From Furfural to Reprocessable Thermosets: Diels–Alder Monomers for Dynamic Polyurethane Networks

Abstract

Covalent adaptable networks (CANs) offer a promising route toward sustainable polymers by addressing the inherent conflict between thermoset performance and thermoplastic reprocessability through exchangeable covalent bonds. This review highlights a bio‐based synthetic approach utilizing functionalized furan‐maleimide Diels–Alder adducts, derived from furfural, as preformed monomers for polyurethane CANs. Unlike conventional postpolymerization crosslinking, this adduct‐based strategy allows predictable incorporation of dynamic motifs into the backbone. This enables precise control over network topology and crosslink density while advancing sustainability through renewable feedstocks and recyclable network design. We demonstrate the advantages of the adduct‐based approach and establish structure–property relationships that yield materials combining thermoset‐like mechanical properties with stimuli‐triggered self‐healing, shape memory, and recyclability. Emerging applications in 3D/4D printing, UV‐curable and waterborne coatings, and debondable adhesives are surveyed. Persistent challenges, including thermal and chemical instability of furanic and maleimide components, high activation barriers for network rearrangement, reliance on nonrenewable comonomers, and potential toxicity concerns, are also discussed. Future directions emphasize rationally designed multidynamic networks, robust dienophiles, and nonisocyanate routes to bridge academic innovation and industrial viability toward fully circular polymer systems.

Reversibility for sustainability, but not yet perfect. Furan‐maleimide Diels–Alder (fmDA) chemistry enables self‐healing, shape memory, and reprocessability in polyurethane covalent adaptable networks through catalyst‐free and atom‐economic bond exchange. Nevertheless, the inherent strength‐dynamics trade‐off, a narrow reprocessing window, and degradation upon thermal cycling constrain both recyclability and service life. Realizing the full sustainability potential of furfural‐derived fmDA‐polyurethanes requires rational network design and careful process optimization.

Abbreviations

BDO

1,4‐butanediol

BHMF

2,5‐bis(hydroxymethyl)furan

BMI

4,4′‐bis(maleimido)diphenylmethane

CAN

covalent adaptable network

CMR

carcinogenic, mutagenic, reprotoxic

CNT

carbon nanotube

DA

Diels–Alder

DBTDL

dibutyltin dilaurate

DETA

diethylene triamine

DGEBA

diglycidyl ether bisphenol‐A

DMAc

dimethylacetamide

DMF

dimethylformamide

DMG

dimethylglyoxime

DMPA

2,2‐bis(hydroxymethyl)propionic acid

DMSO

dimethyl sulfoxide

E _b

elongation at break

EG

ethylene glycol

FDCA

2,5‐furandicarboxylic acid

FGNS

functionalized graphene nanosheets

fmDA

furan/maleimide Diels–Alder

GU

gloss units

H₁₂MDI

bis(4‐isocyanatocyclohexyl)methane

HDI

1,6‐hexamethylene diisocyanate

HEA

hydroxyethyl acrylate

HEMA

hydroxyethyl methacrylate

HIFU

high‐intensity focused ultrasound

HMA

5‐(2‐hydroxyethyl)‐6‐methyl‐2‐aminouracil

HMF

5‐hydroxymethylfurfural

HPA

hydroxypropyl acrylate

HPED

N,N,N′,N′‐tetrakis(2‐hydroxypropyl)ethylenediamine

HPMA

hydroxypropyl methacrylate

HUBs

hindered urea bonds

IPDI

isophorone diisocyanate

LCA

life cycle assessment

M _c

molar mass between cross‐links

MDI

methylene diphenyl diisocyanate

NIPU

nonisocyanate polyurethane

NIR

near‐infrared

PCDL

polycarbonate diol

PCL

poly (ε‐caprolactone) diol

PDMS

polydimethylsiloxane

PEG

polyethylene glycol

PETMP

2‐bis(((3‐mercaptopropanoyl)oxy)methyl)propane‐1,3‐diylbis(3‐mercaptopropanoate)

PGA

polyglycolic acid

pHDI

poly(hexamethylene diisocyanate)

PLA

polylactic acid

PPG

polypropylene glycol

PTMEG

polytetramethylene ether glycol

PU

polyurethane

rDA

retro‐Diels–Alder

R _f

shape fixity ratio

R _r

shape recovery ratio

SLS

selective laser sintering

SMASH

shape memory assisted self‐healing

TAD

1,2,4‐triazoline‐3,5‐dione

TEA

triethanolamine

THF

tetrahydrofuran

TMPTMP

trimethylolpropane tris(3‐mercaptopropionate)

tri‐HDI

hexamethylene diisocyanate trimer

UA

urethane acrylate

UPy

2‐ureido‐4[1H]‐pyrimidinone

UV

ultraviolet

1. Introduction

The transition toward sustainable manufacturing and, more specifically, toward a circular economy demands materials that combine high performance with recyclability, thereby extending service lifetime, reducing waste, and facilitating end‐of‐life recovery [1, 2, 3]. These goals are achieved through adaptability and stimuli responsiveness, which provide self‐healing, reprocessability, shape memory, and debonding‐on‐demand [4, 5, 6, 7, 8, 9, 10, 11]. Conventional polymers fall into two extremes, which arise from their underlying molecular structure and manifest as the core conflict between performance and recyclability. Linear thermoplastics are recyclable but lack mechanical strength and thermal stability. In contrast, thermosets offer greater mechanical strength, chemical resistance, and thermal stability but cannot be reprocessed or recycled because their permanent covalent crosslinks prevent melting or dissolution. Covalent adaptable networks (CANs) have emerged as a transformative solution to the fundamental barrier to thermoset recyclability by incorporating exchangeable covalent bonds into cross‐linked architectures [12, 13, 14, 15, 16, 17, 18, 19, 20, 21]. This design retains desirable thermoset attributes while enabling thermoplastic‐like reprocessability. However, the value of CANs lies not in replacing traditional polymers for conventional uses but in enabling entirely new functionalities that are fundamentally unattainable with either polymer class alone. These capabilities span key industrial sectors, including aerospace, automotive, biomedical, electronics, and additive manufacturing [22, 23, 24, 25, 26, 27], where functionality, long‐term durability, reprocessability, and dynamicity are equally critical.

The development of these materials often relies on petroleum‐derived feedstocks, which compromises their life‐cycle sustainability. Accordingly, full compliance with green chemistry principles dictates that CANs must be derived exclusively from renewable resources [28, 29, 30, 31, 32, 33, 34]. Among the various reversible covalent bonds employed in stimuli‐responsive networks, the Diels–Alder reaction between furan and maleimide (fmDA) is particularly attractive for this purpose. This is due to its thermal reversibility, click chemistry characteristics, catalyst‐free bond exchange, and compatibility with renewable furans [35, 36, 37, 38, 39]. Biobased furanic platform chemicals such as furfural [40, 41, 42, 43, 44, 45, 46] and 5‐hydroxymethylfurfural (HMF) [47, 48, 49, 50, 51, 52, 53], derived from pentose and hexose sugars, respectively, afford reversible covalent bonds while replacing petroleum‐based components with renewable ones. Produced industrially at ~400,000 tonnes per year, furfural is a pragmatic choice for developing sustainable polymer systems [54, 55, 56, 57, 58]. Because of the presence of more functional groups than furfural, HMF offers greater synthetic potential, enabling a more powerful synergy between sustainability and functionality. However, despite commercialization efforts and the fact that the cost of HMF‐derived monomers like FDCA can be comparable to that of fossil‐based analogs, HMF remains a niche chemical with limited availability [59, 60, 61, 62].

Polyurethanes (PUs) are among the most commercially important polymers, with annual production exceeding 20 million tonnes and applications ranging from flexible foams to rigid insulation [63, 64, 65]. Among these well‐established applications, PUs are particularly suited for CANs because of their unique microphase‐separated morphology: A self‐assembled structure comprising hard and soft segments [66, 67, 68, 69]. The hard segments, derived from diisocyanates, chain extenders, and low‐molecular‐weight diols, act as permanent physical crosslinks and impart mechanical strength, thermal properties, and solvent resistance. The soft segments, based on long, flexible polyether or polyester polyols, afford the molecular mobility essential for dynamic bond exchange and deliver elasticity and toughness. This combination of structural morphology and segmental mobility makes PUs an ideal platform for self‐healing materials. While the hard domains maintain structural integrity, the soft segments, together with reversible cross‐links, ensure the chain mobility necessary for macroscopic flow and repair [66, 70, 71].

The advantages of fmDA chemistry translate into several concrete performance benefits that are unattainable in conventional cross‐linked PUs. These include intrinsic self‐healing and shape‐memory behaviors, enabling autonomic repair of microcracks and extending service life in coatings and elastomers [63, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81]. Thermal reprocessability enables reshaping of crosslinked parts at elevated temperatures [36, 82, 83, 84, 85], while debonding‐on‐demand is useful for adhesives and composite disassembly [63, 86]. Biocompatibility and tunable degradation of fmDA‐cross‐linked PUs also allow biomedical uses [87, 88, 89], while their integration with conductive fillers has been explored for flexible/wearable electronics [90, 91, 92, 93, 94].

The implementation of fmDA chemistry in CANs can follow two strategies: postpolymerisation crosslinking of preformed polymer chains or the use of preformed fmDA adducts as monomers [95]. The adduct‐based approach, despite its advantages in structural control and predictive design, has received considerably less systematic attention. This gap is the central focus of the present review. Unlike existing reviews that separately address biobased polyurethanes [96, 97, 98, 99, 100], polyurethane‐derived CANs [101, 102, 103, 104, 105, 106, 107], or broader classes of fmDA‐based dynamic systems [39, 75, 95, 108, 109, 110, 111, 112], this work deliberately bridges renewable feedstocks, Diels–Alder chemistry, and the polyurethane platform into a single conceptual framework. We systematically compare the adduct‐based strategy using preformed 7‐oxanorbornenes from furfural‐derived intermediates with the more common prepolymer cross‐linking method. Quantitative structure–property relationships, key challenges, and perspectives are also highlighted. Our aim is to provide a focused, application‐oriented analysis that guides rational design of sustainable, high‐performance PU CANs.

Throughout this review, the term dynamic/reprocessable thermoset refers to any crosslinked polymer network whose topology can be rearranged via dynamic covalent bonds, encompassing CANs, dynamic elastomers, and reprocessable networks. While illustrated here through polyurethanes, the adduct‐based fmDA approach is modular and applicable to other cross‐linked platforms. Polyurethanes were chosen for several reasons. They offer the richest literature for systematic analysis and wide property tunability that allows testing across diverse mechanical requirements. Their microphase‐separated morphology provides unique insights into how reversible covalent bonds and physical cross‐linking jointly govern network dynamics, and their established industrial footprint ensures practical relevance.

