Chinese lacquer was purchased from Xi’an Institute of Lacquer, China. Urushiol was extracted from Chinese lacquer using ethanol according to the literature. Ferric chloride, copper chloride and aluminum chloride, n-octylamine, formaldehyde (37 wt% in H₂O), 1,4-dioxane, chloroform, dichloromethane, xylene, methanol, and anhydrous sodium sulfate were obtained from Sinopharm Chemical Reagent Co. All chemicals were used as received without further purification.

¹H Proton nuclear magnetic resonance (¹H-NMR) spectra were recorded on a Bruker AV400 NMR spectrometer using CDCl₃ as the solvent and tetramethylsilane (TMS) as the internal standard at a proton frequency of 400 MHz. Fourier transforms infrared (FTIR) spectra were recorded on a United States. Nicolet Magna 5700 spectrometer at room temperature (20°C). The spectra were collected at 32 scans with a spectral resolution of 4 cm⁻¹. Reflex spectra were obtained from the method of ATR. Differential scanning calorimetry (DSC) was performed on a METTLER DSC3 instrument under a nitrogen atmosphere at a heating rate of 10°C min⁻¹ in the range of 30–250°C. Thermogravimetric analysis (TGA) was performed on a METTLER TGA/SD-TA851 instrument under a nitrogen atmosphere at a heating rate of 10°C min⁻¹ in the range of 30–600°C. X-ray photoelectron spectroscopy (XPS) recorded on a VG MultiLab 2000 spectrometer to Mg K_α (1,253.6 eV) as X-ray radiation and by being able to 20 eV, using the surface contamination carbon C_1s binding energy (284.8 eV) as the internal standard calibration of other elements binding energy. The peak area was obtained by integrating the spectral characteristics of the elements and the sensitivity factor to calculate the composition of each surface element in the catalyst.

To a 100 ml three-necked round bottom flask equipped with a thermometer, a reflux condenser, and a dropping funnel, 10 ml of 1,4-dioxane, formaldehyde (37 wt%) (0.05 mol, 4.05 g), n-octylamine (0.025 mol, 3.23 g), and 10 ml of dioxane were added and stirred at below 4°C for 40 min. After adding a solution of urushiol (0.025 mol, 7.85–8.00 g) in 10 ml 1,4-dioxane dropwise within 20 min, the solution was gradually heated to 90°C, and the brown mixture was refluxed for 5 h. The solvent was removed by distillation under reduced pressure, and the residual was dissolved in 100 ml of dichloromethane. The solution was washed many times with distilled water and dried with anhydrous magnesium sulfate for 12 h. The solvent was removed by rotary evaporation, and the brown product was dried under vacuum at room temperature for 24 h. Further purification was conducted by column chromatography and a light red liquid was obtained (yield 96%).

A certain amount of urushiol-based benzoxazine monomer (BZ, 0.01 mol) and anhydrous xylene (10 ml) were added to a 100 ml three-necked round bottom flask equipped with a thermometer and a reflux condenser. The solution was stirred under nitrogen for 20 min. Metal chloride (MCl₃ = FeCl₃, AlCl₃ and CuCl₂) with different molar ratio to BZ (see Table 1) in anhydrous methanol (5 ml) was added dropwise. After vigorous stirring at 80°C for 3 h, PBZ-M (PBZ-Fe, PBZ-Al, and PBZ-Cu) were obtained. PBZ-M (2 ml) was cast onto the clean glass slide or the iron plate and then cured at ambient temperature for 3 h. PBZ cured without metal ions at 120°C (2 h), 140°C (2 h), 160°C (2 h), and 180°C (2 h) was also prepared for comparison. PBA and PBZ-M membranes with a thickness of ∼60 μm were obtained.

To calculate the gel fraction of the polymers, the polymers were immersed in toluene for 2 h and then dried in the oven. The gel fraction was calculated as:

Gel fraction = m_t/m₀×100%,

where m₀ is the initial mass of the polymeric film and m_t is the mass after immersion.

