size dependence of the
oscillations after snap-through allows for reaching faster
responses with smaller specimens if needed.
Finally, we demonstrate the potential of these mechanosensors for large structure monitoring by assembling a fourelement sensing array on an aircraft skin (Figure 4D). In this
example, the sensors were connected by a thin layer of flexible
silicone to enable the overall structure to conform to the
curved surface of an airplane wing. Each element could be
activated independently of each other, allowing for the
generation of 16 distinct possible mechanical states. By
extending this concept to larger arrays and networks, it should
be possible to build structures that can mechanically sense the
environment and perform logic operations that are later
translated into specific actuation responses (further details in
Figure S5 and supplementary text). Since the bistable
mechanosensing elements can be restored independently,
such actuation may even emerge from macroscopic morphing
effects induced by the internal stresses generated upon
snapping.
CONCLUSION
In summary, our concept of bistable mechanosensing offers a
promising route toward low-power adaptive structures with
sensing and possibly computing capabilities directly programmed within the hierarchical design of the material. With
the help of machine learning algorithms, such intelligent
structures can potentially be trained to sense, communicate,
4757
DOI: 10.1021/acsnano.9b01095
ACS Nano 2019, 13, 4752−4760
Article
ACS Nano
The dynamic displacement of the shell was recorded using DIC with
the help of two cameras (Fastcam MiniUX) and a dedicated software
(Vic 3D). Images were acquired simultaneously with the recording of
the electrical resistance with an oscilloscope (Agilent L1251A, USA).
To track the dynamics of the snapping event using DIC, the
mechanosensor was covered with a white and opaque thin tape layer,
which was further covered by a layer of black paint. For the final
demonstrator, bistable shells were connected using a membrane of
PDMS (Sylgard 184) to form a 2 × 2 array. The shells were first
heated at 100 °C above the glass transition temperature of the epoxy
to drive them flat. Then, a 5 cm width aluminum square was placed at
the center of each shell to maintain them in position, while 1 mm of
liquid PDMS was deposited on top and cured for 30 min at 100 °C.
After curing, the four connected shells were turned upside down, and
the process repeated on the other side to create a symmetrical coating
of PDMS. After final curing, the aluminum plates were removed, and
electrical cables were connected to the shell by simple contact.
compute, and actuate in an autonomous fashion with
minimum energy consumption in a range of applications in
robotics, transportation, and medicine.47−53
EXPERIMENTAL SECTION
Fabrication of Microstructured Composite Plates. Alumina
platelets (Ronaflair, WhiteSapphire, diameter 8.9 ± 2.9 μm and
thickness 0.30 ± 0.13 μm) were magnetized according to a procedure
described elsewhere using superparamagnetic iron oxide nanoparticles
(EMG-705, Ferrotec, Germany).37 Magnetized alumina platelets and
nickel flakes (Novamet, USA, diameter 29 ± 5.6 μm and thickness
0.58 ± 0.24 μm) were co-mixed using a mechanical mixer in a liquid
epoxy in predetermined ratios. Typically, the epoxy composite
containing 6 vol % of magnetized alumina and 4 vol % of nickel
flakes was composed of 0.916 g of resin (Araldite GY250,
Hunstmann, Belgium), 0.783 g of hardener (Aradur 917 CH,
Hunstmann, Belgium), 0.0412 g of polyether diamine (Jeffamine D230, Hunstmann, Belgium), 0.014 g of catalyst (DY070, Hunstmann,
Belgium), 0.472 g of magnetized alumina, and 0.684 g of nickel flakes.
The mixture was homogenized at 60 °C using a mechanical mixer
before being cast into a mold prewarmed to 100 °C. The mold
consisted of an aluminum plate covered with a release plastic film and
spacers of 100 μm thickness made from aluminum tape (Griffon,
Germany). The suspension was deposited into the mold, covered, and
pressed by another aluminum plate covered with release film, and
placed under a rotating magnetic field on a heating plate. The
aforementioned field was generated by a permanent neodymium
magnet (300 mT, Supermagnete, Switzerland) rotating at 400 rpm.
The film was finally cured for 10 h at 100 °C.
