Hyperbranched Dynamic Crosslinking Networks Enable Degradable, Reconfigurable, and Multifunctional Epoxy Vitrimer

Abstract Degradation and reprocessing of thermoset polymers have long been intractable challenges to meet a sustainable future. Star strategies via dynamic cross‐linking hydrogen bonds and/or covalent bonds can afford reprocessable thermosets, but often at the cost of properties or even their functions. Herein, a simple strategy coined as hyperbranched dynamic crosslinking networks (HDCNs) toward in‐practice engineering a petroleum‐based epoxy thermoset into degradable, reconfigurable, and multifunctional vitrimer is provided. The special characteristics of HDCNs involve spatially topological crosslinks for solvent adaption and multi‐dynamic linkages for reversible behaviors. The resulting vitrimer displays mild room‐temperature degradation to dimethylacetamide and can realize the cycling of carbon fiber and epoxy powder from composite. Besides, they have supra toughness and high flexural modulus, high transparency as well as fire‐retardancy surpassing their original thermoset. Notably, it is noted in a chance‐following that ethanol molecule can induce the reconstruction of vitrimer network by ester‐exchange, converting a stiff vitrimer into elastomeric feature, and such material records an ultrahigh modulus (5.45 GPa) at −150 °C for their ultralow‐temperature condition uses. This is shaping up to be a potentially sustainable advanced material to address the post‐consumer thermoset waste, and also provide a newly crosslinked mode for the designs of high‐performance polymer.

analysis of small molecular model reaction was performed on a Bruker Advance Neo 500 MHz spectrometer.The concentration of the -OH group was determined by titrimetric test, in accordance with GB/T 7383-2007 (Chinese standard).The molecular weight information was determined by Gel Permeation Chromatography-Eighteen Angle Laser Scatterer (GPC-MALLs, Waters), equipped with a DAWN HELEOS II system (LS) differential refraction detector (dRI) with tetrahydrofuran (THF, SP) as mobile phase.
The mechanical flexural and impact strengths were measured using an electro-mechanical tester in accordance with GB/T 2567-2008 (Chinese standard).The sample dimensions were 80 × 15 × 4 mm 3 (flexural strength) and 80 × 10 × 4 mm 3 (impact strength).Dynamic thermomechanical analysis (DMA, NETZSCH DMA 242E) was performed using a three-point bending model at the frequency of 1 Hz.The storage and loss modulus were continuously recorded from 50 o C to 300 o C, and the samples were monitored at a wider temperature range of -150 o C to 200 o C after solvent treatment.Differential scanning calorimetry was conducted under N2 atmosphere using a Mettler-Toledo DSC3.The rheological measurements of the samples were performed on a stress-controlled rheometer (ThermoFisher Mars) using a 25 mm parallel plate under isothermal time-sweep mode at 150 o C with a constant frequency of 1 Hz.
Thermogravimetric analysis was performed under nitrogen atmosphere using a STA 449F3 thermal analyzer with a heating rate of 15 °C•min -1 .The impact fracture surface was photographed using a Scanning Electron Microscope equipped with a tungsten filament gun (TESCAN VEGA 3 LMH), while Raman spectra were obtained using an Alpha300R spectroscopy equipped with a 532 nm TEM00 laser.
The fire-retardant performance was evaluated using a Cone calorimeter test at a heat flux of 35 kW/m 2 , according to ISO 5660 standard.The sample dimensions were 100 × 100 × 4 mm 3 .The char residue after the cone test was examined using a field emission Scanning Electron Microscope (FEI Verios G4).The limited oxygen index (LOI) was measured using samples of size 80 × 10 × 4 mm 3 , in accordance with GB/T 2406.2-2009(Chinese standard).
The UL 94 vertical burning test was carried out according to ASTM D3801, with specimen dimensions of 125 × 13 × 4 mm 3 .The transparency was measured using a UV-vis spectrophotometer (Hitachi U-3900), and the average transmittance over the wavelength range was obtained by integrating the measured data.