Figure 1 outlines the overall conceptual framework for this integrated approach. The sustainable roadmap begins with the catalytic conversion of lignocellulosic biomass into furfural, followed by its modification into functionalized furan‐maleimide DA adducts. These compounds serve as key components for the synthesis of polyurethanes with a thermoreversible network. The sustainability of these polymers stems from three important features: (i) the renewable furfural feedstock, (ii) the extended service life enabled by self‐healing, and (iii) the circularity ensured by reprocessability. This three‐level framework, which is rarely addressed comprehensively in other dynamic covalent platforms, provides a holistic lens for evaluating the true circularity potential of fmDA polyurethane CANs. A more comprehensive sustainability evaluation requires life cycle assessment and ecotoxicity studies; these aspects are discussed and addressed as open challenges in the Conclusions section. Conceptually, this monomer‐based roadmap aligns with the recent evolution of covalent adaptable network research from simple proof‐of‐reversibility toward structure‐guided control of exchange kinetics, topology, creep resistance, and circularity. In this broader context, preformed 7‐oxanorbornenes are important because they decouple dynamic‐bond design from the subsequent polyurethane preparation stage and thus offer a more sophisticated and predictive approach to tailoring material properties [21, 113].

FIGURE 1.

Sustainable roadmap for polyurethane CANs: from furfural‐derived fmDA adducts to adaptive materials for advanced manufacturing (PU = polyurethane and FG = functional group).

In the following sections, we critically evaluate the reported fmDA‐polyurethane CANs obtained by the adduct‐based approach against several key criteria. These include mechanical performance and self‐healing efficiency, reprocessability and recycling depth, bio‐based content and sustainability implications, toxicity and ecotoxicity aspects, as well as scalability and industrial relevance. Guided by this framework, our analysis moves beyond simple performance reporting to provide a balanced assessment of the current state of the art.

2. Furfural‐Derived Diels–Alder Monomers: A Versatile Bridge From Biomass to Dynamic Polyurethanes

2.1. Advantages and Limitations of fmDA‐Based Polyurethane CANs

The reversible nature of dynamic covalent bonds in CANs translates into a range of functional behaviors that collectively underpin the sustainability of these materials: self‐healing, shape memory, reprocessability, and debonding‐on‐demand. Self‐healing extends service lifetime by autonomously repairing damage, reducing replacement frequency and material demand [111, 114, 115], while shape memory enables reversible shape changes [116]. A common challenge for efficient self‐healing in CANs is the requirement for an external force to close cracks prior to healing. To address this, researchers integrate self‐healing and shape‐memory functions in the so‐called shape memory assisted self‐healing (SMASH), which autonomously closes macroscopic cracks using shape recovery, eliminating the need for external intervention [117, 118, 119, 120, 121]. This SMASH technology is pivotal for time‐responsive materials, including those designed for 4D printing and programmable actuators [122, 123, 124, 125, 126, 127]. Reprocessability allows material recovery through thermal decrosslinking and reshaping, which thereby supports recycling [82, 128]. Debonding‐on‐demand facilitates disassembly of bonded assemblies, easing component separation [129, 130]. Together, these dynamic properties address the core challenges of durability, circularity, and advanced manufacturing, translating directly into lower raw material consumption, reduced manufacturing energy, and decreased waste generation.

CANs can be classified according to their exchange pathway (dissociative vs. associative), network behavior, and complementarity of reactive units (Figure 2, Table 1). In dissociative systems, such as those based on the fmDA reaction, imine condensation and hydrolysis, or hindered urea bonds (HUBs), bond cleavage precedes reformation, enabling sharp temperature responsiveness but temporarily reducing network connectivity [14, 135]. In associative counterparts, often referred to as vitrimers, which include those relying on transesterification, disulfide metathesis, imine transimination/metathesis, or vinylogous urethane exchange, bond formation and cleavage occur simultaneously, preserving cross‐link density throughout the exchange process [148, 149, 150, 151, 152]. Associative and dissociative mechanisms differ fundamentally in their kinetics and thermodynamics [21, 82, 135]. Dissociative CANs typically operate at lower temperatures and exhibit faster bond exchange rates due to the temporary reduction in network connectivity, which enhances polymer chain mobility during the cleavage step. In contrast, associative mechanisms preserve cross‐link density throughout the exchange process, which generally results in higher activation barriers and slower dynamics at equivalent temperatures. Consequently, while both mechanisms enable reprocessability and healability, associative networks are advantageous for creep resistance but require higher temperatures and longer times to achieve sufficient dynamics compared to dissociative networks. This distinction largely determines their practical niches: dissociative CANs are well‐suited for low‐temperature (0–100 °C, low T) self‐healing and reprocessing, while associative ones excel in high‐temperature (>150 °C, high T) structural applications where network integrity must be maintained. However, the temperature window for bond exchange is not solely a function of the mechanism type; rather, it is largely determined by the specific chemistry of each system [21, 143, 153]. Several dynamic linkages, including disulfides, urea/urethane, boronic esters, and imines, can operate via either dissociative or associative pathways, depending not only on temperature but also on moisture and the presence of catalysts or nucleophiles [82, 135, 143, 131].

FIGURE 2.

Classification of CANs by bond exchange mechanism (dissociative vs. associative) and their typical reprocessing temperature ranges, based on the literature data referenced in Table 1.

TABLE 1.

Comparison of CAN systems based on different dynamic covalent chemistries. Diss. mean dissociative; ass. mean associative.

No	CAN system/mechanism	Typical T reprocess , °C	Key advantages	Key limitations	Primary applications	Ref.
1	Disulfides (temperature‐dependent diss. radical/ass. metathesis)	RT to ~180	Redox and photo responsiveness, fast dynamics (seconds–minutes), catalyst–free bond exchange, RT self‐healing, biocompatible	Oxidation/reduction sensitivity, creep due to spontaneous exchange	Biomedical, flexible electronics	[82, 131, 132]
2	Imines (pH‐dependent diss. hydrolysis/ass. transimination)	RT to ~120	pH responsiveness, catalyst‐free synthesis and bond exchange, fast exchange (minutes, RT), easy synthesis	Hydrolysis sensitivity	Hydrogels, flexible electronics, sensors, fire‐retardant materials	[82, 133, 134]
3	Thiol‐ene/thiol‐Michael (ass./diss. depending on conditions)	RT–120	Fast kinetics, catalyst‐free options available, high selectivity, photo‐ and thermo‐activation, "click” chemistry, wide monomer scope	Some precursors toxic, unpleasant thiol odor, limited high‐temperature stability, oxygen sensitivity	Hydrogels, self‐healing coatings and adhesives, 3D/4D printing, biomedical materials	[20, 135, 136]
4	Boronic esters (diss. condensation in water or acid/ass. transesterification or metathesis in bulk)	RT‐120 (hydrogels); up to 150 (dry networks)	Fast exchange kinetics, catalyst‐free bond exchange, pH/glucose responsive, biocompatible, high synthetic tunability	Moisture sensitivity, hydrolytic instability in some systems, limited long‐term stability in aqueous environments	Hydrogels, biomedical, sensors	[137, 138]
5	HUBs (diss.)	RT‐130	Catalyst‐free synthesis and bond exchange, simple synthesis, nonisocyanate routes available	Hydrolysis sensitivity, limited thermal stability	Biomedical, low‐T reprocessable thermosets and self‐healing elastomers/coatings	[101, 139, 140, 141]
6	fmDA cycloaddition (diss.)	100–140	High mechanical strength, catalyst‐free bond exchange, bio‐based, "click” chemistry, modularity	Maleimide side reactions, color stability issues, mid‐T operation only	Self‐healing coatings/adhesives, reprocessable composites, 4D printing	[39, 82, 142]
7	Vinylogous urea/urethane (ass. transamination)	100–170	Fast exchange (seconds–minutes), catalyst‐free options available, high synthetic tunability	Amine and moisture sensitivity, color stability issues, H‐bonding in ureas can slow dynamics, side reactions	3D/4D printing, high‐performance coatings and adhesives, recyclable composites	[21, 101, 142, 143]
8	Transcarbamoylation/urethane exchange (ass. or diss. depending on catalyst and OH groups)	120–200	Industrial significance (commodity polyurethanes), enables reprocessing of thermoset foams and elastomers, catalyst‐free variants (PHUs, internal catalysis)	Often requires catalyst (metal or amine); degradation risk over multiple cycles; side reactions (urea formation); catalyst leaching	Polyurethane recycling, coatings, elastomers, foams, composites, adhesives	[21, 144]
9	Transesterification (ass./rare dis. depending on system)	100–250	High thermal stability, industrial scalability, catalyst‐free systems emerging	Catalyst required, hydrolysis sensitivity in some systems	Structural composites, automotive, aerospace, high‐performance coatings and adhesives, recyclable composites	[20, 82, 135, 142]
10	Siloxane exchange (ass. equilibration)	150–250	Exceptionally high thermal and thermo‐oxidative stability, catalyst‐free variants exist	Catalyst often required, catalyst leaching/deactivation, slow exchange kinetics without catalyst, creep risk at high temperatures	High‐performance engineering plastics, aerospace composites, sealants, coatings, membranes, actuators	[20, 82, 145, 146, 147]

The choice of dynamic covalent chemistry is not merely a matter of mechanistic and thermal preference but is also governed by the reactivity‐stability balance, which critically determines the role for which a material is suited. Low‐T CANs rely on highly reactive chemistries with low activation energy and fast dynamics (up to seconds at room temperature). This high reactivity enables uses where rapid, ambient‐temperature self‐healing is essential, such as biomedical devices, flexible electronics, and soft robotics. However, this high reactivity comes at a cost: these networks are highly susceptible to side reactions (oxidation for disulfides and hydrolysis for imines and HUBs), which compromise long‐term stability and durability. Moreover, some low‐T systems require catalysts to achieve practical exchange rates, introducing concerns related to toxicity, cost, and deactivation. Transesterification‐based and many other types of vitrimers offer excellent thermal stability and industrial scalability, making them suitable for demanding structural applications where creep resistance and long‐term durability are paramount, such as automotive composites, aerospace components, and engineering plastics [154]. However, their lower intrinsic reactivity results in slow kinetics and again necessitates catalysts for efficient bond exchange, limiting their use in temperature‐sensitive components and adding processing complexity.

The fmDA‐based networks occupy the mid‐T range (100–150 °C), where the reactivity and stability are optimally balanced. The dissociative rDA mechanism provides sufficiently fast dynamics for efficient healing and reprocessing (up to minutes at 120–140 °C) without requiring any catalyst [39, 82, 142]. While most CANs operate in the mid‐T range, including those based on disulfides [131], imines [133], boronic ester chemistry [137], urethane/urea dissociation (such as oxime‐urethanes [155], HUBs [139], or polythiourethanes [21, 156]), as well as less studied adducts between furans and maleate/fumarate dienophiles (such as those formed from furan‐functionalized crosslinkers and unsaturated polyesters) [157, 158], N‐hydroxyphthalimide‐urethane bonds [159], and certain β‐amino esters [160, 161, 162], fmDA occupies a unique position. This uniqueness stems from three key advantages. First, fmDA chemistry is synthetically simple and well‐established: it proceeds via atom‐economic cycloaddition, and decades of research have provided a comprehensive understanding of its kinetics, thermodynamics, and structure–property relationships. Second, fmDA offers inherent sustainability, combining renewable furfural feedstock with catalyst‐free C—C bond formation, extended service lifetime through self‐healing, and end‐of‐life circularity through reprocessability. Third, fmDA is highly modular, allowing independent tuning of the diene, dienophile, cross‐link architecture, and activation temperature through electronic and steric effects, enabling precise control over network properties. This combination of simplicity, sustainability, and modularity is rarely achieved by other dynamic covalent platforms.

