Reconfigurable Magnetic Slime Robot: Deformation, Adaptability, and Multifunction

Abstract

Magnetic miniature soft‐bodied robots allow non‐invasive access to restricted spaces and provide ideal solutions for minimally invasive surgery, micromanipulation, and targeted drug delivery. However, the existing elastomer‐based (silicone) and fluid‐based (ferrofluid or liquid metal) magnetically actuated miniature soft robots have limitations. Owing to its limited deformability, the elastomer‐based small‐scale soft robot cannot navigate through a highly restricted environment. In contrast, although fluid‐based soft robots are more capable of deformation, they are also limited by the unstable shape of the fluid itself, and are therefore poorly adapted to the environment. In this study, non‐Newtonian fluid‐based magnetically actuated slime robots with both the adaptability of elastomer‐based robots and reconfigurable significant deformation capabilities of fluid‐based robots are demonstrated. The robots can negotiate through narrow channels with a diameter of 1.5 mm and maneuver on multiple substrates in complex environments. The proposed slime robot implements various functions, including grasping solid objects, swallowing and transporting harmful things, human motion monitoring, and circuit switching and repair. This study proposes the design of novel soft‐bodied robots and enhances their future applications in biomedical, electronic, and other fields. Non‐Newtonian fluid‐based magnetically actuated slime robots with both the adaptability of elastomer‐based robots and the reconfigurable significant deformation capabilities of fluid‐based robots are proposed. This non‐Newtonian fluid‐based robot with large deformation, reconfigurability, self‐healing, and conductivity is expected to be of great value in wearable devices and biomedical applications.

Full text

www.afm-journal.de
© 2022 Wiley-VCH GmbH
2112508 (1 of 13)
L. Zhang
Department of Surgery
The Chinese University of Hong Kong
Hong Kong SAR 999077, China
L. Zhang
CUHK T Stone Robotics Institute
The Chinese University of Hong Kong
Hong Kong SAR 999077, China
L. Zhang
Chow Yuk Ho Technology Center for Innovative Medicine
The Chinese University of Hong Kong
Hong Kong SAR 999077, China
L. Zhang
Multi-Scale Medical Robotics Center
Hong Kong Science Park, Shatin NT, Hong Kong SAR 999077, China
ReseaRch aRticle
Reconfigurable Magnetic Slime Robot: Deformation,
Adaptability, and Multifunction
Mengmeng Sun, Chenyao Tian, Liyang Mao, Xianghe Meng, Xingjian Shen, Bo Hao,
Xin Wang, Hui Xie,* and Li Zhang*
Magnetic miniature soft-bodied robots allow non-invasive access to restricted
spaces and provide ideal solutions for minimally invasive surgery, microman-
ipulation, and targeted drug delivery. However, the existing elastomer-based
(silicone) and fluid-based (ferrofluid or liquid metal) magnetically actuated
miniature soft robots have limitations. Owing to its limited deformability,
the elastomer-based small-scale soft robot cannot navigate through a highly
restricted environment. In contrast, although fluid-based soft robots are more
capable of deformation, they are also limited by the unstable shape of the
fluid itself, and are therefore poorly adapted to the environment. In this study,
non-Newtonian fluid-based magnetically actuated slime robots with both the
adaptability of elastomer-based robots and reconfigurable significant defor-
mation capabilities of fluid-based robots are demonstrated. The robots can
negotiate through narrow channels with a diameter of 1.5 mm and maneuver
on multiple substrates in complex environments. The proposed slime robot
implements various functions, including grasping solid objects, swallowing
and transporting harmful things, human motion monitoring, and circuit
switching and repair. This study proposes the design of novel soft-bodied
robots and enhances their future applications in biomedical, electronic, and
otherfields.
DOI: 10.1002/adfm.202112508
far, most magnetically actuated soft-bodied
robots have been fabricated from soft elas-
tomers mixed with hard magnetic parti-
cles.[26–43] Such elastomer-based soft robots
perform multiple movement modes,[26–28]
adapt to complex interface environ-
ments,[31,32] and enter confined spaces for
robotic manipulation applications.[33–35]
However, the functionality of elastomer-
based soft robots is limited by their prede-
signed shapes and cannot be reconfigured
in situ. Moreover, elastomer-based soft
robots possess limited deformation capa-
bilities and cannot pass through narrow
spaces that are significantly smaller than
theirdimensions.
In contrast, recent studies have demon-
strated that small fluid-based robots, such
as those based on liquid metal or ferro-
fluid,[44–50] behave more gently and softly.
Fluid-based soft robots exhibit better
deformability than elastomer-based soft
robots owing to their fluid flow proper-
ties that allow them to easily pass through
extraordinarily narrow and restricted
spaces and avoid damaging surrounding biological tissues.[51,52]
For instance, by constructing an electromagnet array, an intel-
ligent deformable and cooperative ferrofluid-based soft robot
that could pass through 1.5 mm diameter narrow channels
and perform various functions[53] was constructed. In addi-
tion, researchers have also realized the control of electric cir-
cuits and fluid pumping using liquid-metal-based soft robots.[54]
However, ferrofluid- and liquid metal-based robots require very
demanding operating environments; for instance, oil-based
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/adfm.202112508.
1. Introduction
Miniature robots that respond to external stimuli have the
advantage of being less invasive and more accessible,[1–6]
making them exciting candidates for biomedical applica-
tions[7–12] such as targeted drug delivery,[13,14] minimally inva-
sive surgery,[15] and cell transplantation.[16,17] Owing to its safety,
precision, and fast response, an external magnetic field is a
promising choice for actuating small-scale robots.[18–25] Thus
M. Sun, B. Hao, X. Wang, L. Zhang
Department of Mechanical and Automation Engineering
The Chinese University of Hong Kong
Hong Kong SAR 999077, China
E-mail: lizhang@cuhk.edu.hk
C. Tian, L. Mao, X. Meng, X. Shen, H. Xie
State Key Laboratory of Robotics and Systems
Harbin Institute of Technology
2 Yikuang, Harbin 150001, China
E-mail: xiehui@hit.edu.cn
Adv. Funct. Mater. 2022, 2112508