Characterization of the Physical Parameters of Single
Microstructured Plates. Thermal expansion coefficients (CTE)
were measured using an optical dilatometer (DIL806, TA Instruments, Waters, Germany) by heating the composite specimen from 12
to 100 °C at a rate of 5 °C/min. The glass transition temperature (Tg)
was measured by differential scanning calorimetry (DSC822, Mettler
Toledo, Switzerland) using a heating rate of 10 °C/min. Saturated
magnetization of the samples was measured using a Physical Property
Measurement System (PPMS, Quantum Design, USA) at room
temperature. Electrical conductivity was measured using a portable
multimeter (VC230, Voltcraft, Germany) and two points probes.
Characterization of the Microstructure of the Composites.
The samples of interest were fractured, coated with Pt (Safematic,
Switzerland), and observed with a scanning electron microscope (Leo
1530, Zeiss, Germany).
Characterization of the Functional Response of Bistable
Shells. The curvature of the shells was measured from optical
micrographs by fitting the sample contour with a circle using the
software ImageJ. The mechanical force required for snapping was
measured in a compression machine (Shimadzu, Japan) with a 8 mm
diameter screw acting as the loading part and a 100 N load cell
measuring the applied force. The strain rate applied was fixed at 1
mm/min. A video was recorded while the specimen was mechanically
loaded. The magnetic restoring effect was performed by approaching
the bistable plate close to a 300 mT magnet. A video was recorded to
measure the curvature of the plate as a function of the distance to the
magnet. Three sensing−restoring experiments were performed to
ensure reproducibility of the presented data. The videos obtained
were analyzed using iMovie and ImageJ. The mechanical force
generated during the snap-through event was measured by manually
snapping the composite shell inside a plate−plate compression device
and correlating the force measured with the curvature. The electrical
resistance variation during snapping was measured using two-points
probe electrodes fixed in the center of the shell. The composites were
snapped manually while recording the electrical resistance.
Integration of the Shape-Changing Sensors into Electrical
Circuits. Copper electrical cables where taped at the surface of the
composite shells over a length of 5 mm. The shell was then connected
via an insulating board (Elenco precision model 9440, USA) to a red
LED, a resistor of 333 Ω, and a power supply (Nobatron, Sorensen).
ASSOCIATED CONTENT
S Supporting Information
*
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acsnano.9b01095.
Theoretical estimations of the coefficients of thermal
expansion in the two orthogonal directions, the
theoretical fit of the electrical conductivity using the
percolation theory, the theoretical magnetization of the
composite layer, details on composite relaxation and
cyclic response as well as an example of bistable
mechanosensing for robotic applications. Additional
figures comprise magnetization curve, electron micrograph of the cross-section of the composite, fit of the
percolation, curvature and resistance in along multiple
cycles, and schematics of a practical setup for large-scale
mechanosensing (PDF)
AUTHOR INFORMATION
Corresponding Authors
*E-mail: aarrieta@purdue.edu.
*E-mail: andre.studart@mat.ethz.ch.
ORCID
Hortense Le Ferrand: 0000-0003-3017-9403
André R. Studart: 0000-0003-4205-8545
Andres F. Arrieta: 0000-0003-4641-5220
Present Address
§
School of Mechanical and Aerospace Engineering, Nanyang
Technological University, 50 faculty avenue, Singapore 639798
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
We thank D. Koulialias for the measurement of the magnetic
field of the composite. We acknowledge funding from ETH
Zurich and Purdue University. H.L.F. and A.R.S. were partially
funded by the European Office of Aerospace Research and
Development (EOARD, grant FA9550-16-1-0007).
REFERENCES
(1) Stieve, H. Sensors of Biological Organisms- Biological Transducers. Sens. Actuators 1983, 4, 689−704.
(2) Beccai, L.; Lucarotti, C.; Totaro, M.; Taghavi, M. Soft robotics:
Trends, Applications and Challenges. In Biosystems & Biorobotics;
Laschi, C., Rossiter, J., Iida, F., Cianchetti, M., Margheri, L., Eds.;
4758
DOI: 10.1021/acsnano.9b01095
ACS Nano 2019, 13, 4752−4760
Article
ACS Nano
Springer International Publishing AG: Switzerland, 2011; Vol. 17, pp
11−34.
(3) Maheshwari, V.; Saraf, R. F. High-Resolution Thin-Film Device
to Sense Texture by Touch. Science 2006, 312, 1501−1505.
