Article https://doi.org/10.

1038/s41467-023-36909-y

Single-vat single-cure grayscale digital light

processing 3D printing of materials with
large property difference and high
stretchability

Received: 17 November 2022 Liang Yue 1, S. Macrae Montgomery 1, Xiaohao Sun1, Luxia Yu1, Yuyang Song2,
Tsuyoshi Nomura3, Masato Tanaka2 & H. Jerry Qi 1
Accepted: 23 February 2023
1234567890():,;
1234567890():,;

Multimaterial additive manufacturing has important applications in various

Check for updates emerging fields. However, it is very challenging due to material and printing
technology limitations. Here, we present a resin design strategy that can be
used for single-vat single-cure grayscale digital light processing (g-DLP) 3D
printing where light intensity can locally control the conversion of monomers
to form from a highly stretchable soft organogel to a stiff thermoset within in a
single layer of printing. The high modulus contrast and high stretchability can
be realized simultaneously in a monolithic structure at a high printing speed
(z-direction height 1 mm/min). We further demonstrate that the capability can
enable previously unachievable or hard-to-achieve 3D printed structures for
biomimetic designs, inflatable soft robots and actuators, and soft stretchable
electronics. This resin design strategy thus provides a material solution in
multimaterial additive manufacture for a variety of emerging applications.

Additive manufacturing (AM), or 3D printing, allows the fabrication of that require high stretchability. There have also been studies that mix
structures with geometric and material complexities far beyond what multiple materials prior to extrusion5,14,15 or employ multiple
can be created using traditional manufacturing techniques. New 3D nozzles16–18 in direct ink write (DIW) printing, but DIW printing is
printing capabilities have started to be used for functional applications relatively slow and has limited accuracy compared to other AM tech-
in deployable structures1, soft robotics2, flexible electrical niques. When multiple materials cannot be combined using one AM
components3,4, and biomimetic designs5,6. Many of these applications process, the desired structures are often achieved by using multiple
require the use of materials with vastly different properties, such as fabrication methods2,19,20. For example, DIW printing was used to write
nature-like structures7, airless tires (https://michelinmedia.com/ stiff fibers on prefabricated (by molding or 3D printing) soft elastomer
michelin-uptis/)8, multistable absorbers9,10, and 4D printing8–10. But structures16,21. Previously reported programmable curvatures of tube-
fabricating highly stretchable soft materials and stiff materials in one like actuators with multimaterials were fabricated similarly by using
AM process with high efficiency is very challenging. multi-step processing with different techniques16,19,22. Overall, the
Inkjet printing (IJP) is one of the most popular methods to achieve multi-ink or multi-method AM techniques are often not efficient and
multimaterials in a single printed part11,12 and have been extensively have limited capability to fabricate structures with multimaterial
used in the past, although the cost for IJP printers with multimaterial properties with complicated distributions.
capabilities is high. In addition, most of the IJP printable materials fail Digital light processing (DLP) is a high-speed and high-resolution
at low to moderate strain13, making them unsuitable for applications printing method and has become increasingly popular in recent years.

1
The George W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, GA 30332, USA. 2Toyota Research Institute of North
America, Toyota Motor North America, Ann Arbor, MI 48105, USA. 3Toyota Central R&D Laboratories, Inc., Bunkyo-ku, Tokyo 112-0004, Japan.
e-mail: [email protected]

