Introduction

Hierarchically porous materials, distinguished by their multi-level pore and channel structures, integrate low density with high specific performance and multifunctionality1,2,3. These materials exhibit advantageous properties including low density, a high strength-to-weight ratio, extensive surface area, effective sound and thermal insulation, and excellent permeability. Such characteristics render them highly suitable for diverse applications such as energy absorption, vibration damping, electromagnetic shielding, flexible sensing, separation catalysis, energy and environmental management, and tissue engineering4,5,6,7. Significant research efforts have been directed toward developing innovative techniques for fabricating these materials, with a particular emphasis on utilizing 3D printing to engineer complex three-dimensional structures, such as lattices and honeycombs. Techniques such as the incorporation of sacrificial sodium chloride particles8,9, polymerization-induced phase separation10,11, gas foaming12,13,14,15, and emulsion templating16 have been employed to create micro- and nanoscale porous architectures. Nonetheless, conventional pore-forming methods like solids leaching, phase separation, and chemical foaming frequently necessitate the use of substantial amounts of organic solvents or toxic small molecules, which can result in the persistence of harmful residual pore-forming agents.

In recent years, supercritical fluid (SCF) foaming technology has emerged as a pivotal method for fabricating microporous polymer materials, owing to its non-toxicity, environmental compatibility, versatile material adaptability, and precise control over foaming ratios and structure17. Researchers have synergistically combined supercritical fluid foaming with advanced 3D printing techniques, including fused filament fabrication (FFF)18,19,20,21,22, selective laser sintering (SLS)23, and direct ink writing (DIW)24 to engineer materials with enhanced energy absorption and vibration-damping capabilities through biomimetic hierarchically porous structures20,21,22. These advancements have also facilitated the development of hierarchically porous scaffolds for bone tissue engineering24,25, and materials with improved piezoelectric properties23. However, thermoplastic polymers employed in FFF, SLS, and DIW struggle to form strong chemical crosslinks between printed layers, causing weak bonding and potential delamination during foaming. This leads to significant anisotropy in mechanical properties, limiting broader applicability18,22.

Compared to FFF and other processes, vat photopolymerization (VPP)-based 3D printing offers better interlayer performance. The new layer of resin chemically bonds with the partially cured resin from the previous layer, creating a crosslinked network that improves adhesion26. The integration of vat photopolymerization 3D-printed elastomers with residue-free supercritical fluid foaming processes shows potential for developing customized, lightweight, high-performance porous materials. However, integrating vat photopolymerization 3D printing with supercritical fluid foaming remains challenging due to the high crosslinking density of thermosetting polymers formed from photocurable resins. This high density results in inferior tensile properties and restricts the ability of the supercritical fluid to diffuse into the polymer matrix, thereby impeding effective microporous foaming.

Recently, dynamic reversible covalent bonds have been introduced into polymers to form covalent adaptive networks. Unlike traditional crosslinked networks, covalent adaptive networks can alter the polymer’s topology dynamically through reversible bond breaking and reformation in response to stimuli, enabling thermoset remolding. Researchers have used dynamically dissociative bonds like Diels-Alder27 and hindered urea bonds (HUBs)28. These bonds enhance self-healing and remolding in vat photopolymerization 3D-printed elastomers. However, most dynamic HUBs remain stable at high temperatures29, maintaining high crosslinking density and leading to poor tensile properties. This causes cracks in foamed materials and low expansion ratios. Thus, integrating vat photopolymerization 3D printing with supercritical fluid foaming for hierarchically porous materials remains challenging.

Herein, we report a strategy to modulate the crosslinking density and tensile properties of polyurethane elastomers using a dynamically crosslinked-interpenetrating network structure. This method involves fabricating polyurethane elastomers suitable for supercritical fluid foaming via vat photopolymerization 3D printing (Fig. 1a). First, a polyurethane acrylate (PUA) oligomer containing HUBs was synthesized and mixed with amine-based curing agents in varying proportions. Acrylic monomer 2-ethylhexyl acrylate (2-EHA) served as a reactive diluent to reduce resin viscosity, yielding a photocurable resin with enhanced fluidity (Fig. 1b). During vat photopolymerization 3D printing, radical-initiated copolymerization of monofunctional 2-EHA and diacrylate-functionalized polyurethane acrylate oligomer generated elastomers with dynamically crosslinked networks (Fig. 1c). Subsequently, a temperature-controlled batch supercritical fluid foaming protocol was applied. During heating, HUBs dissociated into isocyanate groups (NCO), which reacted with the amine-based curing agent, 4,4’-methylenebis(cyclohexylamine) (PACM), leading to the formation of high-molecular-weight polyurethane/polyurea chains. These chains, combined with the dynamically crosslinked network, form a precisely controllable dynamically crosslinked-interpenetrating network structure that enhances gas solubility and enables controlled depressurization foaming for hierarchically porous materials (Fig. 1c). As shown in Fig. 1d–f, this methodology facilitated the successful fabrication of lightweight and highly resilient elastomeric foams with hierarchically structured porosity.

