It also retained 81.3% of its original tensile strength after three cycles of cutting, hot pressing and remolding. The experimental results support a molecular design in which dynamic boronic ester and β-hydroxy ester bonds provide different, complementary routes for changing the network topology. Conventional epoxy thermosets form permanent crosslinked networks, which give them strength, stiffness, chemical resistance and dimensional stability but also prevent conventional remolding and chemical recovery.

Epoxy vitrimers retain a crosslinked structure while incorporating dynamic covalent bonds that can exchange under selected conditions, enabling self-healing, welding and reprocessing. The researchers combined boronic ester exchange, which the source describes as catalyst-free and capable of occurring at moderate temperatures, with β-hydroxy ester exchange. Their stated aim was to ease the catalyst and high-temperature requirements often associated with ester-only vitrimers while retaining mechanical strength and thermal stability.

The resulting dual-bond network was designed to rearrange without being dissolved. The researchers prepared diglycidyl furan-2,5-dicarboxylate, abbreviated DGF, from 2,5-furandicarboxylic acid through a two-step synthesis involving formation of an allyl ester intermediate and subsequent epoxidation. They also synthesized di-BONH2, an aminophenylboronic acid-derived crosslinker containing dynamic boronic ester bonds, by reacting 3-aminobenzeneboronic acid with di-trimethylolpropane.

For network preparation, DGF and predetermined quantities of di-BONH2 were dissolved in dimethylformamide, poured into molds, dried at 80 °C for 12 hours and cured successively at 100 °C for six hours and 120 °C for six hours. Four materials, DGFB-0.7 through DGFB-1.0, differed in the molar ratio of active hydrogen atoms in the crosslinker to epoxy groups in DGF. The authors calculated that their bio-based carbon contents ranged from 25.0% in DGFB-1.0 to 29.4% in DGFB-0.7.

Fourier-transform infrared spectroscopy and nuclear magnetic resonance spectroscopy supported the assigned structures of the synthesized compounds and cured networks. After curing, the characteristic epoxy band at 850 cm−1 and primary-amine bands at 3,460 and 3,367 cm−1 diminished, while new bands assigned to tertiary amine and hydroxyl groups appeared at 1,225 and 3,416 cm−1. This spectral pattern was consistent with epoxy–amine ring opening and formation of tertiary amines capable of catalyzing subsequent transesterification within the network.

Swelling tests provided separate support for crosslinking. All formulations remained intact and insoluble after seven days in dimethyl sulfoxide, dimethylformamide, methanol, acetone, ethyl acetate and tetrahydrofuran at room temperature. DGFB-1.0 showed no observable swelling or discoloration, whereas the least crosslinked DGFB-0.7 sample showed slight swelling.

Calculated crosslink densities increased from 1.03 × 10−2 mol cm−3 to 1.10 × 10−2 mol cm−3 across the series. Mechanical testing quantified the effect of increasing crosslinker content. Tensile strength rose monotonically from 48.5 ± 1.4 MPa for DGFB-0.7 to 75.4 ± 2.3 MPa for DGFB-1.0, while Young’s modulus increased from 137.3 ± 5.6 MPa to 512.5 ± 5.2 MPa.

Elongation at break declined from 45.6 ± 1.9% to 26.2 ± 1.1% as the networks became denser and contained more aromatic material. All samples displayed brittle fracture. The researchers associated the strength and stiffness increases with greater crosslink density and aromatic-ring content, while the lower elongation reflected restricted chain mobility.

The DGFB-1.0 tensile strength exceeded the 10–65 MPa range the authors cited for most previously reported bio-based epoxy vitrimers and the 10–56 MPa range they cited for furan-based vitrimers relying only on β-hydroxy ester exchange. These comparisons concern previously published laboratory systems rather than industrial-scale demonstrations. Thermogravimetric analysis showed that the networks retained substantial thermal stability.

The temperature at which 5% of the mass decomposed increased from 273.3 ± 2.4 °C for DGFB-0.7 to 286.7 ± 4.3 °C for DGFB-1.0. The calculated heat-resistant index, Ts, rose from 156.4 ± 2.2 °C to 163.3 ± 2.1 °C, and the carbonaceous residue at 800 °C increased from 23.9 ± 0.47% to 26.2 ± 0.63%. Differential scanning calorimetry placed the glass-transition temperatures between 62.8 and 72.7 °C, with DGFB-1.0 at the upper end.