Nature Communications | (2023)14:1251 1

Article https://doi.org/10.1038/s41467-023-36909-y

It uses a projector to irradiate a thin layer of resin with images of the Results
cross-sections of a solid part. In a typical DLP printing, a single resin vat Multi-properties of single vat g-DLP printing
is used, and only z-direction motion is needed to move the build plate. The photocurable resin ink is rationally designed. As illustrated in
Photopolymerization (or photocuring) can occur in a few seconds. Fig. 1a, it consists of the monofunctional isobornyl acrylate (IBOA) and
These make DLP one of the fastest 3D printing techniques. None- 2-hydroxyethyl acrylate (2-HEA) as the linear chain builders as well as
theless, the use of a single resin vat makes DLP, in general, not suitable the aliphatic urethane diacrylate (AUD) as the crosslinker and the
for printing parts with multiple-material properties. Methods using reactive diluent (Fig. 1b–d and Supplementary Information (SI) Fig. 1 for
multiple vats have been developed to print two or more materials by chemical structures). The formulated ink has a low viscosity of around
transferring a part between multiple vats23,24. Great care must be taken 0.05 Pa∙s (Supplementary Fig. 2). AUD is a viscous oligomer with high
to avoid cross-contamination. Switching resin vats and cleaning sig- molecular weight aliphatic chains and urethane units. It forms H-N∙∙∙O
nificantly slow down the printing speed. In addition, the number of hydrogen bonds when interacting with 2-HEA and IBOA monomers,
available material properties in multi-vat approaches is limited by the meanwhile, 2-HEA has abundant -OH groups that form additional O-
number of vats, making it hard to obtain a continuous transition of H∙∙∙O hydrogen bonds as shown in Supplementary Figs. 3, 4. At low
material properties. Boydston group developed a single-vat multiple- DoC, the covalent network with the prevalent hydrogen bonds provides
material printing method by using two different initiators and two high stretchability at the soft state, as shown in Fig. 1b. At high DoC, the
different light sources5. However, the quality of printed parts relies on stiff IBOA brings up the glass transition temperature (Tg) to be well
the alignment of two light sources and the tunable property range is above room temperature, yielding the glassy behaviors with the high
limited in a single vat. modulus (Fig. 1c). Furthermore, the DoC can be conveniently controlled
Recently, grayscale digital light processing (g-DLP) has emerged by light intensity, thus enabling the g-DLP printing to achieve drastic
as a promising strategy to obtain a wide variety of material mechanical property differences within a single layer of printing.
properties25,26 and color27 from a single resin material. In g-DLP print- We use a bottom-up DLP printer where light is projected from the
ing, the local degree of monomer conversion is controlled by light bottom of the vat (Fig. 1e). As a demonstration, the design of a snail is
intensity, which can be readily manipulated at pixel level by the input sliced and processed into 2D grayscale images (Fig. 1e) using a
grayscale image. To obtain a large range of material properties, a MATLAB script (MathWorks, Natick, MA, USA), which controls the
special ink that contains bisphenol A ethoxylate diacrylate(BPADA), local property using the grayscale (or the ultraviolet (UV) light inten-
glycidyl methacrylate(GMA), and Jeffamine was developed in combi- sity) of each pixel. The grayscale images are then projected onto the
nation with a two-stage cure (or dual cure) strategy25. The first pho- ink-vat window from the bottom, initiating free radical polymerization.
tocuring with controlled grayscale during 3D printing fixes the shape The oxygen-permeable Teflon (PTFE) window enables the easy
with a slight difference in the mechanical properties (modulus differ- separation of the cured polymer. This bottom-up continuous liquid
ence of 10–20 times). The second stage, thermally curing the printed interface production (CLIP)30 approach allows for a rapid printing
part, boosts the differences in the material properties. By this method, speed of 1 mm/min (3 s per 0.05 mm layer). The locally modulated UV
Young’s modulus, which spans nearly 1000 times, was achieved. It intensity results in different DoCs, and thereby different mechanical
requires a long second-stage thermal curing. In addition, the final properties, throughout the printed part. As shown in Fig. 1f, the printed
printed material has low stretchability (less than 35% in the soft state), snail possesses an integrated stiff shell (using 100% light intensity) and
thus limiting their functional applications. Besides, single-cured g-DLP a soft body (using 40% light intensity). The shell holds 1 kg weight
was also used to modulate the printed properties26,28,29. It generally without visible deformation, while the body can be easily stret-
relies on varying the monomer conversion with different light to ched by 400%.
achieve the properties difference. However, it still remains a significant The different material properties are controlled by the grayscale
challenge to achieve high property difference and high stretchability level of the UV light, which varies from 0% (full intensity, labeled as G0)
simultaneously in efficient single-step printing. to 100% (full dark, labeled as G100). The correspondence between
In this paper, we present a design strategy of resins for single- grayscale and light intensity is presented in Supplementary Fig. 5. The
stage g-DLP printing that possesses a very wide range of modulus as photopolymerization kinetics is studied with a photopolymerization
well as a very high stretchability. We use three different monomers in (PP) model31 to analyze the correlation between the depth-dependent
the resin where a stiff monomer helps to increase the stiffness at a DoC at light does (SI, Section 1). Based on our slicing thickness of
high degree of cure (DoC) and two soft monomers with abundant 50 μm and using the theoretically predicted correlation (Supplemen-
side groups for hydrogen bonds facilitate high stretchability at the tary Fig. 6) and experiments, we use a grayscale range from G0 level to
low DoC. The DoC can be readily manipulated by the light intensity in G70 (70% darkness) to ensure a rapid printing speed and good shape
the g-DLP printing. The single-vat single-cure printed polymer can fidelity. At G0, the light intensity is 24.4 mW/cm2 and the DoC is 94%
have a modulus ranging from 0.016 to 478 MPa (nearly 30,000 times (determined by FTIR measurement; Supplementary Fig. 7); at G70, the
difference) with a stretchability of up to 1500% at the soft state. light intensity is 1.4 mW/cm2, and the resulting DoC is 55%.
Besides, the printing can be conducted at high printing speed (3 s per The mechanical and thermomechanical properties of printed
0.05 mm layer, giving a build rate of 1 mm/min), and multiple parts samples with different grayscale levels are evaluated by uniaxial tensile
can be printed simultaneously. To our best knowledge, this cap- tests and dynamic mechanical analysis (DMA; see Materials and
ability of rapid printing with large property differences and high Methods section for details). As shown in Fig. 2a, the printed polymer
stretchability has never been achieved before. The wide range of gradually becomes softer from G0 down to G50, with Young’s modulus
properties outperforms most inks used in inkjet 3D printing, multi- of 478 MPa at G0. The stiff state shows superior toughness, which is
vat, as well as single-vat DLP printing. The soft state stretchability is around 109 J/m3. The fracture toughness is also measured by the
among the best performance in single-material single-property tearing test and ranges from 650 to 10,000 J/m2 (Supplementary
printing. We further demonstrate such superiorities for the fabrica- Fig. 8). The polymer in the rubbery state using G60 and G70 possesses
tion of various monolithic structures from anisotropic composites, a modulus of 0.12 and 0.016 MPa, respectively, and can be stretched by
biomimetic designs, inflatable structures, and customized wearable near 1500% (Fig. 2b). The extensive presence of hydrogen bonding
sensors. This capability for rapid prototyping of parts with dramatic with crosslinked network makes the printed part in a stable organogel
property difference and superior stretchability thus has the potential state with a gel fraction around 40 and 25% for G60 and G70 (Sup-
for applications in pre-surgical models, soft robots, wearable plementary Figs. 9, 10). The soft polymer also exhibits excellent elastic
electronics, etc. properties and resilience even after 10,000 cycles of cyclic stretching

Nature Communications | (2023)14:1251 2

Article https://doi.org/10.1038/s41467-023-36909-y

a
Low light intensity High light intensity
Soft, rubbery Stiff, glassy

55% Conversion 94%

b c

d
AUD: 2-HEA: IBOA:

Hydrogen bond

e f
Grayscale

1 Kg

Fig. 1 | Using light to control mechanical properties in g-DLP printing. bottom-up g-DLP 3D printing; f A g-DLP printed snail with a hard shell that can
a Schematic illustration of the crosslink density and properties with different support 1 kg weight and soft neck that can be stretched by 400%. The scale
conversions. b, c Schematic illustrations of the hydrogen bonding and crosslinking bar is 1 cm.
of the polymer network. d Chemical structure of the monomers; e Schematic of

at a high strain between 200 and 300% (Fig. 2c, d). To validate the DoC, we test a resin by replacing AUD and 2-HEA with PEGDA (poly(-
contribution of 2-HEA monomer to the elastic stretchability, we print a ethylene glycol) diacrylate, Mw ~700) and BA (butyl acrylate). In this
reference sample without 2-HEA with the same DoC. The reference material, low DoC results in fewer crosslinking points and shorter lin-
sample (with only AUD and IBOA) exhibits similar rubbery behavior ear chains. Without hydrogen bonding between the uncured mono-
but much lower elongation at break (~500%, Supplementary Fig. 11). mers and the crosslinked network, the samples are weak with low
After the fatigue test with 10,000 cycles in the same strain range, the stiffness and stretchability (Supplementary Fig. 13). We also observe
residual strain is observed (Supplementary Fig. 11). There is an appar- the same results in previously reported g-DLP resin that low DoC leads
ent stress decrease due to network relaxation (Supplementary Fig. 12), to weak mechanical properties as expected25.
and the network is unable to fully recover in the absence of 2-HEA The stiff monomer IBOA brings up the Tg at high DoC and makes
monomer. To further validate the function of hydrogen bonding at low the network stiff, which ensures the high modulus contrast at different

Nature Communications | (2023)14:1251 3

Article https://doi.org/10.1038/s41467-023-36909-y

a b 0.3
G60
G70

Stress (MPa)
0.2

0.1

0.0
0 500 1000 1500
Strain (%)
c d
G60 printed
0.20
0.12 Strain=200% to 300%
20 mm

