Atom-economy upcycling of commodity thermoset polyurethane into photocuring 3D printing resins based on selective cleavage—crosslink strategy

Abstract

Polyurethane waste poses significant environmental challenges due to its large bulk volume and thermoset nature. Current recycling methods suffer from high reagent consumption, urethane byproduct generation, and limited economic value. Here, we well address these issues through a selective cleavage—crosslink strategy using food-grade ethyl acetoacetate (EAA) to break biuret crosslinks while preserving urethane/urea bonds in non-crosslinked regions virtually intact. This strategy minimizes reagent consumption and avoids urethane byproduct. Importantly, the deconstructed product can be fully upcycled to high value-added photocuring 3D printing products with an exceptionally high waste content (up to ~90%) and excellent mechanical properties. Beyond 3D printing, the deconstructed product also holds potential as raw material for adhesives, PU synthesis, and epoxy modifiers. This strategy uses inexpensive commercial reagents and requires no modification to existing polyurethane infrastructure, establishing a commercially implementable pathway for polyurethane waste recycling.

Current recycling methods for polyurethane waste suffer from high reagent consumption, urethane byproduct generation, and limited economic value. Here, the authors demonstrate a selective cleavage—crosslink strategy using food-grade ethyl acetoacetate to break biuret crosslinks while preserving urethane/urea bonds in non-crosslinked regions intact.

Introduction

Polyurethane materials, due to their excellent mechanical properties and processability, are widely utilized in various forms such as foams, fibers, and elastomers across numerous sectors, including furniture, textiles, construction engineering, aerospace, transportation, etc. They have become indispensable materials in modern life. However, their large-scale application entails significant waste pollution challenges at end-of-service life for these reasons: (1) Their annual consumption is very large and has reached 22 million tons in 2023¹. (2) Most of them (67%) exist as foam with extremely low density (20–80 kg/m³), generating 0.4 billion cubic meters of annual waste volume—exceeding that of most other plastics^2,3. (3) The majority are thermosetting materials⁴ whose cross-linked network structure prevents recycling via melt extrusion like thermoplastics. Traditional disposal approaches like landfilling and incineration remain dominant, yet these methods generate secondary pollution that poses risks to both the environment and human health^5,6.

Numerous recycling strategies have been explored in recent decades. Physical recycling involves pulverizing polyurethane waste for direct use as fillers in new products or consolidating fragments via binding agents under high-temperature compression. Despite operational simplicity and low cost, this approach yields products with inferior performance and limited market value, significantly restricting large-scale implementation⁷. In contrast, chemical recycling depolymerizes polyurethane into new synthetic feedstocks, producing recycled materials with performance comparable to virgin counterparts. Recovering polyol from polyurethane through glycolysis exemplifies this method (Fig. 1a). However, this process necessitates cleaving all the polymerization bonds, consuming substantial alcoholysis reagents and subsequent isocyanate re-bonding agents, thereby increasing costs. Critically, while polyols are recoverable from depolymerized products, recycling urethane-based byproducts remains challenging, preventing full resource recovery from polyurethane waste^8–10. Other chemical methods, such as aminolysis and enzymatic degradation, can avoid discrete urethane-based byproducts^11–13, but still face challenges in selectivity, efficiency, cost, or scalability. Alternatively, employing dynamic bonds can enable thermoplastic reprocessability through molecular network rearrangement^14–17. This strategy can also avoid urethane-based byproducts generation, but requires either specialized molecular designs for dynamic bond incorporation^14,15 or significant catalyst loading to activate the dynamics of inherent urethane bonds^16,17. Moreover, dynamic bonds intrinsically compromise material stability, and reprocessing exposes polymers to heat, shear, oxygen, and moisture—inducing chain scission and performance degradation^18,19. Furthermore, since the molecular network of regenerated products remains identical or similar to virgin materials, their economic value can not exceed that of the original products.

Fig. 1. Strategies for polyurethane foam (PUF) recycling.

a Traditional chemical recycling via glycolysis, characterized by excessive reagent consumption, significant byproduct generation, and limited economic value. b New chemical recycling strategy based on selective cleavage of crosslinking sites, offering high atom economy (minimal reagent consumption and no byproduct), and high-value-added products.