S1.3. Synthesis details and structural information for HBPPB a. Synthesis details of HBPPB
In this study, a hyperbranched phosphate/borate hybrid polymer (HBPPB) was synthesized via a one-pot A2+B3+C3 transesterification.The resulting polymer, as depicted in Figure S1, shows a representative model of the hyperbranched ternary structure.In such structure, the macromolecular backbone is randomly linked with phosphorus and boron sites.The feed ratio was strictly controlled in accordance to Flory-Carothers law to avoid over-crosslinking.The surplus PDO served as the donor monomer to terminate abundant hydroxyl groups in resulting polymer.

Figure S1
Synthesis of HBPPB Specifically, 0.35 mol of 1,3-propanediol (PDO) was mixed with 0.08 mol tributyl borate (TBB) and 0.08 mol of triethyl phosphate (TEP) in a three-necked flask, equipped with a mechanical stirrer, an N2 inlet, and a condenser.The reaction process comprised of three stages, each occurring within a different temperature range.Throughout the polymerization, the byproduct distillate can be continuously separated from reaction system by the condenser to push the reaction forward, and the reaction distillate was collected for FT-IR detection.Firstly, the mixture was heated from 120 °C to 160 °C at a heating rate of 5 °C h -1 .Subsequently, the temperature was raised from to 180 °C for maintaining 2 hours.Finally, the elevated temperature was allowed for 190 °C and maintained until no more distillate generated.The product was purified through dialysis to remove redundant small molecules, followed by pressure-reduced drying.This yielded a yellowish liquid product, namely HBPPB, with a yield of approximately 57.2% and molecular weight of Mn = 64.2kDa (Table S1), calculated based on the mass of the final product relative to the mass of initial feed.

b. Structural information of HBPPB
The molecular structure of HBPPB was investigated using several techniques including by fourier transform infrared (FT-IR) spectroscopy, 1 H, 13 C and 31 P nuclear magnetic resonance (NMR) spectroscopy.Initially, FT-IR was employed to determine the functional groups in the monomers (TBB, TEP and PDO) and the HBPPB (Figure S2a).The IR profiles of TBB and TEP showed similar peaks at 1050 cm -1 , corresponding to similar ester groups of B-O-C and P-O-C, respectively, which also appeared in the spectrum of HBPPB.The peak at 1450 cm -1 observed in PDO, attributed to the stretching vibration of C-OH, was also present in the spectrum of HBPPB.This indicated that the chemical structure of HBPPB contained B-O/P-O groups as well as C-OH and/or C-O-B and/or C-O-P stretching vibrations from PDO, suggesting the occurrence of the reaction. [1]The spectrum of HBPPB also exhibited abundant hydroxyl groups (-OH) in the range of 3000 to 3500 cm -1 , with a peak at 2850 cm -1 to 3000 cm - 1 assigned to the methylene group.Further evidence of the reaction was obtained from the IR profiles of the distillate collected at different temperatures (160°C and 190°C).In Figure S2b, the distillate at 160°C primarily consisted of 1-butanol, while the distillate at 190°C mainly contained ethanol.This indicated that transesterification had occurred between TBB and PDO before the reaction with TEP.
The 1 H-NMR and 13 C-NMR spectra of TBB, TEP, PDO, and HBPPB provided additional structural evidence (Figure S3).DMSO-d6 was selected for TEP and HBPPB, while CDCl3 were chosen for TBB and PDO.It's is worth noting that the proton and carbon from H2 and C2 in PDO are separated into different modes in the spectrum of HBPPB, corresponding to the various link mode and site of boron and phosphorus within hyperbranched structure, indicating the proceeded polymerization.Additionally, a broad signal in the range of 4.1-4.6 ppm corresponds to the active proton of the hydroxyl group. [2]