Stress (MPa)
After 10000 cycles
20 mm 0.15
0.10

0.10
0.08

100 1000 2000 4000 8000 10000

Number of cycles
e 10 3 100
f
10 3 Hard thermosets
Y oung` s modulus (MPa)

80
Y oung` s modulus (MPa)

10 2
10 2
60 2-stage g-DLP
10 1 Stiff, glassy
Tg (oC)

40 101

100 100
This work
20

10-1 0 10-1
Organo/hydrogels
Soft, rubbery Elastomers
10-2
-20 10-2
0 10 20 30 40 50 60 70 0 400 800 1200 1600
Grayscale (%) Elastic elongation (%)

Fig. 2 | Characterization of the printed material. a Stress-strain curves of G50 to e Correlation of Young’s modulus and glass transition temperature with grayscale
G0; b Stress-strain curves of G60 and G70; c Schematic of recovering of the net- levels; f A comparison of the mechanical property range between the single vat
work and photo comparison of G60 sample before and after 10000 fatigue cycles; g-DLP printing with general reported or commercially available DLP ink20,25,35–49.
d Fatigue test of G60 sample with an applied strain between 200 to 300%;

DoC. Figure 2e summarizes Young’s moduli and Tgs from DMA (Sup- soft states, which significantly limited the functionalities of the printed
plementary Figs. 14, 15) at different grayscales, showing Young’s mod- structures. Our resin overcomes these limitations by offering much
ulus contrast between stiff G0 and soft G70 nearly 30,000 times. higher modulus contrast and, more importantly, much higher
Figure 2f shows a comparison of our resin with reported DLP materials stretchability through a single stage and single ink printing.
in literature, including two-stage cured hard thermoset, soft elastomers, Combining highly stretchable elastomers and stiff glassy poly-
hydrogels, and organogels20,25,32–46. As can be seen, our single resin mers into one structure can significantly enhance DLP 3D printing for
covers the widest range of modulus and elastic elongation. It was also structures and devices that not only require complex geometries but
notable that the transition between the stiff thermoset and rubbery gel also have integrated heterogeneous properties to meet demanding
state was not continuous in this plot. This was because the printing material requirements in applications such as soft robots, actuators,
properties were too sensitive to control at this threshold DoC range. flexible electronics, biomimetic, pneumatic structures, etc.
In our previous two-stage g-DLP printing, the modulus difference
was less than 20 times after printing and required further thermal g-DLP printing for composites structures
curing to reach close to 1000 times modulus contrast25. Additionally, To verify the capability of the single vat multi-properties g-DLP printing,
the stretchability of the two-stage cured sample was less than 35% in we print a sample with continuous gradient grayscale from G0 to G70.

Nature Communications | (2023)14:1251 4

Article https://doi.org/10.1038/s41467-023-36909-y

Fig. 3 | g-DLP printed composite structures. a Fiber-embedded composites with number refers to the turn number of the helical fibers and M refers to the matrix
anisotropic behaviors; b Airless tire with rubbery outer circumference and rigid without any fiber); d Composite designed with sequential deformation. All scale
inner hub; c Collagen structure with helical fibers with different pitch numbers (the bar is 1 cm.

The sample is expected to have a continuous Tg transition from ~90 to ~ Such a J-shape curve is important for the functions of artery tissues in
−10 °C. Thus, we exploit the shape memory feature of thermoset poly- regulating the blood flow: at low pressure, the artery can expand
mer to program the sample into a helix shape at 90 °C and fix the shape easily to allow more blood flow; as it expands beyond a certain limit,
by rapid cooling in ice water. Afterward, the programmed sample is it stiffens to restrict the amount of the flow. This behavior is due to
dropped into an 80 °C water bath for free recovery. The sample exhibits the structure of an artery tissue, which is a soft elastin matrix
sequential shape recovery from the low Tg region to the high Tg region, embedded with stiff and tortuous collagen fibers. At a low stretch of
which reflects the smooth transition of multi-properties. This experi- the tissue, the tortuous collagen fibers provide little deformation
ment is presented in Supplementary Movie 1 and Supplementary Fig. 16. resistance (due to bending) until they are straightened and become
Moreover, we further examine the interface and adhesion between the axially stretched at a large tissue stretch ratio49,50. Here, we mimic the
stiff thermoset and rubbery organogel. As the different curing states are structure of artery tissue by using the soft elastin matrix (G60) and
all generated from the same resin, the adhesion is expected to be good. embedded stiff helical fibers (G0) (Fig. 3c). Cylindrical samples with
Meanwhile, the inherent transition between different curing states due straight (T0), helical fibers of different pitches (T2, T6, T8, and T10;
to diffusion enhances the adhesion between different domains of the the digital value refers to the number of turns of the helical fibers),
g-DLP printed structures. As presented in Supplementary Fig. 17, the and without fibers (M) are printed. Uniaxial tensile tests show that the
energy needs to break the interface between the soft (G60) and stiff printed structures with helical fibers exhibit the J-shape behavior, i.e.,
(G0) domains is greater than the energy needs to break the soft part it initially has low stiffness, which gradually increases, then has high
itself, demonstrating good adhesion. stiffness, corresponding to the straightening of the stiff fibers.
Taking advantage of the drastic mechanical property contrast and Adjusting the pitch size of fibers (equivalent to tortuosity) can tune
large stretchability, we first 3D print structures that demonstrate the strain range for fiber straightening. For example, T10 shows the
composite-like behaviors. Figure 3a shows a printed composite with most obvious transition (Fig. 3c). These results indicate that g-DLP
stiff fibers (G0) embedded in a flexible matrix (G60), which exhibits a printing can be used for prototyping artificial tissues that mimic real
high degree of anisotropy. Perpendicular to the fiber direction, the tissue behaviors.
composite is too soft to hold its own weight (about 1 g), and a 200 g We can also use multiple grayscale levels to achieve sequential
weight stretches the composite more than two times. In the parallel deformation responses under increasing force levels. Figure 3d
direction, the composite holds the same 200 g weight without obser- shows an example of spring-shape stiff fibers (G0) embedded in three
vable deformation. Figure 3b shows the prototyping of an airless tire sections of a rubbery matrix with different stiffness assigned (G46,
with a compressible rubbery (G60) outer circumference for shock- G56, and G66) and connected in series, which can achieve the
absorbing. The inner hub structure (G0) is rigid. Such an integrated sequential deformation of three sections under the uniaxial tension.
structure allows the tire to deform when a force is applied in the ver- The left graphic displays the overall strain change of the entire
tical direction (mimicking a bump on the ground) and to return structure and the right one shows the strain change of each section. A
completely to its original state after the force is removed. clear time delay of deformation for each section is observed (Sup-
Next, we design composites that mimic artery tissues, whose plementary Movie 2), with G46 (the stiffest) showing deformation at
stress-strain behaviors exhibit a characteristic J-shaped curve47,48. a later stage.