www.afm-journal.dewww.advancedsciencenews.com
2112508 (2 of 13) © 2022 Wiley-VCH GmbH
ferrofluids require hydrophilic surfaces surrounded by water-
based solutions to maintain the shape of the spherical droplets
without adhering to the substrate,[53] whereas liquid metals
require alkaline or acidic solutions to preserve the form of
spherical droplets without adhering to the substrate.[55,56]
Therefore, it is necessary to combine the characteristics of
large deformations of liquid-based robots with the complex
interface adaptability of elastomer-based robots to create novel
softrobots.
To satisfy this demand, we propose multifunctional mag-
netic slime robots with large deformations and adaptability. We
report the fabrication process of a magnetic slime and its fun-
damental mechanism of shape transformation. In comparison
with existing elastomer-based soft robots, our proposed slime
robot has better deformability, for instance, through narrow
pipes (1.5 mm) and complex maze environments. In addition,
when compared to existing fluid-based soft robots, the slime
robot has greater environmental adaptability, allowing move-
ment not only in two-phase fluids but also in air, and even on
various surfaces such as hydrogel, metallic, and plastic sur-
faces. Thereafter, we demonstrate that this slime robot can
have multiple reconfigurable functions, such as grasping and
delivering objects through the curling mode and wrapping and
transporting harmful things through the endocytosis mode. In
addition, the ability to self-heal and conduct electricity allows
the slime robot to be used as an electrical device, for instance,
as a piezoresistive strain sensor to monitor human movement
and as a circuit control switch or circuit breaker repair agent.
Thus, our proposed non-Newtonian fluid-based robot with a
large deformation, reconfigurability, self-healing, and conduc-
tivity is expected to be of great value in wearable devices and
biomedicalapplications.
2. Results and Discussion
2.1. Preparation and Characterization of the Magnetic Slime
The preparation scheme of the magnetic slime is illustrated in
Figure1a, where magnetic particles (NdFeB) and borax were
sequentially added to a polyvinyl alcohol (PVA) solution to
obtain the magnetic slime. The inset in Figure 1a shows the
critical reactions necessary for the formation of the magnetic
slime. Magnetic slime is formed primarily through the inter-
action of tetrafunctional borate ions with the −OH group of
PVA, where tetrafunctional borate ions are generated through
the hydrolysis of borax. Notably, magnetic slime consists
of >90 wt% water, implying that it is essentially a hydrogel
(Figure1b). As shown in (Figure1c), scanning electron micros-
copy (SEM) images indicate that the microstructure of the
magnetic slime is a 3D porous network of magnetic particles
cross-linked with immobilized polymers. The porous structure
is favorable for the extensibility and fast response of the mag-
netic slime. In addition, we investigated the cytotoxicity of the
slime. As shown in Figure S1A, Supporting Information, at
concentrations up to 400 μg mL−1, non-magnetic slime is non-
toxic to NIH 3T3 cells, indicating its excellent biocompatibility.
However, NdFeB particles are toxic, thus making the magnetic
slime (NdFeB: 30%) non-biocompatible. Therefore, the NdFeB
microparticles were coated with SiO2 via the Stober method.
Transmission electron microscopy images demonstrated that
the NdFeB microparticles were coated with a layer of 35 nm
SiO2 (FigureS1B, Supporting Information). Subsequently, they
were added to the PVA solution to prepare the magnetic slime.
Cytotoxicity tests showed that the 400 μg mL−1 magnetic slimes
prepared using NdFeB@SiO2 were not toxic tocells.
Magnetic slime has both active and passive deformation capa-
bilities, and its shape can be changed by altering the external
magnetic field or it can be adapted to environmental bounda-
ries using rheological properties. The rheological and magne-
torheological properties of the slime were measured using a
rheometer (Anton Paar MCR302) with a 25 mm parallel plate
setup. A dynamic strain sweep ranging from 0.1% to 100% at
ω= 10.0 Hz was first performed, and the storage modulus (G′)
was recorded to define the linear viscoelastic region (LVR) in
which the storage modulus is independent of the strain ampli-
tude (Figure1d). A strain of 1.0% was selected in subsequent
oscillation tests to ensure that the dynamic oscillatory defor-
mation of each sample was within the LVR. Figure 1e shows
the changes in the storage (G′, solid symbols) and loss moduli
(“G”, hollow symbols) as a function of angular frequency for
non-magnetic slime (NdFeB: 0%) and magnetic slime (NdFeB:
50%). Both slimes are in a solid state, with the storage modulus
exceeding the loss modulus over the entire frequency range. It
is observed that the presence of NdFeB microparticles raises
the moduli and enhances the elastic response of the slime. In
addition, Figure1f shows the dependence of G′ on the magnetic
field at a fixed strain amplitude (0.2%) and angular frequency
(6 rad s−1). The change in storage modulus for the magnetic
slime is significant in increasing the magnetic field strength,
indicating a significant magnetorheological eect. Applying a
magnetic field enhances the mechanical properties of the mag-
netic slimes owing to the magnetic dipole–dipole forces. Mag-
netic-field-dependent dynamic viscoelasticity is mainly attrib-
uted to the interaction between adjacent particles when they are
magnetized. Such interactions condense the polymer chain and
make the structure more rigid. The magnetorheological eect
causes magnetic slimes to exhibit tunable stiness by applying
a magnetic field. Therefore, it is possible to control the stiness
of the magnetic slime to adapt it to dierent working environ-
ments. In Figure1g, we first investigate the active deformation
behavior of magnetic slimes, which can deform into complex
shapes, such as circles, hexagons, and rings, under the config-
ured magnetic field. Both experiments and simulations reveal
that the morphology of the magnetic slime is similar to that of
the permanent magnet at the bottom because the unmagnet-
ized magnetic particles in the slime tend to move to the position
with the lowest magnetic field strength.[53] This allows the mag-
netic slime to change its morphology depending on the needs
of dierent tasks. Figure1h shows that the magnetic slime can
be driven by a permanent magnet (NdFeB) and stretched more
than seven times its original length along the direction of the
magnet motion. At t= 0 s, a magnetic slime with a diameter
of 9 mm was placed on a polymethylmethacrylate substrate,
and a circular permanent magnet with a diameter of 20 mm
and height of 10 mm was set at a distance of 4 mm below
the substrate. As the circular permanent magnet moved at a
speed of 2 mm s−1, the slime started to elongate and stretched
Adv. Funct. Mater. 2022, 2112508

www.afm-journal.dewww.advancedsciencenews.com
2112508 (3 of 13) © 2022 Wiley-VCH GmbH
Figure 1. Synthesis process and deformability of magnetic slime. a) Schematic diagram of the magnetic slime fabrication process. Inset illustrates the
crosslinking reaction between PVA and a tetrafunctional borate ion. b) Image of the magnetic slime. c) SEM images of the freeze-dried unmagnetized
slime. Scale bar, 100 μm. d) Strain dependence of G’ and G” for non-magnetic slime and magnetic slime, respectively, measured at ω= 10.0 Hz. e) Fre-
quency dependence of G’ and G” for non-magnetic slime and magnetic slime, respectively, measured at a strain of 1.0%. f) Magnetic field dependence
of the storage modulus G’ of magnetic slime with varying weight percentages of the NdFeB particles measured at 0.2% deformation and oscillation
frequency of 6 rad s−1. g) Programming complex shapes of the magnetic slime using ferromagnets in disk-, hexagon-, and ring-shapes. The simulation
results demonstrate the magnetic field distribution in the plane 2 mm above the magnet. Scale bar, 15 mm. h) Experimental frames of the stretchable
behavior of the magnetic slime. Scale bar, 10 mm. i) Simulation of controlled deformation behavior of magnetic slime.
Adv. Funct. Mater. 2022, 2112508