The synthesis of urushiol-based benzoxazine precursor is shown in Figure 1A. The novel benzoxazine monomer was prepared via the Mannich reaction of urushiol, n-octylamine, and formaldehyde (37 wt% in H₂O). The reaction time was determined by thin-layer chromatography (TLC). The benzoxazine monomer was reacted with the Lewis acid (FeCl₃, AlCl₃, and CuCl₂) to obtain PBZ-Fe, PBZ-Al, and PBZ-Cu, respectively. The benzoxazine monomer was also thermally cured (PBZ) for comparison. It is worth noting that the polybenzoxazine/MCl₃ can be cured at ambient temperature. For better performance, the molar ratio of the reactants was optimized. The results are shown in Table 1. The amount of MCl₃ has a direct impact on the performance of the film. When the amount of MCl₃ is large, the product is precipitate which could not be coated. When the amount of MCl₃ is small, the product is viscous liquid while the surface drying time is too long. The suitable molar ratio for film coating is preferable from Table 1 when the ratio of BZ and MCl₃ including n (BZ): n (FeCl₃) = 6:1, n (BZ): n (AlCl₃) = 3:1, n (BZ): n (AlCl₃) = 6:1, n (BZ): n (CuCl₂) = 4:1. The molar ratio of BZ and MCl₃ was kept at 6:1 in the following experiments.

The structure of the benzoxazine monomer was confirmed by ¹H-NMR and FTIR. Supplementary Figure S1 shows the ¹H-NMR and ¹³C-NMR spectrum of urushiol. Figure 1B and Supplementary Figure S2 show the ¹H-NMR and ¹³C-NMR spectrum of BZ. The peaks in the range 5.40–6.60 ppm are assigned to the protons in–CH = CH–CH = CH–. The peaks from 1.29 to 3.00 ppm are assigned to protons of the saturated bond in the alkyl side group of the urushiol. Two characteristic resonances centered at 4.01 and 4.99 ppm are attributed to the protons in Ar–CH₂–N and O–CH₂–N, respectively, which is clear evidence for the formation of benzoxazine. It is worth noting that the proton resonance peak of the oxazine ring (O–CH₂–N) is overlapped with the phenol hydroxyl. Figure 1C shows the FTIR spectra (Xu et al., 2012) of urushiol and BZ at room temperature. The bands at 947 and 983 cm⁻¹ in the spectra of BZ and urushiol are assigned to the bending vibration of the conjugated double bond (-CH = CH-CH = CH-), indicating that the side chain of the urushiol was not changed during the reaction. The bands at 1,621 cm⁻¹, 1,595 cm⁻¹ in the spectra of urushiol and BZ are assigned to the skeleton vibration absorption peak of the benzene ring (-C=C-). The band at 3,009 cm⁻¹ is assigned to the stretching vibration of the isolated double bond (-CH = CH-CH₂-CH = CH-). The broad band at 3,200–3,600 cm⁻¹ was due to the (–OH) vibration. The band at 1,140 cm⁻¹ was the asymmetric stretching vibration absorption peak of (C-N-C). In comparison with urushiol, the absorption peaks at 983–947 cm⁻¹ shift into one spike, and the absorption is enhanced. BZ exhibited a band at 965 cm⁻¹ corresponding to the out-of-plane (-C-H) vibration of the benzene ring to where the oxazine ring is attached and a band at 1,253 cm⁻¹ due to the asymmetric stretching of (-C-O-C-) of benzoxazine. In the fingerprint area of the spectra, the band 965 cm⁻¹ is the characteristic benzene ring mode of benzoxazine. That broad band at 3,200–3,600 cm⁻¹ assigned to the (–OH) vibration is clearly weakened. Both the ¹H-NMR and FTIR spectra indicate the successful synthesis of the urushiol-based benzoxazine monomer.

The curing behavior of BZ/MCl_x was studied by DSC and the curves are shown in Figure 2 and Table 2. As shown in Figure 2A, the exothermic peak temperature (T_p) of BZ is 192.8°C. For PBZ-M, broad exothermic peaks are observed. The exothermic peak temperatures decreased from 192.8 to 134.2°C, 124.5, and 109.4°C for BZ/CuCl₂, BZ/FeCl_3, and BZ/AlCl₃, respectively. The results indicate that MCl_x can effectively catalyze the polymerization of BZ. Furthermore, no isothermal DSC curves were used to study the curing kinetics of BZ and BZ/MCl_x at different rates (β = 5, 10, 15°C min⁻¹). The results are shown in Figures 2C,D and Supplementary Figures S3,S4. The activation energy (E_a) is calculated using Kissinger (Kissinger 1957) and Ozawa (Takeo 1965) models. The activation energy of BZ/MCl_x is dramatically decreased compared with BZ. This means that BZ/MCl_x is easier to activate and polymerize than BZ.