Nature Communications | (2023)14:1251 5

Article https://doi.org/10.1038/s41467-023-36909-y

Fig. 4 | g-DLP printed biomimetic structures. a Artificial heart valve, scale bar is bar is 5 mm; d Comparison of fracture toughness, insert are samples after test,
1 cm; b Fish fin structure, scale bar is 1 cm; c Nacre and Bouligand structure, scale green line: nacre; blue line: Bouligand.

g-DLP printing for biomimetic structures significantly improved fracture toughness by mimicking nacre and
Biomimetic structures consisting of soft and hard phases are difficult Bouligand structures (Fig. 4c). In our design, stiff flakes or fibers (G0)
to fabricate by conventional DLP methods with a single vat. Here, we are held by the soft interface (G70) that elongates the crack propa-
exploit the capability of printing such biomimetic structures in a gation pathways and dissipates energy. Specific parameters of the
monolithic piece through our single-vat g-DLP printing. Figure 4a structures like flake/fiber size, aspect ratio or detailed arrangement of
shows a printed artificial human heart valve consisting of a rigid sup- the two phases, etc., can be tuned to optimize the toughness of the
port (G0) with three soft valve flaps (G60). The still images in Fig. 4a printed composites (Supplementary Figs. 18, 19). Figure 4d shows the
show the transition between the closed and open states under differ- three-point bending results of the printed biomimetic structures, with
ent fluid flow conditions. The hydrodynamic performance of the g-DLP the nacre-like structure having the dimension of 48 mm (L) by 6 mm
printed heart valves is presented in Supplementary Movie 3. We also (W) by 6 mm (t) and 51% of soft materials and the Bouligand structure
simulate the hydrodynamic performance of the valves by finite ele- with the dimension of 56 mm (L) by 6 mm (W) by 6 mm (t) and 48% of
ment analysis (FEA) using a pressure load on the inside face of the flaps. soft materials. The three-point bending apparatus has a span of 32 mm.
More details can be found in the SI, Section 2. The simulation results The fracture toughness is 166 and 146 kPa m0.5 for the nacre and the
match well with the experiments. This fast, customizable printing Bouligand structure, respectively, which are three to four times higher
method can be an efficient approach to printing patient-specific heart than the direct cured bulk sample (42 kPa m0.5). The g-DLP printing
valve models for pre-surgical planning. Figure 4b shows a printed fish thus demonstrates the capability for concept proofing of various
fin structure with stiff bony rays (G0) attached to the common fin biomimetic structures.
base (G0) and held by the flexible tissue membrane (G60). Fishes can
change the shape of their fins by pushing or pulling the base of the g-DLP printing of inflatable structures
bony rays by muscles and tendons51,52. Our printed model can mimic Next, we demonstrate the rapid fabrication of complex inflatable
this shape change, as shown in Fig. 4b and Supplementary Movie 4. structures, which have found important applications in soft robots in
The fish fin actuation is modeled by FEA using a pressure load normal recent years54. The design space for inflatable structures is often
to the base to cause the opening and closing. The out-of-plane limited by the use of only one material. To achieve more complicated
bending of the membrane at the folding state and in-plane bending of shape change using one material, one has to rely on complex
the bony rays at expanding state are well-matched with the experi- geometries55. Using multiple materials to achieve complex inflations
mental results. The g-DLP printed fish fin structure accurately repli- via 3D printing typically requires multiple-material reservoirs or even
cates such a mechanism, facilitating the rapid biomimetic soft a combination of printing technologies56–58. Using this resin for-
robotic structure fabrication. mulation, stiff inclusions can easily be added anywhere in a soft,
We further design biomimetic structures with complex multi- stretchable matrix in a sing-vat printing. This opens the door to a
material architectures. Nacre and Bouligand structures are composites wide array of possible inflatable structure designs that cannot be
that are well known for their intelligent designs to effectively improve easily achieved using other techniques. Examples of these inflatable
fracture toughness by using hard and soft materials53. Here, we present structures are presented in Fig. 5. The pressure used in these exam-
the capability of our resin with g-DLP to print structures that exhibit ples is 30 kPa.

Nature Communications | (2023)14:1251 6

Article https://doi.org/10.1038/s41467-023-36909-y

Fig. 5 | g-DLP printed pneumatic inflatable structures. a Real puffer fish com- c–f Extension, bending, contraction, and twisting actuator; g Tentacle-like actuator,
pared with printed structure; b Inflatable membranes with different designs; inserts are imitating tentacle grabbing. All scale bars are 1 cm.

Figure 5a shows the inflation of a 3D-printed biomimetic puffer- bottom right panel, the same four concentric one-third circles are
fish. The dotted black line represents the material transition region. printed on the membrane. The left side of the membrane with fibers
The parts above that line are printed with G0 so that it is very stiff, causes it to take on a more linear shape, while the right side is free to
while the material below is the much softer G60. When the fish is take on the usual spherical shape, leading to asymmetric inflation. The
subjected to internal pressure, the rigid body of the fish undergoes inflation experiment agrees well with the simulations (Supplementary
only a small expansion while the soft belly expands significantly (more Movie 6). Although we present only four designs here, many other
than 30 times its initial volume; Supplementary Movie 5.), mimicking inflatable membrane manipulations can be easily achieved.
the behavior of a real pufferfish. This complex biological design can be Soft pneumatic actuators have been widely used in robotics due
easily achieved using this resin with g-DLP printing. to their great adaptability and flexibility than conventional linkage-
Figure 5b shows the actuation of several patterned membrane joint robots59. The general approach to fabricating soft actuators is
structures. Each design consists of a shallow rigid cylinder with a rigid incorporating strain limiters with an airtight soft matrix. Recently
cap at the bottom and a soft membrane on the top (the details design reported methods include wrapping inextensible fibers with varied
can be found in SI, Supplementary Fig. 20, and Supplementary angles, sealing airtight fabric bags, or printing stiff fibers on sealed soft
Movie 6). In the top left design, there is no rigid pattern added to the structures60–62. But these methods require tedious multi-step manu-
membrane. This represents the default inflation case, which is often facturing processes. The g-DLP printing with this resin provides a
the limit of single-material inflatable membranes. However, we can simple, one-step way to fabricate soft pneumatic actuators with
easily add rigid fibers to alter the inflated shape of the membrane. The greater adaptability and flexibility than the previous methods.
bottom left panel of Fig. 5b shows a design with a single stiff ring Figure 5c–f shows stiff fibers embedded in a stretchable airtight
placed in the soft membrane. Internal pressure causes inflation of the cylinder matrix that achieve the four basic motions: extension, torsion,
soft material between the edge of the cylinder and the outer radius of contraction, and bending when inflated. Stacking sequences and
the stiff ring, which makes the ring pop up. The ring itself does not combinations of those basic motions may give rise to more sophisti-
deform due to its much higher modulus. At the inside of the ring is cated deformation. Figure 5c shows the extension case where stiff
another stiff membrane, which also pops up due to internal pressure. rings are placed along the length of a soft tube. Without the rings, the
The top right design features four thin concentric rings. These stiff tube would experience mainly an outward deformation when inflated,
rings constrain the outward expansion of the membrane, but they do but the thin rings prevent the outward expansion. Because the stiff
not prevent the expansion of the membrane upward. This causes the rings are not connected, the actuator is much more compliant in the
final deformation to be more conical in stark contrast to the plain longitudinal direction, causing its length to increase significantly
membrane to the left, that takes on a spherical shape. Finally, in the under pressure. Figure 5d is a modification of the extension design