(4) Sasagawa, G.; Zumberge, M. A. A Self-Calibrating Pressure
Recorder for Detecting Seafloor Height Change. IEEE J. Oceanic Eng.
2013, 38, 447−454.
(5) Boutry, C. M.; Nguyen, A.; Lawal, Q. O.; Chortos, A.; RondeauGagné, S.; Bao, Z. A Sensitive and Biodegradable Pressure Sensor
Array for Cardiovascular Monitoring. Adv. Mater. 2015, 27, 6954−
6961.
(6) Polygerinos, P.; Correll, N.; Morin, S. A.; Mosadegh, B.; Onal, C.
D.; Petersen, K.; Cianchetti, M.; Tolley, M. T.; Shepherd, R. F. Soft
Robotics: Review of Fluid-Driven Intrinsically Soft Devices:
Manufacturing, Sensing, Control, and Applications in Human-Robot
Interaction. Adv. Eng. Mater. 2017, 19, 1700016.
(7) Lentink, D.; Müller, U. K.; Stamhuis, E. J.; De Kat, R.; Van
Gestel, W.; Veldhuis, L. L. M.; Henningsson, P.; Hedenström, A.;
Videler, J. J.; van Leeuwen, J. L. How Swifts Control Their Glide
Performance with Morphing Wings. Nature 2007, 446, 1082−1085.
(8) Kuder, I. K.; Fasel, U.; Ermanni, P.; Arrieta, A. F. Concurrent
Design of a Morphing Aerofoil with Variable Stiffness Bistable
Laminates. Smart Mater. Struct. 2016, 25, 115001.
(9) Barbarino, S.; Bilgen, O.; Ajaj, R. M.; Friswell, M. I.; Inman, D. J.
A Review of Morphing Aircraft. J. Intell. Mater. Syst. Struct. 2011, 22,
823−877.
(10) Boadi-Danquah, E.; MacLachlan, D.; Fadden, M. Cyclic
Performance of a Lightweight Rapidly Constructible and Reconfigurable Modular Steel Floor Diaphragm. Key Eng. Mater. 2018, 763,
541−548.
(11) Chi, M.; Plaza, A.; Benediktsson, J. A.; Sun, Z.; Shen, J.; Zhu, Y.
Big Data for Remote Sensing: Challenges and Opportunities. Proc.
IEEE 2016, 104, 2207−2220.
(12) Saudabayev, A.; Varol, H. A. Sensors for Robotic Hands: a
Survey of State of the Art. IEEE Access 2015, 3, 1765−1782.
(13) Tata, U.; Deshmukh, S.; Chiao, J. C.; Carter, R.; Huang, H.
Bio-Inspired Sensor Skins for Structural Health Monitoring. Smart
Mater. Struct. 2009, 18, 104026.
(14) Yi, X.; Cho, C.; Cooper, J.; Wang, Y.; Tentzeris, M. M.; Leon,
R. T. Passive Wireless Antenna Sensor for Strain and Crack Sensing −
Electromagnetic Modeling, Simulation and Testing. Smart Mater.
Struct. 2013, 22, 085009.
(15) Ozbey, B.; Erturk, V. B.; Demir, H.; Altintas, A.; Kurc, O. A
Wireless Passive Sensing System for Displacement/Strain Measurement in Reinforced Concrete Members. Sensors 2016, 16, 496−513.
(16) Yao, J.; Tjuatja, S.; Huang, H. Dynamic Interrogation of
Wireless Antenna Sensor. IEEE Sens. J. 2015, 15, 4338−4345.
(17) Murmann, B. Digitally Assisted Analog Circuits. IEEE Micro
2006, 26, 38−47.
(18) Poon, C. C. Y.; Lo, B. P. L.; Yuce, M. R.; Alomainy, A.; Hao, Y.
Body Sensor Networks: in the Era of Big Data and Beyond. IEEE Rev.
Biomed. Eng. 2015, 8, 4−16.
(19) Carruthers, A. C.; Thomas, A. L. R.; Taylor, G. K. Automatic
Aeroelastic Devices in the Wings of a Steppe Eagle Aquila Nipalensis.