It is important to distinguish between catalyst–free bond exchange, which is characteristic of the fmDA reaction, and catalyst‐free synthesis of CAN, which would require that no catalyst is used at any stage of material preparation. For conventional fmDA polyurethanes, the formation of urethane linkages typically relies on organotin catalysts such as dibutyltin dilaurate (DBTDL), oxide, or diacetate [163]. Although isolated examples of catalyst‐free polyurethane synthesis from diisocyanates under typical conditions have been reported [164, 165], such approaches remain exceptions. While the development of catalyst‐free routes remains a desirable goal for fmDA polyurethanes, it would require either significantly higher processing temperature and time, or highly reactive monomers, which may trigger premature retro‐Diels–Alder dissociation and other side reactions.

The fmDA platform does not hold the absolute record for any single property, as disulfides and imines heal faster at room temperature, certain vitrimers exhibit superior thermal stability exceeding 200–250 °C, and some dynamic networks achieve higher tensile strengths over 120 MPa and Young's moduli above 2 GPa [21, 166, 167, 168]. Nevertheless, it consistently approaches these benchmark values [169]. The optimal balance of properties, rather than any single outstanding parameter, defines the niche of fmDA‐polyurethane CANs: applications where rapid, thermally triggered healing, mid‐T processing, sustainability, and mechanical robustness are important. Such areas include self‐healing coatings, debondable adhesives, additive manufacturing, and recyclable thermosets [10, 39, 75, 108, 167, 170].

An important consideration when designing fmDA‐based polyurethane CANs is that polyurethane networks themselves can exhibit dynamic behavior through transcarbamoylation (or transurethanisation) reactions, particularly in the presence of suitable catalysts (Table 1, entry 8) [171, 172, 173]. Such intrinsic dynamicity adds a layer of complexity: while catalysts such as DBTDL can accelerate stress relaxation and facilitate reprocessing, they can also promote unwanted side reactions, including the formation of allophanate or urea linkages. This can lead to irreversible structural changes and compromised mechanical performance over multiple cycles. Therefore, when designing fmDA‐polyurethane CANs, both the properties of the adducts and the influence of catalysts on the polyurethane matrix must be considered to avoid network degradation and ensure reliable performance.

Commercial thermoset and other crosslinked polyurethanes display a broad spectrum of mechanical properties tailored to their specific applications. For instance, PU elastomers widely used in industrial rollers, seals, and automotive components exhibit tensile strengths of 20–40 MPa, with high‐performance cast elastomers reaching up to 50 MPa or more [64, 65, 174, 175]. Two‐component polyurethane clearcoats in automotive applications typically have a tensile strength of ~10–20 MPa and an elongation at break of 50%–150%, while structural adhesives achieve lap‐shear adhesion strengths of ~5–10 MPa [63, 176]. This versatility, combined with durability and chemical resistance, has established commercial polyurethanes as indispensable materials across virtually every industrial sector. The fmDA‐type polyurethane CANs achieve fully comparable or even higher values: tensile strengths around 70–80 MPa in rigid systems with elongation at break ranging from 50% to >1600% and Young's modulus spanning around 10–500 MPa [70, 177, 178, 179, 180, 181, 182]. Coating properties remain competitive: UV‐curable fmDA‐urethane acrylates exhibit tack‐free times of 3–6 s, pendulum hardness of 58–190 s, gloss of 50–200 GU, and cross‐cut adhesion ratings of 0–2, values that are typical of commercial UV‐curable systems [183, 184, 185]. Acrylate double bond conversion (up to >95%) and gel content (>90%) also fall within the range of conventional PU‐based coatings, adhesives, and UV‐curable systems [182, 183, 185, and 186]. Collectively, these data confirm that fmDA incorporation does not degrade room‐temperature mechanical performance, curing efficiency, or durability; these parameters are governed primarily by the hard/soft segment balance rather than by the dynamic bonds themselves.

However, these competitive characteristics are maintained only below the rDA onset, which typically begins above 80–100 °C, depending on the fmDA adduct structure. At elevated temperatures (>100–150 °C), the reversible dissociation of fmDA cross‐links leads to a reduction in cross‐link density, resulting in a pronounced drop in modulus, creep resistance, and mechanical strength [13, 21, 187]. In other words, the presence of the reversible Diels–Alder bonds led to an inherent strength‐healing trade‐off: dynamicity comes at the cost of thermal stability. The thermal sensitivity distinguishes fmDA‐based and other CANs from conventional thermosets and imposes a fundamental limitation on their upper service temperature, particularly in load‐bearing applications. It also reflects the key conflict, where conventional CANs often sacrifice robustness for healing efficiency or vice versa (the strength‐healing trade‐off) [188, 189, 190]. This dilemma represents a specific case of conflicts between strength and toughness [191]. Accordingly, designing elastomers that combine high mechanical performance with efficient self‐healing remains a challenge as these properties are often dictated by mutually exclusive molecular mechanisms [102, 115, 192]. Network connectivity is governed by the combined contributions of DA adducts, urethane cross‐links, and any additional cross‐linking agents, while soft segment composition controls chain mobility. High cross‐link density, while enhancing tensile strength, modulus, and creep resistance, restricts chain mobility and thus limits the ability of DA bonds to dissociate and reform efficiently during healing. Conversely, low crosslink density promotes chain diffusion and rapid bond exchange, favoring high healing efficiency but at the expense of mechanical robustness and dimensional stability [71]. Viewed more broadly, the strength‐healing trade‐off is now recognized as one of the central design bottlenecks across self‐healing polymers rather than a limitation specific to polyurethanes alone. The most competitive systems increasingly rely on hierarchical energy‐dissipation pathways, spatially nonuniform dynamicity, or coupled actuation‐healing mechanisms; this broader perspective supports the relevance of multidynamic fmDA polyurethane architectures discussed here and highlights why simple bond reversibility, by itself, is rarely sufficient for high‐performance materials [114, 193].

Beyond this strength‐healing dilemma, the practical implementation of fmDA‐based dynamic polyurethanes is also constrained by a combination of thermochemical and topological factors, together with synthetic, economic, and ecological challenges. In addition to the temperature limitations discussed above, the kinetic and thermodynamic characteristics of the fmDA reaction introduce further constraints on processing and long‐term stability [194]. The forward Diels–Alder cycloaddition, while thermodynamically favorable, exhibits moderate reaction rates that typically require several hours to achieve complete healing at elevated temperatures [71, 195]. This slow kinetics, combined with the high activation barrier for the rDA process, critically restricts the application of fmDA‐based CANs in scenarios where rapid, ambient‐temperature self‐healing is essential. For example, consumer electronics with self‐healing display coatings, artificial skin for soft robotics, and wearable sensors that require instantaneous or near‐instantaneous healing at room temperature remain largely inaccessible to current fmDA systems [92, 196, 197, 198]. Similarly, protective coatings for automotive interiors or medical devices, where thermal healing would damage sensitive substrates or components, cannot benefit from fmDA technology.

The thermodynamic reversibility of the DA equilibrium imposes yet another critical challenge: complete conversion of furan and maleimide groups is difficult to achieve. Therefore, high‐temperature reprocessing shifts the equilibrium toward the reactants, causing a temporary reduction in network connectivity that may never be fully restored upon cooling. This effect limits the maximum achievable cross‐link density and compromises the long‐term performance of reprocessed materials. At temperatures above the rDA onset, mechanical properties degrade substantially due to thermal reversibility [71]. For instance, the storage modulus can drop by several orders of magnitude near the gel point [187]. This effect is attributed to the decrease in cross‐link density with increasing temperature [199]. Such degradation is more pronounced in densely crosslinked networks, where restricted chain mobility hinders the reversible process [71].

Another critical issue is the low stability of furan and maleimide functional groups at healing temperatures typically ranging from 120 to 150 °C. Maleimide groups can undergo irreversible homopolymerization and the Michael reaction with residual amines at these temperatures [36, 108, 200, 201]. Thermally induced radical homopolymerization occurs at the same temperature range as the rDA process, introducing permanent cross‐links that cannot be cleaved and thus progressively compromising the network reversibility and recyclability over multiple healing cycles [119, 202]. Quantitative studies show that the efficiency of mechanical property recovery diminishes upon repeated thermal cycling; for example, tensile strength recovery in a PCL‐based fmDA thermoset PU crosslinked with an HDI trimer (tri‐HDI) decreased from 88% after the first healing cycle to 39% after four cycles at 65 °C, indicating progressive loss of network integrity [70]. Another important practical concern is low chemical stability of the electron‐rich (donor‐type) furans [203, 204], leading to the intrinsic discoloration of furanic products upon thermal or UV exposure [87, 205]. This discoloration, while not impairing bulk mechanical performance, limits the applicability of PUs in optically clear coatings and esthetic consumer products.

While strategies such as stoichiometric imbalance, radical inhibitors, spacer length, or bulkier maleimide precursors can mitigate this side reaction, the complete suppression remains elusive under typical reprocessing conditions (>150 °C) [201]. A particularly promising alternative dienophile is 1,2,4‐triazoline‐3,5‐dione (TAD). This dienophile undergoes an ultrafast DA reaction with 2,5‐disubstituted furans at room temperature, overcoming the elevated temperatures and longer reaction times required for conventional maleimide‐based systems while retaining thermoreversibility at 130 °C [206]. Beyond TAD, a recent systematic screening of various diene–dienophile pairs by MacKinnon et al. explored citraconimide (methyl‐substituted maleimide) and acrylamide as potential dienophiles for thermoreversible DA systems [207]. While citraconimide offered enhanced resistance to irreversible side reactions compared to conventional maleimides, it exhibited very poor reactivity with furan derivatives, with the furfuryl alcohol–citraconimide pair failing to reach more than 20% conversion even at elevated temperatures [207]. This observation aligns with fundamental structure–reactivity relationships. Electron‐withdrawing substituents on the dienophile or unfavorable substitution patterns on the furan (e.g., 2‐substitution with electron‐withdrawing groups) can render the DA reaction not only kinetically sluggish but also thermodynamically unfavorable, shifting the equilibrium strongly toward the starting materials [208]. Nevertheless, the enhanced stability of citraconimide against homopolymerization makes it an attractive candidate for other diene partners, and further structural modifications could potentially unlock its reactivity with furans.