Figure S5
The 31 C-NMR assignment of TEP and HBPPB Generally, the chemical shifts of tetra-coordinated phosphorus(V)-oxygen compounds range from about 0 ppm to -30 ppm, including phosphate and their compounds in ester state.
After polymerization, a small offset from -0.9 ppm to -1.2 ppm is recognized in the 31 P-NMR spectrum of HBPPB, demonstrating that HBPPB still retains the phosphate structure (Figure S6).Additionally, the phosphorus signal in HBPPB splits into five spikes in a closer range, corresponding to the phosphate units at different sites adjacent to boron sites within the hyperbranched structure. [3]gure S6 The 31 P-NMR spectra of TEP and HBPPB The gel permeation chromatography gives the molecular weight information of HBPPB (Table S1).The Gel Permeation Chromatography-Eighteen Angle Laser Scatterer (GPC-MALLs) equipped with DAWN HELEOS II system and differential refraction detector, with tetrahydrofuran (THF, SP) as mobile phase.The number-average molecular weight (Mn) was determined as 64.2 kDa, proving the macromolecular structure of HBPPB.The concentration of hydroxyl group within HBPPB was determined using a titration experiment following the procedure outlined in GB/T 7383-2007 (Chinese standard).Firstly, approximately 1.00 g of HBPPB was mixed with 25 ml phthalic anhydride pyridine solution (140.00 g to 1 L pyridine).The mixture was heated at 115 o C for 60 minutes under a condensing reflux.Next, 4 ~ 5 drops of phenolphthalein indicator were added, and the solution was titrated with 0.5 M sodium hydroxide solution (repeat 3 times) until the pink solution maintaining for more than 15 s.A control trial tset without HBPPB was carried out for three times.The average Q(OH) can be calculated using Equation (1) [4] , as given in Table S2.
Where c is the concentration of sodium hydroxide solution (0.50 mol•L -1 ); V0 and V1 are the volume of sodium hydroxide solution consumed by controlled trial and HBPPB samples, respectively; m0 is the mass of HBPPB.

S1.4. Synthesis details and characterizations for HBPB and HBPP
In parallel, two comparable hyperbranched structures, namely hyperbranched polyborate (HBPB) and hyperbranched polyphosphate (HBPP), were synthesized as controls, each featuring single dynamic bonds of BO3 and PO3 (Figure S7).The synthesis followed the same molar ratio as that of HBPPB.For the synthesis of HBPB, 0.16 mol TBB was mixed with 0.35 mol of PDO in a 100 mL three-necked flask equipped with a mechanical stirrer, an N2 gas inlet and a thermometer.The mixture was stirred and heated from 120 °C to 190 °C at a heating rate of 5 °C•h -1 with continuous N2 blowing-in.The by-product was separated from reaction system by a condenser until no more distillate generating.The product was collected and dialyzed to remove small molecular substances.For the synthesis of HBPP, 0.35 mol of PDO was mixed with 0.16 mol of TEP using the same reactors.The mixture was heated from 120 °C to 190 °C under the heating rate of 5 °C•h -1 .Until no more distillate generating, the reaction product was collected and dialyzed.Finally, the two dialysate liquids were rotary evaporated at 45 °C and dried in a 60 °C vacuum for 6 h, yielding viscous liquid product, denoted as HBPB and HBPP.

Figure S7 Synthesis route of HBPB and HBPP
The structure evidence of HBPB is referred to the previous study of our group. [5]Since the poor compatibility between HBPP with epoxy matrix, the structural information of HBPP is not discussed in the current context, and only exploratory experiments are carried out.