Nature Communications | (2023)14:1251 7

Article https://doi.org/10.1038/s41467-023-36909-y

where stiff rings arranged vertically cause an extension of the actuator, examples demonstrate the capability for fabricating challenging
but adding a single stiff fiber on one side places an asymmetric con- designs for various sensor applications.
straint on the actuator, leading to a bending motion. Although only a
half rotation is shown here, a full rotation can be achieved at a higher Discussion
pressure. Figure 5e presents the retraction case whose design features Combining materials with highly contrasted properties, from stretch-
stiff fibers that extend straight between the two end caps. Under able elastomers to hard thermosets in 3D printing enables the fabri-
pressure, the stiff fibers bend outward but resist a change in length. cation of highly complex 3D functional structures. Soft and stretchable
This causes the two end caps to be pulled closer together by the gel materials consisting of polymeric networks with various solvents,
bending fibers. The design in Fig. 5f shows the torsion case, which has including hydrogel (water), ionogel (ionoic liquid), and organogel
two helical rings that spiral between the top and bottom of the (organic solvent), have been widely reported for 3D printing. These
actuator. Again, the stiff spirals constrain the outward radial expansion gels generally rely on hydrogen bonding between solvent molecules
of the tube. The vertical extension is achieved by the unwinding of the and polymers. Here, the g-DLP ink formulation is inspired by these
spiral structures, which causes the whole actuator to twist, as shown by gels. At low conversion, monomers are able to perform like solvent
the attached arrow. Experiments of these actuators are presented in molecules in organogel via hydrogen bonding, while at high conver-
Supplementary Movie 7. The combination of these actuators can be sion, monomers crosslink into a stiff thermoset. The UV-curable AUD
used for complex applications such as soft robotics and biomimicry. oligomers and 2-HEA monomers are both favorable to form hydrogen
Here, we further designed a tentacle-like actuator consisting of bonding resulting in an elastic and stretchable organogel at low con-
bending and twisting, as shown in Fig. 5g. Upon inflation, the actuator version. Additionally, monomer IBOA ensured the stiffness of the hard
gradually wraps into two circles as designed, imitating the tentacle thermoset at high conversion. The g-DLP printing with this rationally
grabbing. The insert shows that it (weight 5 g) can grasp a plastic designed ink provides the simple single vat approach to realize the
centrifuge tube (weight 14 g) or a marker (weight 8.5 g) (Fig. 5g inser- multi-properties printing of stretchability and stiffness from both
ted), functioning like an elephant trunk. The experiments are pre- states. Besides the used resin in this work, more UV-curable resin can
sented in Supplementary Movie 8. Compared to the previous design of be designed with the same strategy that uncured monomer can
grippers, which typically require two or three fingers, the gripper interact with the crosslinked network through not only hydrogen
enabled by this ink allows a much simple design. We can easily vary the bonding but also ion-dipole or π−π stacking interaction to form a
fiber orientation and density to further change the tentacle’s inflated stretchable gel at low DoC (an exemplary resin formulation is pre-
shape and motion. This demonstrates this ink’s versatility for g-DLP sented in Supplementary Fig. 21).
fabrication customized and programmable deforming soft pneumatic The current ink does not require revising the DLP printing and
actuators. thus retains all the advantages offered by the DLP technique. For
example, multiple objectives can be printed in the same batch at a high
g-DLP printing for elastomer sensors printing speed (1 mm/min). Supplementary Fig. 22 shows that 12
The broad range of properties, from a stretchable rubbery state to a tentacle-like actuators can be printed simultaneously. This can
stiff glassy state, is also ideal for the application of flexible sensors with greatly increase the production rate.
large deformation. Here, we demonstrate the single-vat fabrication of Despite the significant advantages of single vat g-DLP printing,
different types of strain sensors via g-DLP printing. The wave-shaped some limitations still need to be addressed further. Firstly, the g-DLP
microfluidic channel is printed within the soft matrix and filled with a printed rubbery part is an organogel with liquid phase (uncured
liquid metal (eutectic gallium–indium alloy (EgaIn, Sigma-Aldrich)). monomers) immobilized in the hydrogen bonding dominated gel
Conductive leads are glued into the ends of the channel for resistance network. Just like general organogels, it is sensitive to high tempera-
measurement. Figure 6a shows a simple extension strain gauge with a tures since it weakens the hydrogen bonding and makes the liquid
microfluidic channel diameter of 0.8 mm. As the sensor is stretched, phase volatile. Secondly, we observe between the interface of the stiff
the microfluidic channels increase in length and become thinner, part and the rubbery part, there is a pixel scale transition layer
dramatically increasing their resistance, which is measured by a mul- (50–100 μm, Supplementary Fig. 23), which limits the g-DLP printing
timeter. Figure 6a shows the change in resistance recorded over 24 feasibility for micron-scale structures. Although such a transition layer
cycles of stretching to a large deformation (200% strain). Figure 6b is negligible for all the demonstrated printing size scales in this work, it
shows the design of a pressure sensor with a liquid metal-filled could be optimized further with different approaches like using a
microfluidic channel (0.8 mm) in the soft membrane. When internal projector with smaller pixel size, algorithm-corrected grayscaled
pressure is applied, the membrane is deformed upwards, increasing images, or resin with more photo absorber63. Moreover, the low
the resistance in the liquid metal. The rest of the sensor is printed at a grayscale will result in size distortion on the x-y plane, which is a
higher grayscale (G0) so that deformation is confined to the mem- general problem facing g-DLP printing and will affect the printing
brane alone. It can accurately capture the different pressure levels with resolution for micro-structures. Using an increased projecting area for
relatively small deformation, as shown in Fig. 6b. A unique feature of lower grayscale (Supplementary Fig. 24) or developing computational
the proposed grayscale resin is that the membrane’s stiffness can be models can address the challenge. But finding a more general solution
adjusted to alter the sensitivity of the sensor without the need to to this limitation is beyond the scope of this work. Last, the single-vat
change the thickness. grayscale cured samples have varying conversion degrees for multi-
The g-DLP 3D printing with this ink also has unique advantages for properties. Consequently, those uncured acrylate groups make the
fabricating customized human wearable electronic devices and sen- printed structure UV-sensitive. Further exposure under UV will reduce
sors. The rubbery state covers the mechanical pliability range of the property contrast until it becomes a uniformly stiff structure. Our
human skin (Young’s modulus ≈130–657 kPa)49. Also, the stiff state previous work has demonstrated that using a second stage curing
enables intimate conformability to mount the body without extra fix- through amine-acrylate Michael addition reaction could consume the
tures. Figure 6c shows a g-DLP-printed finger-mounted sensor for unreacted acrylate groups64. Polyamine with long and flexible back-
human-machine interfaces. The smaller microfluidic channel (0.4 mm) bones can be used to compose a hybrid two-stage resin, which is part
results in a higher sensitivity that can detect a small bending of the of our following work. Nevertheless, single vat printing with instantly
finger, as shown in Fig. 6c. Similar sensors can easily be designed to fit multi-properties function capability has its unique advantage and
different users or joints for both humans and robots. Overall, these values for many application scenarios.

