In situ formation of benzoxazines in polyoxymethylene: a simple approach for retarding formaldehyde generation and tuning mechanical properties under a semi-interpenetrating network

Abstract

Polyoxymethylene (POM) is an engineering plastic which tends to release formaldehyde under a melt-mixing process. The present work proposes a simple way to control the formaldehyde generation by a reactive blending with bisphenol-A and amine to form in situ benzoxazines via the Mannich reaction. Further thermal treatment leads to partial conversion of benzoxazines (BA-a) to polybenzoxazines (poly(BA-a)) and the POM–poly(BA-a) obtained is under a semi-interpenetrating network. At that time, the BA-a and poly(BA-a) reached the amorphous phase to induce an increase in elongation at break. When POM blends were electrospun, the molecular orientation of POM was induced and this synergistically functions with the poly(BA-a) network, resulting in an increase in tensile strength which never occurs in the bulk material. The present work shows, for the first time, not only reactive blending as a way to control the formaldehyde generation of POM but also an in situ semi-interpenetrating network as an approach to fine tune the properties of POM.

---

Introduction

Polyoxymethylene (POM) is an important engineering plastic for the moving parts in automobiles and electronic appliances.^1–3 POM consists of an oxymethylene (–CH₂O–) repeat unit, which can be easily thermally decomposed to give formaldehyde (HCHO) and further oxidized to formic acid. As the deformaldehyde gives off toxic gases one must be aware of,⁴ methods to control formaldehyde generation to stabilize POM during the melt-mixing process have been variously reported, as seen in the cases of the copolymerization with an oxyethylene chain,^4,5 blending with other thermoplastics, etc.^6–10 In fact, those approaches are based on the method of avoiding the chain scission by extending the methylene to ethylene units or by the induction of chain flexibility to disperse the thermal energy by blending with other aliphatic polymers.

This comes to our idea about the formation of other polymers which require formaldehyde, such as phenolic resins in POM, as the way to control the formaldehyde generation of POM. In this viewpoint, bisphenol-A based benzoxazines (BA-a), which can be prepared from bisphenol-A, amine and formaldehyde via the Mannich reaction, are good candidates. In other words, by simply mixing bisphenol-A and amine with POM, the thermal processing of POM which leads to the formaldehyde generation might allow an in situ BA-a formation. In fact, after BA-a was formed, thermal treatment (150 °C) led to thermoset polybenzoxazines (poly(BA-a))^11,12 (Scheme 1). In this way, the two steps of oxazine ring formation and oxazine ring opening polymerization allow us to prepare a reactive blending with POM after the blending with bisphenol-A and amine. When the benzoxazine ring is formed by consuming the formaldehyde generated from POM, the thermal stability of POM can be expected.

Scheme 1 Reactive blend of bisphenol-A and aniline via melt-mixing process to form in situ BA-a and poly(BA-a) after POM deformaldehyde.

The present work, therefore, aims to demonstrate a reactive blending to form BA-a and poly(BA-a) in POM matrices as an approach to retard the formaldehyde generation. The work extends to studies on the packing structure of the blend to clarify an in situ semi-interpenetrating network of poly(BA-a) where poly(BA-a) might play a role in the amorphous phases and tune the mechanical properties of POM, especially the elongation at break. The work also covers comparative studies between bulk sheets and electrospunfibers of the POM blends to clarify how the POM molecular orientation in nano-confinement synergistically functions with poly(BA-a) networks to further fine tune the mechanical properties.

Experimental

Materials

Polyoxymethylene (POM) with a dioxolane (DOL) content of 1.5% is the product of Thai Polyacetal Co. Ltd., Thailand. Bisphenol-A and aniline were purchased from Sigma-Aldrich Co. LLC, Germany. All chemicals were analytical grade and used as received without further purification.

In situ reactive blending of bisphenol-A-aniline benzoxazines (BA-a) in POM

POM (70 g, 2.33 mol) was mixed with bisphenol-A (0.023 mol, 1% mol) and aniline (0.046 mol, 2% mol) and allowed to blend in an internal mixer (Brabender® OHG Duisburg, Germany) at 200 °C with a screw speed of 20 rpm for 2 h to obtain POM–bisphenol-A-aniline benzoxazine (POM–BA-a) blend as POM–BA01a02. Similarly, a series of blends, i.e. POM–BA02a04, POM–BA03a06, POM–BA05a10, POM–BA07a14, and POM–BA10a20, were prepared by applying 2, 3, 5, 7, and 10% mol of bisphenol-A and the corresponding 4, 6, 10, 14, and 20% mol of aniline, respectively (Table S1†).