S1.5. Synthesis details and characterizations for LPPB
Concurrently, a linear poly-phosphate/borate hybrid (LPPB) was synthesized to provide a point of comparison with the hyperbranched structure.Two monomers including a methylboronic acid (MBA) and a methylphosphonic acid (MPA) were selected as substitutes for the ternary monomers in HBPPB, allowing for esterification with PDO.
In brief, the reaction followed the feed ratio of 0.35 mol of PDO, 0.12 mol of MBA and 0.12 mol of MPA, thereby ensuring an equivalent molar amount of dynamic units compared to HBPPB.Dehydration catalyst in the form of 0.05 g of p-toluenesulfonic acid was added.Since the synthesis relied on the dominance of kinetic route for polymerization rather than intermolecular cyclization, the reaction was therefore conducted at a relatively low temperature (120 o C) for long-time polymerization until no more distillate generated.The distillate was indentified as water since an anhydrous copper sulfate quickly turns blue, indicating the asproceeded polymerization.The reaction product was initially purified using a dialysis bag (D2000) to eliminate the small molecules, yielding a yellowish liquid product with a yield of approximately 70.6%.

Figure S8 Synthesis route of LPPB
Figure S9 gives the IR information of the reaction monomers and the synthesized LPPB.
In the spectrum of LPPB, a methylene stretching vibration appears at 2900 cm -1 from the esterification reaction of PDO with MBA and MPA.A broad peak at 2300 cm -1 corresponds to -P(CH3)-, while the peaks at 1050 cm -1 belongs to the stretching vibration of B-O-C/C-O-C/P-O-C, proving the presence of BO2/PO2 units.Additionally, a relative decrease in the intensity of the hydroxyl group signal in LPPB indicates the consumed hydroxyl groups, proving the proceeded reaction.

Figure S9
The FTIR spectra of the reaction monomers and the as-synthesized LPPB

S1.6. Synthesis details of HDCNs
The synthesis of HDCNs employed conventional thermosetting workflow to cast vitrimer samples.The resin formulations are outlined in Table S11, comprising of a hyperbranched  In parallel, epoxy vitrimers containing different mass fractions of HBPB, HBPP and LPPB were fabricated as controls, respectively, following the same route as the HDCNs.Notably, both HBPPB and HBPB exhibited uniform resin solutions.However, HBPP shows poor compatibility within the epoxy matrix over 6% mass content (Figure S11a), which can observe deposit and bulk agglomerations.This outcome proves that HBPP alone cannot serve as an effective crosslinker in such a system.100°C for 30 minutes with stirring.Subsequently, 24 g of MTHPA and 0.24 g of DMP-30 were added following the same thermosetting procedure to fabricate the post-cycled EP for testing.
The mechanical strengths of the cycled sample were assessed using an electro-mechanical tester.
In the context of reconfigurable experiments, we subjected the samples into a closed container upon heating in ethanol at 95 °C.After 6 hours, the samples were cooled to room temperature.While the thermoset EP retained its rigid form, EP-9 and EP-12 shows significant transformation from a stiff to an elastomer nature (video S1).The samples were evaluated using FT-IR, XPS detection, temperature-dependent IR, dynamic thermomechanometry, and so on.These results substantiate the presence of ester derivatives of methyl tetrahydrophthalic anhydride originating from the solvent residue, supporting the occurrence of ester exchange during vitrimer reconfiguration.However, it should be noted that it is theoretically difficult to undergo complete depolymerization of polymer network, since the transesterification reaction is an equilibrium process with limited equilibrium conversion under kinetic-dominant conditions without catalysts. [6]Additionally, similar solvent-assisted dynamic reconfigurability in ethanol has also been observed under transesterification conditions. [7]gure S14 Characterization of the dissolved substance in ethanol after drying: a) FTIR spectrum, and b) 1 H-NMR compared to that of MTHPA.
To elucidate the role of ethanol as a plasticizer, the post-treated vitrimer sample was weighed after complete drying in a vacuum at 60 °C overnight.The sample weight was 6.78 g, derived from an original sample weight of 5.96 g, with 0.27 g being dissolved in the solvent.
This observation indicates that the sample underwent swelling, with ethanol molecules penetrating the polymer network and potentially acting as a plasticizer to enhance chain mobility and flexibility.Indeed, the solvation effect plays a key role in this process. [8,9]rthermore, we investigated the effects of other commonly used solvating-type plasticizers by immersing EP-9 in various solvents (Table S3).It was observed that reconfigurability was consistently accompanied by the swelling of sample in cases of alcohol-or ester-type solvents.
However, the swelling did not necessarily lead to reconfigurability when acetone was used as the solvent.Hence, we considered that plasticizing effect might involve in reconfigurable process, but the ester exchange should be the main mechanism to drive the reconfigurable behavior of the material.