Electron‐rich furans (e.g., furfuryl alcohol and 2‐methylfuran) exhibit high Diels–Alder reactivity, favoring efficient network formation, but suffer from relatively poor thermal and oxidative stability, as well as low controllable diastereoselectivity. A comprehensive stability study of furanic platform chemicals by Kolykhalov et al. demonstrated that electron‐withdrawing substituents (such as ester or aldehyde groups) significantly enhance the chemical stability of the furan ring toward acids, bases, and thermal degradation [204]. In contrast, electron‐donating substituents (e.g., hydroxymethyl groups) render the ring highly susceptible to polymerization and humification [204]. At the same time, such acceptor‐substituted furans preferentially form the endo Diels–Alder adduct with maleimides [209, 210, 211]. This combination of properties offers a practical design strategy for fmDA‐based CANs. Acceptor‐substituted furans provide a more homogeneous, stereochemically defined network topology (dominated by the endo adduct) and improved resistance to oxidative degradation and discoloration that are particularly valuable for optically clear coatings and applications requiring long‐term color stability. The penalty is a modest reduction in the retro‐Diels–Alder (rDA) temperature of the endo adduct (of the order of 10–40 °C) [212], which is often acceptable when chemical robustness and predictable network architecture are prioritized over maximum thermal resistance. This design principle expands the utility of fmDA chemistry beyond traditional furfuryl alcohol‐based systems, providing a pathway toward more chemically robust and optically transparent dynamic polyurethanes. However, despite these advantages, CANs based on acceptor‐substituted furans remain rare [213, 214, 215, 216, 217].

Besides the fundamental chemical limitations, the practical implementation of fmDA‐based CANs faces significant synthetic and economic hurdles. The formyl groups of furfural and HMF strongly deactivate the furan ring toward Diels–Alder cycloaddition [194, 218]. This makes direct use of furanic aldehydes difficult without prior chemical modification, typically by reduction to furfuryl alcohol, oxidation to acid, acetalization, or imine/hydrazone formation [128, 194, 219, 220, 221, 222]. Furthermore, the reliance on petrochemical comonomers such as polyols (PCL, PTMEG, PPG, and PEG) as well as diisocyanates (IPDI, MDI, HDI, and H₁₂MDI) remains a significant sustainability concern. Most existing systems contain less than 40% bio‐based carbon content, even when the furanic fragment is fully renewable because the major components remain fossil‐derived.

Given that the high cost of maleimides is unlikely to decrease significantly in the foreseeable future, the commercial success of furan–maleimide polyurethanes will be determined not by price competition but by the unique value they offer in high‐added‐value niches, such as adaptive materials and high‐performance engineering applications. Synthesis complexity also poses a practical barrier: the preparation of high‐purity fmDA precursors adds cost and limits scalability. Thus, while fmDA chemistry offers a powerful platform for smart, sustainable polyurethanes, its practical implementation requires careful molecular design to balance stimuli responsiveness, mechanical robustness, durability, and processability.

The ecological footprint of fmDA‐based polyurethane CANs extends beyond the renewable origin of furanic fragments. According to the European Union CLP Regulation (EC) No 1272/2008 and ECHA C&L Inventory, both furfural (CAS 98‐01‐1) and furfuryl alcohol (CAS 98‐00‐0) are classified as Carc. 2 (H351) substances (suspected of causing cancer). According to IARC, furfural is classified as Group 3 (not classifiable as to its carcinogenicity to humans), whereas furfuryl alcohol is classified as Group 2B (possibly carcinogenic to humans). This raises legitimate concerns about worker exposure and potential release of degradation products during service life or disposal. This is compounded by the widespread use of organotin catalysts (e.g., dibutyltin dilaurate), which are bio‐persistent and environmentally toxic, and the reliance on polar aprotic solvents (THF, acetone, DMF, and DMAc) for synthesis and processing. On the other hand, fmDA adducts incorporated into polymer networks can exhibit good biocompatibility [88, 223]. However, this is not a universal property of fmDA‐based materials; it depends critically on the specific formulation, the degree of conversion, and the presence of unreacted monomers or degradation products. Therefore, while biomedical applications remain a promising direction, particularly for systems based on less toxic building blocks such as BHMF [224], biocompatibility must be rigorously validated for each new material system rather than assumed.

Another limitation is that thermal healing or reprocessing requires energy‐intensive conditions, contributing to the overall energy demand. Closed‐loop recycling (defined as depolymerization to the original monomers or equivalent chemical building blocks, followed by repolymerization) has been demonstrated for some furan‐based Diels–Alder polyurethanes [128]. However, reports remain scarce, and to the best of our knowledge, this approach has not yet been extended to furfural‐derived adduct‐based systems. While life cycle assessment (LCA) studies on the production of furfural and HMF have been reported [225], they primarily focus on feedstock cultivation, solvent selection, catalyst use, and energy consumption during monomer synthesis. Comparative LCA studies of fmDA polyurethane CANs covering feedstock sourcing, synthesis energy, solvent use, and end‐of‐life scenarios remain largely absent. These caveats underscore that the presence of a renewable furan fragment alone does not guarantee sustainability; rigorous quantitative assessment of the entire material lifecycle is essential to validate actual environmental benefits over conventional thermosets.

A promising long‐term direction to address the toxicity and sustainability concerns associated with conventional isocyanate‐based PUs is the development of nonisocyanate polyurethane (NIPU) routes. Unlike conventional isocyanate‐based methods, which typically require organotin catalysts, NIPU synthesis by cyclic carbonate‐amine polyaddition is inherently catalyst‐free [97, 226]. This makes NIPU‐based CANs particularly attractive from the perspective of green chemistry as they eliminate both toxic isocyanates and the associated catalyst‐related environmental concerns [227, 228, 229]. However, despite its promise, the NIPU approach currently faces several limitations. The ring‐opening of cyclic carbonates with diamines exhibits significantly slower kinetics than the isocyanate‐alcohol reaction, typically requiring elevated temperatures (>100 °C) and extended reaction times (hours to days) to achieve high conversion in bulk, which can promote side reactions [97, 229, 230]. Scalability is also constrained by the limited availability and higher cost of bio‐based cyclic carbonates and the challenges of solvent‐free processing for high‐viscosity formulations [228, 229, 230]. Nevertheless, the elimination of toxic isocyanates and catalysts and the compatibility with renewable feedstocks make NIPU routes a critical long‐term direction for sustainable dynamic polyurethanes.

While the NIPU approach addresses the feedstock and toxicity challenges, fmDA chemistry provides the reversible bond exchange necessary for dynamic functionality. The combination of these two approaches offers a particularly compelling platform for sustainable dynamic polymer systems. Overall, the fmDA polyurethane CAN systems offer a triple balance that is rarely achieved by other dynamic covalent platforms to the same extent: (i) high mechanical performance comparable to conventional thermosets; (ii) enhanced sustainability, combining catalyst‐free NIPU synthesis, reversible C—C bond exchange (DA/rDA), renewable furfural feedstock, extended service lifetime through self‐healing, and end‐of‐life material recovery via reprocessability; and (iii) smart functionality, including shape memory, stimuli responsiveness, and compatibility with 3D/4D printing. This triple convergence is precisely why furfural‐derived fmDA monomers have emerged as a distinctive platform for next‐generation adaptive and circular materials. Realizing this full potential, however, will require addressing several outstanding challenges. These include prioritizing NIPU routes, developing benign catalyst alternatives (e.g., bismuth carboxylates) [172], replacing petrochemical comonomers with renewable building blocks, and systematically evaluating the ecotoxicity of furanic intermediates and degradation products. Addressing these challenges is essential to transition fmDA polyurethane CANs from promising laboratory concepts to truly sustainable and industrially relevant materials.

2.2. Synthetic Strategies: The Adduct‐Based Approach Versus Prepolymer Crosslinking

Two synthetic strategies are widely recognized in the literature as the primary routes for incorporating fmDA chemistry into polyurethanes: (1) cross‐linking of linear furan‐containing prepolymers with bis‐maleimides (Figure 3a) and (2) introduction of preformed functional Diels–Alder adducts as comonomers (Figure 3b) [75, 95, 110, 231]. In the first case, linear macromolecules can contain the furan ring either as a substituent (furan‐grafted species) or as part of the main chain. Here, we focus on analyzing the second strategy, which was defined as the “adduct approach” [75]. The key distinction between these two strategies lies in when and how the Diels–Alder adduct is formed. In the prepolymer cross‐linking approach, the furan and maleimide groups are introduced separately into the prepolymer and crosslinker (e.g., 4,4’‐bis(maleimido)diphenylmethane, BMI), and the DA adduct must form during network assembly. Conversely, the adduct‐based approach uses a preformed functional DA adduct as a building block, which is then incorporated into the polymer backbone through subsequent polyaddition or polycondensation. The prepolymer cross‐linking offers synthetic simplicity as it does not require a separate adduct synthesis. The furan‐containing prepolymers and bis‐maleimide cross‐linkers are relatively inexpensive, making this approach more cost‐effective and scalable than the adduct‐based route, which requires additional synthetic and purification steps. However, the cross‐linking approach suffers from several significant limitations related to poorer control over network structure and properties, arising from the reversibility of the DA reaction and the instability of maleimides, as well as constraints in molecular design.

FIGURE 3.

General approaches to the synthesis of PU CANs using furfural: (a) cross‐linking of linear furan‐containing prepolymers with bis‐maleimides and (b) adduct‐based approach using preformed functional fmDA adducts.

The aforementioned differences translate into three main advantages for the adduct approach: (i) superior control over network topology and cross‐link density; (ii) simpler process control and minimized side reactions; and (iii) enhanced design flexibility and material performance. These advantages do not eliminate the inherent limitations of fmDA chemistry, such as maleimide side reactions or the strength–healing trade‐off. However, they provide a level of control over network architecture that is unattainable when the dynamic motif is introduced by postmodification of the furanic prepolymers.

The key differences between the two approaches are summarized in Table 2. It should be noted that these comparisons represent general trends; the actual performance of each approach depends on specific monomer structures, reaction conditions, and formulation parameters [75, 95, 110, 231]. In the discussion below, we examine how the adduct approach enables superior control over network homogeneity, stereochemistry, and side‐reaction suppression. The fmDA adduct can be functionalized with different reactive groups such as hydroxyl, isocyanate, or acrylate without altering the dynamic DA core (see Section 2.3). This modularity enables compatibility with diverse polymerization chemistries: step‐growth polyurethane synthesis, UV‐curable systems, and NIPU routes. It is particularly evident in the work of Moyano‐Vallejo et al., who demonstrated that a preformed isocyanate‐functionalized fmDA adduct 9 could be incorporated into a commercial thermoplastic polyurethane adhesive formulation simply by mixing [232]. Such modularity provides decoupling of adduct synthesis from polymer formulation, which is a decisive advantage for industrial adoption as it allows the dynamic motif to be introduced without reengineering the entire polymer synthesis.

TABLE 2.

Comparison of the adduct‐based and prepolymer crosslinking strategies for fmDA‐derived polyurethane CANs.