J. Exp. Biol. 2007, 210, 4136−4149.
(20) Barth, F. G. Learning from Animal Sensors: the Clever “Design”
of Spider Mechanoreceptors. Proceedings from the SPIE Smart
Structures and Materials + Nondestructive Evaluation and Health
Monitoring, March 11, 2012, San Diego, CA; SPIE: Bellingham WA,
2012; Bioinspiration, Biomimetics, and Bioreplication, Vol. 8339, pp
1−11.
(21) Studart, A. R. Biologically Inspired Dynamic Material Systems.
Angew. Chem., Int. Ed. 2015, 54, 3400−3416.
(22) Fratzl, P.; Barth, F. G. Biomaterial Systems for Mechanosensing
and Actuation. Nature 2009, 462, 442−449.
(23) Yang, Y.; Nguyen, N.; Chen, N.; Lockwood, M.; Tucker, C.;
Hu, H.; Bleckmann, H.; Liu, C.; Jones, D. L. Artificial Lateral Line
with Biomimetic Neuromasts to Emulate Fish Sensing. Bioinspiration
Biomimetics 2010, 5, 016001.
(24) Giri, A.; Kottapalli, P.; Asadnia, M.; Miao, J.; Venkatraman, S.
S.; Triantafyllou, M. S. Nanofibril Scaffold Assisted MEMS Artificial
Hydrogel Neuromasts for Enhanced Sensitivity Flow Sensing. Sci. Rep.
2016, 6, 19336.
(25) Sterbing-D’Angelo, S.; Chadha, M.; Chiu, C.; Falk, B.; Xian,
W.; Barcelo, J.; Zook, J. M.; Moss, C. F. From the Cover: Bat Wing
Sensors Support Flight Control. Proc. Natl. Acad. Sci. U. S. A. 2011,
108, 11291−11296.
(26) Brown, R. E.; Fedde, M. R. Airflow Sensors in the Avian Wing.
J. Exp. Biol. 1993, 30, 13−30.
(27) Suh, J. W.; Darling, R. B.; Donald, B. R.; Baltes, H.; Kovacs, G.
T. A. CMOS Integrated Siliary Actuator Array as a General-Purpose
Micromanipulation Tool for Small Objects. J. Microelectromech. Syst.
1999, 8, 483−496.
(28) Evans, B. A.; Shields, A. R.; Carroll, R. L.; Washburn, S.; Falvo,
M. R.; Superfine, R. Magnetically Actuated Nanorod Arrays as
Biomimetic Cilia. Nano Lett. 2007, 7, 1428−1434.
(29) Bora, M.; Kottapalli, A. G. P.; Miao, J.; Triantafyllou, M.
MEMS Biomimetic Superficial and Canal Neuromasts for Flow
Sensing. Bioinspiration Biomimetics 2018, 13, 025002.
(30) Herzog, H.; Klein, A.; Bleckmann, H.; Holik, P.; Schmitz, S.;
Siebke, G.; Tätzner, S.; Lacher, M.; Steltenkamp, S. μ-Niomimetic
Flow-Sensors − Introducing Light-Guiding PDMS Structures into
MEMS. Bioinspiration Biomimetics 2015, 10, 036001.
(31) Kottapalli, A. G. P.; Bora, M.; Asadnia, M.; Miao, J.;
Venkatraman, S. S.; Triantafyllou, M. From Biological Cilia to
Artificial Flow Sensors: Biomimetic Soft Polymer Nanosensors with
High Sensing Performance. Sci. Rep. 2016, 6, 19336.
(32) Kottapalli, A. G. P.; Asadnia, M.; Miao, J.; Triantafyllou, M.
Touch at a Distance Sensing: Lateral-Line Inspired MEMS Flow
Sensors. Bioinspiration Biomimetics 2014, 9, 46011.
(33) Sen, S.; Chatterjee, S.; Har, C. Design and Impedance
Estimation of a Biologically Inspired Flexible Mechanical Transmission with Exponential Elastic Characteristic. Proceedings from the
IEES/RSJ. International Conference on Intelligent Robots and Systems,
Tokyo, Japan, November 3−8, 2013; IEEE: Piscataway, NJ, 2013; pp
5425−5430.
(34) Fratzl, P.; Weinkamer, R. Nature’s Hierarchical Materials. Prog.
Mater. Sci. 2007, 52, 1263−1334.
(35) Erb, R. M.; Sander, J. S.; Grisch, R.; Studart, A. R. Self-Shaping
Composites with Programmable Bioinspired Microstructures. Nat.