b. Synthesis and characterization of two model esters
The methodology of model reactions was employed for further investigation. [10]In principle, to verify the actions of hyperbranched structure, the synthesis of model compounds should involve the utilization of hyperbranched macromonomer, instead of small molecules while applicable. [11,12]Two model esters (denoted as M1 and M2) were synthesized as shown in Figure S15 and S17, each featuring specific ester structures and molecular weights (Table S4).The synthesis subjected to the same stoichiometric ratio as the preparation of EP-9 vitrimer.The model reactions were conducted by heating at 95 o C in ethanol.Specifically, 10 g of the as-synthesized M1 and M2 were individually mixed with 10 g of absolute ethanol.The mixtures were then placed in a flask equipped with a condenser and heated at 95 o C for varying durations of 0.5 h, 2 h, and 4 h.After the reactions, excess ethanol was removed by employing a 60°C rotary evaporator, resulting in the final viscous products (denoted as M1-R and M2-R, respectively).The molecular weights of M1-R and M2-R were both reduced as given in Table S4, indicating depolymerization by ester exchange between ethanol and the model esters.

S2. Material performance test S2.1 The calculation of network crosslinking density
Based on the rubber elasticity theory, the crosslinking density (dcrosslink) of a polymer network can be calculated using the dynamic thermomechanometry method.The calculation is based on Equation (2), with the results are listed in Table S5.
Where E' is the rubbery plateau storage modulus at Tg + 40 o C; γ is Poisson's ratio, generally assumed to be 0.5 when the crosslinked network is incompressible in case of a thermoset; R is the gas constant and T is absolute temperature.It should be noted that this equation is applicable for lightly crosslinked materials and therefore is used only to qualitatively compare the level of crosslinking in the casting resins. [13,14]The result for dcrosslink is presented in the text as Figure 4c.Tg: glass transition temperature; E': storage modulus at three-bending mode; dcrosslink: crosslinking density of polymer