www.afm-journal.dewww.advancedsciencenews.com
2112508 (4 of 13) © 2022 Wiley-VCH GmbH
to 60 mm after 71 s. The elongation process of the slime was
reproduced using COMSOL Multiphysics software (Figure1i;
Movie S1, Supporting Information). In addition, the ability to
deform allows the slime to elongate, whereas the friction with
the interface and the configuration of the magnetic field allow
the slime to remain locked in the path travelled. As shown in
Figure S2, Supporting Information, the magnetic slime can
actively deform into the shape of “CUHK” along the direction
of motion of the permanentmagnet.
2.2. Environmental Adaptability
The passive and active deformation capabilities of magnetic
slimes give them a remarkable advantage, not only for adapting
to complex terrain environments but also for entering con-
fined spaces in a controlled manner while maintaining their
mobility and integrity. The shape-shifting nature allows the
slime to adapt to changing external environments to satisfy the
demands of multitasking. As illustrated in Figure2a, under
the control of a magnetic field, the slime can navigate freely
through various terrains such as narrow channels, confined
tubes, complex mazes, and uneven substrates (Movie S2, Sup-
porting Information). First, an array of channels with diameters
of 6, 4.5, 3, and 1.5 mm were prepared to test the deformation
ability of the magnetic slime. According to the experimental
results, although it took a long time (≈180 s) for the magnetic
slime to pass through the 1.5 mm diameter channel, eventually,
the slime could successfully pass through each channel in turn.
Subsequently, we demonstrated that 1 mL magnetic slime can
flow smoothly in a liquid-free tube with an inner diameter of
5 mm. Deformability allows the slime to negotiate the curved
tube structure quickly and stretch its shape along the direction
of movement within the tube. In addition, a maze with varied
inner diameters (minimum gap: 1.5 mm) and complex branches
was also prepared to demonstrate the deformability and con-
trollability of the slime. Finally, we demonstrate that slimes
can migrate over uneven terrain with a width of 6.28 mm and
height of 3 mm. The magnetic slime can maneuver unobstruct-
edly over multiple terrains by actively and passively changing
its shape, dramatically reducing its damage to the surrounding
environment and expanding its applicationscenarios.
To achieve optimally controlled deformation of the mag-
netic slime, we systematically investigated the aspect ratio of
slime when varying the content of magnetic particles and the
strength of the applied magnetic field. As shown in Figure2b,
the aspect ratio of the magnetic slime deformation is propor-
tional to the magnetic particle content when the applied mag-
netic field is the same. Under an external magnetic field, the
increase in the magnetic powder content dramatically enhances
the magnetic response of the slime, causing it to be subjected
to a more significant deformation force. In addition to the
magnetic particle content, the outer magnetic field strength
also aects the deformation behavior of the magnetic slime.
Figure 2c shows that the deformation ability of the magnetic
slime is positively correlated with the strength of the external
magnetic field. The motion behavior of the internal particles
of the slime is determined by the magnetic field strength that
aects its overall deformation properties. To characterize the
mobility of the magnetic slime in a restricted environment, we
also investigated the minimum diameter of the tube through
which the slime could pass when the strength of the external
magnetic field was varied. The experimental results demon-
strate that the minimum diameter of the channel through
which the slime can pass is 1.5 mm for a fixed content of mag-
netic particles and slime volume. The minimum diameter of
the tube through which it can pass increases as the external
magnetic field strength decreases (Figure2d). In addition, we
systematically compared the deformability of ferrofluid droplet,
magnetic liquid metal, and magnetic slime robots under equal
magnetic field strength. Ferrofluids with dynamic viscosities
of 8 mPa s (EMG 901; Ferrotec Corporation) were used in this
experiment, and the deformation environment was water. In
addition, a magnetic liquid metal robot was constructed using
a previously reported method,[54] and its deformation environ-
ment was a 2 mol L−1 hydrochloric acid solution. As shown in
Figure S3A, Supporting Information, under a low magnetic
field strength, the deformation ability of the ferrofluid droplet
robots is the best; however, as the magnetic field strength
increases, it becomes dicult for the ferrofluid droplet robots
to remain intact. The maximum elongation length increased
linearly as the magnetic field strength increased for the slime
and liquid metal robots. Under equal magnetic field strength,
the elongation length of the slime robot was greater than that
of the magnetic liquid metal robot. In addition, we compared
the maximum elongation lengths of ferrofluid droplet, mag-
netic liquid metal, and slime robots with equal volumes. The
results demonstrate that the elongation of the slime robot was
the largest when the three volumes were equal (Figure S3B,
Supporting Information).
In addition to its excellent terrain adaptability, magnetic
slimes can move across multiple interfaces, which is impos-
sible for conventional liquid-based robots. The structure of the
slime gives it the adaptability of an elastomer-based robot and
the ability to deform substantially as a liquid-based robot, but
without the disadvantages of either form. Thus far, the move-
ment of liquid-based soft robots, such as ferrofluids or liquid
metals, requires hydrophilic interfaces or alkaline solutions,
which significantly limits their application. However, our
proposed magnetic slime can be adapted to a wide range of
substrates. We prepared a total of eight commonly used sub-
strates, including hydrogel, metal, plastic, glass, silica, silicon,
polydimethylsiloxane, and paper substrates, and cut them into
10 cm × 10 cm pieces. As illustrated in Figure2e, when actu-
ated by a permanent magnet, a 500 μL slime can follow the
trajectory of “SLIMEBOT” on these substrates. In addition to
the demonstrated controlled deformation behavior of slime
in air, it can also work stably underwater and meet the needs
of various tasks (Figures S4 and S5, Supporting Information).
The deformation ability of the slime underwater is better than
that in air, which takes only 50 s to pass through the narrow
channel with a diameter of 1.5 mm because the liquid envi-
ronment dramatically reduces the adhesion between the slime
and the channel (Movie S3, Supporting Information). The PVA
component endows the slime with a self-adhesive ability. As
shown in Figure S6A, Supporting Information, the hydroxyl
groups in the molecular chain of PVA combined with water
molecules, resulting in strong hydrogen bonds. When the
Adv. Funct. Mater. 2022, 2112508

www.afm-journal.dewww.advancedsciencenews.com
2112508 (5 of 13) © 2022 Wiley-VCH GmbH
slime fills the gap and is in complete contact with the sub-
strate, it also adheres to the substrate through van der Waals
forces. To further understand the adhesion properties of slime,
we experimentally investigated the peeling strength between
the slime and the substrate (FigureS6B, Supporting Informa-
tion). As shown in Figure S6C, Supporting Information, the
Figure 2. Environmental adaptability analysis. a) Overlapped sequential snapshots of videos show that the magnetic slime is actuated to navigate
through the channel, tube, maze, and uneven surfaces in air. b) The deformation ability of magnetic slime is a function of the content of the inner
magnetic particles. c) Relationship between the strength of the permanent magnet and deformation ability of magnetic slime with varying magnetic
power content. d) Minimum diameter of the magnetic slime that can pass through the thin tube versus the magnetic field strength. e) Overlapped
sequential snapshots of videos demonstrate that the magnetic slime is actuated along the “S”, “L”, “I”, “M”, “E”, “B”, “O”, and “T” trajectories on
varying substrate surfaces in air. Scale bars, 10 mm.
Adv. Funct. Mater. 2022, 2112508