In this work, we develop a recycling strategy that can not only upcycle full of commodity thermoset polyurethane foam (PUF) waste to high value-added 3D printing products but also be highly atom-economic, a term defined here as minimizing reagent consumption and avoiding the generation of stoichiometric byproducts. Clearly, this innovative strategy will make the recycling more economic and eco-friendly. We think it may be realized through selective cleavage of crosslinking sites—the key bonds preventing thermosetting materials from reusing—while maximizing preservation of chemical bonds in non-crosslinked regions. This strategy delivers three key advantages: (1) Selective cleavage primarily at crosslinking sites minimizes bond scission, enabling minimum reagent consumption. (2) Depolymerized products retain reusable polyol-isocyanate block polymer, eliminating urethane-based byproducts and achieving complete resource recovery from PU waste. (3) Abundant preserved unbroken bonds in the non-crosslinked regions can serve as potential grafting/functionalization sites, facilitating value-added material reuse. With this strategy, we hereafter show our successful effort in high-atom-economy upcycling of commodity thermoset polyurethane foams into photocuring 3D printing products.

Results

Design principle for high-atom-economy upcycling of PUF

The key point of this paper is achieving selective cleavage of crosslinking sites. For commodity polyurethane foam, its molecular network is mainly constructed by the reaction of diisocyanates with polyols, with water serving as both a reactant and a blowing agent. These reactions generate a crosslinked network with urethane, urea, and biuret bonds (Supplementary Fig. 1)³. It is noteworthy that allophanates, another type of crosslink formed by the reaction of an isocyanate with a urethane group, could theoretically be present^20,21. However, fourier-transform infrared (FTIR) analysis of the commodity PUF used in this study reveals three dominant and overlapping carbonyl peaks at 1730, 1710, and 1661 cm⁻¹ (Supplementary Fig. 2), assigned to biuret, urethane, and urea bonds, respectively³, while a distinct allophanate peak is absent. This indicates negligible allophanate content, consistent with the literature reports that where the reactivity of isocyanate with active hydrogen to form urethane/urea/biuret is over two orders of magnitude greater than its addition to urethane to yield allophanate²². Nonetheless, due to the structural analogy between allophanate and biuret crosslinks, the selective cleavage strategy described herein is expected to be equally applicable to PUF formulations containing non-negligible amounts of allophanates, thereby extending the generality of the approach. Among these bonds, the biuret bonds function as the crosslinking sites, while the urethane and urea bonds reside predominantly in the non-crosslinked regions. Consequently, the core strategy focuses on the selective cleavage of biuret bonds while maximizing the preservation of intact urethane and urea bonds.

Here, we find that ethyl acetoacetate (EAA), a widely used and inexpensive food additive with FDA approval under 21CFR172.515 and Flavis No. 9.402^23,24, meets the requirements for this selective bond-cleavage reaction. Employing EAA as the cleaving agent enables rapid, efficient, and complete scission of the crosslinking biuret sites while leaving urethane and urea bonds in non-crosslinked regions virtually intact (Fig. 1b). Consequently, the thermoset polyurethane foam can be decrosslinked into reworkable polymer with abundant active sites, bearing two types of abundant active sites: (i) preserved main-chain urea and urethane bonds, and (ii) newly generated terminal groups (acetoacetyl-formed amide, vinylogous urethane, and hydroxyl). This dual reactivity, spanning both the main chain and chain ends, provides versatile handles for chemical modification. Specifically, these sites readily react with isocyanate (NCO) groups, enabling efficient grafting and conversion of the recycled polymer into high-value-added photocurable 3D printing resins. Besides, the selective cleavage of biuret crosslinks minimizes bond scission and reagent consumption, and allows complete reuse of PUF waste, enabling high atom economy. In summary, this work demonstrates food-grade ethyl acetoacetate (EAA)-enabled selective scission of biuret crosslinks in thermoset polyurethane foam, achieving high-atom-economy upcycling to photocuring 3D printing products. It is worth noting that the selective cleavage strategy described herein is particularly effective for commodity polyurethane foams, which predominantly utilize diols and derive biuret crosslinks mainly from water-blowing reactions. For polyurethane crosslinked extensively via higher functional alcohols (≥3), the efficacy of this specific method may vary.