S2.3. Stress relaxation test
Stress-relaxation analyses (SRA) were performed on a Dynamic Mechanical Analyzer Q800 (TA Instrument, Waters Ltd.) with the sample dimensions of 20 mm length, 5 mm width, and 1 mm thickness under the tension mode.After reaching the testing temperature, the samples were allowed to equilibrate at the testing temperature for 10 minutes, then stretching by 5% and the relaxation of stress was monitored.
It is widely acknowledged that classical vitrimers behave linear-viscoelasticity that readily relaxes the stress above their topology-freezing transition temperature (Tv). [15]In case of a purely elastic solid, the strain (force) is instantly responsive regardless of how the stress (change) varies with time, thus we found constant curves in stress relaxation (Figure S24) and creep-recovery test (Figure S26) regarding the thermoset sample.
Nevertheless, Figure S24b illustrates the vitrimer partially relax to a plateau regime (approximately G/G0 to 70%) while fails to entirely relax to the characteristic relaxation time (τ*, defined as the time required for G/G0 = 1/e).This fact indicates that the hyperbranched dynamic crosslinking networks (HDCNs) can behave liquid-like viscoelastic properties due to the activation of dynamic bonds above Tv (describing by Arrhenius and Williams-Landel-Ferry law), but still remains their resilience and recovery ability subject to Hooke's law.
Consequently, the material not only behave like a dynamic vitrimer, albeit with reduced dynamic capabilities compared to classical vitrimers, while demonstrate exceptional dimensional stability and mechanical strength even over Tg and Tv, which might attribute to permanent crosslinked sites and potentially relate to the topological crosslinking architecture.In order to determine the activation energy (Ea) and topology-freezing transition temperature (Tv), we could assign this exception as a linear-viscoelastic model combining with an ideal elastic.Considering the complexity and limitations associated with solving these nonlinear viscoelastic issues, [16,17] we simplify the stress-relaxation curves by dividing them at the boundary of the stress-relaxation plateau (Figure S24b to S24a).The upper region was assigned as viscoelastic relaxation, while the region below the plateau is approximately denoted as an ideal elastic polymer.Based on the equivalence, we can assume the sample fully undergoes its viscoelastic relaxation once they reach the plateau.The characteristic relaxation time (τ*) can be therefore determined using a modified Maxwell equation until G/G0 relax to 1/e' (Equation 3), [18,19] where e' is modified by a correction factor k in Equation 4.
Where k is a correction factor (value at which G/G0 reaches to the relaxation platform), assuming to 0.7 in this system.The activation energy (Ea) could be determined from the slope in an Arrhenius-plot (Figure S25a) using Equation 5-6 as follows.
* =  0     (5)   ln( * ) = ln( 0 ) + The topology-freezing transition temperature (liquid-to-solid transition temperature, Tv) is defined as the point at which a vitrimer exhibits a viscosity (η) of 10 12 Pa s, which may therefore be determined using Maxwell's relation (Equation 7) and E' determined from DMA. [20] τ* was determined as ca.1.9 × 10 5 at Tv.The Arrhenius relationship was then extrapolated to determine Tv as shown in Figure S25b.

S2.5 Comparison of degradation and reconfigurable properties
In cases of three similar structures (LPPB, HBPP, and HBPB), their degradation and reconfigurable performance were evaluated and compared with that of the hyperbranched crosslinker (HBPPB).
a. Degradation and reconfigurable properties of LPPB/EP For the degradation (Figure S27), LPPB/EP also underwent swelling and fragmentation when exposed to DMAc, while the weight retention of 12LPPB/EP is higher than that of 12HBPPB/EP, and its E' and Tg value is significantly lower than the thermoset epoxy and 12HBPPB/EP.For the reconfigurable performance (Figure S28), notably, the ethanol-induced dynamic network reconfiguration properties were not well reflected in LPPB/EP.Not only was its storage modulus inferior to 9HBPPB/EP sample at low temperature (-100 o C, Figure S27b), but also the network reconstruction has not been well recognized, as its tan δ curves shows two distinct peaks (Figure S28c).For the judgement, the same amount of BO3/PO3 and OH groups are linearly introduced, the degradation and reconfigurable properties is not good as the hyperbranched crosslinked one.The flexural strain (ε) at the outer surface of materials occurs at mid-span is calculated as follows: where:    S8.

S2.8. XPS surveys of degradation and reconfigurable vitrimer
In conjunction with the FTIR results, X-ray photoelectron spectroscopy (XPS) was employed to assess the bonding condition following sample degradation. [21]The sample was fully dried and ground into powder for detection.Informed by XPS spectrum (C1s is corrected to 284.8 eV for each spectrum), the beta-hydroxy ester was partially broken after soaking in  To validate the reconfigurable nature of samples, the dynamic ester exchange was further confirmed using XPS analysis.The high-resolution O1s spectrum exhibited a notable shift following after the samples heating in ethanol, which is attributed to the ester exchange between ethanol and the ester moieties for resulting network reconstruction.

S3. Supporting Movie
Movie S1 After heating in ethanol, EP-9 vitrimer sample shows transformative property in material nature from stiff vitrimer to elastomer that can be casually deformed and recovered as will at room-temperature.