Fig. 1: Fabrication of hierarchically porous foams by supercritical fluid foaming (SCF) of vat photopolymerization (VPP) 3D-printed polyurethane elastomers.
Fig. 1: Fabrication of hierarchically porous foams by supercritical fluid foaming (SCF) of vat photopolymerization (VPP) 3D-printed polyurethane elastomers.
Full size image

a Schematic of the process for preparing hierarchically porous foams via supercritical fluid foaming of vat photopolymerization 3D-printed elastomers. b Chemical structures of the photocurable resin components. c Mechanism of formation of photopolymerized dynamically crosslinked networks and their structural evolution during supercritical fluid foaming. d Photographs of foams with different lattice structures before and after foaming. e Photographs depicting the compression and recovery process of a Gyroid lattice foam. f Photograph of a Diamond lattice foam supported by a flower. Scale bar: 10 mm.

Results and discussion

Vat photopolymerization 3D Printing of polyurethane elastomers and annealing behavior

As illustrated in Figure S1, the polyurethane acrylate oligomer was synthesized via a two-step procedure. In the first step, poly(tetramethylene glycol) diol (PTMG) and isophorone diisocyanate (IPDI) were reacted under the catalysis of dibutyltin dilaurate (DBTDL) to form an isocyanate-terminated prepolymer. The reaction process of the polyurethane acrylate oligomer was monitored in real-time using Attenuated Total Reflection Fourier Transform Infrared (ATR-FTIR) spectroscopy. Figure S2 displays the absorption peak of the NCO group at 2270 cm−1 and the N-H stretching vibration peak at 3326 cm−1, which result from the reaction between the NCO groups and the hydroxyl groups of poly(tetramethylene glycol) diol. Subsequently, 2-(tert-butylamino)ethyl methacrylate was added as a capping agent to react with remaining NCO groups in the prepolymer, forming dynamic HUBs and incorporating photocurable methacrylate groups. FTIR confirmed the disappearance of the NCO peak and the appearance of a double bond peak at 810 cm−1. The structure of polyurethane acrylate oligomer was further confirmed by 1H NMR spectroscopy (Figure S3), and GPC analysis showed an average molecular weight of 7693 g mol−1 (Figure S4). As shown in Table S1, the photocurable resin formulation for vat photopolymerization 3D printing of polyurethane elastomers consists of a base resin composed of polyurethane acrylate oligomer (73.45 wt%), a reactive diluent (2-EHA, 24.50 wt%), a photoinitiator (2.00 wt%), and a photoabsorber (0.05 wt%). An amine-based curing agent, PACM, was added at varying weight percentages relative to this base resin. The resulting resin system is designated as PUE-X, where X represents the weight percentage of PACM. The photocurable resin exhibits excellent flowability and automatically levels in the printing window after each layer. It exhibits Newtonian fluid behavior with minimal viscosity increase as PACM content rises (Fig. 2a). Specifically, at a shear rate of 10 s-1, the viscosity rises from 12.1 Pa s for PUE-0 to 13.9 Pa s for PUE-6. The photocuring rate of the resin is also a critical parameter in determining the vat photopolymerization printing speed and success rate. Photorheological tests were performed at the same exposure power (3.2 mW cm−2) as used in the printer (Fig. 2b). Upon UV light activation at 20th s, both the storage modulus (G’) and loss modulus (G”) of the photocurable resin exhibited rapid increases. Following the crossover of G’ and G”, both moduli stabilize quickly. The crossover time of G’/G”, commonly referred to as the estimated gel point of the resin30, was observed to increase with higher PACM content. Specifically, the measured gel point times were 4.4 s for PUE-0, 6.5 s for PUE-2, 9.5 s for PUE-4, and 11.4 s for PUE-6. This phenomenon may be attributed to the absence of photopolymerizable groups in PACM. Double bond conversion confirms this trend (Fig. 2c); with PACM content below 4%, photocuring remains efficient, achieving over 80% conversion in 15 seconds. The characteristic penetration depth (hp) of the resin is a key parameter governing resolution in vat photopolymerization 3D printing, as it reflects the ultraviolet light absorption capacity of the resin. Specifically, a smaller hp indicates stronger light absorption. Dyes act as photoabsorber, enhancing the absorption capability of resin and reducing light scattering effects, thereby improving printing resolution31. The relationship is described by the Beer-Lambert law (Eq. 1):

$${z}_{c}={h}_{{{{\rm{p}}}}}{{\mathrm{ln}}}\left(\frac{{t}_{{{{\rm{c}}}}}}{{T}_{0}}\right)$$
(1)
Fig. 2: Utilization of vat photopolymerization 3D printing with photocurable resins for fabricating complex 3D polyurethane elastomer objects.
Fig. 2: Utilization of vat photopolymerization 3D printing with photocurable resins for fabricating complex 3D polyurethane elastomer objects.
Full size image

a Shear-dependent viscosity profiles of resins with varying PACM content. b Photorheological response curves of resins with different PACM concentrations. G’ and G” represent the storage modulus and loss modulus, respectively. c Double bond conversion in resins as a function of PACM content. d Curing depth versus exposure time with and without photoabsorber addition. Error bars represent the standard deviation of 3 trials. e SEM image of a Gyroid lattice elastomer. f Photograph of vat photopolymerization 3D-printed elastomers with various lattice structures. Scale bar: 10 mm.

the hp value was determined by measuring the curing depth (zc) at different exposure times (tc), corresponding to the slope obtained from linear regression of Eq. 1. Addition of dye (0.05 wt%) reduced the hp of the resin from 0.722 mm to 0.474 mm (Fig. 2d). As demonstrated in Fig. 2e, the Gyroid lattice structure with rod thickness of 225 μm can be printed. The above results demonstrate that photocurable resin possesses low viscosity, a high photocuring rate, and a low curing depth. Using a liquid crystal display (LCD) vat photopolymerization 3D printer, a single print cycle can produce 15 porous lattice elastomers, each 25 mm × 25 mm × 25 mm, demonstrating the high-throughput capability of this photocurable resin (Fig. 2f).