Electrospinning of POM–BA-a blends

The POM–BA-a blends were dissolved in 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP) at a concentration of 5 wt%. The solutions obtained were electrospun by using a NANON Electrospinning Setup (MECC Co., Ltd., Japan) equipped with a metallic disc collector rotating at a linear velocity of 1890 m min⁻¹. The needle-to-collector distance, electrical voltage, and volumetric flow were fixed at 10 cm, 15 kV, and 0.5 mL h⁻¹, respectively. All experiments were carried out at 25 °C under 70% relative humidity.

Instruments and equipment

Fourier transform infrared spectra (FTIR) were recorded by an Equinox 170 Bruker FT-IR spectrometer in the range 4000–650 cm⁻¹ with 64 scans and a resolution of 4 cm⁻¹. The POM and BA-a blends were characterized in the solid-state by a CMX 300 Infinity Plus/300 MHz: Varian-Chemagnetics NMR. The specimen was placed in a ø4 mm sample tube and rotated at an MAS of 15 000 Hz. The standard used to determine the chemical shift was silicon rubber for which the methyl group was at 0 ppm. The samples were heated at 200 °C for 2 h under nitrogen atmosphere. The formaldehyde produced was absorbed by aqueous sodium sulfite (Na₂SO₃) solution. The solution obtained was titrated with hydrochloric acid (HCl) (0.1 N) and the formaldehyde emission amount (FEA) was calculated by using the following equation:

FEA (ppm) = [(V₂ − V₁) × C_HCl × 30.03 × 1000]/M (1)

where M is the weight of sample, C_HCl is the molar concentration of HCl, V₁ is the volume of HCl to neutralize the blank absorbent solution, and V₂ is the volume of HCl to neutralize the absorbent solution which absorbed the formaldehyde during the measurement. Thermo-gravimetry-Fourier transform infrared spectroscopy (TG-FTIR) measurements were carried out by using a TA TGA-Q50 (USA) interface with a Thermo Nicolet Nexus 670 FTIR (USA). TG measurements were performed using 20 mg of the sample at a heating rate of 10 °C min⁻¹ under nitrogen in the temperature range 30 °C to 600 °C. Each spectrum was recorded by FTIR every 60 s with a 4 cm⁻¹ resolution. For isothermal degradation, each sample was heated from 30 °C to 200 °C for 2 h at a heating rate of 10 °C min⁻¹ under nitrogen. Differential scanning calorimetry analysis was carried out by using a 200F3 NETZCH (Germany) under a nitrogen flow rate of 50 mL min⁻¹ and heating rate of 5 °C min⁻¹ from −90 °C to 200 °C. The degree of crystallinity (χ_c) was estimated by assuming that the heat of melting per unit mass of crystalline material is identical to that of melting of a 100% crystalline POM sample, i.e., 317.93 J g⁻¹ as reported by Iguchi et al.¹³ Dynamic mechanical analysis (DMA) was performed with an Explexor 100N (Gabo QualimeterTestanlagen GmbH, Germany) in the tensile mode. Measurements of the storage moduli (E′) along with the ratio of loss modulus (E′′) to storage modulus (E′) (tan δ) were carried out at a frequency of 1 Hz and the strain of 0.1%. All measurements were determined at a heating rate of 2 °C min⁻¹ from −100 °C to 100 °C. The morphology of the samples was observed by an S-4800, Hitachi ultra-high resolution cold field emission scanning electron microscope (FE-SEM) (Japan). Number-averaged molecular weight (M_n), weight-averaged molecular weight (M_w), and polydispersity index (PDI) were determined by a Shimadzu LC-20AD and CTO-20A system equipped with four Shodex GPC K-802.5, 803, 804, and 805 columns connected in series and a Shimadzu RID-10A refractive index detector. Chloroform (CHCl₃) was used as an eluent at a flow rate of 1.0 mL min⁻¹. Polystyrene standards were used and the measurements were performed at 40 °C. The 2D-SAXS measurements were carried out at the Synchrotron Light Research Institute (Public Organization), Suranaree University of Technology, Thailand. The X-ray wavelength (λ) was tuned at 0.1549 nm, and the scattering vector (q) defined by q = (4π/λ)sin(θ/2) (θ: the scattering angle), was calibrated by silverbehenate (AgBH). The packing structure was characterized by a Rigaku RINT 2000 wide angle X-ray diffraction (WAXD) using Cu Kα line as the incident X-ray beam. The diffraction profiles were measured in a reflection mode at 5–60° (2θ) at a scanning rate of 2° min⁻¹. Tensile modulus was measured according to ASTM D638 by using a 4320 Instron (USA) universal testing machine. The single nanofiber tensile test was performed at a strain rate of 0.1 mm min⁻¹ using a Nanomechanical Testing System (Nano UTM™, MTS Systems Corporation, USA).

