Introduction

Fossil resource crisis motivates world scientists to explore other substitutes. Renewable resources from plants and animals are inexhaustible to some extent due to their possible involvement in ecological system [1,2,3]. In fact, some common renewable resources, such as cellulose and lignin, have been extensively studied and applied in recent decades [4, 5]. Among various bio-mass resources, furfural and its derivatives can be readily produced from agricultural feedstock, such as wheat bran, corncobs, and sawdust [6, 7]. A variety of useful organic chemicals and polymers have been prepared from furfural and its derivatives [8,9,10]. For example, Zhang’s research group reported synthesis of bis(hydroxylmethylfurfuryl) amines (BHMFAs) from 5-hydroxymethylfurfural (5-HMF) through reductive amination using Ru(II) catalysts, thus avoiding use of cost and toxic Na(CN)BH3 reductant [11]. BHMFAs are expected to be a new group of inspiring furan-based monomers that can be used to synthesize biopolymers, such as polyesters and polyurethanes. More recently, renewable polyurethanes were provided by Kieber III and coworkers by utilizing isohexides and 5-hydroxymethylfurfural [12].

Exploration and application of atom-economic and highly efficient reactions are always the goals of chemists [13]. Multi-component reaction (MCR) is a powerful tool to meet the requirements of this object due to its feature of high efficiency, atom-economy, molecular diversity, cheap reactants, etc. [14,15,16]. MCRs were previously applied to synthesize small organic molecules [17,18,19], yet, preparation of polymers through this methodology has also attracted extensive attention in recent years [20,21,22]. Passerini three-component reaction (Passerini 3-CR) and Ugi four-component reaction (Ugi 4-CR) are undoubtedly two of the most investigated MCRs both in small molecular and macromolecular syntheses [23,24,25]. The popular application of these two MCRs lies in that α-acetoxyamides and α-acetamidoamides with multi-functional groups can be formed in one step without using any catalyst. For instance, a pioneering work reported by Zhang and coworkers indicated that polypeptoids can be prepared from natural amino acids by Ugi reaction [26]. Using their method, structurally diverse and functional biocompatible polypeptoids, including γ- and δ-, and poly(ε-peptoid)s, may be obtained under mild conditions (e.g., room temperature, open to air, and catalyst free). Another interesting work is an efficient preparation of redox-responsive poly(ester-amide)s that contain phenylboronic acid esters by Passerini MCR by Li and Du’s research group, meaning that these polymers have potential application as H2O2-responsive delivery vehicles [27].

Smart polymers are special functional materials which can respond to environmental or factitious stimuli, such as light, heating, redox, pH, and magnetic and mechanical forces [28,29,30]. Response of smart materials may be changes of volume or shape, sol–gel transformation, solid–liquid conversion, reversible or irreversible cleavage and coupling, hydrophobic-hydrophilic transitions, etc. In recent years, stimulus-responsive self-healing polymers have drawn much attention for their interesting properties and potential applications in electronic devices, soft materials, biomimetic materials, composite materials, etc.[31,32,33].

Taking all the antecedents into consideration, the aim of present work was to synthesize novel linear polymers from bio-based furfural and other renewable stocks through Passerini three-component reaction, followed by subsequent Diels–Alder crosslinking reaction to prepare thermosets with covalent adaptable networks. The functional polymers showed thermal and/or UV-triggered self-healable properties. Our work features renewable stocks, atom-economic reaction and functional polymers.

Experimental

Materials

1,6-Diaminohexane, diisopropyl amine, phosphorous oxychloride, N,N´-4,4´-diphenylmethyene bismaleimide (BMI), furfural, adipic acid, 3,3´-dithiodipropionic acid, ethyl format, 1,4-dioxane, methanol, ethyl acetate, petroleum ether (60–90), dichloromethane (DCM), tetrahydrofuran (THF), toluene, ethanol, and NaOH were purchased from Shanghai Aladdin Biochemical Technology Co. Ltd, China and were used straightly without any purification. Silica gel for column chromatography was obtained from Qingdao Haiyang Chemical Co. Ltd, China.