Metal ionic catalytic polymerization of PBZ-M was examined by NMR and ATR-FTIR. Figure 3A and Supplementary Figure S5 show the comparison of NMR spectra of PBZ-M and BZ. Compared to BZ, the peaks assigned to protons in Ar–CH₂–N and O–CH₂–N in PBZ-M disappeared and a new peak assigned to the Mannich bridge appeared. Figures 3B,C show the ATR-FTIR spectra of PBZ-M, PBZ, and BZ. The band at 968 cm⁻¹, characteristic absorption peaks of benzoxazine oxazine ring, dramatically decreases. In addition, the stretching vibration absorption peak of Ar-O shifts from 1,350 cm⁻¹ to 1,295 cm⁻¹ because of that the structure of Ar-O-C is converted into the structure of Ar-O-H during the ring-opening reaction of the oxazine. The abovementioned changes indicate that the MCl_x can promote benzoxazine’s ring-opening reaction. The band at 3,009 cm⁻¹ assigned to the stretching vibration of the isolated double bond is weakened, indicating that the isolated double bond in the alkyl side group is also involved in the polymerization reaction. The band at 3,200–3,600 cm⁻¹ assigned to free (–OH) in the monomer is weak. After polymerization, the shoulder at 3,200–3,600 cm⁻¹ is still weak in PBZ-Cu and PBZ-Fe. In contrast, a broad band at 3,200–3,600 cm⁻¹ appeared in PBZ-Al. The possible explanation for this unexpected difference between PBZ-Al and PBZ-Cu and PBZ-Fe is that Cu²⁺ and Fe³⁺ are oxidizing agents. The phenolic hydroxyl groups generated during polymerization were oxidized to carbonyl groups. It can also be further demonstrated by the stronger intensity of the peak at 1,630 cm⁻¹ that is assigned to the carbonyl groups in PBZ-M than that of PBZ. Another reason for this difference may be that the coordination situation of the metal and the hydroxyl group in PBZ-M is different. We boldly assume that the Al³⁺ in PBZ-Al is not mainly coordinated with hydroxyl groups while the Cu²⁺ and Fe³⁺ is.

To reveal the interaction of the metal ion and the polymers, XPS was conducted to show the element distribution in the polymer. Figure 4 shows the XPS spectra of PBZ, PBZ-Al, PBZ-Cu, and PBZ-Fe. In the XPS spectra of PBZ, there are only three peeks relating to O1s (532.1 eV), N1s (399.8 eV), and C1s (284.6 eV). In the XPS spectra of PBZ-Al (Kurdi et al., 2002), there are four peeks relating to O1s (532.1 eV), N1s (400.7 eV), C1s (284.6 eV), and Al2p (74.2 eV). In the XPS spectra of PBZ-Cu(Dong et al., 2011), there are four peeks relating to O1s (532.7 eV), N1s (403.1 eV), C1s (284.6 eV), and Cu2p (932.9 eV). In the XPS spectra of PBZ-Fe (Deng et al., 2017), there are four peeks relating to O1s (532.1 eV), N1s (401.1 eV), C1s (284.6 eV), and Fe2p (711.4 eV). The feature peak position and the change of the polymer electron binding energy of elements are summarized in Table 3. Compared to the binding energy of the metal elements in MCl_x, the binding energy of the metal elements in PBZ-M are all changed (△_Al = −0.6 eV, △_Al = −0.2 eV, △_Cu = −1.3 eV). It indicates the interaction between metal ions and the main structure of benzoxazine after the polymerization. In addition, the change of the binding energy of Cu²⁺ is larger than Fe³⁺ and then Al³⁺, indicating a stronger interaction between Cu²⁺ and benzoxazine. The binding energy of N is also changed (△_N-PBZ-Al = 1.3 eV, △_N-PBZ-Cu = 1.5 eV, and △_N-PBZ-Al = 0.9 eV). As for the element of oxygen, the binding energy of PBZ-Al and PBZ-Fe is not changed compared with that of PBZ. The binding energy of PBZ-Cu increased from 532.1 to 532.7 eV (△_O-PBZ-Cu = 0.6 eV). The abovementioned analysis indicates Cu²⁺ is coordinated with both oxygen and nitrogen atoms while the Fe³⁺ and Al³⁺ are coordinated with only nitrogen atoms.