Deconstruction of PUF and revelation of the chemical mechanism

PUF deconstruction methodology is detailed in the Methods section. The industrial-grade PUF sample used in this work features a crosslinking network characterized by urea/urethane/biuret bonds at a molar of 3:1:0.2 (Supplementary Fig. 1). Briefly, the PUF sample is mechanically ground and blended with EAA. After heating for a defined period, complete dissolution of the PUF is achieved (Fig. 2a). At a fixed EAA/PUF weight ratio of 1, the requisite deconstruction time decreases from 280 min to 25 min with increasing temperature from 140 °C to 180 °C (Fig. 2b). Further the increasing deconstruction temperature would exceed the boiling point of ethyl acetoacetate, therefore the relevant experiment was not conducted. By fixing the temperature at 180 °C, the deconstruction time does not change notably as the EAA/PUF weight ratio increases (Fig. 2c). We note that when the EAA/PUF weight ratio falls below 1, the excessively high solid content significantly deteriorates the fluidity of the deconstruction solution at room temperature, which adversely affects subsequent 3D printing. Based on the above results, we picked EAA/PUF weight ratio of 1 at 180 °C for 25 min as the standard deconstruction condition for further investigation. Gel permeation chromatographic analysis (Supplementary Fig. 3) suggests that the deconstructed PUF has a weight-average molecular weight (M_w) of 41,146 g mol⁻¹, a number-average molecular weight (M_n) of 15,952 g mol⁻¹, and a polydispersity index (PDI) of 2.58. The high molecular weight is primarily due to the minimized bond scission achieved through our selective cleavage-crosslink strategy during deconstruction. The relatively broad molecular weight distribution (PDI > 2), manifested as multiple overlapping peaks in the chromatogram, is attributed to the structural heterogeneity of the original PUF, where biuret crosslinks form at varying sites.

Fig. 2. Deconstruction of polyurethane foam (PUF) and revelation of the chemical mechanism.

a Photographs of the PUF sample before and after deconstruction. b, c The impact of temperature and ethyl acetoacetate (EAA)/PUF weight ratio on the deconstruction time. d ¹H NMR (500 MHz) analyses of the cleavage reactions of biuret, urethane, and urea bonds by EAA in DMSO-d6. The total intensity of the peaks corresponding to aromatic hydrogens is set as 1, and the comparison of the intensity of the peak corresponding to NH in biuret/urethane/urea is used to calculate the conversion. e Cleavage reaction kinetics of biuret, urethane, and urea from NMR analyses. f Chemical deconstruction of PUF into reprocessable non-crosslinked polymer with abundant main-chain urea and urethane bonds along with small amounts of terminal groups (acetoacetyl-formed amide, vinylogous urethane, etc).

To elucidate the deconstruction mechanism, three model compounds bearing isolated urea, urethane, or biuret bonds are individually reacted with ethyl acetoacetate (EAA) under conditions identical to PUF deconstruction. The reaction products are analyzed by ¹H NMR (see “Model Compound Experiments” in Methods; Fig. 2d, e and Supplementary Figs. 4–13), with kinetic analysis focusing on the well-resolved signals in the 6–11 ppm region, as the methyl and methylene peaks in the 0–6 ppm region exhibit strong overlap due to the presence of EAA, its isomers, and the Z/E isomers of the degradation products. NMR analyses revealed EAA-mediated conversion of these bonds into acetoacetyl-formed amide, vinylogous urethane, alcohol, and urea moieties (Fig. 2d). It should be noted that exchange between the ester group of ethyl acetoacetate (EAA) and urethane bonds is theoretically possible^25,26. Previous studies indicate it requires stringent conditions (≥190 °C, often with catalysis) and proceeds with low efficiency. Given our deconstruction temperature (180 °C) is below these effective thresholds, such exchange is expected to be negligible. To confirm this, a model reaction between a urethane compound and benzyl benzoate (serving as an ester surrogate for EAA while excluding interference from its active methylene group) was conducted (Supplementary Fig. 14). The results show no detectable spectral changes indicative of exchange within 30 min, confirming that ester-urethane exchange is insignificant under our conditions and does not interfere with the selective cleavage mediated by the active methylene group of EAA. Kinetic studies revealed the bond cleavage order: biuret » urea > urethane (Fig. 2e and Supplementary Figs. 9–13). Consistent with this reactivity order, complete scission of biuret bonds occurs within 10 minutes of reaction initiation, whereas urea and urethane bonds exhibit minimal cleavage conversion of only 3% and 1%, respectively (Fig. 2e), confirming the exceptional bond selectivity. This selective cleavage strategy disassembles the crosslinked PUF network, yielding a reprocessable, non-crosslinked polymer. Crucially, this polymer retains abundant main-chain urea and urethane bonds (due to their low cleavage conversion) along with small amounts of terminal acetoacetyl-formed amide, vinylogous urethane, and hydroxyl groups (Fig. 2f). Owing to the selective scission of biuret crosslinks, the deconstructed product is essentially a linear polymer. The terminal group concentration, calculated from its number-average molecular weight (M_n = 15,952 g mol⁻¹), is approximately 0.125 mmol g⁻¹, quantitatively confirming the limited number of chain ends generated by this highly selective process.