The thermal reaction behavior of polyurethane elastomers during annealing was investigated using differential scanning calorimetry (DSC) (Fig. 3a). Results show that PUE-0 without PACM exhibited no significant exothermic peak during heating. In contrast, when PACM was added to polyurethane elastomers, a clear exothermic peak occurred from 75 to 130 °C. The peak temperature of the exothermic curve suggests that the reaction primarily occurs around 110 °C. As the PACM content increases, the enthalpy increases from 8.3 J g−1 in PUE-2 to 15.5 J g−1 in PUE-6. This trend indicates that as the PACM content increases, more dynamic HUBs dissociate and subsequently react with PACM, thereby facilitating the formation of high-molecular-weight polyurethane/polyurea polymer chains. Equilibrium swelling experiments revealed that the crosslinking density (Ve) of polyurethane elastomers decreased with increasing PACM doping after thermal annealing (Fig. 3b). Correspondingly, the molecular weight between crosslinking points (Mc) increased (Fig. 3c). This further suggests that the highly crosslinked dynamic network of polyurethane elastomer formed via vat photopolymerization 3D printing undergoes a structural transition to a crosslinked-interpenetrating network with reduced crosslinking density after annealing. Figure 3d illustrates the molecular network evolution during annealing. Diacrylate-functionalized polyurethane acrylate oligomers serve as dynamic crosslinkers. These oligomers contain terminal acrylate groups that undergo radical-mediated photopolymerization with monofunctional monomer 2-EHA, forming a dynamic crosslinked network. Theoretically, this network inherently possesses a high crosslinking density and a low molecular weight between crosslinking points, which is determined by the molecular weight of the polyurethane acrylate. Polyurethane acrylate oligomer incorporates dynamic HUBs that dissociate upon thermal annealing, generating prepolymers with terminal NCO groups. These NCO-terminated prepolymers subsequently react with primary amines of PACM to form high-molecular-weight linear polyurethane/polyurea chains. Reaction with PACM decouples polyurethane acrylate chains from the photopolymerized network. Increasing PACM content promotes the formation of longer polyurethane/polyurea chains during annealing, thereby increasing the molecular weight between crosslinking points and reducing overall crosslinking density.

Fig. 3: Thermal annealing of polyurethane elastomers leads to the formation of a dynamically crosslinked-interpenetrating network.
Fig. 3: Thermal annealing of polyurethane elastomers leads to the formation of a dynamically crosslinked-interpenetrating network.
Full size image

a DSC exothermic curves recorded during the heating of polyurethane elastomers with varying PACM content. b Crosslinking density (Ve) and (c) molecular weight between crosslinking points (Mc) of polyurethane elastomers with different PACM contents before and after annealing. Error bars represent the standard deviation of 3 trials. d Schematic of the formation of high-molecular-weight polyurethane/polyurea chains within the dynamically crosslinked network during the annealing process.

Supercritical fluid foaming optimization

The annealing process transforms the highly crosslinked dynamic network of polyurethane elastomers into a low crosslinking density crosslinked-interpenetrating network with PACM, as confirmed by DSC and equilibrium swelling tests. This transformation critically modulates the viscoelastic properties governing supercritical fluid foaming. The viscoelastic behavior of annealed polyurethane elastomers was characterized at 120 °C using oscillatory frequency scanning with a rotational rheometer. As shown in Fig. 4a, the G’ of PUE-0 and PUE-2 remains consistently high and nearly invariant with increasing frequency, indicating a high crosslinking density that preserves a rigid structure even at elevated temperatures (120 °C). With an increase in PACM content, G’ exhibits a significant decrease. This suggests that the incorporation of PACM forms an interpenetrating network structure after annealing, reducing the crosslinking density of polyurethane elastomers. For PUE-4 and PUE-6, the frequency dependence of both G’ and G” becomes more pronounced, with both moduli showing a substantial increase with frequency (Fig. 4b). As shown in Fig. 4c, PUE-0 and PUE-2 exhibit low tan δ values with minimal frequency dependence, reflecting strong elastic characteristics. PUE-4 exhibits a tan δ value approaching 1 at low frequencies, which indicates a balance between elastic (G’) and viscous (G”) moduli—a characteristic feature of optimal network dynamics. In contrast, PUE-6 consistently exhibits tan δ > 1 across 0.1–100 rad s−1, signifying sustained viscous dominance (G” > G’). This behavior originates from enhanced chain mobility, which overrides elastic recovery. The annealed PUE-6, characterized by higher PACM content and reduced crosslinking density, consequently demonstrates superior fluidity at elevated temperatures. The complex viscosity (η*) of polyurethane elastomers decreases with increasing frequency as PACM content increases (Fig. 4d), showing typical shear-thinning behavior. Further increases in PACM content result in a reduction of η*, indicating reduced crosslinking density and enhanced molecular mobility. These findings suggest that the dynamically crosslinked-interpenetrating network formed in PACM-containing polyurethane elastomers after high-temperature annealing effectively lowers the crosslinking density of the photocured elastomer, thereby reducing elasticity while enhancing viscosity.