Nature Communications | (2023)14:1251 8

Article https://doi.org/10.1038/s41467-023-36909-y

Fig. 6 | g-DLP printed sensor devices. a Strain sensor and ΔR/R0 response of 200% strain; b Pressure sensor and ΔR/R0 response with corresponding pressure; c Finger
sensor and ΔR/R0 response with corresponding bending. All scale bar is 1 cm.

In summary, a resin is developed for grayscale digital light pro- used for soft robots and actuators, metamaterials, flexible electronics,
cessing (g-DLP) printing to fabricate parts with drastically varying biomimetic, and pneumatic structures.
properties in a single pot resin. This resin offers the capability of
continuously varying modulus from 0.016 to 478 MPa, nearly 30,000 Methods
times difference. Also, the material printed with low-intensity light is Ink preparation
highly stretchable (~1500%). Using the DLP technique also permits high The photocurable resin was prepared by mixing 2-hydroxyethyl acry-
printing speed (1 mm/min). The ability to fabricate parts with very rigid late (Sigma-Aldrich, MO, USA), isobornyl acrylate (Sigma-Aldrich), and
regions and flexible regions all at once and at high speed is a crucial AUD (Ebecryl 8413, Allnex, GA, USA) with the weight ratio of 20:60:20.
advantage in rapid prototyping applications. The resin formula pre- Then 1 wt% photoinitiator (Irgacure 819, Sigma-Aldrich) and 0.05 wt%
sented here possesses significant advantages over previously reported photo absorber (Sudan I, Sigma-Aldrich) are added.
DLP resins, which allows for unprecedented freedom in printable
designs. With such a wide range of properties, we demonstrate robust Digital light processing 3D printing
and adaptable capability for various applications requiring multiple 3D printing is performed with a bottom-up DLP printer that employs a
properties in a monolithic structure. It provides a rapid manufacturing 385 nm UV-LED light projector (PRO4500, Wintech Digital Systems
approach for prototyping and performance validation, which can be Technology Corp., Carlsbad, CA, USA) and a linear translation stage