No	Criterion	Adduct approach	Prepolymer crosslinking approach
1	Adduct formation	Separate step	In situ during network assembly
2	Polymerization chemistry	Irreversible (urethane formation)	Reversible (DA crosslinking) + irreversible (urethane)
3	Network homogeneity and control over crosslink density	Typically high, formulation‐dependent	Moderate due to rDA equilibrium and statistical crosslinking
4	Stereochemical control	Yes (endo/exo tunable)	Limited (statistical)
5	Modularity and design flexibility	High (functional groups can be varied independently of DA adduct core)	Low (restricted to bis‐maleimide crosslinkers)
6	Process temperature	Generally mild (≤80 °C)	Often elevated (>100 °C)
7	Side reactions	Minimized during network formation (but occur during healing)	Pronounced (maleimide homopolymerization, Michael addition)
8	Reproducibility	Generally high, contingent on adduct purity	Moderate, depends on reaction conditions
9	Cost	Moderate (additional synthetic and purification steps)	Lower (fewer steps, less waste)
10	Scalability	Lower	Higher
11	Areas of applications	Precision materials, structure‐property studies, high‐value products	Cost‐sensitive, large‐volume applications

The number of functional groups on the adduct can be systematically varied. This directly governs crosslink functionality and network topology: a diol adduct yields linear or lightly branched structures, while a triol or tetraol adduct gives rise to more rigid networks with higher crosslink density. Such tunability provides a systematic handle for tuning mechanical properties from elastomeric to glassy by simply varying the functionality and loading of the preformed adduct [117, 233, 234]. Moreover, the prepolymer cross‐linking approach is restricted to bis‐maleimides, which can only provide linear or kinked chain extension between furan‐functionalized prepolymers, further limiting the achievable network topology.

Control over network topology is not exclusive to the adduct‐based approach; in principle, nondynamic crosslinkers such as TEA could be introduced into a prepolymer system to tune network architecture. However, this strategy introduces a potential problem: the presence of permanent, irreversible cross‐links that do not participate in the DA/rDA equilibrium. Even after complete dissociation of DA bonds, these nondynamic crosslinks remain and can restrict chain mobility, impeding the diffusion and recombination of dynamic fragments during healing. In the adduct route, by contrast, for highly functionalized adducts (e.g., triols or tetraols), every crosslink is dynamic. Therefore, the entire network is reversible, and upon heating above the rDA temperature, it can more completely decrosslink, enabling, in theory, higher chain mobility and more efficient healing [233, 234, 235, 236].

Another advantage is that the adduct approach offers flexibility in the choice of maleimide functionality. When a mono‐maleimide is used, the adduct contains a single maleimide unit, simplifying network formation entropically compared to bis‐maleimide crosslinkers. When a bis‐maleimide is used, the preformation step ensures that the adduct is fully characterized and purified before network assembly, minimizing the risk of incomplete conversion and side reactions during polymerization. In both cases, the initial network is more regular and predictable than in the prepolymer route, where furan‐functionalized polyurethane prepolymers are crosslinked with bismaleimides (typically BMI). In the prepolymer approach, the reversible nature of the DA reaction may prevent complete conversion of furan and maleimide moities, leaving a fraction of these functional groups unreacted and compromising network homogeneity [71, 187]. These incomplete DA/rDA processes can further lead to a decrease in healing efficiencies (Table 3, entry 3) [117]. Moreover, the adduct is synthesized and purified in a separate step, where the equilibrium can be driven to high conversion and diastereoselectivity (>95%). The subsequent polyurethane formation relies on irreversible urethane chemistry under mild conditions (≤80 °C) and not on the DA equilibrium, which minimizes premature adduct dissociation. This precision of stereochemical composition is important for CANs, since the endo and exo Diels–Alder stereoisomers exhibit significantly different thermal stability. Their coexistence in the network leads to a heterogeneous distribution of cross‐links with different thermal lability, which complicates the prediction of healing and reprocessing temperatures [71, 195]. Thus, in the prepolymer cross‐linking approach, the endo/exo ratio is determined statistically during network formation, leading to stereochemical heterogeneity.

TABLE 3.

Functional fmDA adducts with maleimides and types of prepared polyurethane CANs.

No	Adduct	Other components	Product, key features	Ref.
1	1	PGA/PLA, glycerol/pentaerythritol/xylitol, H12MDI	Thermoreversible biodegradable PU networks; M c 0.40–2.72 kg/mol; T g −3 to +42 °C; T crossover 40–128 °C	[237, 238]
2		PCL, glycerol, H12MDI	Thermoreversible PU networks; scratch healing; G′ up to 120 kPa	[239]
3	2	TEA, H12MDI	SMASH polyurethane; TS 76.8–87.3 MPa; self‐healing efficiency: 1st eff. 80%–84%, up to 3 cycles (59%–76%); T g ~ 42–46 °C	[117]
4		PLA1000, HPA/TMPTMP/PETMP, IPDI	UV‐curable coatings; thiol‐ene + DA double click; reduced healing T to 90 °C; fast scratch healing 35–45 s (80–100 µm scratches)	[240]
5		TEA/HPED, H12MDI, Carbon fiber	First continuous carbon fiber PU composites with fmDA matrix; interlaminar strength recovery up to 88%; TS ~ 53–87 MPa	[241]
6		PTMEG1000, DMPA, IPDI/HDI/H12MDI, tri‐HDI	Waterborne PUs; systematic study of diisocyanate structure on self‐healing & anticorrosion; PE up to 99.9%	[184]
7		Jeffamine D400, IPDI, pHDI	PUU adhesive for wood debonding; 71% lap shear reduction; film TS 36.4 → 28.8 MPa	[129]
8		PCL, HDI/IPDI/MDI, tri‐HDI	SLS 4D‐printable fmDA‐PU; near‐isotropic properties (Z‐axis 83% of X); NIR‐triggered healing; TS ~ 10.9 MPa	[179]
9		PCL4000, MDI, tri‐HDI	HIFU‐triggered SMASH polyurethane; TS 23.6–28.0 MPa; Eb ~ 1600%–1900%; healing efficiency: ~92% → ~84% → ~70% over 3 cycles	[120]
10		PEG400–1000, HEA/HPA/HEMA/HPMA, H12MDI/IPDI	Healable UV‐curable polyurethane (meth)acrylates; C=C conversion 38–100% (IPDI > HDI; acrylates > methacrylates; PEG400 > PEG1000); IPDI‐based coatings show better self‐healing; acrylates give higher conversion (86%–97%)	[182, 183, 185, 186]
11		PPG, DMPA, TEA, HEMA, IPDI	Healable/reprocessable waterborne UV‐curable poly(UA); photocuring conversion 55%–63%	[242]
12		PTMEG2000, DMG, DMPA, TEA, IPDI	Healable PU with dual dynamics (fmDA + oxime‐carbamate); TS 36.6 MPa; healing eff. 92.9%; recyclable in acetone/ethanol at RT after brief preheating at 120 °C; >90% recovery after three cycles	[178]
13	3	PCL2000, HMA, BDO, H12MDI	Recyclable/healable PU based on fmDA + UPy quadruple H‐bonds; TS up to 46 MPa; E b up to 2196%; toughness 406 MJ/m3; SIC‐enhanced	[233]
14		PCL2000, MDI	Recyclable PU; recycling by 3 methods (hot pressing, injection, solution casting); TS 36–47 MPa	[235]
15	4	Acrylic polyol	Scratch‐healing PU; application in automotive clear coats; 100% scratch depth recovery at 150 °C	[243]
16		Hydroxyl‐terminated polyacrylate graft copolymers	Effect of scratch loading mode (progressive vs. stepwise) on healing; stepwise loading causes irreversible damage even at rDA	[244]
17	5	BDO, isocyanate‐terminated prepolymer	Microcapsules with fmDA shell for controlled isocyanate release; oxime‐urethane microcapsules show superior mechanical strength	[245]

The adduct approach allows the use of sensitive furan derivatives as they can be purified before network formation. This does not eliminate their inherent instability, but it reduces the negative impact of impurities and side reactions on the final network. Crucially, the purity of the DA adduct was found to be critical. A DA‐triol, synthesized by the reaction of tris‐maleimide with furfuryl alcohol, was used as a thermoreversible cross‐linker with a biobased macrodiol and polymeric MDI (Figure 4). The resulting polyurethanes, whose properties were tunable from flexible to rigid, demonstrated thermoreversibility at ~160 °C and excellent reprocessability via compression and extrusion. The materials exhibited tunable mechanical properties from flexible elastomers (E  = 16.7 MPa) to rigid plastics (E  = 476.5 MPa) while maintaining reprocessability with >80% recovery of mechanical properties after recycling [200]. Recycled samples retained over 80% of their original mechanical properties, a performance highly dependent on the purity of the initial tris‐adduct [200, 246]. Higher purity (99%) of this DA‐triol gave better stress relaxation (τ* = 12 min at 160 °C) and higher recovery efficiency compared to lower purity (82%, τ* = 28 min) [200]. In a subsequent study, the same group addressed the inherent limitation of maleimide homopolymerization at elevated reprocessing temperatures by incorporating p‐benzoquinone as a free‐radical scavenger [247]. The inhibitor extended the operational recycling window by 15–20 °C and enabled the thermosetting network to relax nearly to zero stress at 160 °C, whereas the uninhibited analog remained partially locked due to irreversible cross‐linking. After one recycling cycle, the inhibited formulation retained ~90% of its original flexural modulus and strength, while the uninhibited sample showed a significant drop in mechanical performance. Importantly, the inhibitor caused only minimal changes to the pristine mechanical properties (flexural modulus decreased by ~12%, from 3.15 to 2.75 GPa), confirming that radical suppression can be achieved without compromising network integrity [247]. Therefore, the adduct approach ensures that the initial network contains minimal defects and impurities, which should translate into enhanced durability and recyclability. When combined with suitable radical inhibitors, this strategy effectively mitigates the detrimental side reactions that typically limit the reprocessing of furan–maleimide networks, thereby preserving the dynamic character of the material over multiple life cycles.

FIGURE 4.

General scheme of synthesis of polyurethane CANs using DA‐triol. Reproduced with permission [200]. This material is licensed under a Creative Commons Attribution 4.0 International License.

The primary drawback of the adduct approach is the additional synthetic and purification steps required to prepare the fmDA adduct, which increases process complexity and cost. Purification (typically by recrystallisation or column chromatography) must be carefully optimized to achieve the high purity necessary for reproducible network formation, and losses during purification can reduce overall yield. This makes the approach less attractive for cost‐sensitive, large‐volume applications. Additionally, the adduct must be stable under the conditions of polyurethane synthesis (moisture‐free and moderate temperatures), which, while generally manageable, imposes constraints on the choice of solvents and processing conditions. This investment, however, brings decisive benefits because it provides superior control over crosslinking density and network architecture. Moreover, it enables versatile molecular design, allowing independent modification of the Diels–Alder motif, for example, by introducing functional groups or orthogonal reactive sites. The monomer‐based approach is best suited for demanding applications where precise property optimization is critical, for instance, in systematic structure–property studies, multidynamic networks with orthogonal functionalities, or high‐value materials where reproducibility is paramount. It is also the method of choice when the furan component is sensitive or requires special handling or when the dynamic fragment must be independently modified. In contrast, the prepolymer cross‐linking approach remains attractive for cost‐sensitive, large‐volume production, where processing ease outweighs the need for absolute network homogeneity.