Fig. 4: Rheological and morphological analysis of polyurethane elastomers during supercritical fluid foaming.
Fig. 4: Rheological and morphological analysis of polyurethane elastomers during supercritical fluid foaming.
Full size image

a Storage modulus (G’), (b) loss modulus (G”), (c) tan δ, and (d) complex viscosity (η*) as functions of angular frequency for annealed elastomers at 120 °C. e1–e3 SEM images of PUE-0 foams, (f1–f3) PUE-2 foams, (g1–g3) PUE-4 foams, and (h1–h3) PUE-6 foams, obtained after foaming at 110 °C, 120 °C, and 130 °C under 20 MPa N2 pressure. i Photograph of elastomeric foams with various lattice structures. Scale bar: 10 mm. j Photograph of Diamond lattice structure elastomer before and after foaming. Scale bar: 10 mm. k SEM images of the interlayer structure of the Diamond lattice after foaming.

Furthermore, the formation of a dynamically crosslinked-interpenetrating network structure significantly enhances the tensile properties of the polyurethane elastomers. As illustrated in Fig. S5, the tensile strength of PUE-0 is 1.09 MPa, with a fracture elongation of 112.9%. Upon incorporating PACM, the tensile properties of polyurethane elastomers improve markedly. Specifically, the fracture elongation increases with PACM content, reaching 288%, 847%, and 879% for PUE-2, PUE-4, and PUE-6, respectively. The tensile strength initially increases from 1.09 MPa (PUE-0) to a maximum of 19.5 MPa (PUE-4), then declines sharply to 0.43 MPa (PUE-6). This non-monotonic trend can be attributed to the optimal PACM content in PUE-4, which facilitates the formation of an optimized dynamically crosslinked-interpenetrating network with intermediate crosslinking density that provides sufficient reinforcement for high strength while allowing extensive chain mobility for enhanced elongation. Conversely, excessive PACM in PUE-6 promotes the transformation of dynamic crosslinks into linear polyurethane/polyurea chains during annealing, thereby reducing the crosslinking density. Consequently, PUE-6 exhibits reduced resistance to high-temperature deformation, resulting in inferior tensile strength.

Figure 4e–h illustrate the microstructural evolution of foam in PUE-0 to PUE-6 elastomers after 2-hour saturation under supercritical N2 (20 MPa) at 110 °C, 120 °C, and 130 °C. PUE-0 shows sparse foam cells and macroscopic cracks, attributed to its high crosslinking density and rigidity (Figure 4e1–3). These characteristics restrict cell nucleation and growth, resulting in a consistently low foam expansion ratio of ~1.3 across all temperatures (Figure S6). PUE-2 with low PACM content shows moderate cell formation at 110 °C, with the expansion ratio increasing from 1.5 (110 °C) to 3.2 (130 °C). However, its insufficient tensile strength results in rapid matrix biaxial stretching and subsequent cell wall rupture (Figure 4f1–3). PUE-4 exhibits distinct advantages: At 110 °C, it forms crack-free closed-cell foams with uniformly distributed cells (Figure 4g1). Increasing the temperature to 120 °C enhances cell density while preserving structural integrity (Figure 4g2), increasing the expansion ratio from 2.3 to 4.4 (Figure S6). The foam cells in PUE-4 become larger at 130 °C, with noticeable coalescence observed (Figure 4g3). Cell coalescence emerges in PUE-6 even at 110 °C (Figure 4h1) and becomes significantly more pronounced above 120 °C due to excessive chain mobility (Figure 4h3). These results confirm that PUE-4, with its optimized PACM content, forms a dynamically crosslinked-interpenetrating network during thermal treatment. This network exhibits an intermediate crosslinking density that ensures structural stability by maintaining elasticity while allowing sufficient chain mobility for cell growth, synergistically enhancing supercritical fluid foaming performance.

While PUE-4 demonstrates favorable foaming performance under supercritical N2, the limited N2 solubility in PUE-4 (0.58 wt% vs. 3.18 wt% for CO2 at 6 MPa) restricts cell nucleation (Figure S7), limiting expansion ratios. As shown in Figure. S8, increasing CO2 content in the N2/CO2 gas mixture significantly refines the microcellular structure of PUE-4 foams: Average cell diameter decreases from 50.2 μm (pure N2) to 17.5 μm (pure CO2) (Figure S9), while cell density increases from 2.0 × 107 to 9.9 × 108 cell cm−3, achieving an expansion ratio improvement from 4.4 to 13.4. These enhancements stem from the superior solubility and nucleation efficiency of CO2. However, high CO2 content induces severe aging shrinkage (Figure S10), with the expansion ratio dropping to 2.7 after 14 days (Figure S11). This shrinkage arises from two mechanisms: Rapid CO2 outgassing creates a pressure imbalance (external atmospheric pressure > internal cell pressure) due to its high diffusivity32. And PUE-4 exhibits low modulus in its rubbery state at room temperature, making it incapable of withstanding the pressure differential. Additionally, stretched molecular chains undergo spontaneous curling during aging to release internal stresses generated during foaming, further exacerbating shrinkage (Figure S12a)33. To mitigate these effects, we developed a hybrid N2/CO2 system. This approach leverages the high nucleation efficiency of CO2 and the slower diffusion rate of N2 to minimize pressure differentials and reduce shrinkage. An optimal gas ratio of N2:CO2 = 12:8, resulted in a maximum stable expansion ratio of 4.0 (Figure S11). Subsequently, when the optimal gas mixture (N2:CO2 = 12:8) was applied as the blowing agent, PUE-4 elastomers with diverse lattice structures can achieve uniform foaming at 120 °C (Fig. 4i, j). Due to the unique characteristics of vat photopolymerization 3D printing, a chemically crosslinked network forms between layers during photocuring. This network evolves into a dynamically crosslinked-interpenetrating structure during the heated supercritical fluid foaming process, exhibiting crack-free interlayer cells with enhanced structural integrity (Fig.4k).