Characterization

Gel permeation chromatography (GPC) was conducted on an HP 1100 HPLC (America), equipped with a Waters 2414 refractive index detector and three Styragel HR 2, HR 4, HR 5 of 300 × 7.5 mm columns (packed with 5 mm particles of different pore sizes). The column packing allowed the separation of polymers over a wide molecular weight range of 500–1,000,000. THF was used as the eluent at a flow rate of 1 mL/min at 40 °C. PMMA standards were used as the reference. Structures of monomers and linear polymers were characterized by 1H NMR spectroscopy on a Bruker AV 400 MHz spectrometer (Germany). CDCl3 was used as the solvent. The FTIR spectra were recorded through the KBr pellet method using a Bruker V70 FTIR spectrophotometer (Germany). Glass transition temperature (Tg) of polymers was determined by a DSC-204F1 differential scanning calorimeter produced by German NETZSCH corporation. Temperature rising rate was 10 °C/min under N2 atmosphere. Thermogravimetric analysis (TGA) was measured by NETZSCH TG 209F3, Germany.

Synthesis of 1,6-diisocyanohexane (1)

This compound was synthesized according to the method reported by Gulevich and coworkers [34]. To a 50 mL three-necked flask were added ethyl formate (17.20 mL, 200 mmol) and 1,6-diaminohexane (1.00 g, 8.61 mmol) to form a solution, which was then refluxed overnight. After cooling to room temperature, the mixture was evacuated at room temperature to remove liquor under reduced pressure, followed by adding DCM (17.5 mL) and diisopropyl amine (8.71 g, 86.1 mmol). Phosphorous oxychloride (3.69 g, 24.1 mmol) was added dropwise to the reaction mixture and cooled by an ice bath. After adding phosphorous oxychloride, the reaction system was stirred for 1 h in the ice bath. Then the ice bath was removed and stirring was continued for 6 h. The mixture was poured into 50 mL of K2CO3 (10.0 g) aqueous solution. The organic layer was separated, and the aqueous phase was extracted with DCM for three times, and the combined organic phase was dried with anhydrous K2CO3 followed by removing DCM to obtain brown oil as a crude product. The oil was purified on a silica gel column with petroleum ether/ethyl acetate (2:1 v:v) as eluent to achieve light yellow oil (0.89 g, 76%). 1H NMR (400 MHz, CDCl3, δ ppm): 3.43–3.37 (m, 4H), 1.75–1.65 (m, 4H), 1.55–1.45 (m, 4H).

Synthesis of linear polymer LP1 through Passerini reaction

The polymer LP1 was prepared by modification of the method from Li’s research group [35]. Furfural (192 mg, 2 mmol), 1,6-diisocyanohexane (1) (136 mg, 1 mmol), adipic acid (146 mg, 1 mmol), and DCM (5 mL) were added to a 50 mL flask. The mixture was stirred at 30 °C for 36 h under N2 protection. Viscous liquid was obtained after distilling off the solvent DCM. The crude viscous liquid was again dissolved in 1 mL of DCM, and 10 mL petroleum ether was added to precipitate brown oligomer LP1 (219 mg, 46.2%). Mn 2189 g/mol (measured by GPC). 1H NMR (400 MHz,CDCl3, δ ppm): 7.42–7.39 (s, 2H), 6.52–6.41 (s, 2H), 6.40–6.35 (s, 2H), 6.23–6.17 (s, 2H), 3.23–3.35 (m, 5H), 2.97–2.87 (m, 5H), 2.52–2.40 (m, 4H), 1.37–1.30 (m, 5H). IR (KBr) ν, 3340, 3081, 2931, 2852, 1754, 1530, 1126 cm−1.

Synthesis of crosslinked polymer CP1 through D–A reaction

The polymer CP1 was prepared according to the method of our previous work [36]. The linear oligomer LP1 (474 mg), N,N´-4,4´-diphenylmethyene bismaleimide (BMI) (360 mg) and 1,4-dioxane (5 mL) were sequentially added to a 25 mL flask, and the mixture was stirred at 78 °C for 10 h under N2 atmosphere. Then part of the mixture was coated on a glass substrate to form a film. The left part was continued stirring at 78 °C for another 38 h to obtain brown gel-like polymer CP1 (530 mg, c.a. 64%). The glass transition temperature (Tg) of CP1 was 14.78 °C. IR (KBr) ν: 2931, 2845, 1773, 1692, 1407 cm−1.