www.afm-journal.dewww.advancedsciencenews.com
2112508 (6 of 13) © 2022 Wiley-VCH GmbH
experimental results demonstrate that paper has the highest
bond strength that decreases with increasing magnetic particle
content. This is mainly due to the better water absorption eect
of the paper and the rough surface where the slime can dif-
fuse into the micropores and be retained as a residue during
peeling. The hydrogen bonding between the PVA molecular
chains enhances adhesion during the tearing of the mate-
rial when the bonding force of the hydrogen bond is greater
than the aforementioned van der Waals force. Furthermore,
magnetic particles can decrease the slime adhesion properties
because they prevent the hydroxyl groups in the PVA mole-
cular chain from combining with water molecules. This adhe-
sion property reduces the deformation rate and overall motion
speed; however, the driving magnetic field can overcome it,
preventing it from aecting the final deformation and reach-
able range. The adaptability of slime to external environments
allows it to meet multitasking needs such as micromanipula-
tion and biomedicalapplications.
2.3. Controlled Manipulation
In addition to adapting to various complex terrains based on
their rheological properties, slimes can also utilize reconfigur-
ability to achieve flexible manipulation. Figure3a shows that
slimes can perform object grasping using the curling mode
after being elongated (Movie S4, Supporting Information). The
mechanism of the curling mode is illustrated in Figure S7,
Figure 3. Multimodal manipulation. a) Slime grasps a single target object (above) and multiple target objects (below) in curling mode. b) Slime grasps
target objects in open environment (above) and complex restricted environment (below) in endocytosis mode. Scale bars, 10 mm.
Adv. Funct. Mater. 2022, 2112508

www.afm-journal.dewww.advancedsciencenews.com
2112508 (7 of 13) © 2022 Wiley-VCH GmbH
Supporting Information. At t= 0 s, the magnetic slime actuated
by the permanent external magnet began to stretch. After 46 s,
the slime elongates to the bottom of the target object. During
the elongation process, the magnetic particles in the slime are
magnetized by a permanent external magnet and maintain
a consistent magnetization direction. Because slime exhibits
non-Newtonian behavior leading to limited dimensional sta-
bility,[57] the orientation of internal magnetic particles with
remanent magnetism does not change for a short time when
the permanent external magnet is withdrawn. At this time, if a
rotating magnetic field is applied, the elongated magnetic slime
starts to curl and hook the target object (t= 83 s). Finally, driven
by the rotating magnetic field, the slime continues to curl until
it eventually wraps the target object completely and can carry
it for rolling motion (t= 140 s). Furthermore, the magnetic
slime can grasp multiple target objects simultaneously using its
deformation characteristics, as shown in Figure3a. When the
manipulated object is present in all three directions, the slime
can extend its tentacles in three directions, such as an octopus
arm (t= 43 s). Thereafter, an external rotating magnetic field is
applied in sequence, and the three tentacles of the slime begin
to curl sequentially. Finally, the three target objects were grasped
using the curl mode (Movie S4, Supporting Information). The
curl operation mode can help accomplish the grasping function
for long-range restricted environmentalspecies.
Owing to its shape-shifting property, the slime can also
grasp and transport objects through endocytosis. As illus-
trated in Figure 3b (above), the slime can swallow the target
object through wrapping, spreading, and curling, consecu-
tively (Movie S5, Supporting Information). At t= 0 s, the mag-
netic slime begins to deform under the action of the bottom-
ring-shaped permanent magnet. The circular slime gradually
becomes C-shaped when the ring-shaped permanent magnet is
tilted, and moving the permanent magnet can induce the slime
to enclose the target object (t= 60 s). Thereafter, the C-shaped
slime becomes a ring when the ring-shaped permanent magnet
is placed parallel to the operating surface. Subsequently, the
permanent magnet was continuously moved outward in a recip-
rocal manner to spread the slime on a flat surface (t= 192 s).
As the slime expands, the permanent magnet magnetizes the
magnetic particles in the slime. Finally, the slime curls under a
rotating magnetic field and swallows the target object placed on
top of it (t= 237 s). When an object is swallowed by the slime,
it is completely isolated from the external environment, which
facilitates the capture of dangerous objects. For instance, mag-
netic slime can be used in clinical interventions for button bat-
tery ingestion, which seriously endangers the life of the patient.
We simulated human ingestion by sandwiching button bat-
teries between the lining of the large intestine of a pig. After
30 min, it caused severe damage to the lining of the intestine
(Figure S8, Supporting Information). To reduce the risk of
battery ingestion, magnetic slime is utilized and controlled
by directing it to the injury site and thereafter to remove the
stuck button battery through endocytosis (Figure 3b (below)).
We prepared a simulated stomach model with the inner lining
filled with folds. When the magnetic slime enters the stomach
interior, it can move along the uneven inner wall, driven by an
external magnetic field (t= 42 s). When the slime reaches the
damage site, the button cell can be wrapped in the slime by
covering and curling to prevent further discharge and damage
to the stomach lining (Movie S5, Supporting Information). In
addition, the magnetic slime can transform into a crescent
shape underwater and accomplish the collection and transport
of solid spherical particles (FigureS9 and Movie S6, Supporting
Information). Magnetic slime can also enter a complex and
narrow maze environment by deforming and extending one
end to achieve the target grasping operation (FigureS10, Sup-
porting Information). Additionally, we quantitatively studied
the relationship between the movement speed of the slime
robot and the strength of the external magnetic field as well
as the relationship between the curling rate and the strength
of the external magnetic field. The experimental results dem-
onstrated that the slime robot could reach the fastest speed of
30 mm s−1 under the drag of a spherical permanent magnet.
When the magnetic field strength was less than 100 mT, the
slime robot could not move as a whole. When a slime robot
curls, the curling rate is proportional to the magnetic field
strength. The higher the magnetic field strength, the faster the
curling rate (Figure S11, Supporting Information). The mag-
netic slime system demonstrates curl and endocytosis modes of
operation that can address multitaskingneeds.
2.4. Self-Healing and Electromechanical Analysis
The non-Newtonian fluid behavior of slime, combined with
the magnetization character of its internal magnetic parti-
cles, endows it with remarkable deformability and reconfigur-
ability. The damage and recovery of hydrogen bonds between
its internal tetrafunctional borate ions and −OH groups exhibit
self-healing capabilities. This crosslinking was easily bound
and dissociated dynamically at room temperature, thus exhib-
iting self-healing at room temperature. This occurs spontane-
ously without the help of external sources such as chemical
reagents. Self-healing is mainly due to the sucient mobility
of the polymer chains within the slime and free tetrafunctional
borate ions, allowing hydrogen bonds at the fracture interface
to rapidly trigger the self-healing process in the absence of
external stimuli.[57] Two slime samples without the addition of
magnetic particles were cut into four pieces and placed together
to demonstrate their self-healing ability. The two samples were
stained with green and blue dyes to distinguish the cut sites;
thereafter, four small pieces of slime were joined together at
intervals to begin healing. Figure4a illustrates that the inter-
face at the cut site is immediately reconnected (less than 1 s).
The healed slime is well connected, maintains excellent plas-
ticity, and withstands major strains even after being stretched
to 8.6 times the original length without damage to the recon-
nected parts. As demonstrated in Figure 4b, a long strip of
magnetic slime can eventually be healed into a circular shape
after being cut into multiple pieces. First, at room tempera-
ture, the striped slime is cut into five small pieces; thereafter,
driven by a magnetic field, the five slimes touch each other one
by one, and finally, the magnetic slime completes the healing
process(Movie S7, Supporting Information). In addition, in the
face of puncture injuries, the slimes can still heal (FigureS12,
Supporting Information). Slime possesses electrical conduc-
tivity properties. Thereafter, we clarified the time of self-healing
Adv. Funct. Mater. 2022, 2112508

