Page 34 - Read Online

Ren et al. Soft Sci. 2026, 6, 6 Page 5 of 33

and insulating inks for DIW of flexible light-emitting devices and co-printed them with soft robotic
structures . By integrating these devices with quadrupedal robots and sensing units, they achieved artificial
[50]
camouflage that instantly adapts its displayed color to the surrounding environment.

In practice, DIW holds significant appeal for 4D printing due to its straightforward hardware and
compatibility with LCEs, SMPs, hydrogels, and related composites, enabling broad applications in
biomedicine, soft robots, and electronics. However, its drawbacks include limited throughput, poor surface
roughness, and restricted resolution. In addition, extra solidification steps may reduce feature fidelity.

FDM
FDM, conceived by S. Scott Crump in the late 1980s and brought to market in the early 1990s through
Stratasys , is an extrusion-based additive process where thermoplastic filament is fed into a heated nozzle,
[51]
deposited layer by layer, and solidified upon cooling [Figure 2B]. Its applicability to 4D printing stems from
the wide range of extrudable thermoplastics and composites, as well as the ability to program internal
stresses during printing. In particular, shape-memory thermoplastics - such as polylactic acid (PLA) ,
[52]
thermoplastic polyurethane (TPU) , and acrylonitrile butadiene styrene (ABS) - have been extensively
[53]
[54]
studied due to their ability to melt when heated, bond with adjacent materials, and rapidly solidify upon
cooling to form robust solids. The internal stresses originate from the asynchronous cooling of sequentially
deposited filaments .
[55]
Kačergis et al. systematically varied FDM parameters such as print speed, build plate temperature, and active
layer count to preprogram deformation in PLA and TPU hinges, demonstrating that deformation can be
preprogrammed through process and design choices . To overcome the rigidity of traditional 4D-printed
[56]
SMP hinges below their glass transition temperature, Yamamura et al. proposed a hybrid hinge design .
[57]
This design placed an elastic hinge adjacent to a rigid self-folding segment, enabling large elastic folding with
high durability over hundreds of cycles. Extending the idea of process-encoded strain, Wang et al. proposed
an economical multi-speed FDM strategy . This approach embedded graded built-in strain into flat
[58]
precursors to enable rapid, repeatable formation of complex 3D shapes upon heating while reducing build
time and workflow complexity. From hinges to actuators, Dezaki et al. employed TPU to fabricate soft
pneumatic actuators whose tip deflection, tip force, stiffness, and strength can be tuned via lattice topology
and printing parameters . The performance of wearable rehabilitation devices based on this actuator was
[25]
validated through modeling and closed-loop pneumatic actuation.

FDM is a mature, cost-effective platform now widely adopted across automotive, aerospace, defense,
biomedicine, and packaging . However, it still faces key challenges including anisotropic interlayer
[59]
adhesion, relatively low precision and surface quality, extended build times, and risks of delamination,
shrinkage, and thermal degradation. Advancements in in-situ sensing with feedback, localized reheating,
post-processing, and improved feedstocks are steadily addressing these shortcomings and propelling FDM
toward the reliable fabrication of functional 4D components. Notably, rotary 4D printing has recently
emerged as an extension of FDM toward continuous manufacturing, in which filaments are deposited on a
rotating mandrel to directly encode circumferential or helical toolpaths, anisotropies, and gradient
architectures within tubular or curvilinear structures. Compared with planar layer stacking, the rotary format
can improve throughput and repeatability for axisymmetric geometries and enables efficient programming of
ring- or helix-like reinforcements that are difficult to realize by conventional FDM, as demonstrated in
rotary-printed programmable metamaterials and architected phase-change artificial muscles with
[60]
[61]
encoded dynamic behaviors.