Results and discussion

In general, the melt-mixing process induces thermo-oxidative degradation of POM. It is known that formaldehyde is oxidized and becomes formic acid, accelerating the degradation of POM via acidolysis.⁴ As the present work proposes the reactive blending of POM with phenol and amine, it is important to clarify the amount of formaldehyde released from POM in the reactive blending conditions applied in this work, i.e. 200 °C for 2 h. Here, POM resin (20 mg, 0.67 mmol) was kept in isothermal conditions (200 °C for 2 h) and the weight loss was observed by the TGA technique. Assuming that the weight loss at 200 °C for 2 h reflects the amount of formaldehyde, it was found to be about 1 wt% (0.0233 mol) (Fig. S1†). Based on this amount, the stoichiometric ratios of bisphenol-A and aniline were varied (Table S1†) to obtain a series of POM–BA-a blends.

Structural characterization of POM–BA-a blends

As POM can be degraded to release formaldehyde, especially during melting at 160 °C,⁴ it is important to clarify whether or not the reactive blending by melt-mixing of phenol and amine with POM results in benzoxazine monomer (BA-a) in POM matrices. In fact, if the BA-a is formed in POM in the blending process, the continuous heat may also lead to the step of polymerization of BA-a to obtain poly(BA-a) in the POM (Scheme 1). After blending of bisphenol-A and aniline with POM in the molten stage using a Brabender mixer, a yellowish blend was obtained.

In order to identify the existence of BA-a in the blend, the following procedures were carried out. Each blend was carefully dissolved in 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP) to obtain a clear yellowish solution. The HFIP was evaporated to obtain a porous product. The porous product was washed thoroughly with CHCl₃ to dissolve BA-a. At this point, the CHCl₃ phase was evaporated and the precipitates obtained were characterized by FTIR.

The precipitates showed the characteristic peaks of benzoxazines, i.e. 920–950 cm⁻¹ and 1500–1510 cm⁻¹ (trisubstituted benzene ring), 1327 cm⁻¹ (CH₂ wagging of oxazine), 1233 cm⁻¹ and 1031 cm⁻¹ (ether linkage in benzoxazine ring) (Fig. S2†). This confirmed the formation of BA-a as a consequence of the in situ reactive blending in POM.

An attempt to confirm BA-a in the bulk POM blend sample by using CP-MAS NMR was also carried out. In the case of the pristine POM, the CH₂ resonance of POM at 88.5 ppm is observed (Fig. 1A). For POM–BA-a such as POM–BA05a10 (Fig. 1B), the spectrum was curve-fitted to identify each peak clearly. Here, the peaks at 31 ppm and 41 ppm assigned to methyl group (carbon k) and quaternary carbon (carbon j), respectively, are observed. The peak at 151 ppm assigned to the aromatic carbon bonded to the oxygen atom of the oxazine ring (carbon e) and the peak at 95 ppm assigned to the aromatic carbon between the oxygen atom and nitrogen atom (carbon b) are also identified. It is important to note that the peaks at 126 ppm and 158 ppm which belong to carbon i′ and e′ of poly(BA-a) are also identified. This reveals that not only BA-a but also poly(BA-a) are formed in the POM matrix. The confirmation of BA-a and poly(BA-a) in POM after reactive blending confirms the reaction between bisphenol-A, aniline, and formaldehyde which was generated from POM during the melt-mixing process, as in Scheme 1.

Fig. 1 CP-MAS spectra of (A) pristine POM and (B) POM–BA05a10.

Since BA-a was further polymerized to give poly(BA-a), the question here is which factors relate to the control the polymerization in POM.

As the disappearance of the oxazine ring, i.e. the C–O–C peak at 1233 cm⁻¹, reflects the polymerization of poly(BA-a), the integral ratio between the C–O–C peak at 1233 cm⁻¹ and the CH stretching of the CH₃ group (2965 cm⁻¹) determined the conversion of BA-a to poly(BA-a) as in the following equation:

As shown in Fig. S3,† the conversion of BA-a to poly(BA-a) increased from 19% to 76% with an increase in bisphenol-A content from 0.023 mol to 0.23 mol. This confirmed that the bisphenol-A content was one of the main factors in the polymerization. This can be explained from the mechanism of the ring opening for which the hydrogen bond between phenol and oxazine ring is required in the initial step.¹⁴

Optimal condition for in situ BA-a and poly(BA-a) formation in POM

A series of POM–BA-a blends with different mole contents of bisphenol-A and aniline were prepared so that the maximal formaldehyde consumption could be identified. In other words, it is expected that at the optimal amount of bisphenol-A and aniline, the formaldehyde generated from thermal decomposition of POM would be completely consumed.