www.afm-journal.dewww.advancedsciencenews.com
2112508 (8 of 13) © 2022 Wiley-VCH GmbH
of the magnetic slime by measuring the resistance of the slime
over time during the cutting–healing process. Figure4c shows
repeated cut-repair ten times at the same position of the mag-
netic slime. An open circuit is formed once the magnetic slime
is completely cut and the resistance becomes infinite. When
the broken parts were combined, the resistance dropped rapidly
and returned to near the initial value within 1 s. Magnetic slime
has significant and reproducible electrical repair properties and
exhibits excellent self-healing eciency during each cut and
repair process (FigureS13, Supporting Information). Therefore,
this self-healing ability may also cause electrical healing, that
is, restored conductivity after the damage-healing cycle. Con-
sequently, we connected magnetic slime to the circuit of the
light-emitting diode (LED) bulb to test its electrical self-healing
capability (Figure 4d). As expected, when the magnetic slime
was cut in half, the LED turned o immediately. After putting
the two separated parts back together, the circuit heals quickly
and the light intensity returns to the initiallevel.
The deformation and self-healing capabilities of magnetic
slime allow it to be used as a motion sensor adapted to bending
and dynamic mechanical environments. Considering the pat-
terns of human motion (stretching, bending, and twisting),
the electromechanical properties can be characterized. First,
we encapsulated the magnetic slime between two pieces of
VHB tape and connected the two ends using copper elec-
trodes (Figure4e). The VHB tape can create a strong connec-
tion between the magnetic slime and the copper electrodes to
ensure repeatability during the test and to avoid evaporation
Figure 4. Electromechanical and self-healing analysis. a) The cut damage of slime samples (I and II) is healed in less than 1 s. The healed slime sample
can withstand at least up to 8.6 times the stretch deformation, and the healed areas are as strong as the pristine areas. b) Sequence of images showing
the healing process of magnetic slime. c) Cycling of the electrical self-healing capacity examination. d) Snapshots demonstrating the electrical healing
capacity of magnetic slime. e) A layer of magnetic slime is sandwiched between two layers of VHB tape to create a piezoresistive strain sensor, which
is thereafter connected to two Cu electrodes for tensile, bending, and twisting tests. f) Relative resistance of the magnetic slime as a function of strain.
g) Normalized resistance versus bending angle ranging from 0° to 180°. The inset shows the definition of the bending angle. h) Variation of normalized
resistance versus twist angle. Scale bars, 10 mm.
Adv. Funct. Mater. 2022, 2112508

www.afm-journal.dewww.advancedsciencenews.com
2112508 (9 of 13) © 2022 Wiley-VCH GmbH
of water from the ion-conductive magnetic slime, which could
alter the quality of the electromechanical signal. Figure4f shows
the relative resistance change versus strain, where the resist-
ance increases with increasing strain. The relative resistance
change was calculated as follows: (R-R0)/R0=ΔR/R0, where R
denotes the resistance at the tension state and R0 denotes the
resistance at the initial state. The relative resistance of the mag-
netic slime was 1600% at a strain of 500%, and this response
was reproducible. In addition, a clear hysteresis exists within
the large deformation (500%) stretch–release cycle. However,
negligible hysteresis occurred during 100% strain (FigureS14,
Supporting Information). In both cases, the initial resistance of
the slime completely recovered after its release from the strain.
The response of the magnetic lime-based strain transducer to
deformation with bending and twisting was thereafter tested.
Figure 4g reveals the change in resistance of the magnetic
slime-based strain sensor as a function of the bending angle.
The experimental results demonstrate that as the bending
angle increased from 0° to 180°, the resistance increased
from the original value to 220%. When the lime-based strain
sensor is twisted, the variance in resistance obeys a parabolic
equationfor twist angles less than 540° (Figure4h). However,
at larger torsion angles (over 540°), the magnetic slime sepa-
rates from each other around the torsion point, and the resist-
ance increases rapidly. After three turns (1080°), the resistance
increased from its value in the untwisted state to 1200%. In con-
clusion, the magnetic slimes can be modified into stable and
repeatable piezoresistive strain sensors to maintain their opera-
tion in various mechanically demanding human bodyareas.
2.5. Applications of Magnetic Slime in the Electronic Field
Self-healing and deformable conductive slimes are promising
tools for various electronic devices. First, the electroconduct-
ibility of the magnetic slime allows it to act as a circuit switch,
as shown in the schematic in Figure5a. A three-way circuit is
fabricated, and processing each circuit will cause the red, blue,
and yellow LED bulbs to light up (Movie S8, Supporting Infor-
mation). In the initial state, the three LED bulbs were discon-
nected. Driven by an external magnetic field, the magnetic
slime was controlled to light up the blue, yellow, and red LED
bulbs in sequence. Interestingly, the slime can also transform
into the shape of an octopus and turn on the three circuits
simultaneously to light the three LED bulbs. In addition to
acting as a circuit switch on a 2D plane, slimes can also be used
to repair damaged circuits. Electronic devices may fail because
of partial disconnection if they are corrupted by a long-term
environment or are scratched by sharp objects. As illustrated
in Figure5b, the circuit repair process comprises three steps:
stretching, spreading, and curling. The repaired circuit can
function properly and light up the bulb (Movie S9, Supporting
Information). As shown in Figure5c, the magnetic slimes can
achieve 3D open-circuit connections. In addition, sensors are
an essential component of soft robots. Therefore, we explored
the possibility of a slime robot acting as a sensor. We applied
magnetic slimes encapsulated in VHB tape directly to the skin
of the human body to detect bending and extension behavior,
such as fingers, wrists, shoulders, and elbows. As shown in
Figure 5d, we monitored the variation in the resistance with
respect to the bending angle of the index finger, wrist, shoulder,
and elbow joints. When the index finger was repeatedly bent
from the relaxed state (0°) to the bent state (90°) at a frequency
of ≈1 Hz, the slime resistance changes monitored by the
impedance analyzer exhibited a fast and repeated response to
the finger motion. To detect wrist flexion, the resistance evolu-
tion of the sensor was measured at 45° and 90° bending angles,
respectively. Interestingly, the resistance of the slime-based
sensor increases with increasing bending angle (≈1.15 at 45° and
≈1.35 at 90°), thus allowing dierentiation between dierent
bending angles of the wrist joint. The sensor can also measure
shoulder and elbow movements of the body separately. The pre-
sented sensor signal exhibited good stability during the cyclic
measurements. Magnetic slime can use external deformation
to obtain signals of human motion, and in the future, it may
also use deformation signals to infer the position or stiness
of the robot. Theoretically, the magnetic slime robot combines
motion and sensing, and will have a wider range of application
scenarios in the future. In conclusion, this deformable, electri-
cally conductive magnetic slime can be used in a number of
applications, including wearable devices and softrobots.
3. Conclusions
We proposed reconfigurable multifunctional magnetic slime
robots using spatiotemporally programmed external magnetic
fields. In comparison with existing elastomer-based soft robots,
our approach allows for greater deformation in these slime
robots, which are capable of adapting to extremely constrained
environments, such as passing through a narrow channel of
1.5mm. In comparison with existing liquid-based robots, mag-
netic slime robots can adapt to complex interface environments,
such as water or air, and can also achieve on-demand complex
shape deformations and programmable behaviors such as
curling operations. Various functions have been demonstrated
for these slime robots, including navigation in narrow channels
much smaller than their size, object capture operations via the
curl or endocytosis modes, and circuit repair and controlled
switching using their own conductive properties that can even
be reconfigured as self-healing strain sensors for monitoring
human motion. The widely applicable working environment
of the presented magnetic slime robots, as well as their largely
deformable, reconfigurable, self-healing, and conductive prop-
erties, make them promising for future applications in biomed-
ical and wearabledevices.
4. Experimental Section
Preparation of Magnetic Slime: First, 1 g of PVA was added to 20 mL
of deionized water, and stirred at 90°C for 3 h. Then 10 g of NdFeB
powder (MQP-15-7, Magnequench, Inc.; average diameter 5 μm) was
added to the above solution, and the stirring was continued for 30 min
under sonication. Meanwhile, 1 g of sodium tetraborate was dissolved
with 20 mL of deionized water. Finally, the borax aqueous solution and
the NdFeB/PVA solution were mixed with vigorous stirring at a volume
ratio of 1:4 until a magnetic slime was obtained. The preparation method
of magnetic slime with dierent magnetic particle content was similar,
Adv. Funct. Mater. 2022, 2112508

