Morphological adjustment determines the properties of cationically polymerized epoxy resins

Abstract

The morphological adjustment of a cationically polymerized cycloaliphatic epoxy resin is presented. For this purpose, the epoxy resin is polymerized both with a reactive poly(ε-caprolactone) (PCL-diol) and a unreactive one (PCL-diester) leading to a copolymer and a polymer alloy of different morphologies, respectively. Small-angle X-ray scattering (SAXS) shows that the formed nanostructures are the result of the reaction-induced microphase separation (RIMS) mechanism. The activated monomer mechanism (AM) proceeded faster than the activated chain end mechanism (ACE) leading to a preferred integration of PCL-diol into the epoxy network so that segregation is partially inhibited, especially for small polyester contents. It was shown that esterified PCL did not react with the epoxy resin so that it was forced to segregate greatly in the form of spherulitic domains. This was shown by STEM imaging, by the occurrence of crystallinity, as well as by enhanced glass transition temperatures compared to its comparable copolymers.

---

Introduction

Thermosetting epoxy resins exhibit good mechanical and thermal properties as well as excellent chemical resistance.¹ For this reason, epoxy resins have been established for technical applications such as printing inks, adhesives, electrical insulators, composites, and coatings. Despite these outstanding properties, thermosetting epoxides are brittle. To overcome this disadvantage, there are several toughening and structuring methods described in the literature like rubber inclusion,² incorporation of inorganic nanoparticles,^3,4 nano-structuring by block copolymers and block ionomers,^5–16 reaction with hyperbranched polymers,^17–19 plasticization by for example a partially crystalline polyester,^20–23 and recently also by twin copolymerization.²⁴ Some of these methods are based on widening of the three dimensional epoxy network resulting in a lower network density. Therefore, one has to consider the trade-off between toughening on the one hand and the glass transition temperature T_g as well as moduli on the other hand. To minimize the disadvantages of toughening, a nanoscopic and heterogeneous morphology has been reported to be crucial as observed by the naturally occurring materials like spider silk,^25,26 nacre,^27,28 and human enamel²⁹ which exhibit superior mechanical properties. To obtain nanosized structures, self-assembly and reaction-induced microphase separation (RIMS) has been pointed out as feasible concepts.^12,30–37 The subtle distinction between these two concepts is the miscibility of components before and after the polymerization reaction. Self-assembled mixtures perform phase separation before curing whereas samples appearing homogenous before curing, segregate into separated phases during the polymerization reaction.

In case of cationically polymerized epoxy resins, toughening with polyester polyols or alcohols was observed to be an effective method leading to a partially segregated structure.^20,22,38 Thereby, the polyol is integrated into the polymer network due to a chain transfer reaction, named activated monomer mechanism (AM).^38–40 Thus, a stronger phase separation might occur by prevention of the AM mechanism due to blocking the reactive hydroxyl group of the partially crystalline polyester polyols. In general, the AM reaction accelerates the polymerization of the epoxy resin due to reaction between the epoxy group and the hydroxyl group and is therefore faster than the ACE as evidenced in the literature.^41,42 To prevent this kind of reaction in the early stage of polymerization progress, the hydroxyl groups of the polyester polyol can be esterified which make them unreactive towards the AM mechanism and potentially more suitable for phase separation processes.

In this work we studied the reaction behavior as well as the phase separation mechanism of a cationically polymerized epoxy resin in presence of both a reactive poly(ε-caprolactone) bearing hydroxyl end groups and a unreactive poly(ε-caprolactone)diester. Furthermore, we investigated the morphological differences of the resulting copolymers and polymer alloys.

Experimental

Materials

All chemicals were used as received from commercial suppliers. 3,4-Epoxycyclohexyl-3′,4′-epoxycyclohexane carboxylate (1) (Omnilane OC1005) was purchased from IGM resins (Krefeld, Germany). Phosphotungstic acid hydrate and acetic anhydride (>98.0%) were obtained from Sigma-Aldrich (Steinheim, Germany). Dichloromethane was purchased from Merck (Darmstadt, Germany). Poly(ε-caprolactone) (2) bearing hydroxyl end groups (PCL, Capa™ 2402, M_n = 4000 g mol⁻¹) was supplied by Perstorp (Warrington, UK). The thermolatent initiator benzyl tetrahydrothiophenium hexafluoroantimonate was prepared according to Endo et al.⁴³

Physico-chemical characterization

¹H NMR spectroscopy. ¹H NMR spectra were recorded with a Bruker DPX-200 spectrometer (200 MHz). Tetramethylsilane was used as external standard. The spectra were recorded in CDCl₃ at room temperature.