Reconstruction of network and upcycling to photocuring 3D printing products

Photocuring 3D printing effectively addresses customization demands, and its commercially available resins command prices several to dozens of times higher than conventional plastics. Converting PUF waste as described above into photocurable 3D printing resins would substantially enhance its economic value-added potential. We herein present our value-added conversion methods.

As described in the above section, PUF waste has been successfully transformed into a decrosslinked product with abundant main-chain urea and urethane bonds along with small amounts of terminal acetoacetyl-formed amide, vinylogous urethane, and hydroxyl groups. Given the reactivity of these chemical bonds/groups with isocyanates (Fig. 3a–e)^{20–22,27,28}, methacryloyl isocyanate (MAI) is employed to graft-modify the deconstructed PUF waste, converting it into a photocurable resin (Fig. 3f). The grafting reaction between the deconstructed PUF and MAI proceeds efficiently without added catalyst, owing to residual polymerization catalysts present in the industrial PUF feedstock²⁹. FTIR analysis shows complete consumption of the isocyanate groups after reaction, confirming complete grafting of MAI (Supplementary Fig. 15). It should be noted that although reactions between isocyanates and urethane/urea bonds are typically reported to occur with low reactivity²², our model compound experiments confirm that these reactions can also take place under our grafting conditions (80 °C, 2 h), albeit with low conversion rates (2–3%) (Supplementary Fig. 16). Therefore, the observed complete grafting of MAI reflects the collective contribution of multiple reactive sites reacting with isocyanate groups. Subsequent photocuring enables network reconstruction, which is confirmed by dissolution experiments (Supplementary Fig. 17). Gel fraction tests suggest that all the regenerated products possess high gel content (above 80%, Supplementary Fig. 18), indicating the high crosslinking network. It should be noted that the gel content does not reach the theoretical value of 100%, which is attributed to the inherent structural and compositional heterogeneity of the deconstructed PUF. Its broad molecular-weight distribution and the varied reactivity of the different active sites toward isocyanate²² lead to a heterogeneity of grafting efficiency. This results in a small soluble fraction of polymer chains that are richer in low-reactivity sites and are thus less effectively incorporated into the crosslinked network. Notably, this architectural feature carries a positive impact on performance, as the partially soluble fraction enhances the material’s flexibility and elongation while the predominant crosslinked network ensures elasticity and mechanical strength. While the resulting regenerated product is a solid, photocured network rather than a re-foamed material, the chemical integrity of the urethane/urea bonds and polyether soft segments from the original PUF is largely preserved. This retention of the core chemical structure enables the material to inherit key properties such as flexibility and toughness from the PUF backbone.

Fig. 3. Reconstruction of network and upcycling to photocuring 3D printing products.

The reactions of isocyanate (NCO) with a urea, b urethane, c hydroxyl group, d vinylogous urethane, and e acetoacetyl-formed amide. f Chemical conversion of the deconstructed PUF into pohotcurable resin by employing methacryloyl isocyanate (MAI) as grafting reagent. g Stress-strain and h DMA curves of the photocured products with different polyurethane foam (PUF) waste proportion. i Cyclic loading-unloading curves of 89% PUF waste formulation in tensile test at 90% strain. Photocuring 3D-printed products with j 80% and k 89% PUF waste in the shape of snowflake, lattice structure, and shoe sole. l–m The repeated shape reconfiguration of the regenerated products by triggering the dynamic bond exchange (100 °C for 20 min) in the presence of external forces.

Here, the polyurethane waste content in the recycled product was set at 67–89%, with corresponding MAI content at 33–11%. By modulating the PUF waste ratio, we can regulate the concentration of flexible segment, grafting ratio, and molecular network crosslinking density, thereby tuning mechanical properties (Fig. 3g). With 67% PUF waste, the regenerated product exhibited tensile strength of 21.37 ± 1.87 MPa, breaking elongation of 75.28 ± 4.18%, elastic modulus of 135.16 ± 26.9 MPa, and toughness of 12.19 ± 1.16 MJ m⁻³. Increasing PUF waste to 80% and 89% would decrease the tensile strength to 11.76 ± 0.35 MPa and 7.72 ± 1.11 MPa, respectively. However, it significantly enhances the breaking elongation (197.69 ± 12.63% for 80% PUF waste, and 513.37 ± 114.48% for 89% PUF waste) and toughness (13.98 ± 0.96 MJ m⁻³ for 80% PUF waste, and 19.75 ± 5.63 MJ m⁻³ for 89% PUF waste). More importantly, the elastic modulus can be lowered to 23.65 ± 0.25 MPa for 80% PUF waste and 2.24 ± 0.18 MPa for 89% PUF waste, which can better align with flexible material requirements.