Commun. 2013, 4, 1712.
(36) Schmied, J.; Le Ferrand, H.; Ermanni, P.; Studart, A. R.;
Arrieta, A. F. Programmable Snapping Composites with Bio-Inspired
Architecture. Bioinspiration Biomimetics 2017, 12, 026012.
(37) Erb, R. M.; Segmehl, J.; Charilaou, M.; Loffler, J. F.; Studart, A.
R. Non-Linear Alignment Dynamics in Suspensions of Platelets under
Rotating Magnetic Fields. Soft Matter 2012, 8, 7604−7609.
(38) Riley, K. S.; Le Ferrand, H.; Arrieta, A. F. Modeling of
Snapping Composite Shells with Magnetically Aligned Bio-Inspired
Reinforcements. Smart Mater. Struct. 2018, 27, 114003.
(39) Arrieta, A. F.; Kuder, I. K.; Waeber, T.; Ermanni, P. Variable
Stiffness Characteristics of Embeddable Multi-Stable Composites.
Compos. Sci. Technol. 2014, 97, 12−18.
(40) Hyer, M. W. Calculations of the Room-Temperature Shapes of
Unsymmetric Laminates. J. Compos. Mater. 1981, 15, 296−310.
(41) Celzard, A.; Marêché, J. F. Non-Universal Conductivity Critical
Exponents in Anisotropic Percolating Media: a New Intepretation.
Phys. A (Amsterdam, Neth.) 2003, 317, 305−312.
(42) Le Ferrand, H.; Bolisetty, S.; Demirors, A. F.; Libanori, R.;
Studart, A. R.; Mezzenga, R. Magnetic Assembly of Transparent and
Conducting Graphene-Based Functional Composites. Nat. Commun.
2016, 7, 12078.
(43) Zhang, Z.; Wu, H.; He, X.; Wu, H.; Bao, Y.; Chai, G. The
Bistable Behaviors of Carbon-Fiber/Epoxy Anti-Symmetric Composite Shells. Composites, Part B 2013, 47, 190−199.
(44) Li, L.; Kessler, M. R. Creep-Resistant Behavior of SelfReinforcing Liquid Crystalline Epoxy Resins. Polymer 2014, 55,
2021−2027.
4759
DOI: 10.1021/acsnano.9b01095
ACS Nano 2019, 13, 4752−4760
Article
ACS Nano
(45) Starkova, O.; Buschhorn, S. T.; Mannov, E.; Schulte, K.;
Aniskevich, A. Creep and Recovery of Epoxy/MWCNT Nanocomposites. Composites, Part A 2012, 43, 1212−1218.
(46) Handwerker, H. O. General Sensory. In Human Physiology;
Schmidt, R. F., Thews, G., Eds.; Springer-Verlag: Berlin Heidelberg,
1983; pp 176−195.
(47) Lum, G. Z.; Ye, Z.; Dong, X.; Marvi, H.; Erin, O.; Hu, W.; Sitti,
M. Shape-Programmable Magnetic Soft Matter. Proc. Natl. Acad. Sci.
U. S. A. 2016, 113, E6007−E6015.
(48) Menges, B. A. Material Computation: Higher Integration in
Morphogenic Design. Archit. Des. 2012, 82, 14−21.
(49) Tibbits, S.; Cheung, K. Programmable Materials for
Architectural Assembly and Automation. Assembly Automation 2012,
32, 216−225.
(50) Fang, Y.; Yashin, V. V.; Levitan, S. P.; Balazs, A. C. Pattern
Recognition with “Materials that Compute". Sci. Adv. 2016, 2,
e1601114.
(51) McEvoy, M. A.; Correll, N. Materials that Couple Sensing,
Actuation, Computation and Communication. Science 2015, 347,
1261689.
(52) Jiang, Y.; Korpas, L. M.; Raney, J. R. Bifurcation-Based
Embodied Logic and Autonomous Actuation. Nat. Commun. 2019,
10, 128.
(53) Treml, B.; Gillman, A.; Buskohl, P.; Vaia, R. Origami
Mechanologic. Proc. Natl. Acad. Sci. U. S. A. 2018, 115, 6916−1921.
4760
DOI: 10.1021/acsnano.9b01095
ACS Nano 2019, 13, 4752−4760