Because fully crosslinked thermosets are insoluble, highly soluble low‐molecular‐weight impurities (unreacted monomers, catalyst residues, and oligomers) can be efficiently removed by repeated solvent washing or Soxhlet extraction without dissolving the network [222, 248, 249]. In principle, the reversible nature of the Diels–Alder bond could offer an additional purification pathway. After washing to remove soluble impurities, heating the material above the rDA temperature would temporarily decrosslink the network into soluble furan‐ and maleimide‐terminated fragments. Insoluble impurities such as products of maleimide homopolymerization could then be separated by filtration. However, this approach has not yet been demonstrated for fmDA polyurethane CANs, and deeper purification by temporary decrosslinking remains an unexplored but conceptually attractive direction.

In summary, the monomer‐based design philosophy thus converts the dynamic fragment from a passive cross‐linking element into an independent molecular variable. This is especially valuable for high‐impact material design, where the central question is no longer whether a network can heal but how precisely bond placement, local mobility, and network heterogeneity can be engineered to balance stiffness, toughness, healing rate, and dimensional stability [114, 193]. In the following section, we demonstrate how this key advantage has been realized in practice by systematically examining the growing library of functional furan‐based adducts and their integration into polyurethane platforms (Tables 3 and 4). We trace the evolution from simple reversible networks to sophisticated multidynamic systems capable of shape‐memory, 4D printing, and self‐healing behaviors. This molecular‐level control over the type, density, and spatial distribution of reversible bonds unlocks the tailored balance between mechanical properties and healing efficiency, which remains elusive in networks where cleavable motifs are introduced by statistical postmodification.

TABLE 4.

Functional fmDA adducts with bis‐maleimides and types of prepared polyurethane CANs.

No	Adduct	Other components	Product, key features	Ref.
1	6	PCL2000, TMPMP, HDI	PCL‐based elastomer; TS > 32 MPa; stress‐free two‐way shape memory; healing of wide scratches (80–100 µm) in 6 min at 110 °C (programmed sample); >100% TS recovery after healing	[177]
2		PTMEG2000, MDI, CNT‐Ag	Healable PU/CNT‐Ag composite; λ = 1.65 W/mK; TS 23.45 → 15.97 MPa; healing efficiency 90.1% (TS)/92.9% (strain) after 130 °C, 5 min + 65 °C, 2 h; interfacial R c 1.37 × 104 K/W	[250]
3		PTMEG2000, MDI, FGNS	Healable PU composite for flexible electronics; ultrafast IR laser healing (1 min, eff. >96%); TS 36 MPa (0.5% FGNS)	[251]
4		PTMEG, MDI, TiO2	Healable dielectric TiO2/PU nanocomposites; ε 13.7 (1 kHz, 5 wt% TiO2); tan δ < 0.06; TS 10.99 MPa (5 wt% TiO2); E′ 10.04 MPa; healing efficiency 61.9% (5 wt%)	[252]
5		PCL2000, EG, TDI, lignin	Lignin‐based healable PU with dual dynamics (fmDA + H‐bonds); TS up to 28.93 MPa (up to 100% recovery upon healing)	[253]
6		HEA, IPDI	DLP 3D‐printable PU acrylates; TS ~32–45 MPa; shape memory (R f ~ 96%, R r ~ 99%, σ rec = 10.1 MPa); recycling efficiency >98% (TS retention)	[254]
7		PCL, HDI, DGEBA, Jeffamine	Shape memory EP‐PU conetworks; two synthetic routes (preformed vs. in situ DA); E modulus ~1–350 MPa; shape fixity up to 99%	[234]
8		PTMEG, BHPS, IPDI	Self‐healing PU; healing of large‐scale damage, mm at 60 °C; G′ up to 8.15 × 107 Pa	[255]
9		PCL‐triol, PEG, IPDI	Dynamic thermoset PU; 4D‐printable (FDM, 130 °C), self‐healing (94.8%, 37 °C, 12 h), shape memory (fixity/recovery >90%), swelling‐stiffening, biocompatible	[256]
10	7	Polyol, IPDI, HEA, AgNWs	UV‐curable PU composite; transparency ~88%; low roughness; flexible	[257]
11	8	PCDL500–2000, HEMA, cyclophosphazene, IPDI	UV‐curable coatings with dual networks (fmDA + H‐bonds); pencil hardness up to 3H; self‐healing at 120 °C in ≤60 s (optimized); gloss recovery up to 100% after multiple cycles.	[258]
12	9	DETA	NIPU network with full thermo‐healing (complete scratch healing; 100% dissolution in DMSO at 130 °C in 30 min vs. ~83% for epoxy counterpart)	[259]
13	10	BDO‐polyadipate‐MDI PU	Thermoplastic PU adhesive with reversible DA crosslinking; rDA at ~130 °C; on‐demand debonding; adhesion ~8.5 N/mm (SBR); compatibility with industrial formulations	[232]
14	11 or 12	Cyclic/acyclic PCLs, HDI	Shape‐memory and self‐healing PU; cyclic topology: higher shape fixity (~95% vs. ~80%); self‐healing after 130 °C, 4 h + 60 °C, 48 h	[260]

Before turning to the properties of adduct‐derived fmDA polyurethanes, it is worth noting an important distinction in feedstock choice. While 5‐hydroxymethylfurfural (HMF) has also been explored as a starting building block for the preparation of exchangeable cross‐linked polyurethanes, existing reports primarily rely on the cross‐linking of linear prepolymers via hydroxymethyl‐functionalized HMF‐based products such as BHMF [248, 261, 262, 263, 264, 265, 266, 267, 268], cyclic acetals [128, 269], or other derivatives [270, 271, 272, 273]. To the best of our knowledge, the direct incorporation of preformed adducts of such HMF‐derived compounds with maleimides for polyurethane CANs has not been reported. This does not, however, diminish its scientific interest: the additional hydroxymethyl group of HMF offers a valuable handle for introducing orthogonal functionality, branching, or postpolymerisation modifications within the monomer‐based framework. In the context of preformed fmDA adducts, HMF‐derived building blocks could enable the design of polyurethane CANs with more sophisticated network topologies or dual‐dynamic behavior, provided that cost‐effective production routes become available in the future. From the viewpoint of biomass valorization, this distinction between furfural‐ and HMF‐derived monomers is also strategically important. Recent analyses of renewable polymer feedstocks emphasize that the future impact of biomass‐derived networks will depend not only on bio‐based carbon content but also on whether the platform molecule can be diversified into families of building blocks with orthogonal reactivity and application‐specific functions; this makes furfural/HMF platform expansion highly relevant for dynamic polyurethane design [32, 274].

2.3. Properties of Adduct‐Derived fmDA‐Polyurethane CANs: Bridging Sustainability With Smart Functionality and Performance

This section focuses exclusively on properties and applications of fmDA‐based polyurethane thermosets obtained by the adduct‐based approach. A direct comparison with prepolymer‐derived systems is not attempted as the literature is too diverse and properties are highly dependent on formulations and conditions. The systematic structure‐property study by Terryn et al. provides a quantitative framework for designing fmDA‐based polyurethane CANs [236]. By independently varying the DA bond concentration, monomer functionality, and stoichiometric ratio, they demonstrated that mechanical properties (from elastomeric to glassy) are governed by crosslink density, the gelation temperature can be tuned via functionality without affecting room‐temperature performance, and healing kinetics can be accelerated by an excess of furan groups.

The discussion below demonstrates how temporal decoupling, together with precise control over network topology, enables the construction of sophisticated multicomponent systems that balance mechanical robustness with dynamic functionality. First, we examine the interplay between the three key structural elements, hard segments, soft segments, and fmDA adducts, which jointly govern the mechanical and dynamic responses of fmDA‐based polyurethane CANs, as summarized in Table 5. Hard segments, derived from diisocyanates, low‐molecular‐weight diols, and chain extenders, act as physical crosslinks through hydrogen bonding and form rigid domains that provide mechanical strength, solvent resistance, and creep resistance. Their content and structure (aromatic vs. aliphatic diisocyanates) directly influence the glass‐transition temperature of the hard phase, the degree of microphase separation, and the overall stiffness of the material. Higher hard segment content enhances tensile strength and modulus but restricts chain mobility, which can limit self‐healing efficiency. The soft segment structure (e.g., PCL, PTMEG, and PPG) plays an equally critical role: it governs chain mobility and thus the glass transition temperature, which in turn determines the temperature window for efficient healing and shape recovery. For example, systems with crystallisable PCL soft segments combine the shape‐memory effect (fixity > 95%, recovery > 99%) with high healing efficiencies [177] as the melting transition of PCL provides an additional mechanism for crack closure prior to DA bond reformation [260]. The fmDA adduct introduces reversible covalent cross‐links, enabling dynamic functionalities such as self‐healing, reprocessability, and shape‐memory reprogrammability while also contributing to network stiffness and cross‐link density. This interplay between permanent physical crosslinks (hard segments), reversible covalent crosslinks (fmDA adduct), and chain mobility (soft segments) underpins both the mechanical performance and the dynamic responsiveness of fmDA‐based polyurethane CANs. A notable example of this synergy is shape‐memory, where hard segments fix the temporary shape, soft segments enable elastic recovery, and the fmDA adduct provides reprogrammability via reversible DA/rDA crosslinking.

TABLE 5.

Structure–property relationships in fmDA‐based polyurethane CANs.

No	Component	Molecular level	Result	Mechanical contribution	Dynamic contribution
1	Hard segments	Physical crosslinks (H‐bonds), rigid aromatic/aliphatic domains	Rigidity	Mechanical strength, solvent resistance, creep resistance, hardness, high modulus	Shape memory (shape fixation); indirectly limits self‐healing (restricts chain mobility)
2	Soft segments	Flexible high‐molecular‐weight chains	Mobility and diffusion	Elasticity, toughness, low Tg	Facilitates self‐healing (chain mobility), shape memory (recovery), reprocessability
3	fmDA adduct	Reversible covalent crosslinks (DA/rDA), rigid bicyclic core	Reversibility and network integrity	Tensile strength (covalent crosslinks), stiffness, crosslink density	Self‐healing, reprocessability, shape memory (switching/reprogrammability)

The data outlined in Table 5 also capture the strength‐healing dilemma: the same crosslinks that impart mechanical strength also restrict chain mobility and limit healing efficiency. The incorporation of additional energy‐dissipation mechanisms can effectively decouple mechanical robustness from healing efficiency. A range of complementary strategies have been developed to push the boundaries of material performance while preserving dynamic functionality. These approaches can be grouped into four main categories depending on the underlying mechanism: (i) architectural control and topological engineering (spatial placement of DA crosslinks), (ii) multidynamic network design (introduction of supplementary reversible bonds), (iii) nanocomposite reinforcement (filler‐induced energy dissipation), and (iv) SMASH (crack closure via shape recovery). Together, these strategies demonstrate that mechanical strength and healing efficiency can be reconciled through rational network design.