Dynamic mechanical analysis (DMA) reveals that the regenerated materials exhibit two glass-transition temperatures (T_g, Fig. 3h). The first (around −40 °C) corresponds to the polyether soft segments of the PUF, endowing the material with superior flexibility. The second (0–200 °C) arises from the combined contributions of the urethane/urea hard segments and the acrylic grafts, enhancing the mechanical strength. In the 89% PUF waste formulation, the acrylate content is relatively low. Both the acrylate grafts and the native hard segments are dispersed as discrete domains within the continuous soft segment matrix. This morphology leads to insufficient phase separation and a blurred interface between the phases, which merges the two transitions and produces the observed gradual increase in tan δ above 0 °C. As the PUF waste content decreases and the acrylic fraction increases, the acrylic/hard segment domains evolve from a dispersed to a more continuous morphology, enhancing phase separation and giving rise to a distinct and higher temperature transition peak. In addition to the transition behavior, DMA is used to characterize the operational temperature range of the regenerated products. As shown in Supplementary Fig. 19, the storage modulus (G’) remains higher than the loss modulus (G”) across the entire tested temperature range (−100 to 200 °C), confirming their elastic-dominant and dimensionally stable behavior. Based on the criterion of a 50% retention of the room-temperature G’, the upper service limit is estimated to be 61 °C for the 67% PUF waste formulation and 41 °C for the 89% PUF waste formulation. These results demonstrate that the upcycled materials maintain sufficient mechanical integrity over a temperature window suitable for flexible elastomer applications. Successive cyclic tensile tests suggest the regenerated products possess excellent resilience (Fig. 3i). The nearly complete recovery stems from the material’s high elasticity and low T_g, which also enable temporary shape-fixation below T_g and elastic recovery once above T_g, a behavior indicative of glass-transition-based shape-memory properties (Supplementary Fig. 20). Notably, the resulting elongation at break (exceeding 500% for the 89% PUF waste formulation) and resilience are competitive with and even better than those of typical commercial photocurable resins for 3D printing^30–32, demonstrating that high-value mechanical performance can be achieved even with a very high waste content. With the above chemistry, geometrically complex products can be produced by photocuring 3D printing (Fig. 3j, k). Furthermore, the dynamic exchange reactions of urethane/urea bonds, which can be activated by the residual polymerization catalysts under heating²⁹, impart the reconfiguration of the newly formed network^3,33, enabling solid-state plasticity. To validate this, the iso-strain stress relaxation test at 100 °C was conducted. As expected, almost 50% stress was relaxed within 40 min (Supplementary Fig. 21), which allows the regenerated products to be reformed into diverse geometries without predefined molds, simply by annealing at 100 °C under applied deformation loads (Fig. 3l, m). Such direct geometric reprogramming offers a significant advantage over conventional cross-linked elastomers, whose shape possibilities are inherently constrained by molding processes.

Atom economic analysis

Fully utilizing materials/ reagents achieves atom economy, which not only reduces costs but also advances sustainable development. For waste recycling, we posit that the ideal objective is to minimize byproduct generation and reagent consumption.

Regarding byproducts, the selectively deconstructed PUF yields reworkable polyol-isocyanate block polymer rather than isolated polyol and isocyanate derivatives, eliminating urethane byproducts typically generated in conventional glycolysis. Thus, our strategy produces virtually no residues and enables full reutilization of PUF waste.

Concerning reagent consumption, our process employs only two reagents: the bond-cleaving agent (EAA) and the grafting reagent (MAI). The EAA consumption is negligible due to the selective cleavage of biuret crosslinks, which requires minimal bond scission. This was verified through thermogravimetric analysis (TGA), fourier-transform infrared spectroscopy (FTIR), gravimetric measurements, and nuclear magnetic resonance (NMR) spectroscopy. With collecting and weighing, around 97.6% of EAA is recovered (Fig. 4a), with purity matching the pristine material as confirmed by identical ¹H NMR (Fig. 4b). In addition, the deconstructed PUF product exhibits distinct EAA characteristic bands at 1037 cm⁻¹ and 1149 cm⁻¹ corresponding to C-O bond, whereas these bands are absent in the regenerated products, which again confirms the near-complete removal of EAA for the system. For MAI consumption, it is inevitable as introducing photoreactive groups is essential for valorizing recycled products into photocurable 3D printing materials. However, its usage is minimal and can be further reduced to 11% via partial grafting strategy. Furthermore, the FTIR and TGA curves of the regenerated products are largely similar to those of the original PUF in their major profile (Fig. 4c, d). While the newly incorporated acrylate grafts cause a slight decrease in thermal stability, this retention of the major profile reflects the high proportion of preserved PUF backbone and the low reagent consumption. Consequently, the entire recycling process exhibits high atom economy. With this high-atom-economy upcycling feature, the material cost for converting PUF waste into photocurable 3D printing resin is estimated as low as US$2.1/kg (Supplementary Table 1), which is over an order of magnitude below commercial photocuring elastomer resins. Here, it should be noted that this analysis assumes a near-zero acquisition cost for end-of-life PUF feedstock, consistent with its status as a virtually valueless waste. The relevant logistical and indirect costs are difficult to estimate, but are anticipated to be minimized in a scaled-up and integrated recycling process.