Synthesis of linear polymer LP2 through Passerini reaction

The polymer LP2 was synthesized by the method similar to that of LP1 [35]. Furfural (192 mg, 2 mmol), 1,6-diisocyanohexane (1) (136 mg, 1 mmol), 3,3'-dithiodipropionic acid (210 mg, 1 mmol), and DCM (5 mL) were added to a 50 mL flask. The mixture was stirred at 30 °C for 36 h under nitrogen protection in a dark place. Viscous liquid was obtained after distilling off the solvent DCM. The crude viscous liquid was again dissolved in 1 mL of DCM, and 10 mL petroleum ether was added to precipitate brown oligomer LP2 (299 mg, 55.6%). Mn 5464 da (measured by GPC). IR (KBr) ν, 3315, 3087, 2923, 2845, 1747, 1535, 1426, cm−1. 1H NMR (400 MHZ, DMSO, δ ppm): 7.60 (s, 2H), 6.47 (m, 4H), 5.92 (s, 2H), 3.06 (s, 4H), 2.86 (m, 8H), 1.34 (m, 4H), 1.17 (m, 6H).

Synthesis of crosslinked polymer CP2 from the linear LP2 and N,N’-4,4’-diphenylmethyene bismaleimide (BMI) through Diels–Alder [4 + 2] cycloaddition reaction

The polymer CP2 was prepared by the method similar to that of CP1 [36]. The linear oligomer LP2 (528 mg), N,N´-4,4´-diphenylmethyene bismaleimide (BMI) (360 mg) and 1,4-dioxane (5 mL) were sequentially added to a three-necked 25 mL flask, and the mixture was stirred at 80 °C for 16 h under N2 atmosphere in a dark place. Then, part of the mixture was coated on a glass substrate to form a film. The left part was continued stirring for another 32 h to obtain brown gel-like polymer CP2 (471 mg, c.a. 53%). The glass transition temperature (Tg) of CP2 was 7.00 °C. IR (KBr) ν: 3307, 2931, 2845, 1774, 1718, 1401 cm−1.

Determination of medium resistance of the crosslinked polymers

The film mass of the crosslinked polymer was measured on a balance, and the film was dipped in different media, 1 mol/L aqueous HCl, 1 mol/L aqueous NaOH, THF, toluene, EtOH, and deionized water for 5 days. The film sample was dried in an oven and its mass was measured again. Medium resistance of the crosslinked polymer can be evaluated according to mass loss of the film. The lower mass loss of the film, stronger is the medium resistance of the crosslinked polymer.

Determination of gel fraction of the crosslinked polymers

Gel fraction was measured by modifying the method reported in literature [37, 38]. About 350 mg of a sample CP1 or CP2 was weighed accurately, and the mass was noted as W0. Then the sample was packed by a steel wire mesh (120 mesh) and extracted with xylene in Soxhlet for 10 h. The sample was taken out and washed with ethanol, followed by drying at 140 °C for 4 h. The sample was weighed accurately again, and the mass was noted as W. Gel fraction was calculated as G (%) = (W/W0) × 100.

Determination of self-healing performance of the crosslinked polymers

Method 1: a film of crosslinked polymer was cut to form a crack by a razor. The crack of the film was photographed by a Japanese JSM-6701F cold field emission scanning electron microscope. Then the injured film was heated at 50 °C in an oven for 3 h, and photographed again to observe self-healing state. Thus, the film was at 50 °C in an oven for another 5 h, and photographed again to observe self-healing state. The self-healed film was cut again in the same place and placed in an oven at 50 °C for 12 h, and photographed again to observe re-self-healing performance. Method 2: stress–strain curves of original film and self-healed film were compared.

Results and discussion

Synthesis and characterization of polymers

Synthesis of 1,6-diisocyanohexane (1)

There are usually two steps for the synthesis of isocyanides: the first step is preparation of N-substituted formamides from ethyl formate and free amines; the second is synthesis of isocyanides by appropreiate acidic dehydrating agens, such as phosphorous oxychloride and para-tosyl chloride [39]. Thus, 1,6-diaminohexane was converted to the corresponding diformamides, followed by adding phosphorous oxychloride to obtain 1,6-diisocyanohexane (1) in 76% yield. The 1H NMR spectrum of 1 is shown in Fig. 1. The peak at chemical shift of 3.43–3.37 ppm (marked as “a”) is signal of protons of two methylene groups adjacent to isocyanate groups. The other two peaks at 1.75–1.65 ppm and 1.55–1.45 ppm represent eight protons signals of the rest four methylene groups (marked as “b” and “c”, respectively).