Mechanical performance of foamed elastomers

To systematically evaluate the mechanical behavior of elastomer foams, we fabricated dumbbell-shaped tensile specimens using vat photopolymerization 3D printing followed by supercritical fluid foaming (Fig. 5a). The foamed PUE-4 demonstrates superior mechanical performance compared to its as-printed counterpart, achieving a tensile strength of 5.5 MPa and an elongation at break of 510.8%. The stress-strain behavior of the foamed PUE-4 exhibits a characteristic J-shaped curve (Fig. 5b), consistent with classical elastomeric materials34. Real-time imaging during tensile testing further confirms the exceptional stretchability of the PUE-4 foam (Fig. 5c). The remarkable mechanical properties stem from structural features induced by supercritical fluid foaming. In situ formation of high-molecular-weight polyurethane/polyurea within a densely crosslinked matrix creates a dynamically crosslinked-interpenetrating network. This structure fosters a highly entangled polymer network within the foam. During tensile deformation, the entangled molecular chains progressively disentangle and align, enabling extensive elongation. Moreover, the supercritical fluid foaming process imparts a distinctive ‘turtle shell’-like morphology, featuring a closed-cell foam core encapsulated by a solid outer layer. This robust outer shell significantly enhances the tensile strength of the PUE-4 foam. Comparative analysis with existing hierarchically porous elastomers fabricated via extrusion-based 3D printing (DIW/FFF) highlights the superiority of PUE-4 foam (Fig. 5d and Table S2). Although a prior study achieved stretchable porous materials via vat photopolymerization and emulsification16, their tensile strength remains substantially inferior to PUE-4 foam. In addition, the static creep behavior of PUE-4 foam was evaluated (Figure S13). The sample was stretched at 50 mm min−1 to a stress of 0.5 MPa (approximately 10% of its tensile strength) and held for 900 s. Under constant stress, a significant creep response was observed, characterized by an increase in strain from 103.4% to 183.1%, indicating pronounced creep due to viscous sliding of polymer chains in the high-elastic state. Upon unloading, the foam exhibited excellent elastic recovery, with the strain rapidly decreasing to 15.3% within 20 minutes. This behavior can be attributed to the stability of the dynamic HUBs at room temperature35, maintaining the crosslinked network. Although PUE-4 foam exhibits excellent shape recovery, its limited dimensional stability under sustained loading may limit its application in precision components. Beyond static creep behavior, polyurethane elastomer foams are widely utilized across industries due to their exceptional resilience and energy absorption capabilities. A critical performance metric is their ability to endure cyclic loading without significant degradation. In this study, we assessed the resilience and durability of PUE-4 foams through cyclic compression testing. Figure 5e–h presents the results of 50 cycles of 50% compressive strain testing for solid blocks, foamed blocks, solid Kelvin structures, and foamed Kelvin structures. The foamed block (Fig. 5f) exhibited markedly improved resilience over the solid block (Fig. 5e), as evidenced by the reduced area within the hysteresis loops. Notably, after foaming, the energy loss coefficient decreased dramatically from 74.3% to 12.2% (Fig. 5i), and residual strain was reduced from 10.2% to 1.7% (Fig. 5j). Additionally, the foamed block achieved a drop ball resilience rate of 67.5% (Movie S1), comparable to values reported for thermoplastic polyurethane elastomer foams36. This substantial enhancement is attributed to the closed-cell structures formed during supercritical fluid foaming, which function as microscopic energy storage units. The elastic potential energy stored in the cell walls and the energy from compressed gas facilitate rapid recovery upon unloading. Furthermore, the uniform cellular architecture promotes homogeneous stress distribution, resulting in exceptional compressive fatigue resistance, with the foam retaining 96% of its initial maximum compressive stress after 50 cycles (Fig. 5k). We also investigated lattice-based designs, specifically Kelvin structures. The solid Kelvin lattice design significantly enhanced resilience and reduced both residual strain and energy loss compared to solid blocks (Fig. 5g). However, due to its open-cell configuration, solid Kelvin lattice exhibited lower energy storage capacity than closed-cell foams, leading to 44.6% energy loss in the first compression cycle (Fig. 5i). Following foaming, the foamed Kelvin lattice exhibited a narrowed hysteresis loop (Fig. 5h), reduced residual strain (decreasing from 6.0% in the solid state to 4.0% in the foamed state), and lowered energy loss (from 44.6% to 33.9%). These results demonstrate that supercritical fluid foaming, by forming closed-cell structures, significantly enhances the resilience of elastic materials. Moreover, the foamed Kelvin lattice displayed excellent fatigue durability, maintaining structural integrity without interlayer cracking after 1000 compression cycles (Figure S14).