Nature Communications | (2023)14:1251 9

Article https://doi.org/10.1038/s41467-023-36909-y

(LTS150 Thorlabs, Newton, NJ, USA). A homemade container with an 12. Tee, Y. L., Peng, C., Pille, P., Leary, M. & Tran, P. PolyJet 3D printing of
oxygen-permeable window (Teflon AF-2400, Biogeneral Inc., CA, USA) composite materials: experimental and modelling approach. JOM
is used as the resin vat. The designed 3D structures are sliced into 72, 1105–1117 (2020).
image files with a thickness of 0.05 mm and then converted into 13. Khalid, G. A., Bakhtiarydavijani, H., Whittington, W. R., Prabhu, R. &
grayscaled image files with a MATLAB script. The continuous liquid Jones, M. D. Material response characterization of three poly jet
interface production (CLIP) approach is utilized at the optimized printed materials used in a high fidelity human infant skull. Mater.
speed of 3 s/layer to print the designed 3D structures. All samples are Today Proc. 20, 408–413 (2020).
raised with isopropyl alcohol after printing and wrapped in aluminum 14. Ren, L. et al. 3D printing of materials with spatially non-linearly
foil for further testing or demonstration. The light intensity of the varying properties. Mater. Des. 156, 470–479 (2018).
printer is calibrated with a photometer (ILT1400-A Radiometer, 15. Miriyev, A., Xia, B., Joseph, J. C. & Lipson, H. Additive manufacturing
International Light Technologies Inc., MA, USA) before printing. of silicone composites for soft actuation. 3D Print. Addit. Manuf. 6,
309–318 (2019).
Characterizations 16. Schaffner, M. et al. 3D printing of robotic soft actuators with pro-
The uniaxial tension tests are performed with a universal test machine grammable bioinspired architectures. Nat. Commun. 9, 878 (2018).
(Insight 10, MTS Systems Corp., Eden Prairie, MN, USA) with a cross- 17. Xu, C., Quinn, B., Lebel, L. L., Therriault, D. & L’Espérance, G. Multi-
head speed of 5 mm/min. Dynamic thermomechanical properties are material direct ink writing (DIW) for complex 3D metallic structures
conducted on a DMA machine (Q800, TA Instruments, New Castle, DE, with removable supports. ACS Appl. Mater. Interfaces 11,
USA) with a temperature ramped at a rate of 10 °C/min. The degree of 8499–8506 (2019).
curing is decided by normalized FTIR (Nicolet iS50 spectrometer, 18. Skylar-Scott, M. A., Mueller, J., Visser, C. W. & Lewis, J. A. Voxelated
Thermo Fisher Scientific, Waltham, MA, USA) peak intensity of the soft matter via multimaterial multinozzle 3D printing. Nature 575,
acrylate group present at 809 cm−1. Multiple tests are conducted for 330–335 (2019).
each sample to guarantee reproducibility. 19. Bas, O. et al. Ultrafast, miniature soft actuators. Multifunct. Mater. 4,
045001 (2021).
Data availability 20. Wallin, T. J. et al. 3D printable tough silicone double networks. Nat.
The authors declare that the data supporting the findings of this study Commun. 11, 4000 (2020).
are available within the article and its Supplementary Information file. 21. Zhou, L.-y et al. Multimaterial 3D printing of highly stretchable sili-
Raw data of the resin properties generated in this study are provided in cone elastomers. ACS Appl Mater. Interfaces 11, 23573–23583
the Source Data file with this paper. All other data underlying the (2019).
results, including the digital designs of those printed structures, are 22. Connolly, F., Polygerinos, P., Walsh, C. J. & Bertoldi, K. Mechanical
available from the corresponding author upon request. Source data programming of soft actuators by varying fiber angle. Soft Robot. 2,
are provided with this paper. 26–32 (2015).
23. Yuan, C., Wang, F. & Ge, Q. Multimaterial direct 4D printing of high
References stiffness structures with large bending curvature. Extreme Mech.
1. Zhang, Q. et al. Rapid volatilization induced mechanically robust Lett. 42, 101122 (2021).
shape-morphing structures toward 4D printing. ACS Appl. Mater. 24. Ge, Q. et al. 3D printing of highly stretchable hydrogel with diverse
Interfaces 12, 17979–17987 (2020). UV curable polymers. Sci. Adv. 7, eaba4261 (2021).
2. Lee, H. et al. 3D-printed programmable tensegrity for soft robotics. 25. Kuang, X. et al. Grayscale digital light processing 3D printing for
Sci. Robot. 5, eaay9024 (2020). highly functionally graded materials. Sci. Adv. 5, eaav5790 (2019).
3. Sekitani, T. et al. Stretchable active-matrix organic light-emitting 26. Peterson, G. I. et al. Production of materials with spatially-controlled
diode display using printable elastic conductors. Nat. Mater. 8, cross-link density via vat photopolymerization. ACS Appl Mater.
494 (2009). Interfaces 8, 29037–29043 (2016).
4. Parida, K. et al. Extremely stretchable and self-healing conductor 27. Peng, X. et al. Multi-color 3D printing via single-vat grayscale digital
based on thermoplastic elastomer for all-three-dimensional printed light processing. Adv. Funct. Mater. 32, 2112329 (2022).
triboelectric nanogenerator. Nat. Commun. 10, 2158 (2019). 28. Zhang, Z., Corrigan, N., Bagheri, A., Jin, J. & Boyer, C. A versatile 3D
5. Schwartz, J. J. & Boydston, A. J. Multimaterial actinic spatial control and 4D printing system through photocontrolled RAFT poly-
3D and 4D printing. Nat. Commun. 10, 791 (2019). merization. Angew. Chem. Int. Ed. 58, 17954–17963 (2019).
6. Sydney Gladman, A., Matsumoto, E. A., Nuzzo, R. G., Mahadevan, L. 29. Valizadeh, I., Al Aboud, A., Dörsam, E. & Weeger, O. Tailoring of
& Lewis, J. A. Biomimetic 4D printing. Nat. Mater. 15, 413–418 functionally graded hyperelastic materials via grayscale mask ste-
(2016). reolithography 3D printing. Addit. Manuf. 47, 102108 (2021).
7. Gu, G. et al. A soft neuroprosthetic hand providing simultaneous 30. Tumbleston, J. R. et al. Continuous liquid interface production of 3D
myoelectric control and tactile feedback. Nat. Biomed. Eng. (2021). objects. Science 347, 1349–1352 (2015).
https://doi.org/10.1038/s41551-021-00767-0. 31. Vitale, A., Hennessy, M. G., Matar, O. K. & Cabral, J. T. Interfacial
8. Roach, D. J. et al. The m4 3D printer: a multi-material multi-method profile and propagation of frontal photopolymerization waves.
additive manufacturing platform for future 3D printed structures. Macromolecules 48, 198–205 (2015).
Addit. Manuf. 29, 100819 (2019). 32. Kuang, X. et al. High-speed 3D printing of high-performance ther-
9. Ren, Z. et al. SMP-based multi-stable mechanical metamaterials: mosetting polymers via two-stage curing. Macromol. Rapid Com-
from bandgap tuning to wave logic gates. Extreme Mech. Lett. 42, mun. 39, e1700809 (2018).
101077 (2020). 33. Shah, D. M., Morris, J., Plaisted, T. A., Amirkhizi, A. V. & Hansen, C. J.
10. Che, K., Yuan, C., Qi, H. J. & Meaud, J. Viscoelastic multistable Highly filled resins for DLP-based printing of low density, high
architected materials with temperature-dependent snapping modulus materials. Addit. Manuf. 37, 101736(2021).
sequence. Soft Matter 14, 2492–2499 (2018). 34. Caprioli, M. et al. 3D-printed self-healing hydrogels via digital light
11. Gu, G. X., Chen, C.-T., Richmond, D. J. & Buehler, M. J. Bioinspired processing. Nat. Commun. 12, 2462 (2021).
hierarchical composite design using machine learning: simulation, 35. Peng, S. et al. 3D printing mechanically robust and transparent
additive manufacturing, and experiment. Mater. Horiz. 5, polyurethane elastomers for stretchable electronic sensors. ACS
939–945 (2018). Appl. Mater. Interfaces 12, 6479–6488 (2020).