Fig. 1
figure 1

1H NMR spectrum of 1,6-diisocyanohexane (1)

Full size image

Synthesis and characterization of the linear polymer LP1

Passerini three-component reaction (Passerini 3-CR) can be conducted easily by mixing an appropriate amine, carboxylic acid and isocyanide without any other catalysts or additives, which is unambiguously an atom-economic reaction. In order to utilize renewable materials from biomass or renewable platform chemicals, we deliberately chose furfural, adipic acid, and 1,6-diisocyanohexane (1) as reactants to synthesize LP1 (Scheme 1). Furfural can be produced directly from biomass such as corn cob or rice husk [40, 41]. Adipic acid may be potentially prepared from renewable platform chemicals such as glucaric acid or mucic acid [42,43,44]. 1,6-Diisocyanohexane (1) has been synthesized from 1,6-diaminohexane, while 1,6-diaminohexane may be prepared from adipic acid. Thus, after furfural, adipic acid, and 1,6-diisocyanohexane (1) were mixed and stirred for hours, the resulting linear polymer LP1 could be collected through precipitation and filtration by adding poor solvent petroleum ether. The 1H NMR spectrum of LP1 is illustrated in Fig. 2. The peaks (marked as “c, e and d”) at 7.41 ppm, 6.37 ppm, 6.19 ppm are signal of protons on aryl ring of furans, and the signal at 6.50 ppm (designated as “f”) can be attributed to protons of methylene group of furfural. Ratio of the four kinds of protons is almost 1:1:1:1, indicating that Passerini reaction product was formed. The peak at 3.30 ppm (marked as “h”) may be the protons of methylene groups connected with nitrogen atoms, while that at 2.93 ppm (marked as “a”) may be contributed by protons of methylene groups close to ester groups. Number-average molecular weight of LP1 (MnLp1) is 2189 da. The molecular weight of one repeating unit of LP1 is 474.5, so degree of polymerization (DP) is about 4.6, i.e., n = 4.6.

Scheme 1
scheme 1

Synthesis of linear polymer LP1 through Passerini reaction

Full size image
Fig. 2
figure 2

1H NMR spectrum of linear polymer LP1

Full size image

The FTIR spectrum of the linear polymer LP1 is shown in Fig. 3A-a. The peak at 3340 cm−1 is stretching vibration of amide N–H, and the strong peak at 1754 cm−1 means stretching vibration of carbonyl groups. At 3081 cm−1, the weak peak may be C–H stretching vibration of furan rings. The peaks at 2931 and 2852 cm−1 represent C–H stretching vibration of methylene groups.

Fig. 3
figure 3

A FTIR spectra of a LP1 linear polymer, and b CP1 crosslinked polymer; B FTIR spectra of: a LP2 linear polymer, and b CP2 crosslinked polymer

Full size image

Synthesis and characterization of the crosslinked polymer CP1

There are side furan rings in the linear pre-polymer LP1 as shown in Schemes 1 and 2. These furan rings in LP1 as electron-efficient diene can easily proceed Diels–Alder [4 + 2] cycloaddition reaction with electron-deficient dienophile such as BMI, which produces the crosslinked polymer CP1 as illustrated in Scheme 2. When the molar ratio of BMI and repeating unit of LP1 is being controlled to be 1:1, gelation phenomenon would occur after the reaction had been conducted for 48 h, causing inconvenience for manipulation of samples. Therefore, when the reaction proceeded for 10 h, the mixture was coated on a glass substrate to form a film for measuring other properties. The images of gel sample and film sample of CP1 are demonstrated in Fig. 4.

Scheme 2
scheme 2

Synthesis of crosslinked polymer CP1 from LP1 and BMI through Diels–Alder reaction

Full size image
Fig. 4
figure 4

Two forms of the crosslinked polymer CP1: (a) gel state, and (b) a film on glass substrate

Full size image

The structure of CP1 is further confirmed by FTIR spectrum (Fig. 3A-b). A very weak peak appears at 1775 cm−1, indicating the formation of new carbon–carbon double bonds in the previous furan rings through Diels–Alder reaction [36]. The peaks at the wavenumber of 2931 and 2845 cm−1 are C–H stretching vibration of methylene groups. The strong peak at 1741 cm−1 corresponds to the stretching vibration of carbonyl groups. Several peaks from 1548 to 1505 cm−1 stems from benzene skeleton vibration. At 1407 cm−1, there appears the signal of amide C–N stretching vibration.