Fig. 5: Mechanical properties of foamed PUE-4.
Fig. 5: Mechanical properties of foamed PUE-4.
Full size image

a Photograph of tensile specimens before and after foaming. b Tensile stress-strain curves of the as-printed PUE-4 and foamed PUE-4. c Real-time images showing the tensile deformation of PUE-4 foam. d Comparative analysis of tensile strength and elongation at break between foamed PUE-4 and hierarchically porous materials reported in the literature (see Table S2 for details and references). e Compressive stress-strain curves of solid block, (f) foamed block, (g) solid Kelvin lattice, and (h) foamed Kelvin lattice under cyclic compression. i Energy dissipation ratio, (j) residual strain, and (k) retention of max stress of solid block, foamed block, solid Kelvin lattice, and foamed Kelvin lattice.

Recyclability of hierarchically porous foams

The dynamic HUBs within the dynamically crosslinked-interpenetrating network structure impart excellent processability to PUE-4 (Fig. 6a). To quantify the kinetic characteristics of the dynamic exchange reaction, stress relaxation experiments were conducted at various temperatures. As shown in Fig. 6b, the exchange reaction rate of the HUBs increases significantly with temperature, leading to accelerated stress relaxation. By fitting the characteristic relaxation time (τ) to an Arrhenius plot, the activation energy (Ea) was determined to be 133.2 kJ mol−1. Furthermore, after fragmenting the PUE-4 foam, it can be reprocessed into a solid elastomer block through repeated hot pressing (130 °C, 2 MPa, 20 min) (Fig. 6c). This reprocessed elastomer exhibited exceptional tensile properties, recovering its original shape during rapid stretching and release (Fig. 6d, Movie S2). It maintained a tensile strength of 8.9 MPa and an elongation at break of 965.5% (Fig. 6e), primarily due to the dynamically crosslinked-interpenetrating network structure. The interpenetrating polyurethane/polyurea chains form a highly entangled network, providing the reprocessed elastomer with superior mechanical properties. Notably, the elastomer retained stable mechanical performance after multiple thermal reprocessing cycles, owing to the inherent thermal stability of PUE-4 (Figure S15) and the relatively low reprocessing temperature (130°C), which minimizes molecular weight degradation. Additionally, the reprocessed elastomer was successfully re-foamed using supercritical fluid foaming, resulting in a uniform closed-cell structure (Fig. 6f) with excellent resilience, as it preserved the dynamically crosslinked-interpenetrating network. Cyclic compression testing of the re-foamed elastomer (Fig. 6g) revealed a low residual strain of 5.1% and an energy loss coefficient of 11.7% in the first cycle, with a maximum compressive stress retention rate of 93.8% after 50 cycles (Fig. 6h).

Fig. 6: Reprocessing performance of foamed PUE-4.
Fig. 6: Reprocessing performance of foamed PUE-4.
Full size image

a Exchange reaction of hindered urea bonds (HUBs) within the dynamically crosslinked-interpenetrating network. b Stress relaxation curves of PUE-4 at various temperatures. (Inset: Corresponding Arrhenius plot). c Closed-loop recycling demonstration of a sectioned Kelvin lattice foam undergoing thermal reprocessing and supercritical fluid re-foaming. d Photographs of a reprocessed tensile specimen stretched to 700% strain and rapidly recovering to its original shape upon release. e Stress-strain curves of PUE-4 foam during repeated reprocessing cycles. f SEM image of supercritical fluid re-foamed PUE-4. g Cyclic compression stress-strain curves of supercritical fluid re-foamed PUE-4. h Residual strain, energy dissipation ratio, and retention of max stress of supercritical fluid re-foamed PUE-4 under cyclic compression.

This work presents an approach to fabricating hierarchically porous elastomeric materials by integrating vat photopolymerization 3D printing with supercritical fluid foaming, overcoming limitations of traditional pore-forming methods that often rely on toxic agents and struggle with structural integrity. By synthesizing a polyurethane acrylate oligomer with HUBs and engineering a dynamically crosslinked-interpenetrating network, we achieve precise control over crosslinking density and tensile properties, enabling the creation of lightweight, resilient foams with uniform porosity. Optimized through systematic tuning of PACM content and a hybrid N2/CO2 foaming system, the resulting elastomeric foams exhibit high tensile strength (5.5 MPa) and elongation (510.8%) through stress-distributing closed-cell architectures, along with 96% compressive stress retention after 50 cycles and a 67.5% drop ball resilience rate. Moreover, the dynamic HUBs-enabled network grants the foam exceptional recyclability. Through thermal reprocessing, the thermoset elastomer maintains a tensile strength of 8.9 MPa and an elongation of 965.5%. This innovative approach provides fresh perspectives for advancing research on lightweight, high-performance, and recyclable hierarchically porous elastomeric materials.