Fig. 4. Atom economic analysis.

a Recovery efficiency of ethyl acetoacetate (EAA) with theoretical and experimental values from three parallel batches. The theoretical value is set as 100%. b ¹H NMR (500 MHz) spectra of the initial and recovered ethyl acetoacetate in DMSO-d6. c FTIR spectra of EAA, original polyurethane foam (PUF), deconstructed PUF, and regenerated products with 80% and 89% PUF waste. d TGA curves of the original PUF and regenerated products with 80% and 89% PUF waste.

Exploration and discussion of the application potential of the deconsturcted PUF

In the above sections, we have discussed the deconstruction of PUF by employing selective cleavage—crosslink strategy using EAA, and the subsequent upcycling towards 3D printing products. Although 3D printed products hold high economic value and are developing rapidly, current market demand remains very limited. Therefore, exploring and discussing the application potential of this recycling method will facilitate its promotion. Here, we discussed its potential as pressure-sensitive adhesive, feedstock for polyurethane synthesis, and curing/toughening agent for epoxy resins.

Pressure-sensitive adhesives (PSAs) are widely used in tapes, bandages, and sticky notes owing to their temporary bonding capability via rapid and low-strength adhesion. Their bonding mechanism relies on pressure-activated noncovalent interactions, eliminating the need for thermal or radiative activation. Herein, we hypothesized that the flexible polyether segments in the deconstructed PUF could facilitate surface wetting at ambient conditions and enable pressure-sensitive adhesion. To test this hypothesis, a series of pressure-sensitive adhesion experiments was performed by bonding multiple substrates over a surface area of 100 mm² with the deconstructed PUF using manual application. As a result, the PUF-based PSA can adhere to all the tested substrates, including steel, aluminum, paper, and Teflon (Fig. 5a, b). Peel tests suggest that the peel strength of the PUF-based PSA to steel, aluminum, paper, and Teflon is 30.0 ± 8.3 N/m, 9.5 ± 2.2 N/m, 10.9 ± 1.0 N/m, and 5.6 ± 1.2 N/m, respectively (Fig. 5c, d). In addition, the PSA also exhibits excellent rebonding capability and reusability. Repeated peel tests suggest that the peel strength of the PUF-based PSA on steel shows no obvious degradation after running three cycles of peeling and repressing of the adhesion joint (Fig. 5e). The aforementioned adhesion features make it a promising candidate for PSA applications.

Fig. 5. Application in pressure sensitive adhesive (PSA).

a, b Photographs showing pressure-sensitive adhesion of the deconstructed polyurethane foam (PUF) with Teflon, aluminum, paper, and steel. c Schematic of peel tests. d Peel strength of the PUF-based PSA on four common substrates. e Peel strength of the PUF-based PSA on steel showing no obvious degradation after running three cycles of peeling and repressing of the adhesion joint. Data are presented as mean values ± standard error of the mean (s.e.m.). Centre values denote average. Error bars denote s.e.m. n  ≥  3 for the samples.

Given the abundant active sites in the deconstructed PUF and their reactivity with isocyanates (Fig. 3a–e), this material can serve as a feedstock to synthesize polyurethanes through reaction with diisocyanates. Alternatively, owing to the reactivity of amine groups, which can be generated from thermal dissociation of dynamic urea bond, toward epoxy groups (Supplementary Fig. 22), and the presence of flexible polyether segments, the material also holds potential as a possible curing and toughening agent for epoxy resins.