The formaldehyde generation and consumption were traced by the formaldehyde emission amount (FEA) (Table S2†). The sample (10–20 g) was added to a test tube and heated at 200 °C for 2 h under nitrogen atmosphere. The pristine POM showed the FEA for 105.3 ppm. When the pristine POM was melted in a Brabender at 200 °C for 2 h, for so-called thermally treated POM, the FEA became as much as 8 times as high. In this work, it should be noted that the preparation of the in situ reactive blends was similar to that of the thermally treated POM but with addition of bisphenol-A and aniline. Bisphenol-A and aniline were added to the test tube containing POM and heated at 200 °C for 2 h under nitrogen atmosphere. It is clear that the FEA was still identified until the content of bisphenol-A was as high as 0.01 mole content (Table S2†). This indicated that above this bisphenol-A mole content, the formaldehyde generated from thermal decomposition of POM was completely consumed. In fact, when POM, bisphenol-A and aniline were reactively blended in a Brabender at 200 °C for 2 h, e.g. POM–BA05a10, the sample obtained did not show any FEA. This implied that the formaldehyde generated was completely consumed once the reactive blending was accomplished.

This comes to the question of how much formaldehyde was consumed in the in situ reactive blends. To answer this, the weight loss and char yield of POM–BA-a blends, which represented BA-a and poly(BA-a), were considered from TGA. The amount of both in moles also reflected the formaldehyde consumed. In fact, the pristine POM and the thermally treated POM are completely degraded at 420 °C without char yield (Fig. S4†) since there is no thermoset poly(BA-a) in the system. In the case of POM–BA-a blends, the weight loss, starting at 450 °C and rising to 600 °C, is observed in the range of 5–17 wt%, referred to the degradation of BA-a (Fig. 2A and S4†). The char yield, which reflected the cross-linked poly(BA-a), is observed from 6 wt% to 18 wt% at 600 °C (Fig. 2A and S4†). From the weight loss and the char yield, one can evaluate the amount of formaldehyde generated in the system. For example, POM–BA01a02 showed the weight loss of BA-a determined from 450 °C to 600 °C to be 5 wt% (0.008 mol). The char yield of poly(BA-a) at 600 °C is 6 wt% (0.008 mol). The result implied that the formaldehyde consumed in POM–BA01a02 was 0.064 mol (Table S1†). When the bisphenol-A content was increased to 7% mol and 10% mol (i.e. POM–BA07a14 and POM–BA10a20), the formaldehyde consumed was found to be around 0.20 mol (Table S1†). As the stoichiometry ratio of formaldehyde: bisphenol-A: aniline to form BA-a is 4 : 1 : 2, the amount of formaldehyde can be calculated. For POM–BA10a20, the formaldehyde consumed is 0.20 mol, which is 85 714 ppm. This number is much higher than the amount identified in the FEA test. The reason behind this might be related to the conditions of the FEA test, in which only a small amount of sample under nitrogen atmosphere was used. Moreover, from the weight loss and the char yield, it can be concluded that the content of poly(BA-a) increased with an increase in bisphenol-A content and this supports the mechanism of benzoxazine ring opening, which requires the phenol.¹⁴

Fig. 2 Reactive blends of POM with different bisphenol-A contents for; (A) weight loss at 450–600 °C (●) and char yield (○), (B) thermal degradation temperature (●) and initial degradation temperature (○), and (C) melting temperature (●) and crystallinity degree (○).

In order to trace the degradation temperature and the degradation component at each stage, TG-FTIR was applied. In the case of pristine POM (Fig. 3A), the degradation with HCHO peaks at 1745 cm⁻¹ and 2800 cm⁻¹ can be observed from 289 °C. From the HCHO peaks, the initial degradation temperature of pristine POM is identified at 289 °C. For the thermally treated POM, the HCHO peak starts at 251 °C (Fig. 3B), which is much lower than the pristine POM. This implies the ease of releasing formaldehyde if the pristine POM is once thermally treated. However, in the cases of POM–BA-a blends, for example POM–BA05a10, the release of formaldehyde can be observed at a temperature as high as 338 °C (Fig. 2B and 3C) which is 87 °C higher than those of POM and the thermally treated POM. It is also important to mention that the degradation temperature (T_d) of POM–BA05a10 is 431 °C, which is higher than that of pristine POM (378 °C). This suggests that the formaldehyde released during the in situ reactive blending was reacted with bisphenol-A and aniline to form BA-a. As a consequence, the free formaldehyde to form formic acid was drastically reduced.