Figure S2
Figure S2 FTIR spectra of a) the monomers and the as-synthesized HBPPB, b) reaction distillate comparing with standard 1-butanol and ethanol
macromonomer (HBPPB), a commercial petroleum-based epoxy resin (DGEBA, E51), an anhydride-type curing agent (MTHPA), and a small quantity of tris(dimethylaminomethyl)phenol (DMP-30, curing catalyst).Specifically, varying mass fractions of HBPPB (3 wt%, 6 wt%, 9 wt%, 12 wt%) was mixed with 80.0 g of DGEBA and stirred in a 250 ml beaker at 100 °C for 10 minutes, yielding in a yellow-transparent resin solution.The mixture was then cooled to 60 °C for the next step.Subsequently, 56.0 g of MTHPA were added and stirred at 60 o C for 15 minutes.Finally, 0.7 g of DMP-30 was dropped into the pre-polymer solution and degassed in a 60 o C vacuum for 30 minutes before casting into the mold.The resulting samples, denoted as EP-x, where x represents the mass fraction of HBPPB, were cured following a temperature procedure of 120 °C for 2 hours, 150 °C for 3 hours, and 180 °C for 2 hours.The typical reactions occurring during polymer crosslinking are described in Figure S10.

Figure S10
Figure S10Typical reactions between HBPPB and epoxy/anhydride system.

Figure S11
Figure S11 Compounded resin solutions showcases the poor compatibility in a) 6% HBPP/EP, and uniform solutions in both b) HBPB/EP and HBPPB/EP.

Figure
Figure S12 FT-IR spectra indicates the consumption of -OH in crosslinking of HDCNs

Figure S13
Figure S13 Photographs showing the samples with varying mass concentrations of HBPPB exposed in deionized water, ethanol, and DMAc at room temperature for several days, and at 95°C for 6 hours.
heated at 120 o C for 1 hour, yielding the target pre-polymer ester M1 (Mn =18561.5 kDa, PDI = 1.241).The molecular weight information was provided in TableS4.The FT-IR evolution of synthetic M1 (FigureS16) revealed an increase in the absorption of C=O and C-O-C signals at 1740 cm -1 and 980 cm -1 , respectively, indicating the formation of carboxyl esters bonds.

Figure S15
Figure S15 Synthesis of model ester M1.

Figure
Figure S16 FT-IR evolution of the synthesis of M1 for varying reaction time

Figure S17
Figure S17 Synthesis of model ester M2

Upon conducting time-dependent 1 H
-NMR analysis, a pronounced shift in the methylene proton from 3.97 ppm (signal b, triplet) to 3.72 ppm (signal b', quartet) was observed in Figure S19b (right part).Besides, the signals of methyl proton (signal a') appear at 1.23 ppm with triplet peak characteristics.The normalized intensity of these signals gradually increased with reaction time, giving another evidence that the exchange reaction has taken place.

Figure
Figure S19 a) Ester exchange of molecular model 1, and b) normalized time-dependent 1 H-NMR of the model molecule upon 95 o C heating in ethanol.

Figure
Figure S20 a) Ester exchange of molecular model 2, and b) normalized time-dependent 1 H-NMR of the model molecule upon 95 o C heating in ethanol.

Figure
Figure S21 a) Tan δ curves, b) storage modulus (E') recorded at 25 o C; c) Differential scanning calorimetry (DSC) displays similar exothermic peaks for all the pre-polymer systems; d) TGA and e) derivative thermogravimetry (DTG) curves show increase in temperature to maximum decomposition rate (Tmax) and char yield with the incorporation of HBPPB.

Figure
Figure S22 a) Time-sweep curves illustrating the increasing viscosity of the sample at 150 °C (The subsequent decrease is attributed to the complete curing of sample); b) Non-isothermal DSC curves indicating fully curing of EP and EP-9 without any exothermic peaks.
at peak initial; TEX : Temperature at peak value; Tf : Temperature at peak final.b )E': Storage modulus at three-bending mode; Tg: glass transition temperature.c )T5%: The initial decomposition temperature where 5% weight loss reaches; Tmax: The temperature to maximum decomposition rate.