In summary, we demonstrate a chemical recycling strategy for commodity thermoset PUF waste based on selective cleavage of crosslinks. This allows us to not only fully upcycle of the waste to high value-added 3D printing products but also achieve high atom economy. This strategy is economically attractive and does not require changing existing PUF products and production infrastructures. All reagents employed are commercially available, low-cost, and require minimal consumption, making it easy to be implemented in large scale. Furthermore, the products regenerated through this selective cleavage process exhibit versatile application potential beyond 3D printing, serving effectively as pressure-sensitive adhesives, feedstock for polyurethane synthesis, and curing/toughening agents for epoxy resins, which points to a promising direction in future sustainable development.

Methods

Materials

Ethyl acetoacetate (EAA) was purchased from Macklin. Methacryloyl isocyanate (MAI) and benzyl benzoate were purchased from Energy Chemical. Dimethyl sulfoxide (DMSO), tetrahydrofuran (THF), and N, N-dimethylformamide (DMF) were purchased from Sinopharm Chemical Reagent. p-Tolyl isocyanate, p-toluidine, and 1-octanol were all obtained from TCI. Phenyl bis(2,4,6-trimethylbenzoyl)-phosphine oxide was purchased from Aladdin. The commercial PUF sourced from UE Furniture Co., Ltd.‘s production line was mechanically ground for 4 min using a crusher (Huangdai 800Y, Zhejiang Yongkang Platinum Metal Products Co., Ltd., China) and subsequently dried prior to use.

General measurements

¹H-nuclear magnetic resonance (¹H NMR) spectra were recorded by Bruker Avance 500 NMR spectrometer. Chemical shifts (δ) were reported in ppm relative to residual solvent signals (DMSO-d6: 2.5 ppm, water in the deuterium reagent: 3.3–3.5 ppm). Tensile tests were performed on an electronic universal testing machine (SUNS) at a tensile rate of 20 mm min⁻¹. Cyclic tensile tests were conducted at a tensile rate of 100 mm min⁻¹ and a recovery rate of 20 mm min⁻¹. Gel permeation chromatography (GPC, Waters 1525) analysis was performed with tetrahydrofuran as the eluent at a flow rate of 1 mL min⁻¹ at 36 °C using polystyrene as the standard. Stress relaxation experiment was performed at 100 °C using DMA Q800 (TA instruments, Germany) in a “stress relaxation” mode with an iso-strain value set at 100%, and glass transition temperatures (T_g) were tested using the same equipment. Thermal gravimetric analysis was performed on a Thermo Electron NICOLET 5700 at a heating rate of 10 °C min⁻¹ in nitrogen. Fourier transform infrared (FTIR) spectra was performed on a Bruker VERTEX 70 v spectrometer. The spectra were collected through 16 scans with a spectral resolution of 4 cm⁻¹. Gel fraction tests were conducted through solvent extraction in DMF for 36 h and vacuum-drying at 50 °C for 2 h.

Upcycling PUF waste to photocuring 3D printing products

4 g PUF containing urea/urethane/biuret bonds (Supplementary Fig. 1) was dried under vacuum at 180 °C for 10 min and subsequently mixed with EAA (4 g). After reaction at 180 °C for 25 minutes, the PUF dissolved, forming a homogeneous solution. This degradation solution was then mixed with MAI (0.5 g, 1.0 g, or 2.0 g). Following reaction at 80 °C for 2 h and addition of photo-initiator (Irgacure 819, 3%), an acrylate-blocked photocurable precursor derived from PUF was obtained.

3D printing fabrication was performed using a desktop 3D printer (Miracle 3D-SLA150ultra) with a layer thickness of 50 µm and an exposure time of 30 s per layer. Post-printing, the samples underwent vacuum drying to remove the unreacted EAA (1 h). To assess the recoverability of EAA and the atom economy of the process, a recovery apparatus based on vacuum distillation was employed. The samples were heated under vacuum to distill off the unreacted EAA, which was then condensed and collected. The masses of the recovered EAA and the remaining product were measured to determine the recovery yield. The high purity of the recovered EAA was confirmed by ¹H NMR with pristine EAA as a reference. The remaining products were then post-cured in a UV chamber (120 s) to yield the final regenerated products. The PUF waste content in these products, depending on EAA dosage, was 89%, 80%, and 67%, respectively.

Model compound experiments

To elucidate the degradation reactions of polyurethane foam (PUF), model compound experiments were performed. Three small molecules containing urea, urethane, and biuret functional groups were synthesized, and their degradation behaviors, including products and kinetics, were subsequently investigated. The molar ratios of model compound to EAA were selected to approximately mirror the relative abundance of each bond type (urethane, urea, biuret) within the PUF network, thereby simulating the local chemical environment during bulk deconstruction.