Infrared spectroscopy. Infrared (IR) spectra were recorded with a Bruker Equinox 55 FTIR spectrometer in attenuated total reflexion (ATR). The spectrometer was equipped with a Golden Gate cell with a resolution of 4 cm⁻¹ (32 scans).

Differential scanning calorimetry. Differential scanning calorimetry (DSC) was carried out in a sealed pan with a Discovery DSC from TA Instruments. A temperature range from 20 °C to 250 °C was used for reactivity investigations (heating rate of 10 K min⁻¹). In case of analysis on crystallization behavior, the following cyclic temperature profile was used (heating rate 5 K min⁻¹): heating from −50 °C to 150 °C, holding for 5 minutes, cooling to −20 °C, holding for 5 minutes, and at last heating again to 150 °C.

Dynamic mechanical analysis. Dynamic mechanical analysis (DMA) was carried out with a DMA 2980 from TA Instruments to determine the glass transition temperature of the polymer samples. A heating rate of 2 K min⁻¹ was used in the temperature range from −150 °C to 270 °C.

Polarization microscopy. Polarization microscopy (PM) was performed with a Zeiss optical microscope Axio Imager.M1 with a polarization filter. The images were recorded in 20 times magnification.

Scanning electron microscopy. Scanning electron microscopy (SEM) was carried out with a LEO1530 Gemini microscope from Zeiss. For this, a fracture surface was prepared by breaking the sample at cryo conditions.

Transmission electron microscopy. Transmission electron microscopy (TEM) was performed with a FEI Tecnai F-20 S-TWIN (Hillsboro, Oregon, USA). The measurements were carried out at 200 kV in scanning mode (STEM) with a spatial resolution of about 1 nm. STEM images were recorded with a Fischione Model 3000 ADF detector in high-angle annular-dark-field (HAADF STEM) mode. The samples were microtomed with a Leica Ultracut UCT at cryo conditions. Furthermore, the sliced samples were stained with phosphotungstic acid according to the literature.²² Energy dispersive X-ray spectroscopy (EDX) was performed with an EDAX r-TEM-EDX-Detector with an energy resolution of 136 eV.

Size exclusion chromatography. Size exclusion chromatography (SEC) was measured at 23 °C with a 1260 Infinity Refractive Index Detector (Agilent Technologies). The used columns were a PLgel 5 μm Guard (50 × 7.5 mm; Agilent Technologies), PLgel 5 μm MIXED-C (7.5 × 300 mm; Agilent Technologies), and PLgel 10⁴ Å (7.5 × 300 mm; Agilent Technologies). PMMA was used as standard for calibration and tetrahydrofuran was chosen as solvent with a flow rate of 1 mL min⁻¹.

Small-angle X-ray scattering. Small-angle X-ray scattering (SAXS) measurements were performed with GALAXI (Gallium Anode Low-Angle X-ray Instrument) at Jülich Centre for Neutron Science (Germany). GaK_α X-rays with a wavelength λ = 0.134 nm were generated by a Bruker AXS Metaljet X-ray source with parabolic Montel mirrors. The Pilatus 1M from Dectris was used as two-dimensional small-angle detector with a signal to noise ratio of 10⁶. The measurements were carried out for two different detector distances and the results were illustrated in one graphic. The beam path is fully evacuated between the X-ray source and the detector which means that also the samples were measured in vacuum. Solid polymer samples with a thickness of 1 mm and uncured mixtures in 1.7 mm thick capillaries were measured at a temperature of 22 °C. The collected data were radially averaged and normalized to the intensity of the transmitted beam. Scattering from empty sample holder or empty capillary was subtracted.

X-ray powder diffraction. X-ray powder diffraction (XRPD) was carried out with a D5000 diffractometer from Siemens with a Bragg–Brentano geometry. CuK_α X-ray radiation with a wavelength of λ = 0.1541 nm was used. A diffraction angle range 10–30° (2θ) was chosen.