Fig. 3 TG-FTIR spectra of gases generated during thermal degradation of (A) pristine POM, (B) thermally treated POM, and (C) POM–BA05a10 as a function of temperature.

Moreover, in the case of POM–BA05a10, an additional T_d at 475 °C was observed, indicating the formation of BA-a (Fig. S4†).

This comes to our question: what if the bisphenol-A and aniline were in excess? For example, in the case of POM–BA01a02, the TGA thermogram and derivative weight loss indicated that unreacted bisphenol-A and aniline were identified at 280 °C, and 167 °C, respectively (Fig. S4†). Therefore, this comes to our idea of extending the in situ BA-a formation by adding more formaldehyde. Table S1† indicates that 0.064 mol of formaldehyde generated was used to react with bisphenol-A and aniline. Thus, an additional 0.028 mol of formaldehyde was added so that the reaction with 0.023 mol of bisphenol-A and 0.046 mol of aniline is stoichiometric, i.e. formaldehyde: bisphenol-A: aniline for 4 : 1 : 2, and the blend was termed POM–F04BA01a02. However, the blend obtained did not show a greater amount of poly(BA-a), as seen from the weight loss (3 wt%) at 450 °C to 600 °C and char yield (6.8 wt%). In fact, the result was no different from the case of POM–BA01a02 (6 wt%) (Table S1†). Surprisingly, POM–F04BA01a02 started to release formaldehyde at a much lower temperature, i.e. 232 °C as compared to the thermally treated POM which shows the release at 251 °C. The results suggest that the additional formaldehyde did not effectively react with bisphenol-A and aniline but rather initiated the degradation of POM. This might be due to the fact that the additional formaldehyde can be further oxidized to formic acid, leading to acidolytic degradation of the POM chain.⁴

Poly(BA-a) in POM as a semi-interpenetrating network

As mentioned earlier, the CP-MAS NMR, FTIR and TG-FTIR results suggested the ring opening polymerization of BA-a. The question is how poly(BA-a) matrices existed in POM matrices. Considering the thermal properties, all types of POM–BA-a blends show decreases in the melting temperature (T_m) and the crystallinity degree (χ_c) with an increase in bisphenol-A content (Fig. 2C). In fact, the T_m and χ_c values are also lower than those of the pristine POM and of the thermally treated POM. This suggests that the poly(BA-a) in POM matrices obstructs the packing of POM, in other words, it induces the amorphous phase in POM.

In fact, the shift in glass transition temperature (T_g) also reflects the molecular chain penetration of two or more polymer chains.¹⁵ Although DSC is a good way to determine the T_g, for POM, the T_gs could not be clearly recognized since the heat capacity change (ΔC_p) of crystalline polymer is very difficult to detect.¹⁶

DMA was, thus, applied to trace the ratio of loss modulus (E′′) to storage modulus (E′) (tan δ) as a function of temperature for pristine POM, thermally treated POM, and POM–BA-a blends (Fig. 4A). In the present work, the DMA was applied to measurements from −100 °C to 100 °C. In fact, a temperature above 130 °C may lead to the breaking of the samples. From the DMA analysis, the tan δ of the pristine POM and the thermally treated POM are at about −68 °C (Fig. 4A(a and b)), which can be referred to the T_g of POM. For the POM–BA-a blends, two tan δ are observed (Fig. 4A(c–h)) at around −68 °C and 20 °C, which referred to the T_gs of POM and poly(BA-a), respectively. In fact, both tend to shift to the lower temperature when bisphenol-A content is increased (Fig. 4B). For example, in the case of POM–BA10a20, the T_g of POM is identified at −80 °C while that of poly(BA-a) is at 17 °C (Fig. 4A(h)). For T_g of poly(BA-a), it was in the range from 17 °C to 26 °C. This might be due to the low molecular weight of poly(BA-a) that existed in the POM matrices (Table S3†). However, the decrease in T_g of POM for POM–BA-a blends implies the in situ formation of BA-a, the polymerization of BA-a, and the semi-interpenetrating network of poly(BA-a) in POM. As shown in Fig. 4A, the baseline at 50 °C to 100 °C is shifting up. This might be because the POM was in the α-relaxation state. This state is associated with the motions of the crystalline phase.^17,18

Fig. 4 Reactive blends of POM for; (A) dynamic loss modulus for the samples as a function of temperature: (a) pristine POM, (b) thermally treated POM, (c) POM–BA01a02, (d) POM–BA02a04, (e) POM–BA03a06, (f) POM–BA05a10, (g) POM–BA07a14 and (h) POM–BA10a20 and (B) glass transition temperature of POM (●) and poly(BA-a) (○) as a function of bisphenol-A content.