The versatility of the fmDA platform is best illustrated by the wide property tunability accessible through rational variation of molecular design parameters (Tables 3 and 4). This ranges from soft, highly extensible elastomers to stiff, rigid plastics through rational variation of molecular design parameters. Among these parameters, crosslink density (which is governed by the concentration of reversible bonds and thus by the functionality of the fmDA adduct) is the most influential factor. In general, higher cross‐link density leads to increased tensile strength, Young's modulus, and glass‐transition temperature, but at the expense of healing efficiency and chain mobility. Conversely, lower cross‐link density favors higher elongation at break, faster healing kinetics, and better shape memory, though with some sacrifice in creep resistance and stiffness [119]. The adduct‐based approach is particularly advantageous in this context as it allows the functionality of the fmDA adduct to be precisely varied (from diols to triols and tetraols) and the crosslink density to be systematically tuned. This molecular‐level control over the concentration and distribution of reversible bonds is the key advantage of the adduct‐based route and largely inaccessible in prepolymer crosslinking strategies.

From a practical synthesis perspective, the construction of dynamic polyurethanes utilizing fmDA adducts typically employs conventional polymer chemistry. Furfuryl alcohol or its derivatives are commonly used as dienes in the synthesis of adducts. Macromolecular diols such as poly(ε‐caprolactone) (PCL), polyglycols, and polytetramethylene ether glycol (PTMEG) are often employed as soft segment components. Hydroxyl‐functionalized 7‐oxanorbornenes are used as monomers, but the use of more reactive isocyanate‐substituted (Table 3, entries 15–17) or acrylic (Table 4, entries 10 and 11) derivatives has also been reported. Common tin‐based catalysts such as DBTDL facilitate the urethane‐forming reactions under mild conditions (typically up to 80 °C), preserving the integrity of thermoreversible cross‐linkers. The synthesis can be conducted either in polar solvents (e.g., THF [129], acetone [254], DMF [239], and DMAc [233]) or in bulk at elevated temperatures [117, 235], with careful protection from moisture to prevent undesirable isocyanate side reactions. As shown in Figure 5, the functionality of fmDA adducts can be systematically varied from diols to triols and higher‐functionality crosslinkers.

FIGURE 5.

Structures of fmDA monomers obtained from maleimides (a) or bis‐maleimides (b) used in the synthesis of CANs (described in Tables 3 and 4). Green highlights the furfural‐derived fragment of the 7‐oxanorbornene core.

Although systematic comparative studies of the adduct‐based and prepolymer cross‐linking strategies are scarce, a few reports have employed both approaches within the same study [237]. For example, Czifrák and coworkers synthesized epoxy‐polyurethane conetworks using either a preformed hydroxyl‐terminated adduct 6 or an in situ formed fmDA network via reaction of furan‐functionalized polyurethane with a furfurylamine‐modified DGEBA oligomer and BMI (Table 4, entry 7) [234]. This side‐by‐side comparison revealed that the adduct‐based approach afforded materials with a broader range of mechanical properties from soft elastomers (Young's modulus, E ~ 1–4 MPa) to rigid plastics (E > 230 MPa) depending on stoichiometric variations. In contrast, the in situ approach yielded consistently stiff networks (E = 294–648 MPa) with limited tunability. Notably, the adduct‐based route enabled the decoupling of fmDA network formation from epoxy cross‐linking, allowing independent optimization of dynamic reversibility and permanent network rigidity. In the adduct‐based strategy, the epoxy component (DGEBA) was introduced separately and could be additionally crosslinked with Jeffamine, providing an extra handle for tuning mechanical stiffness and network density. In the in situ approach, by contrast, the introduction of Jeffamine was precluded because the epoxy functionalities of the furfurylamine‐modified DGEBA oligomer had already been consumed by amine addition to introduce the furan moieties, leaving no reactive epoxy groups available for subsequent cross‐linking. This modularity, which is inherent to the adduct‐based design, facilitates the construction of more complex network architectures and underscores the strategic advantage of this approach for precision materials engineering.

Chen et al. showed that cyclic PCL segments yield more compact networks with higher shape fixity (~95%) compared to linear analogs (~80%) while maintaining fmDA‐mediated self‐healing (Table 4, entry 14). This topological effect arises from the reduced number of chain ends in cyclic architectures, which restricts chain slippage and enhances network cohesion without compromising the reversibility of DA cross‐links [260]. The example illustrates that controlling the placement of fmDA cross‐links (whether at the hard‐soft interface or within cyclic architectures) offers a powerful handle for tuning both mechanical and healing properties without altering the chemical composition.

One of the most powerful strategies for alleviating the trade‐off between strength and healing efficiency involves the integration of complementary dynamic interactions within a single network. By combining reversible covalent fmDA crosslinks with noncovalent motifs (predominantly based on hydrogen bonding), researchers have achieved high mechanical properties while maintaining efficient self‐healing and reprocessability [275, 276, 277, 278]. Representative examples of multidynamic systems obtained using adduct approach are listed in Table 3 (entries 12, 13) and Table 4 (entries 5, 11). Although dynamic covalent bonds can impart self‐healing and reprocessability to PUs, the resulting mechanical strength and toughness may not suffice for certain applications given that the energies of dynamic covalent bonds are typically lower than those of conventional covalent bonds [279]. The incorporation of dynamic noncovalent interactions introduces an additional energy dissipation mechanism and facilitates chain rearrangement at lower temperatures without compromising the covalent network integrity [188, 261, 280]. Consequently, the material assumes both high stiffness (from permanent DA crosslinks and strong supramolecular interactions) and rapid, efficient healing (from reversible bond dissociation and reformation), effectively decoupling these traditionally conflicting properties. This approach was efficiently realized in the synthesis of PU CANs by combining covalent Diels–Alder crosslinking with complementary hydrogen bonding. The seminal work by Li et al. demonstrated this concept through the incorporation of UPy (2‐ureido‐4[1H]‐pyrimidinone) motifs capable of forming quadruple hydrogen bonds as chain extenders in linear polyurethanes subsequently crosslinked via DA chemistry (Figure 6 and Table 3, entry 13). The resulting orthogonal dynamic networks exhibited exceptional mechanical performance: tensile strengths up to 46 MPa, elongations at break reaching 2196%, and toughness values as high as 406 MJ m⁻³ [233]. This remarkable combination arises from a hierarchical energy‐dissipation mechanism: UPy‐based hydrogen bonds act as sacrificial bonds that break and reform under stress, dissipating energy and enhancing toughness, while the permanent DA cross‐links maintain network integrity and provide structural stability. Critically, the presence of supramolecular interactions does not compromise the reversibility of the DA network; rather, it facilitates chain rearrangement at lower temperatures, effectively decoupling strength from healing efficiency [105].

FIGURE 6.

Example of a multidynamic polyurethane network combining covalent fmDA bonds and H‐bonding. Reproduced with permission [233]. Copyright 2018, Elsevier.

Similarly, Liu et al. developed stiff UV‐curable coatings based on double‐reversible networks combining DA crosslinking with hydrogen bonds (Table 4, entry 11) [258]. The incorporation of a rigid cyclotriphosphazene core and flexible PCDL arms yielded coatings with pencil hardness up to 3H, while the hydrogen‐bonding network enabled fast scratch healing (complete recovery within 60 s at 120 °C) without sacrificing mechanical stiffness [258]. Another elegant example is the combination of fmDA bonds with oxime‐carbamate dynamic linkages in waterborne polyurethanes (Table 3, entry 12). The dual dynamic network achieved tensile strengths of 36.6 MPa with healing efficiencies of 93% while enabling green recyclability by dissolution in acetone/ethanol at room temperature [178]. The integration of renewable biomass‐derived components represents both a sustainability imperative and a route to enhanced mechanical properties. Du et al. developed lignin‐based self‐healing polyurethanes combining DA bonds with hydrogen bonds from hydroxylated lignin (Table 4, entry 5). The lignin acts as a partial crosslinker, with optimal formulations achieving tensile strengths of ~29 MPa and healing efficiencies of up to ~102% [253]. Remarkably, the lignin‐based polyurethane outperformed petroleum‐based analogs in both mechanical and healing properties, demonstrating that bio‐derived components can synergistically reinforce the network through hydrogen bonding, while the DA network maintains dynamic responsiveness.

The incorporation of nanofillers represents a particularly attractive strategy for enhancing mechanical properties without sacrificing dynamic functionality as the fillers can reinforce the polymer matrix while the DA network remains responsible for self‐healing and reprocessability. Importantly, certain nanofillers can even accelerate the healing process through photothermal or electrically induced heating. Wu et al. demonstrated that functionalized graphene nanosheets (FGNS) covalently linked to the polyurethane matrix enable ultrafast infrared laser‐induced self‐healing (Table 4, entry 3). At just 0.5  wt% FGNS loading, the nanocomposites achieved tensile strengths of 36 MPa and healing efficiencies exceeding 96% within 1 min of IR laser irradiation. The photothermal conversion of FGNS rapidly raises the local temperature to 150 °C (within 20 s), triggering rDA bond cleavage precisely at the damage site while minimizing thermal exposure of the surrounding material [251]. This approach not only improves mechanical properties but also enables spatially controlled rapid healing that is particularly valuable for flexible electronics and smart coatings. Similarly, Ding et al. employed CNT‐Ag hybrid nanoparticles to create thermally conductive pathways that accelerate healing (Table 4, entry 2). The silver nanoparticles act as “bridges” between carbon nanotubes, reducing interfacial thermal resistance from 8.13 × 104 to 1.37 × 104 K/W and enabling more efficient heat transfer during the healing process. At 1 wt% loading, the composites achieved healing efficiencies of 90.1% (tensile strength) and 92.9% (strain) at 130 °C [250]. Nanoparticles TiO₂ have also been successfully integrated into fmDA polyurethanes, affording dielectric nanocomposites (Table 4, entry 4). The nanocomposites exhibited enhanced dielectric constants (from 8.7 to 13.7 at 1 kHz) and mechanical properties (TS up to 19.85 MPa), with dielectric properties almost fully recovered after healing [252]. This demonstrates that nanofillers can impart multifunctionality, while the DA network retains its dynamic character. Collectively, these studies demonstrate that nanofillers not only reinforce the polymer matrix but can also impart additional functionalities such as photothermal, electrical, or dielectric properties while preserving the dynamic character of the fmDA network.