www.afm-journal.dewww.advancedsciencenews.com
2112508 (10 of 13) © 2022 Wiley-VCH GmbH
just change the ratio of NdFeB. However, the NdFeB particles were
toxic, thus making the magnetic slime not biocompatible. Therefore, the
NdFeB microparticles were coated with SiO2 via the Stöber method.[59]
Typically, 100 mg of NdFeB microparticles were dispersed in 320 mL of
ethanol, 80 mL of DI water, and 4 mL of ammonium hydroxide aqueous
solution (28%) via ultrasonic sonication for 15 min. Then, 0.2 mL
of tetraethyl orthosilicate (98%) was slowly added into the mixture,
followed by vigorous mechanical stirring for 6 h. The microparticles were
then collected using a permanent magnet, washed with DI water five
times, and finally dispersed in 80 mL of DIwater.
Figure 5. Human motion monitoring and circuit control. a) Experimental demonstration of a switch based on the magnetic slime. Inset illustrates
the schematic of employing the magnetic slime as a microswitch. b) Schematic and corresponding experimental results of repairing broken circuit
with magnetic slime. c) Schematic and corresponding experimental results of 3D circuit connections with magnetic slime. d) Movement in dierent
parts of the human body causes changes in resistance. Images illustrating that the magnetic slime sensor can be firmly attached to the fingers, wrists,
shoulders and elbows. Scale bars, 10 mm.
Adv. Funct. Mater. 2022, 2112508

www.afm-journal.dewww.advancedsciencenews.com
2112508 (11 of 13) © 2022 Wiley-VCH GmbH
Characterization: A scanning electron microscope (SEM, JEOL 7600F,
Japan) was used to characterize the microstructure of the freeze-dried
magnetic slime. The magnetic slime was dried by freeze drying (Scanvac
CoolSafe 110−4 PRO 4lt freeze dryer) at −110 °C and 0.091 mbar before
imaging. Rheometer (Malvern Kinexus lab+) was used to measure the
rheological properties ofslime.
Simulation and Computation: Commercial finite element software
COMSOL Multiphysics was used to simulate the shape-changing of
slime robot. The gravity was considered during the simulation, and all
walls were defined as no-slip boundaries. Two electromagnetic coils
of opposite polarity were placed at the two ends of the platform. The
magnetic flux density norm of the left electromagnetic coil was equal
to 200 mT during the whole simulation period, while the magnetic flux
density of the right electromagnetic coil was equal to zero in the initial
state, which gradually increased to 200 mT with time to simulate the
process of a permanent magnet approaching the platform. The phase-
field method was utilized to analyze the deformation of the slime
robot surrounded by another fluid (in this simulation, the surrounding
medium is water). The two-phase flow dynamic was governed by the
Cahn–Hilliard equation, which was divided into two equations
·
1
2
22
tu
φφγλ
εψ
ψε
φφ φ
()
∂
∂+⋅∇=∇∇
=−∇⋅ ∇+ −
(1)
where φ is the dimensionless phase field variable goes from −1 to 1
according to the fluid domain,
u
is the fluid velocity, γ is the mobility, λ
is the mixing energy density, and ε is the interface thickness parameter.
The ψ variable is referred to as the phase-field help variable. The
interface normal between two phases is then calculated as
||
n
φ
φ
=
∇
∇ (2)
The magnetic force exerted on the slime robot was introduced by a
volumetric force as shown in the following equation
FHMB
xHMB
yV
FHMB
xHMB
yV
xxxxyy
zf
yzz
y
yy
y
f
r
r
()
()
()
()
=+
∂
∂++∂
∂