It is important to confirm the semi-interpenetrating network of poly(BA-a) in POM matrices. In order to clarify this, the morphologies of the fracture surfaces of pristine POM and POM–BA-a blends were observed by SEM (Fig. S5†). The micrograph of POM–BA03a06 exhibited a rougher surface than that of pristine POM. This might be due to an incomplete semi-interpenetration of poly(BA-a) in POM matrices. In the case of POM–BA07a14 and POM–BA10a20, their micrographs exhibited smooth surfaces, suggesting the semi-interpenetrating poly(BA-a) in POM matrices.

Micro-structure of POM and POM–BA-a blends

The question arising here is about the change in microstructure of the POM–BA-a blends as compared to the pristine POM. The WAXD patterns of the pristine POM and the thermally treated POM show peaks at 2θ = 22.8°, 34.6°, and 48.2°, corresponding to (100), (105), and (110) lattice plane of hexagonal crystal of POM^19–21 (Fig. 5A(a)). For POM–BA-a blends, although the diffraction peaks appear at the same positions (Fig. 5A(b–d)), there are some amorphous haloes in the range of 2θ = 20.5° to 18.9°. In fact, poly(BA-a) was individually prepared and its WAXD pattern is as shown in Fig. 5A(e). Comparing the diffraction patterns of POM–BA-a blends to that of poly(BA-a), the broad peak implies poly(BA-a) in POM (Fig. 5A(e)). The WAXD patterns also support the semi-interpenetrating network of poly(BA-a) into POM matrices with regard to the changes in the packing structure of POM. Further confirmation to identify the semi-interpenetrating network was carried out. The POM–BA-a sheets (i.e. POM–BA01a02, POM–BA03a06 and POM–BA10a20) were dissolved by HFIP and allowed to evaporate to obtain the white porous sheet. The white porous sheet was dispersed in CHCl₃ at 24 h to extract out BA-a and poly(BA-a). After extraction, the WAXD patterns of the white porous sheet show the same diffraction peaks positions without the halo position at 2θ = 19° (Fig. 5B(a–d)) as compared to those of pristine POM.

Fig. 5 WAXD patterns of (A) before extraction and (B) after extraction: (a) pristine POM, (b) POM–BA01a02, (c) POM–BA03a06, (d) POM–BA10a20 and (e) poly(BA-a).

Moreover, the residues obtained from the CHCl₃ phase show the halo position at 2θ = 19° (Fig. 5B(e)). This confirms that BA-a and poly(BA-a) form a semi-interpenetration network in POM matrices. The micro-structure was further investigated by 2D-SAXS. The circular patterns (Fig. S6†) indicated the random orientation of the lamella in the pristine POM, the thermally treated POM and POM–BA-a blends. The SAXS profile showed a scattering maximum at the relative q_m position, suggesting a lamella stacking structure of the samples (Fig. S6†). The POM–BA-a blends showed slight changes in scattering maxima at the relative q_m position as compared to those of the pristine POM and the thermally treated POM. The q_m values slightly decreased until they could not be observed when the bisphenol-A content was increased. From q_m values, the long period (L) can be calculated from the magnitude of the q_m vector as L = 2π/q_m. It was found that the L increased from 12.57 nm to 22.44 nm with an increase in bisphenol-A content from 0.023 mol to 0.11 mol (Fig. 7A). In the cases of POM–BA07a14 and POM–BA10a20, which contained 0.16 mol and 0.23 mol of bisphenol-A content, respectively, their L values could not be calculated because no circular pattern was observed. This implied the poly(BA-a) disturbed the packing of POM.