Methods

Materials

Isophorone diisocyanate (IPDI) (≥99%), 2-(tert-butylamino)ethyl methacrylate (TBEMA) (≥98%), poly(tetramethylene glycol) diol (PTMG, Mn = 2000 g mol−1) (≥98%), dibutyltin dilaurate (DBTDL) (≥95%), 2-ethylhexyl acrylate (2-EHA) (≥99%) and the photoinitiator, 2,4,6-trimethylbenzoyldiphenylphosphine oxide (TPO) (≥97%) were purchased from Shanghai Aladdin Biochemical Technology Co., Ltd. 4,4’-Methylenebis(cyclohexylamine) (PACM) (97%) was purchased from Shanghai Bide Pharmaceutical Technology Co., Ltd. The photoabsorber selected was the Neozapon Green 975 dye from BASF.

Synthesis of polyurethane acrylate oligomer

The polyurethane acrylate oligomer was synthesized via a two-step approach. Initially, 0.1 mol of polytetrahydrofuran glycol was weighed and placed in a 500 mL three-necked round-bottom flask. The flask was then placed in a vacuum oven at 110 °C for 2 h to remove moisture. Subsequently, the flask was removed from the oven and cooled to 50 °C. Under a nitrogen atmosphere, 0.2 mol of IPDI was slowly added dropwise, accompanied by the addition of a single drop of dibutyltin dilaurate as a catalyst. Upon the completion of the addition, the reaction temperature was elevated to 70 °C and maintained for an additional 2 hours, forming a terminal NCO group prepolymer. The reaction progress was monitored using Attenuated Total Reflectance Fourier Transform Infrared Spectroscopy (ATR-FTIR). Once the NCO characteristic absorption peak stabilized, the reaction temperature was reduced to 50 °C, and 0.2 mol of TBEMA was slowly introduced dropwise. The reaction was then continued for 0.5 h. The disappearance of the infrared characteristic absorption peak of NCO signified the completion of the reaction. Agitation was then stopped, and the resulting polyurethane acrylate oligomer was collected.

Preparation of photocurable 3D printing resin

According to the resin formula (Table S1), polyurethane acrylate (PUA), 2-EHA, photoinitiator, and photoabsorber were added to a 150 mL solder paste tank, placed in a vacuum planetary degassing mixer (ZYMC-200V, ZYE, China), and stirred at 1500 rpm for 3 minutes. Subsequently, PACM was added to the mixture in varying proportions as per the formulation, and stirred at 2000 rpm for 1 min to obtain a homogeneous photocurable 3D printing resin.

Supercritical fluid microcellular foaming

The 3D-printed polyurethane elastomers were placed in a high-pressure foaming autoclave (FP1770, Deyang, China) set at a temperature range of 100 to 130 °C. The autoclave was pressurized with gas at 20 MPa for 2 h. The high-pressure gas was then rapidly vented, and the samples were removed. In this study, various gases were employed as blowing agents for foaming. For instance, a ratio of N2:CO2 = 16:4 indicates that under a total gas pressure of 20 MPa, the partial pressures are 16 MPa for N2 and 4 MPa for CO2. Unless otherwise specified, all elastomeric foams with lattice structures demonstrated in this work were produced under foaming conditions of N2:CO2 = 12:8 at 120 °C.

Recyclability of foams

For the recycling and reprocessing of the foam, a torque rheometer (POTOP, Guangzhou Putong) was first used to mix at 130 °C for 10 min to form a uniform polymer mass, and then a flat plate vulcanizer (LN-251, Guangdong Lina) was used for hot pressing.

Characterization

During the synthesis of polyurethane acrylate, the process was monitored using an ATR-FTIR spectrometer (Nicolet iS5, Thermo Fisher Scientific, USA). The samples were scanned in the wavelength range of 4000 to 400 cm−1 with a resolution of 4 cm−1 over 16 scans. To monitor the photopolymerization rate of the photocurable resin in real time, a UV-LED light with a power of 3 mW cm−2 and a wavelength of 405 nm was positioned above the ATR sample stage. After each 2-second exposure of the resin to light, the change in the acrylic ester double bond absorption peak area at 810 cm−1 (A810) was measured. The area of the ester carbonyl (C = O) absorption peak at 1720 cm−1 (A1720) served as an internal standard. The conversion rate (α) of the double bonds (C = C) in the photopolymerization process of the resin was then calculated using Eq. (2):

$$\alpha=\left(1-\frac{\left(\frac{{A}_{810}}{{A}_{1720}}\right)_{{{{\rm{t}}}}}}{{\left(\frac{{A}_{810}}{{A}_{1720}}\right)}_{0}}\right)\times 100\%$$
(2)

where A0 and At correspond to the absorption peak areas of the resin before and during real-time curing, respectively. The 1H-NMR spectrum was acquired on a nuclear magnetic resonance (NMR) spectrometer (AVANCE-400M, Bruker, Germany) using DMSO-d6 as the solvent. The molecular weight of the prepolymer was determined by gel permeation chromatography (GPC, Breeze-2, Waters).