Figure S23
Figure S23 Dynamic thermomechanometry for EP thermoset and EP-9 vitrimer after ethanol treating: a) Storage modulus of vitrimer rapidly decreases from -150 o C to -50 o C, b) Loss modulus indicates a double relaxion behavior of polymer chains.

Figure
Figure S24 a) Stress-relaxation curves for an equivalence of Maxwell relaxation in viscoelastic regime to relative 1/e'.b) original stress-relaxation of EP-12 with respect to thermoset EP.

Figure S25
Figure S25 Arrhenius analysis of a) ln(τ*) versus 1000/T and b) τ* versus 1000/T to determine Ea and Tv.

Figure S27
Figure S27 Photographs of a) the samples soaked in DMAc at ambient condition for 2 days, b) weight retentions as compared to 12HBPPB/EP; Dynamic thermomechanical analysis for c) storage modulus and d) glass transition temperature.

Figure S28 S2. 6 .S3. 6 . 1 .
Figure S28 Reconfigurable performance of the linear system as compared to the HDCNs: a) photographs showing the samples after heating in 95 o C ethanol for 6 h in a closed container; b) storage modulus and c) tan δ curves.

Figure S30
Figure S30 Impact fracture surface showing a brittle feature (rive-like) transformed to tough feature (dimple-like fracture) with increasing content of HBPPB.

Figure S31
Figure S31Typical flexural stress-strain curves of EP and its blends with HBPPB.

Figure S33
Figure S33Typical flexural stress-strain curves of EP and its blends with LPPB.

Figure S35
Figure S35Typical stress-strain curves for the tensile test of CFRPs and its blend with HBPPB.

Figure S36
Figure S36 Typical flexural force-displacement curves for CFRPs and its blend with HBPPB.

Figure
Figure S37 a) Photographs of the EP-12 samples exposed to 0.1 M HCl, 0.1 M NaOH and different organic solvents at the ambient condition for 2 days, b) weight retention of the final sample being soaked and then dried.
DMAc, as evidenced by its heightened and broader O-H/COOH intensity in O1s spectra, as well as the relatively lower signals of C=O and C-O-C in C1s spectra.Additionally, highresolution XPS revealed very weak signals of B and P in the degraded sample, suggesting the depolymerization of HBPPB.

Figure
Figure S40UV-vis transmission spectra of EP and its blends with HBPPB.

Figure S41
Figure S41 Cone calorimeter test records: a) heat release rate; b) total heat release; c) smoke production release; d) total smoke production; e) fire growth rate and fire performance index; f) CO production.SEM morphology of char surface: g) neat EP and h) EP-9.XPS surveys of char residue of EP-9: i) full spectrum and high-resolution spectrum of j) C1s, k) O1s, l) P2p.Notably, the reduction in THR from 124.8 kJ m 2 to 104.1 kJ m 2 was accompanied by an increase in average effective heat combustion (av-EHC, heat release per unit), attributed to the early decomposition of HBPPB, which induces charring effects.Therein, the condensed mechanism plays a critical role, leading to an increase in char yield from 4.36 % up to 12.94 %.

Figure S42
Figure S42 Raman spectra for the char surface of a) EP and b) EP-9 vitrimer.

Table S3
Weight retentions of EP-9 after 95 o C heating for 6 h in varying solvating plasticizers

Table S4
Molecular weight information of the model esters determined by GPC-MALLs

Table S5
DMA parameters for calculating the crosslinking density of HBPPB/EP

Table S6
Characteristic thermal and thermomechanical parameters o C) TEx ( o C) Tf ( o C) E' at 25 o C (GPa)

Table S7
Flexural , tensile and impact strength of EP and its blends with HBPPB

Table S9
Tensile and flexural test of CFRP and its blend with HBPPB