Possible mechanism of semi-interpenetrating network of poly(BA-a)

From the above results, the semi-interpenetrating network of poly(BA-a) in POM matrices can be confirmed. Thus, a possible mechanism was proposed, as shown in Fig. 6. When POM was heated, the formaldehyde was generated. At that time, formaldehyde reacts with bisphenol-A and aniline, which were added into the system, resulting in BA-a formation. Further thermal treatment leads to oxazine ring opening polymerization of poly(BA-a) in POM matrices. Because poly(BA-a) is thermoset, it is expected that the semi-interpenetrating network between POM and poly(BA-a) was formed in this step.

Fig. 6 Mechanism of semi-interpenetrating network of poly(BA-a) in POM matrices. Blue: POM chain, pink: formaldehyde, purple: bisphenol-A, and brown: aniline.

Mechanical properties

The POM–BA-a blends were prepared by compression molding and the sheets obtained were used to identify the tensile modulus (E), storage modulus (E′) and elongation at break (ε). The pristine POM and the thermally treated POM sheets show the E (∼745 MPa), E′ (∼941 MPa) and ε (∼35%) (Fig. 7B and C). In the cases of POM–BA-a blends, both E and E′ are lower than those of the pristine POM and the thermally treated POM. In addition, an increase in bisphenol-A from 0.023 mol to 0.23 mol in POM–BA-a blends leads to decreases in E (from 729 MPa to 219 MPa) and G′ (from 825 MPa to 199 MPa) while the ε increased from 37% to 173%. The results support the decrease in crystallinity degree and the looseness in packing structure of POM due to the semi-interpenetrating network of poly(BA-a), as discussed earlier. In fact, the POM–BA-a samples were rather soft and tough compared to the pristine POM.

Fig. 7 Reactive blends of POM under various bisphenol-A contents for; (A) long period, (B) elongation at break, and (C) tensile modulus (●) and storage modulus (○).

Controlling POM–BA-a blend orientation via electrospun fiber

Previously, our group declared electrospinning to be the technique to align polymer chain orientation using POM as a model case.^22–27 It is important to point out that, in the above section, the POM–BA-a blends were in sheet form and showed increases in amorphous phases and, as a consequence, the increases in elongation at break. At that time, the tensile modulus decreased and this reflected the obstruction of the POM packing structure by poly(BA-a) semi-interpenetrating network. Here, POM–BA-a blends were electrospun and the nanofibers obtained were collected with a rotational disk collector at a take-up velocity speed of 1890 m min⁻¹ so that the POM orientation can be induced. The E of a single electrospun nanofiber was investigated using a nanotensile tester to investigate whether or not the orientation of POM under the semi-interpenetrating network of poly(BA-a) can be obtained.

It was found that in the case of POM–BA-a in sheet form, the E values decreased from 729 MPa to 219 MPa with an increase in bisphenol-A content from 0.023 mol to 0.23 mol. In other words, the tensile modulus decreases with an increase in poly(BA-a). For the POM–BA-a nanofibers, the E increases with an increase in bisphenol-A content (Fig. 8). For example, the E of the POM–BA10a20 nanofiber (26.7 GPa), to which 0.23 mol of bisphenol-A was added, increases almost 7-fold compared to that of the pristine POM nanofiber (3.8 GPa). This characteristic is attributed to the alignment of fibers which was promoted during the fiber formation process, as reported earlier.²⁴

Fig. 8 Reactive blends of POM under various contents of bisphenol-A for; poly(BA-a) content (●), tensile moduli of sheet (○) and single nanofiber (△).

Conclusions

POM is known for its thermal degradation to produce formaldehyde. The present work proposed the reactive blending with bisphenol-A and aniline so that an in situ Mannich reaction occurs to form benzoxazines. This simple reactive blending consumed formaldehyde generated from POM and effectively retarded the thermal degradation of POM. The ring opening of benzoxazines by bisphenol-A further led to thermoset polybenzoxazines which were in a semi-interpenetrating network with POM. The polybenzoxazines networks obstructed the packing of POM, resulting in increases in the amorphous phase, and led to an increase in the elongation at break. However, the electrospinning technique further induced the POM packing to result in increases in the tensile modulus. The present work demonstrated a simple way, a so-called reactive blend, to stabilize POM, in the form of a semi-interpenetrating network, as well as tuning its mechanical properties, which is never easily obtained in other blending systems.

Acknowledgements

The authors are grateful for the support provided by Thai Polyacetal Co. Ltd., Thailand and appreciative of the financial support from the Ratchadaphiseksomphot Endownment Fund (CU-58-049-AM), Chulalongkorn University, Thailand. The authors would like to acknowledge the Royal Golden Jubilee Scholarship, Thailand Research Fund (PHD/0308/2552, 2.L.CU/52/F.1).