The viscosity, photo-rheological properties, and viscoelastic behaviors of the photocurable resin and 3D-printed elastomers were measured using a rotational rheometer (DHR-2, TA Instruments, USA). The viscosity of the resin was determined by employing a flow sweep method at 25 °C with the use of parallel plates that had a diameter of 40 mm, and the shear rate range was set from 0.01 to 1000 s−1. The photo-rheological behavior of the resin was measured using a UV-LED light source. The upper fixture consisted of a stainless steel parallel plate with a diameter of 20 mm, while the lower fixture was a high-transparency polymethyl methacrylate (PMMA) parallel plate with a diameter of 20 mm. At the 20-second mark, a UV-LED with a power intensity of 3.2 mW cm−2 was activated and irradiated for 100 seconds. The elastic modulus (G’) and loss modulus (G”) were monitored in vibration mode to investigate the reaction characteristics of the photocurable resin during the UV curing process. The viscoelastic behavior of the elastomer was tested by dynamic frequency sweeping test using a DHR-2 rheometer at 120 °C, with a strain of 0.1% and a frequency sweep range of 0.1 to 100 rad s-1. The sample size was a disc with a diameter of 25 mm and a thickness of 1.5 mm.

The annealing behavior of the 3D-printed polyurethane elastomers was assessed using a differential scanning calorimeter (DSC25, TA Instruments, USA) under a nitrogen atmosphere, with a heating rate of 10 °C min−1 over the temperature range of 25 to 180 °C. Thermogravimetric analysis was performed under nitrogen using a thermogravimetric machine (HS-TGA-101, HESON), and the temperature was raised from 30 to 600 °C at a rate of 10 °C min−1.

The crosslinking density (Ve) and the average molecular weight between crosslinks (Mc) of the polymer were determined using the equilibrium swelling method37. Initially, a known mass (m0) of polyurethane elastomer was immersed in toluene solvent for 48 h. The sample was weighed periodically until its mass remained constant over three consecutive measurements, indicating full swelling, with the final swollen mass recorded as m1. Subsequently, the swollen polyurethane elastomer was dried in an oven at 80 °C for 24 h, and the dried mass was recorded as m2. The Flory-Rehner Eqs. (3)-(5) was applied to calculate the desired parameters:

$${v}_{e}=-\frac{{{{\mathrm{ln}}}}\left(1-{V}_{r}\right)+{V}_{r}+\chi {{V}_{r}}^{2}}{{V}_{1}\left({V}_{r}^{\frac{1}{3}}-\frac{{V}_{r}}{2}\right)}$$
(3)

where Vr is the volume fraction of the polymer, V1 is the molar volume of the toluene solvent (106.2 cm3 mol−1), Ve is the crosslinking density, and χ is the interaction parameter between the polymer and the solvent (0.417).

The volume fraction Vr of the polymer is calculated by Eq. (4).

$${V}_{r}=\frac{{m}_{2}/{\rho }_{1}}{{m}_{2}/{\rho }_{1}+\left({m}_{1}-{m}_{2}\right)/{\rho }_{2}}$$
(4)

where ρ1 is the density of polyurethane elastomer, and ρ2 is the density of toluene.

Average molecular weight of polymer between crosslinks (Mc):

$${M}_{c}=\frac{{\rho }_{1}}{{v}_{e}}$$
(5)

The gravimetric method was employed to investigate the gas desorption behavior in the elastomer. Specifically, the elastomer was placed in an autoclave, and 6 MPa of gas was injected at ambient temperature for a saturation period of 6 h. Following this, the gas was gradually released, and the sample was promptly removed. The weight loss curve of the sample over a 90-minute period was recorded using an electronic balance. The expansion ratio (φ) was determined from the density ratio before and after foaming, as given by Eq. 6:

$$\varphi=\frac{{\rho }_{s}}{{\rho }_{f}}$$
(6)

where the density of the sample before (ρs) and after (ρf) foaming was measured directly using a densitometer (model DX-120TS, Qunlong, China).

Scanning electron microscope (SEM, EM-30N, COXEM, Korea) was used to observe the microscopic morphology of the cells. Cell size and density were analyzed from SEM images using Image-Pro Plus6.0 software. At least 100 cells were selected in one image to calculate the average cell size. The average cell size (Da) of the foam was calculated according to Eq. 7:

$${D}_{a}={\left(\frac{1}{n}{\sum}_{i=1}^{n}{d}_{i}^{3}\right)}^{\frac{1}{3}}$$
(7)

where di is the diameter of each cell; n is the number of cells counted in the SEM micrograph.

The cell density N0 is calculated according to the following Eq. 8:

$${N}_{0}=\varphi {\left(\frac{n}{A}\right)}^{\frac{3}{2}}$$
(8)

where A is the area of the SEM micrograph (unit is square centimeter, cm2).

The mechanical properties were assessed using a universal testing machine (AGX-100 plus, Shimadzu, Japan). To evaluate the high-temperature tensile properties of the elastomer, the green tensile specimens were annealed in an oven at 120 °C for 2 h, Subsequently, they were rapidly removed and tested at a loading rate of 1000 mm min−1. The tensile and cyclic compression properties of the foaming materials were evaluated at a loading rate of 50 mm min−1. The dimensions of the 3D printed dumbbell-shaped specimen are 60.0 mm × 12.8 mm × 2.0 mm, and the dimensions of the tensile test specimen of PUE-4 foam are 130.0 mm × 26.0 mm × 3.5 mm. Creep resistance was evaluated through tensile testing of dumbbell-shaped PUE-4 foam specimens at 25 °C. Samples were stretched at 50 mm min−1 to a constant tensile stress of 0.5 MPa and maintained at this stress for 900 s. Subsequently, the load was reduced to 0 MPa at 50 mm min−1, followed by a 2400 s recovery period during which strain evolution was monitored.