Synthesis and characterization of linear polymer LP2 and crosslinked polymer CP2

Healing ability and reliability of materials may be promoted by introducing two or more kinds of dynamic bonds [45]. Dynamic disulfide or multi-sulfide bonds are usually applied in designing functional materials including self-healing or reprocessable polymers [46,47,48]. These inspired us to utilize 3,3´-dithiodipropionic acid, a diacid with a disulfide group, to replace adipic acid in Passerini reaction for preparing the linear polymer LP2 (Scheme 3). The reaction was carried out in a similar way as preparation of LP1. The 1H NMR spectroscopy (DMSO-D6 as the solvent) of LP2 is illustrated in Fig. 5. Two peaks at the chemical shift of 7.60 and 6.47 ppm are signals of six protons on two furan rings. The signals of protons of methyne that is connected with a furan ring, an oxygen and a carbonyl group appear at 5.92 ppm. The peak at 3.06 ppm is attributed to methylene groups connected to nitrogen atom. The four methylene groups from 3,3'-dithiodipropionic acid ester give a signal at 2.84 ppm. Protons of all the other methylene groups produced peaks at 1.34 and 1.17 ppm, respectively.

Scheme 3
scheme 3

Synthesis of linear polymer LP2 through Passerini reaction

Full size image
Fig. 5
figure 5

1H NMR spectrum of the linear polymer LP2

Full size image

The FTIR spectrum of LP2 is demonstrated in Fig. 3B-a. The broad peak at 3315 cm−1 is attributed to stretching vibration of amide N–H bond. The peak at 3087 cm−1 results from the C–H stretching vibration of aromatic furan rings. Signals at 2923 and 2845 cm−1 correspond to the C–H stretching vibration of methylene groups in the main chain. The strong peak at 1747 cm−1 is a signal of carbonyl groups. The C–N stretching vibration of amide groups gives a peak at 1426 cm−1.

The crosslinked polymer CP2 was prepared through Diels–Alder reaction between LP2 and BMI in a similar way as that of CP1. The FTIR spectrum of CP2 is shown in Fig. 3B-b, and the characteristic peak at 1774 cm−1 confirms successful Diels–Alder crosslinking reaction.

Properties of the crosslinked polymers CP1 and CP2

Chemical resistance of CP1 and CP2

Chemical corrosion is sometimes unavoidable during application of polymer materials (e.g., rubber gasket in chemical reaction kettle). Therefore, measurement of chemical resistance is necessary. Herein, we only evaluate the chemical resistance property from relative mass loss after dipping samples into chemical media for a period of time. Related relative mass loss data of CP1 and CP2 are shown in Table 1. We selected five typical chemical media including strong acids, strong bases, and good solvents. The polymer CP1 has relatively good resistance to acid, ethyl acetate and toluene, but relatively poor resistance to base and THF. The polymer CP2 has relatively good resistance to acid and ethyl acetate. By comparison, chemical resistance performance of CP1 is better than that of CP2, which may be due to the existence of relatively weak disulfide bonds in CP2.

Table 1 Chemical resistance of the crosslinked polymers CP1 and CP2
Full size table

Gel fraction of CP1 and CP2

Gel fraction may reflect crosslinking state of thermoset polymers. Gel fraction of CP1 was determined to be 82%, while that of CP2 was 74%. This means that most of linear segments of LP1 and/or LP2 were crosslinked. We think that the measured values may be lower than the real values as it was in favor of reverse reaction (i.e., depolymerization reaction) under heating conditions during measurement. The value of CP1 was higher than that of CP2, which may be explained that CP2 has undergone both reverse D–A reaction and reverse disulfide-bond-cleavage/reshuffling, leading to lower gel fraction values.