Notes and references

1. T. A. Koch and P. E. Lindvig, J. Appl. Polym. Sci., 1959, 1, 164–168.
2. J. Masamoto, T. Iwaisako, M. Chohno, M. Kawamura, J. Ohtake and K. Matsuzaki, J. Appl. Polym. Sci., 1993, 50, 1299–1305.
3. J. M. Samon, J. M. Schultz, B. S. Hsiao, S. Khot and H. R. Johnson, Polymer, 2001, 42, 1547–1559.
4. S. Lüftl, V. M. Archodoulaki and S. Seidler, Polym. Degrad. Stab., 2006, 91, 464–471.
5. K. Matsuzaki, M. Maeda, M. Kondo, H. Morishita, M. Hamada, T. Yamaguchi, K. Neki and J. Masamoto, J. Polym. Sci., Part A: Polym. Chem., 1997, 35, 2479–2486.
6. Y. Hu, L. Ye and X. Zhao, Polymer, 2006, 47, 2649–2659.
7. L. Gan and L. Ye, J. Thermoplast. Compos. Mater., 2010, 23, 543–559.
8. K. Pielichowski and A. Leszczyńska, J. Therm. Anal. Calorim., 2004, 78, 631–637.
9. T. Sun, Y. Lai, L. Ye and X. Zhao, Polym. Adv. Technol., 2008, 19, 1286–1295.
10. Z.-Y. Wang, Y. Liu and Q. Wang, Polym. Degrad. Stab., 2010, 95, 945–954.
11. X. Ning and H. Ishida, J. Polym. Sci., Part A: Polym. Chem., 1994, 32, 1121–1129.
12. V. M. Russell, J. L. Koenig, H. Y. Low and H. Ishida, J. Appl. Polym. Sci., 1998, 70, 1413–1425.
13. M. Iguchi, Macromol. Chem., 1976, 177, 549–566.
14. G. Riess, J. M. Schwob, G. Guth, M. Roche and B. Laude, in Advances in Polymer Synthesis, ed. B. M. Culbertson and J. E. McGrath, Springer, Boston, MA, US, 1985, pp. 27–49,  DOI:10.1007/978-1-4613-2121-7_2.
15. J. Qiu, C. Xing, X. Cao, H. Wang, L. Wang, L. Zhao and Y. Li, Macromolecules, 2013, 46, 5806–5814.
16. R. P. Chartoff, P. T. Weissman and A. Sircar, in Assignment of the glass transition, ASTM International, 1994.
17. S. Siengchin, G. C. Psarras and J. Karger-Kocsis, J. Appl. Polym. Sci., 2010, 117, 1804–1812.
18. R. J. V. Hojfors, E. Baer and P. H. Geil, J. Macromol. Sci., Part B: Phys., 1977, 13, 323–348.
19. G. A. Carazzolo, J. Polym. Sci., Part A: Polym. Chem., 1963, 1, 1573–1583.
20. X. Liu, S. Bai, M. Nie and Q. Wang, J. Polym. Res., 2011, 19, 1–6.
21. C. Lorthioir, F. Lauprêtre, K. Sharavanan, R. F. M. Lange, P. Desbois, M. Moreau and J.-P. Vairon, Macromolecules, 2007, 40, 5001–5013.
22. A. Baji, Y.-W. Mai, S.-C. Wong, M. Abtahi and P. Chen, Compos. Sci. Technol., 2010, 70, 703–718.
23. K. H. K. Chan, S. Y. Wong, X. Li, Y. Z. Zhang, P. C. Lim, C. T. Lim, M. Kotaki and C. B. He, J. Phys. Chem. B, 2009, 113, 13179–13185.
24. T. Kongkhlang, K. Tashiro, M. Kotaki and S. Chirachanchai, J. Am. Chem. Soc., 2008, 130, 15460–15466.
25. X. Ma, J. Liu, C. Ni, D. C. Martin, D. B. Chase and J. F. Rabolt, ACS Macro Lett., 2012, 1, 428–431.
26. E. P. S. Tan and C. T. Lim, Compos. Sci. Technol., 2006, 66, 1102–1111.
27. E. P. S. Tan, S. Y. Ng and C. T. Lim, Biomaterials, 2005, 26, 1453–1456.

---

Footnote

† Electronic Supplementary Information (ESI) available: Table of conditions to prepare in situ reactive blend of BA-a in POM together with weight loss, char yield and utilized formaldehyde, TGA, FTIR, SEM, and 2D-SAXS patterns. See DOI: 10.1039/c6ra18841a

---