=+
∂
∂++∂
∂












(3)
where
V
fr is the volume fraction of the slime robot. Hx and Hy are
the relative components of magnetic field, Mx and My represent the
magnetization, and Bx and By are the magnetic flux density in x and
ydirections.
In Vitro Cytotoxicity: To evaluate in vitro cytotoxicity, NIH 3T3 cells with
a density of 2000 cells per well were seeded in a 96-well plate, followed
by 12 h incubation in 100 μL Eagle’s Minimum Essential Medium with
10% fetal bovine serum. Then the medium was discarded, and 100 μL
fresh medium containing dierent concentrations of non-magnetic
slime (NdFeB: 0%), magnetic slime (NdFeB: 30%), and magnetic slime
(NdFeB@SiO2: 30%) were added to the NIH 3T3 cells. Subsequently,
these dierent samples were cocultured with the cells for 48 h. The
MTS assay quantified cell viability. 10 μL MTS solution was added to
each well, followed by another 2 h incubation. Then nanoparticles
were concentrated on the bottom with a permanent magnet, and
the supernatant solution was transferred to a new 96-well plate. The
absorbance was detected at 490 nm with a microplate reader. All of the
tests were repeated threetimes.
Magnetic-Actuation Systems: The magnetic drive system in our
work mainly consisted of a robotic arm, a stepper motor, and a
spherical permanent magnet with a diameter of 20 mm. As shown
in the Figure S15A, Supporting Information, the 3-degree-of-freedom
robotic arm was loaded with motor-driven permanent magnets, and
the robotic arm could move to the targeted position autonomously
encoded according to the task needs. Moreover, there was a step
motor at the end of the robotic arm, and the motor can adjust the
speed and direction of the permanent magnet on demand. Thus
the whole system could generate a directional gradient magnetic
field and a non-uniform rotating magnetic field. The magnetic field
intensity distribution of a spherical permanent magnet is shown in
the FigureS15B, Supporting Information, the magnetic field intensity
of the surface of the permanent magnet was about 700 mT, and the
magnetic field intensity at a distance of 10 mm from the permanent
magnet was about 200 mT. In this work, the magnetic field strength
to drive the magnetic slime robot through complex terrain, grasping
and transporting objects, was about 200 mT and was marked in both
the article and the videos. In addition, the magnetic slime robot was
driven for circuit switching control, which was cooperatively controlled
by two robotic arm systems, as shown in Figure S15C, Supporting
Information. The qualitative comparison of the three magnetic soft
materials in question 3 was also realized by the cooperative control of
two manipulators carrying permanentmagnets.
Calculation of Water Content: The water content (%) was calculated
using the following equation[58]
wa
tercontent (%
)1
00%
id
i
WW
W
=
−
× (4)
where Wi is the initial weight of the slime before freeze drying and Wd is
the weight of the freeze-driedslime.
Resistance Measurements: Resistance measurements during stretching
with the CH Instruments Model 700E (CH Instruments, Inc.). A fixed
AC excitation voltage of 2 V amplitude and 100 Hz was used for all
measurements. And a sample with length, width, and height of 3, 2, and
0.1 cm, respectively were used for tensile testing. It was then sandwiched
between two pieces of VHB tape (1 mm thick) and copper electrodes
were attached at both ends (distance: 2 cm). Tensile tests (at a speed
of 2 mm s−1) were performed using a home-made linear actuator. In
addition, the length and width of the samples used in the twist and bend
measurements were 4, 2, and 0.1 cm,respectively.
Human Motion Sensing: For human motion monitoring, the the CH
Instruments Model 700E was used. Samples for finger and wrist motion
were 3 cm long, and samples for knee, elbow, and shoulder motion
were 7 cm long. Informed consent was obtained from the volunteer (Xin
Wang) for the motion sensingexperiments.
Supporting Information
Supporting Information is available from the Wiley Online Library or
from the author.
Acknowledgements
This work was partially supported by the National Natural Science
Foundation of China (grant no. 61925304). The research work was
financially supported by the Hong Kong Research Grants Council
(RGC) with project Nos. JLFS/E-402/18 and E-CUHK401/20, the RGC
Collaborative Research Fund (CRF) with Project Nos. C4063-18GF
and C1134-20GF; the ITF project with Project Nos. MRP/036/18X
and ITS/374/18FP funded by the HKSAR Innovation and Technology
Commission (ITC); the Croucher Foundation Grant with Ref. No.
CAS20403, and the CUHK internal grants. The authors also thank the
SIAT-CUHK Joint Laboratory of Robotics and IntelligentSystems.
Conflict of Interest
The authors declare no conflict ofinterest.
Adv. Funct. Mater. 2022, 2112508