Determination of gel yield. The gel content was determined by Soxhlet extraction. The cut polymer specimen was extracted with dichloromethane for 24 hours (and also for 72 hours) and dried in a vacuum oven (10⁻² mbar) at 65 °C to the point of weight constancy.

Syntheses

Poly(ε-caprolactone) with ester end groups (PCL-diester (3)). 34.01 g (8.5 mmol) PCL-diol (2) was placed into a round bottom flask with a reflux condenser and heated to a temperature of 65 °C to melt the PCL. Under stirring, an excess of 1.70 mL (18 mmol) acetic anhydride was slowly dropped into the melt. Afterwards, the mixture was heated at 106 °C for 3.5 hours. At last, the formed acetic acid was removed under vacuum (10⁻² mbar) at 70 °C. The conversion of >93% (detection limit) was determined by ¹H NMR spectroscopy.

¹H NMR (CDCl₃): δ [ppm] = 4.05 (t, 2H, CH₂–O–); 2.30 (t, 2H, CH₂–COO–); 2.03 (s, 3H, CH₃ end group); 1.57–1.71 (m, 4H, CH₂); 1.29–1.44 (m, 2H, CH₂).

Polymerization of epoxy resin (1) with PCL-diol (2) or PCL-diester (3). The cationic polymerizations of the cycloaliphatic epoxy resin (1) with PCL-diol (2) or PCL-diester (3), respectively, were carried out with the thermolatent initiator benzyl tetrahydrothiophenium hexafluoroantimonate. The composition of epoxy resin (1) and PCL species was varied in 10 wt% intervals from pure epoxide to a content of 50 wt% PCL. First, one weight percent of the initiator was dissolved in epoxy resin (1) under magnetic stirring. Then the epoxy resin/initiator solution was filled into a 60 mL plastic cup together with the respective PCL species. The mixture with a mass of 50 g was homogenized in an oil bath by heating at 85 °C for 30 minutes. Subsequently, the homogenized mixtures were transferred into aluminum molds with dimensions of 4 × 1 × 0.3 cm³ for DMA specimens. The polymerization was carried out in a preheated oven under following heating profile: 0.5 hours at 75 °C, one hour at 110 °C, one hour at 125 °C, and one hour at 145 °C. After cooling to 24 ± 2 °C, the specimens were removed from the molds and stored for one month at ambient conditions.

Results and discussion

Copolymerization and polymer alloy formation of a cycloaliphatic epoxy resin and poly(ε-caprolactone)

The morphological adjustment of thermosets is a feasible concept for enhancing mechanical properties. Especially nanoscopic phase separation has emerged as the key for a successful improvement of material properties.^{4,8–15,25–29} In case of cationically polymerized epoxy resins, segregation can be achieved by addition of certain non-reactive additives or by reaction with components like alcohols. Toughening with polyesters bearing alcohol functions, especially poly(ε-caprolactone), has been described as effective method resulted in a partially phase separated morphology.^20–22 To further enhance segregation, especially nanoscopic phase separation, the reactive hydroxyl end groups of poly(ε-caprolactone) are blocked by esterification so that the activated monomer mechanism (AM) as well as a potential covalent linkage with the epoxy network will be prevented.

The cationic polymerization of the cycloaliphatic epoxy resin (1) was investigated each with poly(ε-caprolactone) bearing hydroxyl end groups (PCL-diol (2)) and poly(ε-caprolactone) featuring non-reactive esterified end groups (PCL-diester (3)) in concentrations of 10, 20, 30, 40, and 50 wt% forming copolymers and polymer alloys as depicted in Scheme 1. PCL was chosen with a molecular weight M_n of ca. 4000 g mol⁻¹ and it was expected that the high degree of crystallinity enhance formation of a heterogeneous structure. The used partially crystalline PCL-diol (2) has a M_n of 4316 g mol⁻¹ as determined by ¹H NMR spectroscopy which corresponds to the information of the provider of 4000 g mol⁻¹. The PCL-diester (3) species exhibits a M_n of 4628 g mol⁻¹ which was also examined by ¹H NMR spectroscopy. For comparison, M_n was also determined by size exclusion chromatography (SEC). PCL-diol (2) shows a M_n of 8942 g mol⁻¹ with a narrow polydispersion of 1.25 and in case of PCL-diester (3) the M_n is 9017 g mol⁻¹ with a polydispersion of 1.26. The discrepancy in molecular weight between SEC and ¹H NMR spectroscopy results due to the used PMMA standard for SEC measurements of PCL. The values determined with both methods for the PCL-diol (2) and the PCL-diester (3) reveal that the polymer was not degraded during esterification of the hydroxyl end groups.