Synthesis of octyl p-tolylcarbamate

Octyl p-tolylcarbamate, the urethane model compound, was prepared by dropwise addition of p-tolyl isocyanate (1.33 g, 10 mmol) to a solution of 1-octanol (1.3 g, 10 mmol) in toluene (5 g) under magnetic stirring. The mixture was reacted at 80 °C for 3 h. Following vacuum drying at room temperature for 24 h, the product was obtained. Its structure was verified by ¹H NMR (Supplementary Fig. 4).

Synthesis of 1,3-di-p-tolylurea

1,3-Di-p-tolylurea, the urea model compound, was synthesized by adding p-tolyl isocyanate (1.33 g, 10 mmol) dropwise to a solution of p-toluidine (1.07 g, 10 mmol) in THF (10 g) with magnetic stirring. The reaction proceeded at room temperature for 10 min. The precipitate formed was collected by filtration, washed with THF, and vacuum-dried at room temperature for 24 h to obtain the product. ¹H NMR analysis confirmed its structure (Supplementary Fig. 5).

Synthesis of 1,3,5-tri-p-tolylbiuret

1,3,5-Tri-p-tolylbiuret, the biuret model compound, was obtained by reacting 1,3-di-p-tolylurea (2.4 g, 10 mmol) with excess p-tolyl isocyanate (10.64 g, 80 mmol) at 130 °C for 6 h under magnetic stirring. After the reaction, excess p-tolyl isocyanate was subsequently removed by rotary evaporation at 90 °C for 2 h under vacuum. The product structure was confirmed by ¹H NMR (Supplementary Fig. 6).

Synthesis of acetoacetyl-formed amide

p-Tolyl isocyanate (1.33 g, 10 mmol) and ethyl acetoacetate (EAA) (1.30 g, 10 mmol) were mixed in DMSO-d6 (11 mL). After reaction at 80 °C for 1 h, acetoacetyl-formed amide was obtained. Its structure was confirmed by ¹H NMR (Supplementary Fig. 7).

Degradation of octyl p-tolylcarbamate

Degradation of octyl p-tolylcarbamate was carried out with EAA at 180 °C for predetermined times (urethane/EAA molar ratio was set as 1:17). The proposed degradation reaction pathway is illustrated in Supplementary Fig. 8. The reaction progress was quantitatively monitored via ¹H NMR analysis at different degradation times (Supplementary Fig. 9).

Degradation of 1,3-di-p-tolylurea

Degradation of 1,3-di-p-tolylurea was carried out with EAA at 180 °C for predetermined times (urea/EAA molar ratio was set as 3:17). The proposed degradation reaction pathway is illustrated in Supplementary Fig. 10. The reaction progress was quantitatively monitored via ¹H NMR analysis at different degradation times (Supplementary Fig. 11).

Degradation of 1,3,5-tri-p-tolylbiuret

Degradation of 1,3,5-tri-p-tolylbiuret was carried out with EAA at 180 °C for predetermined times (biuret/EAA molar ratio was set as 0.2:17). The proposed degradation reaction pathway is illustrated in Supplementary Fig. 12. The reaction progress was quantitatively monitored via ¹H NMR analysis at different degradation times (Supplementary Fig. 13).

Ester-urethane exchange reaction

Exchange reaction of octyl p-tolylcarbamate with benzyl benzoate was carried out at 180 °C for predetermined times (urethane/benzyl benzoate molar ratio was set as 1:17). The proposed exchange reaction is illustrated in Supplementary Fig. 14. The reaction progress was quantitatively monitored via ¹H NMR analysis at different degradation times (Supplementary Fig. 14).

Shape reconfiguration experiments

Shape reconfiguration was achieved by deforming the regenerated material to a desired form, followed by thermal annealing under load at 100 °C for 20 minutes. Subsequent cooling to room temperature locked in a new permanent 3D configuration. This reprogramming cycle using distinct deformation forces during identical annealing conditions enabled the material to be successively reshaped into different geometries.

Author contributions

Z.L., J.C., and Y.Z. conceived the concept and supervised the project. Y.H. performed the experiments with assistance from X.G., Y.D., S.Y., and Y.Z. All authors participated in result analysis and discussion. Z.L. wrote the manuscript.

Peer review

Peer review information

Nature Communications thanks Pere Verdugo Fernández, and the other, anonymous, reviewers for their contribution to the peer review of this work. A peer review file is available.

Data availability

The data generated in this study have been deposited in the Figshare database (10.6084/m9.figshare.31552786). All data are available from the corresponding author upon request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-70951-w.