www.afm-journal.dewww.advancedsciencenews.com
2112508 (12 of 13) © 2022 Wiley-VCH GmbH
Author Contributions
M.S. conceived and designed the processing approach. M.S. conducted the
experiments and wrote the manuscript with input from the other authors.
M.S. and X.W. co-analyzed the experimental and calculated data. B.H.
completed the simulation of slime deformation. H.X. and L.Z. mentored
the work and revised the manuscript. All authors wrote thepaper.
Data Availability Statement
The data that support the findings of this study are available from the
corresponding author upon reasonablerequest.
Keywords
circuit control, environmental adaptability, large deformation, magnetic
slime, manipulation, self-healing
Received: December 7, 2021
Revised: March 12, 2022
Published online:
[1] V. K. Bandari, Y. Nan, D. Karnaushenko, Y. Hong, B. Sun,
F. Striggow, D. D. Karnaushenko, C. Becker, M. Faghih,
M.Medina-Sánchez, O. G.Schmidt, Nat. Electron. 2020, 3, 172.
[2] Y.Zhao, C.Xuan, X.Qian, Y.Alsaid, M.Hua, L.Jin, X.He, Sci. Rob.
2019, 4, eaax7112.
[3] X. Qian, Y. Zhao, Y. Alsaid, X. Wang, M. Hua, T. Galy,
H.Gopalakrishna, Y.Yang, J.Cui, N. Liu, M.Marszewski, L.Pilon,
H.Jiang, X.He, Nat. Nanotechnol. 2019, 14, 1048.
[4] S. Tang, F. Zhang, H. Gong, F. Wei, J. Zhuang, E. Karshalev,
B. Esteban-Fernandez de Avila, C.Huang, Z. Zhou, Z. Li, L. Yin,
H.Dong, R. H.Fang, X.Zhang, L. Zhang, J.Wang, Sci. Rob. 2020,
5, eaba6137.
[5] W. Gao, S. Sattayasamitsathit, K. M. Manesh, D. Weihs, J. Wang,
J. Am. Chem. Soc. 2010, 132, 14403.
[6] B. Wang, K. F. Chan, K. Yuan, Q. Wang, X. Xia, L. Yang, H.Ko,
Y. X. J. Wang, J. J. Y.Sung, P. W. Y.Chiu, L. Zhang, Sci. Rob. 2021,
6, eabd2813.
[7] J.Wang, W.Gao, ACS Nano 2012, 6, 5745.
[8] S.Fusco, M. S.Sakar, S.Kennedy, C.Peters, R.Bottani, F.Starsich,
A. Mao, G. A. Sotiriou, S. Pané, S. E. Pratsinis, D. Mooney,
B. J.Nelson, Adv. Mater. 2014, 26, 952.
[9] K. E.Peyer, L.Zhang, B. J.Nelson, Nanoscale 2013, 5, 1259.
[10] J.Li, B.Esteban-Fernández de Ávila, W.Gao, L.Zhang, J.Wang, Sci.
Rob. 2017, 2, eaam6431.
[11] X.Wang, X. H.Qin, C.Hu, A.Terzopoulou, X. Z.Chen, T. Y.Huang,
K.Maniura-Weber, S.Pané, B. J.Nelson, Adv. Funct. Mater. 2018, 28,
1804107.
[12] B.Wang, K.Kostarelos, B. J.Nelson, L.Zhang, Adv. Mater. 2021, 33,
2002047.
[13] M. Sun, X. Fan, X. Meng, J. Song, W. Chen, L. Sun, H. Xie,
Nanoscale 2019, 11, 18382.
[14] D. Jin, K. Yuan, X.Du, Q.Wang, S. Wang, L. Zhang, Adv. Mater.
2021, 33, 2100070.
[15] T. G. Leong, C. L. Randall, B. R. Benson, N. Bassik, G. M. Stern,
D. H.Gracias, Proc. Natl. Acad. Sci. USA 2009, 106, 703.
[16] J. Li, X. Li, T. Luo, R. Wang, C. Liu, S. Chen, D. Li, J. Yue,
S.-H.Cheng, D.Sun, Sci. Rob. 2018, 3, eaat8829.
[17] S. Jeon, S. Kim, S. Ha, S. Lee, E. Kim, S. Y. Kim, S. W. Park,
J. H. Jeon, S. W. Kim, C. Moon, B. J. Nelson, J. Y. Kim, S. W.Yu,
H.Choi, Sci. Rob. 2019, 4, eaav4317.
[18] Y.Dong, L.Wang, N.Xia, Y.Wang, S.Wang, Z. Yang, D.Jin, X.Du,
E.Yu, C.Pan, B.Liu, L.Zhang, Nano Energy 2021, 88, 106254.
[19] J.Yu, D.Jin, K. F.Chan, Q.Wang, K.Yuan, L.Zhang, Nat. Commun.
2019, 10, 5631.
[20] J.Yu, B.Wang, X.Du, Q.Wang, L.Zhang, Nat. Commun. 2018, 9, 3260.
[21] H. Xie, M. M. Sun, X. J. Fan, Z. H. Lin, W. N. Chen, L. Wang,
L. X.Dong, Q.He, Sci. Rob. 2019, 4, eaav8006.
[22] Q. Wang, K. F. Chan, K. Schweizer, X. Du, D. Jin, S. C. H. Yu,
B. J.Nelson, L.Zhang, Sci. Adv. 2021, 7, eabe5914.
[23] M.Sun, X.Fan, C.Tian, M.Yang, L.Sun, H.Xie, Adv. Funct. Mater.
2021, 31, 2011193.
[24] M.Sun, Q.Liu, X. Fan, Y.Wang, W. Chen, C.Tian, L.Sun, H. Xie,
Small 2020, 16, 1906701.
[25] M.Sun, W.Chen, X.Fan, C.Tian, L.Sun, H.Xie, Appl. Mater. Today
2020, 20, 100682.
[26] W.Hu, G. Z.Lum, M.Mastrangeli, M.Sitti, Nature 2018, 554, 81.
[27] T. Xu, J. Zhang, M. Salehizadeh, O. Onaizah, E. Diller, Sci. Rob.
2019, 4, eaav4494.
[28] X.Du, H.Cui, T.Xu, C.Huang, Y.Wang, Q.Zhao, Y.Xu, X.Wu, Adv.
Funct. Mater. 2020, 30, 1909202.
[29] X.Kuang, S.Wu, Q.Ze, L. Yue, Y.Jin, S. M. Montgomery, F.Yang,
H. J.Qi, R.Zhao, Adv. Mater. 2021, 33, 2102113.
[30] T.Wang, Z.Ren, W.Hu, M.Li, M.Sitti, Sci. Adv. 2021, 7, eabf7364.
[31] H.Lu, M. Zhang, Y. Yang, Q. Huang, T. Fukuda, Z.Wang, Y.Shen,
Nat. Commun. 2018, 9, 3944.
[32] Y. Cheng, K. H.Chan, X. Q. Wang, T.Ding, T.Li, C.Zhang, W.Lu,
Y.Zhou, G. W.Ho, Adv. Funct. Mater. 2021, 31, 2101825.
[33] J.Zhang, Z.Ren, W.Hu, R. H.Soon, I. C.Yasa, Z.Liu, M.Sitti, Sci.
Rob. 2021, 6, eabf0112.
[34] Y.Kim, G. A.Parada, S.Liu, X.Zhao, Sci. Rob. 2019, 4, eaax7329.
[35] H.-W.Huang, M. S.Sakar, A. J.Petruska, S.Pané, B. J.Nelson, Nat.
Commun. 2016, 7, 12263.
[36] J. D.Carlson, M. R.Jolly, Mechatronics 2000, 10, 555.
[37] A. G.Olabi, A.Grunwald, Mater. Design 2007, 28, 2658.
[38] J.Kim, S. E.Chung, S. E.Choi, H.Lee, J.Kim, S.Kwon, Nat. Mater.
2011, 10, 747.
[39] M.Zrinyi, L.Barsi, A.Büki, Polym. Gels Networks 1997, 5, 415.
[40] R. Fuhrer, C. M. Schumacher, M. Zeltner, W. J. Stark, Adv. Funct.
Mater. 2013, 23, 3845.
[41] V. Q.Nguyen, A. S.Ahmed, R. V.Ramanujan, Adv. Mater. 2012, 24,
4041.
[42] G. Z.Lum, Z.Ye, X.Dong, H.Marvi, O.Erin, W.Hu, M.Sitti, Proc.
Natl. Acad. Sci. USA 2016, 113, E6007.
[43] Y.Kim, H.Yuk, R.Zhao, S. A.Chester, X.Zhao, Nature 2018, 558, 274.
[44] J.Čejková, T. Banno, M. M.Hanczyc, F. Štěpánek, Artif. Life 2017,
23, 528.
[45] X.Wang, R.Guo, J.Liu, Adv. Mater. Technol. 2019, 4, 1800549.
[46] F. Li, J. Shu, L. Zhang, N. Yang, J. Xie, X. Li, L. Cheng, S. Kuang,
S.-Y.Tang, S.Zhang, W.Li, L.Sun, D.Sun, Appl. Mater. Today 2020,
19, 100597.
[47] S. Chen, H.-Z. Wang, R.-Q.Zhao, W. Rao, J. Liu, Matter 2020, 2,
1446.
[48] X.Zhang, L.Sun, Y.Yu, Y.Zhao, Adv. Mater. 2019, 31, 1903497.
[49] W. Yu, H.Lin, Y.Wang, X.He, N.Chen, K. Sun, D. Lo, B. Cheng,
C. Yeung, J. Tan, D. Di Carlo, S. Emaminejad, Sci. Rob. 2020, 5,
eaba4411.
[50] M.Zhou, Z.Wu, Y. Zhao, Q.Yang, W. Ling, Y.Li, H.Xu, C. Wang,
X.Huang, Adv. Sci. 2019, 6, 1901862.
[51] F.Serwane, A.Mongera, P.Rowghanian, D. A.Kealhofer, A. A.Lucio,
Z. M.Hockenbery, O.Campas, Nat. Methods 2017, 14, 181.
[52] A. Mongera, P. Rowghanian, H. J. Gustafson, E. Shelton,
D. A.Kealhofer, E. K. Carn, F. Serwane, A. A.Lucio, J.Giammona,
O.Campás, Nature 2018, 561, 401.
[53] X.Fan, X.Dong, A.Karacakol, H.Xie, M.Sitti, Proc. Natl. Acad. Sci.
USA 2020, 117, 27916.
Adv. Funct. Mater. 2022, 2112508

www.afm-journal.dewww.advancedsciencenews.com
2112508 (13 of 13) © 2022 Wiley-VCH GmbH
[54] L.Hu, H.Wang, X. Wang, X.Liu, J. Guo, J. Liu, ACS Appl. Mater.
Interfaces 2019, 11, 8685,
[55] F.Li, S.Kuang, X.Li, J. Shu, W.Li, S.-Y.Tang, S.Zhang, Adv. Mater.
Technol. 2019, 4, 1800694.
[56] X.Li, S.Li, Y.Lu, M. Liu, F.Li, H.Yang, S.-Y.Tang, S.Zhang, W.Li,
L.Sun, ACS Appl. Mater. Interfaces 2020, 12, 37670.
[57] G.Cai, J. Wang, K.Qian, J. Chen, S.Li, P. S.Lee, Adv. Sci. 2017, 4,
1600190.
[58] M.Karolina Pierchala, F. B. Kadumudi, M.Mehrali, T. G.Zsurzan,
P. J. Kempen, M. P. Serdeczny, J. Spangenberg, T. L. Andresen,
A.Dolatshahi-Pirouz, ACS Nano 2021, 15, 9531.
[59] W.Stöber, A.Fink, E.Bohn, J. Colloid Inter. Sci. 1968, 26, 62.
Adv. Funct. Mater. 2022, 2112508