Thermal properties of CP1 and CP2

Glass transition temperatures of CP1 and CP2 are TgCP1 = 14.78 ℃ and TgCP2 = 7.00 ℃, respectively, indicating that both polymers are at rubbery state. Heat resistance of the polymers was determined by thermogravimetric analysis (TGA)-derivative thermogravimetry (DTG) (Fig. 6a, b). TGA diagram of the polymer CP1 is demonstrated in Fig. 6a. Td10 value is about 151 ℃, meaning that small molecular impurities such as water are volatile at this temperature. The first decomposition peak value appears at about 210 ℃, indicating that monomer or oligomer residues are decomposed or rapidly volatile at this temperature. Td50 value is about 420 ℃. The TGA diagram of CP2 is shown in Fig. 6b. Its Td10 value is about 131 ℃, and the first decomposition peak occurs at about 150 ℃. Td50 value is about 397 ℃. By comparing CP1 and CP2, heat resistance of CP1 is higher than that of CP2, presumably because the polymer CP2 contains relatively weak disulfide bonds.

Fig. 6
figure 6

TGA-DTG diagrams of: a CP1 polymer and b CP2 polymer

Full size image

Self-healing behavior of CP1 and CP2

Diels–Alder reaction is reversible and can be used to prepare versatile functional polymer materials [49, 50]. Among them, thermally triggered self-healable polymers were also prepared through this strategy [36, 51]. The thermosets CP1 as well as CP2 were also crosslinked by Diels–Alder reaction, and expected to have healable behavior. Thus, firstly, we tested healing property of CP1 under appropriate thermal conditions. Film samples of CP1 were cut by a razor to form scratches, and the “injured” samples were then heated at temperatures of 40, 50, 60, 70, 80, 90, 100 °C for different times. To our joy, the samples showed self-healing property at the temperature of 60 °C or higher. Compared to high temperatures, longer time was required for healing process at the temperature as low as 60 °C. This may be explained as follows: (i) main chain of CP1 is composed of flexible alkyl monomers, which makes the wounded part “flow or move” more easily at temperatures close to or higher than Tg temperature (14.78 ℃); (ii) retro-Diels–Alder reactions were usually endothermic, and it is slow to build dynamic balance between Diels–Alder and retro-Diels–Alder reactions at relatively low temperatures such as 60 °C [36]. Yet, we chose the temperature of 60 °C to explore healing behavior of CP1 in order to save energy. First, a sample of CP1 was cut for the first time by a razor to form a scratch (Image 1, Fig. 7). Then, the scratched sample was heated in an oven at 60 °C for 3 h, and the scratch was partially healed as shown in Image 2 of Fig. 7. Then, the partially healed sample was continued to be heated at 60 °C for another 5 h, and the sample was almost completely self-repaired (Image 3, Fig. 7). The healed sample was cut again at the same site as that of the first time (Image 4, Fig. 7), and then the second cut sample was placed in an oven at 60 °C for 24 h. The sample was self-healed again (Image 5, Fig. 7), meaning that the intrinsic self-healing polymer CP1 could self-heal more than once.

Fig. 7
figure 7

SEM images of self-healing process of the thermoset CP1. Image 1: A film of CP1 sample was cut by a razor to form a scratch; Image 2: the scratched sample (Image 1) was partially healed after being heated at 60 °C for 3 h; Image 3: the partially healed sample (Image 2) was heated at 60 °C for another 5 h and was almost “completely” self-healed; Image 4: the “completely” healed sample (Image 3) was cut by a razor again at the same position of the sample as that of Image 1; Image 5: the cut sample (Image 4) was healed again after being heated at 60 °C for 24 h

Full size image

Next, we evaluated the healing percentage by comparing stress–strain properties of an initial uncut sample and the second-time healing CP1 sample. As shown in Fig. 8a, the largest stress of the initial uncut sample is 0.37 MPa while the largest strain can reach more than 100%, indicating that CP1 has good elastomeric property. Young's modulus of CP1 is 0.35 MPa. For the second-time healing sample, the largest stress is about 0.20 MPa (Fig. 8b). Thus, the healing percentage may be calculated as follows: 0.20/0.37 = 54.1%. Young's modulus of the healed CP1 is 0.23 MPa.