The effect of metal ions on the mechanical properties of PBZ-M is shown in Table 4. Overall, PBZ-Cu has better mechanical property (hardness = 2H, impact resistance = 40 cm, flexibility = 1 mm and adhesion = grade 2) than PBZ-Fe and PBZ-Al. The impact resistance and flexibility of PBZ-Fe and PBZ-Al are similar. PBZ-Fe is harder than PBZ-Al, while the adhesion is poorer than PBZ-Al.

Due to the strong intramolecular hydrogen bonding between the hydroxy groups of polybenzoxazines, its surface energy is even lower than pure polytetrafluoroethylene (PTFE). To reveal the influence of the metal ion on the wettability of the polymer, we measured the water contact angle (WCAs) of PBZ, PBZ-Cu, PBZ-Fe, and PBZ-Al. As shown in Figure 5, compared with PBZ (WCA_PBZ = 103° ± 2°), the introduction of Al³⁺ and Fe³⁺ reduces the WCA (WCA_PBZ-Al = 92° ± 1.9° and WCA_PBZ-Fe = 96° ± 1.3°). The reason for the negative effect of Al³⁺ and Fe³⁺ on the WCAs of PBZ-Al and PBZ-Fe is the chelation between Al³⁺ and Fe³⁺ and the nitrogen atom, destroying the intramolecular hydrogen bonding of the hydroxy groups. In contrast, the intramolecular hydrogen bonding of the hydroxy groups is replaced by the chelation between Cu²⁺ and the nitrogen atom and oxygen atom. The introduction of Cu²⁺ has a positive effect on the WCA (WCA_PBZ-Cu = 106° ± 2°).

The effect of the metal ions catalyst on the thermal stability was studied by using thermogravimetric analysis (TGA). Figure 6 shows the TGA traces and DTG curves of PBZ, PBZ-Fe, PBZ-Al, and PBZ-Cu. The related thermogravimetric results are presented in Table 5. As we can see, the onset degradation temperature of these polymers is almost the same (∼200°C). The char yield of PBZ at 600°C is 19.20%. The char yield of PBZ-M is effectively improved compared to PBZ. The char yields of cured PBZ-Al, PBZ-Fe, and PBZ-Cu are increased from 19.20 to 27.42, 31.23, and 38.34%, respectively. Accordingly, PBZ-Cu exhibits better thermal stability than PBZ-Fe and PBZ-Al as reflected by T _10% , T _50% , and T _max values. The effect of metal ions on improving the thermal stability of benzoxazine can be explained as follows: 1) the mechanical properties (hardness, listed in Table 3) and the gel fraction of PBZ-Al (HB, 84.3), PBZ-Fe (H, 85.3) and PBZ-Cu (2H, 87.4) are increased, indicating that the degree of crosslinking is increased. The higher degree of crosslinking, the more stable the PBZ-M is. 2) Cu²⁺ and Fe³⁺ are variable valence metal ions while Al³⁺ is not, indicating that Cu²⁺ and Fe³⁺ can oxidize the polybenzoxazine. In the oxidation of polybenzoxazine, the Mannich base transforms into an amide and/or imide-like structure, which is a thermally more stable structure. 3) According to the results of XPS analysis, PBZ-Cu has chelation between Cu²⁺ and nitrogen (N) and oxygen (O) atoms. PBZ-Fe and PBZ-Al have chelation between M³⁺ and oxygen (O) atom. The coordination bond strength in PBZ-Cu is higher than that in PBZ-Fe and PBZ-Al. It can be concluded from the abovementioned results that the thermal stability of polybenzoxazines treated with metals can be affected by the chemical structures of the polymers.