Nature Communications | (2023)14:1251 10

Article https://doi.org/10.1038/s41467-023-36909-y

36. Zhang, B. et al. Self-healing four-dimensional printing with an 59. Tawk, C. & Alici, G. A review of 3D-printable soft pneumatic
ultraviolet curable double-network shape memory polymer system. actuators and sensors: research challenges and opportunities. Adv.
Acs Appl. Mater. Interfaces 11, 10328–10336 (2019). Intell. Syst. 3, 2000223 (2021).
37. Zhang, B. et al. Mechanically robust and UV-curable shape-memory 60. Kim, S. Y. et al. Reconfigurable soft body trajectories using uni-
polymers for digital light processing based 4D printing. Adv. Mater. directionally stretchable composite laminae. Nat. Commun. 10,
33, e2101298 (2021). 3464 (2019).
38. O’Bryan, C. S. et al. Self-assembled micro-organogels for 3D 61. Ceron, S., Cohen, I., Shepherd, R. F., Pikul, J. H. & Harnett, C. Fiber
printing silicone structures. Sci. Adv. 3, e1602800 (2017). embroidery of self-sensing soft actuators. Biomimetics 3, 24
39. Patel, D. K. et al. Highly stretchable and UV curable elastomers for (2018).
digital light processing based 3D printing. Adv. Mater. 29, 62. Heiden, A. et al. 3D printing of resilient biogels for omnidirectional
1606000 (2017). and exteroceptive soft actuators. Sci. Robot. 7, eabk2119
40. Thrasher, C. J., Schwartz, J. J. & Boydston, A. J. Modular elastomer (2022).
photoresins for digital light processing additive manufacturing. 63. Montgomery, S. M., Demoly, F., Zhou, K. & Qi, H. J. Pixel-level
ACS Appl Mater. Interfaces 9, 39708–39716 (2017). grayscale manipulation to improve accuracy in digital light pro-
41. Zhang, B. et al. Highly stretchable hydrogels for UV curing based cessing 3D printing. Adv. Funct. Mater. 2213252 (2023).
high-resolution multimaterial 3D printing. J. Mater. Chem. B 6, 64. Zou, W., Lin, X. & Terentjev, E. M. Amine-acrylate liquid single
3246–3253 (2018). crystal elastomers reinforced by hydrogen bonding. Adv. Mater. 33,
42. Hatamleh, M. I. et al. Structural response of 3D-printed rubber lat- 2101955 (2021).
tice structures under compressive fatigue. MRS Commun. 11,
168–172 (2021). Acknowledgements
43. Yang, C. et al. 4D printing reconfigurable, deployable and H.J.Q. would like to acknowledge the support from an AFOSR grant
mechanically tunable metamaterials. Mater. Horiz. 6, (FA9550-20-1-0306; Dr. B.-L. “Les” Lee, Program Manager), Toyota North
1244–1250 (2019). America, the gift funds from HP, Inc. and Northrop Grumman Corpora-
44. Ji, Z. et al. 3D printing of photocuring elastomers with excellent tion. This work is performed in part at the Georgia Tech Institute for
mechanical strength and resilience. Macromol. Rapid Commun. 40, Electronics and Nanotechnology, a member of the National Nano-
e1800873 (2019). technology Coordinated Infrastructure, which is supported by the
45. Zhao, T. et al. Superstretchable and processable silicone elasto- National Science Foundation (ECCS-1542174).
mers by digital light processing 3D printing. ACS Appl. Mater.
Interfaces 11, 14391–14398 (2019). Author contributions
46. Azoug, A. et al. Viscoelasticity of the polydomain-monodomain L.Yue and H.J.Q. conceived the concept. L.Yue, H.J.Q., Y.S., M.T., and
transition in main-chain liquid crystal elastomers. Polymer 98, T.N. designed the experiments. S.M.M. performed the FEA simulations.
165–171 (2016). L.Y., S.M.M., and X.S. conducted the composite structures design. L.Y.
47. Shadwick, R. E. Mechanical design in arteries. J. Exp. Biol. 202, performed the property characterization. L.Yue and H.J.Q. prepared the
3305–3313 (1999). manuscript. All authors discussed the results and commented on
48. Zhalmuratova, D. et al. Mimicking “J-shaped” and anisotropic the paper.
stress–strain behavior of human and porcine aorta by fabric-
reinforced elastomer composites. ACS Appl. Mater. Interfaces 11, Competing interests
33323–33335 (2019). L.Yue, H.J.Q., Y.S., and M.T. are coinventors on two provisional patent
49. Khatyr, F., Imberdis, C., Vescovo, P., Varchon, D. & Lagarde, J.-M. applications. (1) Resins for digital light processing 3d printing (no. 17/
Model of the viscoelastic behaviour of skin in vivo and study of 833,144, filed 6 June 2022); (2) digital light processing 3d-printed
anisotropy. Ski. Res. Technol. 10, 96–103 (2004). monolithic substrates with integrated and embedded sensors (no. 17/
50. Kao, P. H. et al. A microstructurally driven model for pulmonary 977,314, filed 31 October 2022), filed by Toyota Motor Engineering and
artery tissue. J. Biomech. Eng. 133, 051002 (2011). Manufacturing North America Inc. and Georgia Tech Research Cor-
51. Tangorra, J., Phelan, C., Esposito, C. & Lauder, G. Use of biorobotic poration. The authors declare that they have no other competing
models of highly deformable fins for studying the mechanics interests.
and control of fin forces in fishes. Integr. Comp. Biol. 51, 176–189
(2011). Additional information
52. Alben, S., Madden, P. G. & Lauder, G. V. The mechanics of active fin- Supplementary information The online version contains
shape control in ray-finned fishes. J. R. Soc. Interface 4, supplementary material available at
243–256 (2007). https://doi.org/10.1038/s41467-023-36909-y.
53. Wu, K. et al. Discontinuous fibrous Bouligand architecture enabling
formidable fracture resistance with crack orientation insensitivity. Correspondence and requests for materials should be addressed to
Proc. Natl Acad. Sci. USA 117, 15465–15472 (2020). H. Jerry Qi.
54. Gorissen, B. et al. Elastic inflatable actuators for soft robotic appli-
cations. Adv. Mater. 29, 1604977 (2017). Peer review information Nature Communications thanks Yi Yang and
55. Jones, T. J., Jambon-Puillet, E., Marthelot, J. & Brun, P. T. Bubble the other, anonymous, reviewer(s) for their contribution to the peer
casting soft robotics. Nature 599, 229–233 (2021). review of this work. Peer reviewer reports are available.
56. Truby, R. L. et al. Soft somatosensitive actuators via embedded 3D
printing. Adv. Mater. 30, 1706383 (2018). Reprints and permissions information is available at
57. Wang, H., Totaro, M. & Beccai, L. Toward perceptive soft robots: http://www.nature.com/reprints
progress and challenges. Adv. Sci. 5, 1800541 (2018).
58. Hubbard, J. D. et al. Fully 3D-printed soft robots with integrated Publisher’s note Springer Nature remains neutral with regard to jur-
fluidic circuitry. Sci. Adv. 7, eabe5257 (2021). isdictional claims in published maps and institutional affiliations.

Nature Communications | (2023)14:1251 11

Article https://doi.org/10.1038/s41467-023-36909-y

Open Access This article is licensed under a Creative Commons

Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as
long as you give appropriate credit to the original author(s) and the
source, provide a link to the Creative Commons license, and indicate if
changes were made. The images or other third party material in this
article are included in the article’s Creative Commons license, unless
indicated otherwise in a credit line to the material. If material is not
included in the article’s Creative Commons license and your intended
use is not permitted by statutory regulation or exceeds the permitted
use, you will need to obtain permission directly from the copyright
holder. To view a copy of this license, visit http://creativecommons.org/
licenses/by/4.0/.

© The Author(s) 2023

Nature Communications | (2023)14:1251 12