The absence of residual-curing exotherms was consistent with essentially complete consumption of reactive groups. In dynamic mechanical analysis at 120 °C, DGFB-1.0 had a stress-relaxation time of 11 seconds, compared with 23, 33 and 60 seconds for DGFB-0.9, DGFB-0.8 and DGFB-0.7. At 90 °C, the DGFB-1.0 relaxation time increased to 29 seconds.

Arrhenius fitting yielded an activation energy of 28.3 ± 0.7 kJ mol−1. Self-healing was evaluated by cutting grooves approximately 60–70 micrometres wide into specimen surfaces and heating the damaged samples at 90 °C. After 60 minutes, scratch widths in DGFB-0.7, DGFB-0.8 and DGFB-0.9 decreased from initial values of 62.3, 64.1 and 65.4 micrometres to 11.8, 6.9 and 4.1 micrometres, respectively.

Polarized optical images showed that the scratch in DGFB-1.0 was almost eliminated, leaving faint residual traces. Quantitative healing efficiency, defined from recovered tensile strength relative to the uncut material, ranged from 85.4% to 98.2%, with the maximum measured for DGFB-1.0. Two rectangular pieces of that formulation were also joined with a 10 mm overlap by heating them at 90 °C for two hours.

The welded specimens achieved a tensile strength of 71.4 MPa and a welding efficiency of approximately 94.7%. They also supported a static load equivalent to 10 kilograms for 60 minutes without failure in the reported test. Physical reprocessing was demonstrated by cutting DGFB-1.0 into small pieces and hot pressing them at 90 °C under 5 MPa for 40 minutes.

The resulting specimens had smooth surfaces after successive processing generations. Tensile-strength retention was 93.6% after the first cycle and declined to 81.3% after the third. Infrared spectra retained the principal signals associated with hydroxyl, carbonyl, boronic ester and tertiary-amine groups, indicating that key network features remained detectable.

The study did not provide a detailed analysis of how repeated processing affected glass-transition temperature, elongation at break, crosslink density or the complete evolution of chemical structure, so the tensile and infrared results should not be interpreted as proof that every property remained unchanged. For chemical recycling, post-tensile-test DGFB-1.0 waste was cut into fragments and soaked in the E.G. solvent specified by the study at 160 °C for 40 minutes. The network dissolved, and the researchers reported a mass recovery of approximately 96%.

Fourier-transform infrared spectra showed enhanced hydroxyl absorption, reduced B–O absorption and increased B–OH absorption, changes consistent with alcoholysis products from boronic ester and β-hydroxy ester reactions. The degradation products were recured with the original schedule to produce defect-free regenerated specimens whose tensile-strength recovery exceeded 90.3%. The same formulation was used for carbon-fiber-reinforced composites with three layers of woven fabric and a 50% carbon-fiber mass fraction.

After exposure of a composite sample to the E.G. solvent at 160 °C for 40 minutes appeared to substantially decompose the matrix, recovered fibers were used to prepare a regenerated composite. Its tensile strength was 753.8 MPa, compared with 799.3 MPa for the pristine composite, while modulus decreased from 14.4 to 11.2 GPa and elongation increased from 8.91% to 10.1%. Scanning electron micrographs showed clean, smooth recovered fibers, and Raman spectra suggested that their chemical structure was essentially unchanged.

Together, the measurements show that the dual-bond formulation can combine rigidity with several forms of reversible processing, although the evidence remains based on laboratory specimens rather than long-term service testing. The study included chemical characterization, swelling tests, tensile measurements, thermal analysis, stress relaxation, self-healing, welding, repeated hot pressing, chemical recycling and carbon-fiber composite recovery, with at least three independent measurements for the reported mechanical and thermal values. The findings therefore establish a promising molecular design for recyclable epoxy vitrimers, not a validated industrial recycling process or a ready commercial material.

Subject of Research: Laboratory development of recyclable bio-based epoxy vitrimers using dual dynamic covalent bond networks.

Article Title: Robust bio-based epoxy vitrimers with dual dynamic boronic ester and β-hydroxy ester bonds: self-healing, weldability and recyclability

Article References: Yi, Q., Shen, R., Mo, W., & Zeng, Y. (2026). Robust bio-based epoxy vitrimers with dual dynamic boronic ester and β-hydroxy ester bonds: self-healing, weldability and recyclability. Frontiers in Materials, 13, Article 1945972. https://doi.org/10.3389/fmats.2026.1945972

Image Credits: AI Generated

DOI: 10.3389/fmats.2026.1945972

Keywords: epoxy vitrimers, dynamic covalent bonds, boronic esters, beta-hydroxy esters, self-healing polymers, reprocessable materials, chemical recycling, carbon-fiber composites, furan-based polymers, Robust, bio-based, epoxy