Fig. 8
figure 8

Curves of stress–strain: a the initial uncut CP1 sample, and b the second-time healing CP1 sample

Full size image

There are two kinds of dynamic covalent bonds in the polymer CP2, which is expected to own thermo- and light-dual stimuli response behavior. In a similar manner as that of CP1, the CP2 sample was cut and heated at 60 °C for 24 h, and after that the cut mark disappeared completely (Fig. 9). Then, we explored healing behavior of CP2 under UV light irradiation conditions. A sample of CP2 was wounded by a razor (Fig. 10a), followed by irradiation under UV light with 360 nm wavelength. To our surprise, the scratched sample was partially healed within only 3 min (Fig. 10b), and the partially healed sample was completely healed after being irradiated by UV for another 7 min. That is to say, the scratched sample can be completely healed within 10 min (Fig. 10c). The rapid healing behavior under UV light may be contributed to both dynamic disulfide bonds and dynamic D–A bonds, as the temperature of the CP2 sample also rose under UV light. The healed sample was cut again and self-healed for the second time after being irradiated under UV for 30 min (Fig. 10d, e).

Fig. 9
figure 9

SEM images of self-healing process of the thermoset CP2 under heating conditions: (1) A film of CP2 sample was cut to form a scratch, and (2) the scratched sample was “completely” restored after being heated at 60 °C for 24 h

Full size image
Fig. 10
figure 10

SEM images of self-healing process of the thermoset CP2 under UV light (wavenumber 360 nm) irradiation conditions: a a film of CP2 sample was cut to form a scratch; b the scratched sample was partially healed after being irradiated by UV light for 3 min; c the partially healed sample was completely healed after being further irradiated for another 7 min; d the healed sample was cut again at the same place as that of the first time; e the secondly cut sample was healed for the second time after being irradiated under UV for 30 min

Full size image

Mechanical properties of CP2 were also studied by testing stress–strain correlation. As shown in Fig. 11, (a) is the stress–strain curve of an initial CP2 sample, and the maximal stress is 0.33 MPa while the maximal strain can reach 125%. Thus, Young's modulus of CP2 is calculated to be 0.26 MPa; (b) is the stress–strain curve of the UV-triggering self-healed CP2 sample, which provides 0.21 MPa of maximal stress and 60% of maximal strain (Young's modulus of the healed CP2 is 0.35 MPa). Thus, the healing percent is: 0.21/0.33 = 63.6%. In contrast, the healing percent of CP2 (63.6%) is higher than that of CP1 (54.1%), which may be ascribed to dual dynamic bonds of CP2 that causes more efficient self-healing process.

Fig. 11
figure 11

Curves of stress–strain: a the initial uncut CP2 sample, and b the second-time healing CP2 sample

Full size image

Conclusion

Two linear polymers LP1 and LP2 were synthesized by highly efficient Passerini three-component reactions from materials based on platform chemicals. Among them, LP1 was even prepared from fully bio-based platform chemicals, i.e., furfural, adipic acid and 1,6-hexadiamine. LP2 was produced from furfural, 3,3´-dithiodipropionic acid and 1,6-hexadiamine. The structure of LP1 and LP2 was confirmed by 1H NMR as well as FTIR spectra. The number-average molecular weights of LP1 and LP2 are 2189 da and 5464 da, respectively. There are pedant furan side groups in both LP1 and LP2, which make them successfully undergo Diels–Alder crosslinking reaction with BMI. Two dynamically crosslinked networks CP1 and CP2 were prepared by this crosslinking reaction. The structures of CP1 and CP2 were verified by appearance of a weak peak at around 1775 cm−1 in FTIR spectra. Gel fraction of CP1 was determined to be 82%, while that of CP2 was 74%, which indicated that most of linear segments of LP1 and/or LP2 were crosslinked. Both of CP1 and CP2 showed good resistance to acid, and ethyl acetate, but relatively poor resistance to base and THF. The Tg values of CP1 and CP2 are 14.78 ℃ and 7.00 ℃, respectively. From TGA-DTG analysis data, Td10 of CP1 is 151 °C, while that of CP2 is 131 °C, indicating that heat resistance of CP1 is higher than that of CP2, presumably because the polymer CP2 contains relatively weak disulfide bonds. At 60 °C or higher temperatures, both of CP1 and CP2 demonstrated thermo-stimulus self-healing behavior, though the healing process required hours at 60 °C. Furthermore, CP2 showed rapid self-healable property under 360 nm UV irradiation within 10 min, because there are dynamic disulfide bonds in CP2. By determining and comparing maximal stress, the healing percent of CP2 (63.6%) is higher than that of CP1 (54.1%), which may be ascribed to dual dynamic bonds of CP2 that causes more efficient self-mending process. Our work provides a simple method for designing and preparing self-healable polymers from renewable platform chemicals.