The integration of SMASH with fmDA chemistry represents a particularly elegant physically driven strategy for overcoming the healing of large‐scale or wide cracks, where simple thermal activation of DA bonds is insufficient to bring damaged surfaces into contact. The SMASH strategy was demonstrated by Nguyen et al., who incorporated a tris‐DA adduct into a PDMS‐based polyurethane network (Figure 7). The material combined shape‐memory‐assisted crack closure with efficient self‐healing at 60 °C, enabled by the flexibility of PDMS segments and free maleimide groups that facilitated bond mobility [119]. Another example of SMASH approach was demonstrated in a polyurethane elastomer containing DA adducts and crystallisable PCL soft segments (Table 4, entry 1). Upon programming, the material autonomously shrinks at 60 °C due to PCL melting, narrowing wide scratches (80–100 µm), while subsequent heating to 110 °C triggers rDA bond cleavage, enabling chain flow and complete scratch closure [177]. This dual mechanism yields both high initial tensile strength (32.9 MPa) and excellent healing efficiency (113% recovery). Indeed, the system achieved >100% recovery of tensile strength (up to 34.58 MPa), demonstrating that the synergistic effect of shape‐memory‐assisted crack closure and DA bond reconstruction can actually enhance mechanical properties beyond the original state [177]. Heo and Sodano pioneered the use of shape‐memory for autonomous crack closure in fmDA polyurethanes, achieving healing efficiencies of 80%–84% over multiple cycles without external force (Table 3, entry 3).

FIGURE 7.

Schematic representation illustrating an approach to form a thermoreversible network. Reproduced with permission [119]. Copyright 2020, Elsevier.

In summary, each of the four strategies addresses the strength–healing trade‐off through a distinct mechanism: chemical reinforcement (multidynamic networks), structural optimization (architectural control), physical reinforcement (nanofillers), or mechanical actuation (SMASH). Their combination within a single material offers a promising route toward high‐performance, durable, and sustainable fmDA‐based polyurethane CANs. These strategies have enabled fmDA‐based polyurethanes to achieve a level of performance and versatility that translates directly into practical applications across diverse industrial sectors; the adduct‐based approach provides the valuable platform for implementing them. Key representative examples highlighted in Tables 3 and 4 include high‐performance elastomers (Table 4, entry 1) [177], debondable adhesives for recyclable wooden products (Table 3, entry 7) [129], and smart coatings (Table 3, entries 4, 15; Table 4, entry 11) [240, 243, 258]. A visual representation of self‐healable shape‐memory PUs for various applications, including selective laser sintering (SLS) [179], 4D printing [254, 256], and thermally induced shape recovery, is provided in Figure 8. Beyond these established examples, a dominant trend in this field is the shift toward more sustainable systems, as evidenced by the growing number of waterborne formulations (Table 3, entries 6, 11, and 12) [178, 184, 281].

FIGURE 8.

Demonstration of the shape‐memory effect in fmDA‐based polyurethane CANs. (A) 4D‐printed selective laser sintering object before and after shape recovery. Reproduced with permission [179]. Copyright 2022, American Chemical Society. (B) Shape‐memory polyurethane acrylate for DLP 3D printing. Reproduced with permission [254]. Copyright 2022, Elsevier. (C) Thermally induced shape recovery of a reversibly crosslinked shape‐memory material. Reproduced with permission [119]. Copyright 2020, Elsevier.

The integration of urethane bonds with (meth)acrylate functional groups enables the development of photocurable PUs. This class of high‐performance materials that transforms from a liquid to a solid film almost instantaneously upon exposure to ultraviolet (UV) has been gaining increasing attention for photocurable coatings, adhesives, and filaments for 3D printing [282, 283]. Self‐healing UV‐curable urethane (meth)acrylate (UA) coatings were successfully prepared via a photoinitiated radical polymerization of the (meth)acrylate groups located at the termini of furanic oligomer chains. As demonstrated in a series of earlier studies, the chemical structure of the employed DA adduct and UA oligomers significantly influences the photocuring kinetics, process efficiency, and the temperature required for self‐healing (Table 3, entries 4, 10, 11; Table 4, entries 10, 11). This photocurable branch deserves particular emphasis because it connects the chemistry of dynamic polyurethanes with digital manufacturing. Recent high‐level analyses of vat photopolymerization of polyurethanes show that resin latency, cure speed, viscosity window, and postprint network rearrangement are emerging as the key design parameters for 3D/4D‐printable systems. Therefore, dynamic urethane acrylates are especially attractive when reversible network adaptation is required after printing rather than only during printing [284].

This point can be strengthened further by placing fmDA adducts within the rapidly growing nonisocyanate CAN landscape. Recent reports demonstrate that dynamic nonisocyanate networks are increasingly evaluated not merely as safer analogs of conventional PUs but as platforms for chemical recycling, reprocessing, and shape programmability. Accordingly, furfural‐derived monomers may become valuable building blocks for integrating biomass origin, reversible cross‐linking, and isocyanate‐free synthesis in one material concept [140, 156]. The strategy of using functional adducts is directly compatible with the principles of green chemistry through its application in nonisocyanate polyurethane synthesis. Pioneering work by Dolci et al. demonstrated that 7‐oxanorbornenes bearing cyclic carbonate groups could react with diamines to yield remendable poly(hydroxyurethane) networks, opening a sustainable pathway to reversible macromolecules that entirely avoid the use of toxic isocyanates [285]. Subsequently, Karami et al. synthesized a bio‐based NIPU network by reacting a furan‐containing cyclocarbonate (obtained by CO₂ fixation on furfuryl glycidyl ether) with bismaleimide, followed by amine curing (Table 4, entry 12). The resulting network exhibited thermo‐reversible behavior and superior healing compared to its epoxy counterpart [259]. Although NIPU‐based Diels–Alder CANs remain rare [265, 286, 287, 288], this approach offers a promising pathway to eliminate toxic isocyanates and organotin catalysts while maintaining dynamic functionality. While most of these systems rely on bismaleimide crosslinkers, Wang et al. demonstrated an alternative approach using an unsaturated polyester as the dienophile [158]. The resulting polymers achieved self‐healing efficiencies up to 100% at rDA temperatures of 90–100 °C. This convergence of different dynamic covalent chemistries and isocyanate‐free synthesis represents a critical frontier for the next generation of sustainable smart materials, aligning advanced functionality with enhanced environmental compatibility [176].

Overall, current molecular‐level studies indicate that the next advances in Diels–Alder‐based recyclable networks will depend on a finer control of intrinsic reaction thermodynamics and substituent effects rather than on empirical monomer screening alone. For polyurethane CANs, this means that future progress may come from coupling computationally informed diene/dienophile selection with monomer‐based network construction, enabling a more rational optimization of healing temperature, side‐reaction resistance, and service window stability [289].

3. Conclusions

This review has demonstrated that the adduct‐based fmDA platform offers a compelling route to sustainable, high‐performance polyurethane CANs. This technology shows promise for many industrially important areas, such as aviation, automotive, wind energy, corrosion protection, consumer electronics, soft robotics, and biomedical devices. A high‐productivity way to frame this field is to recognize that fmDA‐based polyurethane CANs address three grand challenges simultaneously: decarbonization of feedstocks, extension of service lifetime, and recovery of value at end of life [21, 32]. This triple convergence is increasingly viewed as a defining criterion for next‐generation adaptable polymers, and it is precisely where furfural‐derived monomers can offer a distinctive contribution. However, its practical implementation is constrained by inherent chemical and processing limitations that must be critically addressed.

A balanced assessment can be framed through a SWOT analysis. Strengths include catalyst‐free and atom‐economic bond exchange, compatibility with renewable furans, high mechanical performance, modular control over network topology, and an established library of functionalized adducts with well‐understood structure–property relationships. Weaknesses encompass the intrinsic strength‐healing trade‐off, the narrow dynamic window for rDA activation, maleimide side reactions, furan instability and discolouration, reliance on petrochemical comonomers and toxic tin catalysts, and the CMR classification of furfural and furfuryl alcohol. Opportunities lie in the development of NIPU routes, benign catalysts (e.g., bismuth carboxylates), waterborne and UV‐curable formulations, integration with 3D/4D printing, expansion to other furan platform chemicals (HMF), and the use of acceptor‐substituted furans for improved stability. Threats arise from the high cost of maleimides, competition from cheaper and more thermally stable vitrimers (such as transesterification) and from disulfide‐ or imine‐based systems with milder activation conditions, as well as the absence of closed‐loop recycling and LCA data required for industrial adoption [290].

Addressing these limitations requires a focused research agenda: (1) expand the operational temperature window by reducing rDA temperatures and accelerating rebonding kinetics; (2) suppress irreversible side reactions via steric protection, radical inhibitors, or alternative diene‐dienophile pairs; (3) move beyond single‐cycle healing demonstrations to assess mechanical retention and network integrity under realistic environmental stress; (4) report property retention after ≥5 reprocessing cycles with clear distinction between reprocessing, chemical recycling, and closed‐loop monomer recovery; (5) replace petrochemical polyols, diisocyanates, and organic solvents with renewable alternatives and quantify the overall biogenic carbon content; (6) prioritize NIPU routes and benign catalysts to avoid CMR substances; and (7) conduct comparative LCAs to validate actual environmental benefits.

For this reason, the most promising future studies will likely be those that combine monomer‐level synthesis with mechanistic stability analysis under realistic processing studies, including repeated thermal cycling, oxygen exposure, UV irradiation, and prolonged residence at partial rDA conversion. Such projects would move the field from elegant reversibility demonstrations toward durability‐qualified smart thermosets, which is the level of evidence typically required for broader technological adoption and next‐generation materials [113, 289]. In essence, the trajectory of this field follows a clear logic: molecular‐level control (enabled by monomer‐based design) → multidynamic networks (integrating covalent and supramolecular interactions) → achieving both mechanical robustness and efficient healing → smart polymers for advanced manufacturing (self‐healing, shape‐memory, and 4D‐printable systems). Extending this logic from furfural to other furan platform chemicals, such as HMF, represents an obvious and promising direction for further expanding the structural diversity and functionality of sustainable polyurethane CANs. By unifying dynamic covalent chemistry with renewable feedstocks, novel diene–dienophile pairs, and isocyanate‐free synthesis, next‐generation polyurethane CANs exemplify new polymers combining high performance and sustainability.

Funding

This work was supported by the Russian Science Foundation (grant number 23‐73‐00003) (analysis of chemical aspects) and the P.L. Kapitsa Grant Program (Phase III) at Moscow Polytechnic University (analysis of applications).

Conflicts of Interest

The authors declare no conflicts of interest.

Biographies

Konstantin I. Galkin is a research fellow in the Laboratory of Metal‐Complex and Nanoscale Catalysts at the N.D. Zelinsky Institute of Organic Chemistry, Russian Academy of Sciences (Moscow). In 2009, he graduated from the D.I. Mendeleev University of Chemical Technology of Russia, and in 2014, he received his Ph.D. in organic chemistry from the A.N. Nesmeyanov Institute of Organoelement Compounds, Russian Academy of Sciences. His research interests include the chemistry of biomass‐derived furans and their applications in fine organic synthesis and materials science.

Valentine P. Ananikov received his Ph.D. degree in 1999 and habilitation in 2003, and in 2005, he was appointed Professor and Laboratory Head of the N.D. Zelinsky Institute of Organic Chemistry, Russian Academy of Sciences. In 2008, he was elected a Member of the Russian Academy of Sciences. He was the recipient of a Liebig Lectureship from the German Chemical Society (2010) and Organometallics Distinguished Author Award Lectureship by American Chemical Society (2016). His scientific interests focus on catalysis, the development of new sustainable methods, and molecular complexity.

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.