Scheme 1 Reaction scheme of the cycloaliphatic epoxide (1) each with PCL-diol (2) and PCL-diester (3) in presence of a cationic initiator leading to formation of a copolymer and a polymer alloy, respectively.

The polymerization reaction was investigated with differential scanning calorimetry (DSC). The temperature of maximum exothermic heat flow T_max serves as measure for the curing temperature as designated in Fig. 1 and Table 1. The polymerization reaction of the pure epoxy resin (1) shows a single signal with a T_max value of 140 °C whereas the mixtures of the epoxy resin (1) and the reactive PCL-diol (2) exhibit a second exothermic signal or a shoulder, respectively. The first exothermic signal appears at lower temperature (120–132 °C) compared to the neat epoxy resin (1) and the second one at higher temperature (146–170 °C). The decreased polymerization temperature of the first exothermic signal might be due to the activated monomer mechanism (AM) which accelerates the polymerization reaction.^41,42 This acceleration is also supported by the onset temperature T_onset as seen in Table 1. In this case, the hydroxyl groups of the PCL-diol (2) react with the epoxy resin forming a copolymer and simultaneously release a proton which initiates a new polymerization. The second exothermic signal at higher temperature might be the result of the activated chain end mechanism (ACE) which proceeds more slowly than the AM mechanism. Furthermore, the T_max values increase with increasing PCL-diol (2) content which is in agreement with the observation that this mechanism is retarded by dilution of the epoxide. At last, the reaction of an epoxy resin (1) in presence of a non-reactive PCL-diester (3) confirms the results described before due to the appearance of only one exothermic signal which is shifted to higher temperature (Table 1). Fig. 1 shows exemplarily the DSC of 30 wt% PCL-diester (3) containing sample revealing a T_max value of 151 °C indicating the polymerization of the neat epoxy resin which is comparable to the second signal of the epoxide polymerization with PCL-diol (2). Thus, the polymerization of an epoxy resin (1) in presence of PCL-diester (3) leads to a polymer alloy. Furthermore, the reaction enthalpies Δ_RH (Table 1) decrease linearly for both systems with nearly the same values for PCL-diol (2) and PCL-diester (3). This shows that the reaction enthalpy of both epoxide consuming reactions is almost identical.

Fig. 1 DSC thermograms for the cationic polymerization of epoxide (1)/PCL mixtures in different composition. The depicted temperatures (T_max) are the signal maxima of DSC curves.

Table 1 DSC data of the reaction behaviour of epoxide (1) with PCL-diol (2) and PCL-diester (3), respectively. The reaction enthalpy Δ_RH, the onset temperature T_onset, and the temperature auf maximum heat flow T_max are given for the copolymer and polymer alloy formation

Sample	PCL-diol (2)			PCL-diester (3)
	ΔRH [J g^−1]	Tonset [°C]	Tmax [°C]	ΔRH [J g^−1]	Tonset [°C]	Tmax [°C]
100/0	500	120	140	500	120	140
90/10	472	116	132/146	450	121	138
80/20	420	107	120/149	415	124	142
70/30	374	108	129/147/156	364	136	151
60/40	304	110	128/158	291	135	150
50/50	259	112	131/170	261	138	156

---

Infrared (IR) spectra were recorded to confirm a complete polymerization reaction. This was shown by the disappearance of the epoxide signals at 898 cm⁻¹ and 789 cm⁻¹ for all samples with exception of the pure epoxy resin which exhibits a very small amount of residual epoxy functions. This is most likely due to the very high glass transition temperature. Extraction experiments of the copolymers and polymer alloys were carried out to determine the gel yield of the composites to support the DSC data of reaction behaviour. The geld yields for the epoxide (1)/PCL-diol (2) composites are significantly higher compared to their PCL-diester (3) counterparts (Fig. 2). The difference results from the integration of the PCL-diol (2) into the epoxy network due to the activated monomer mechanism whereas the PCL-diester (3) could not react by this way. However, the gel yields of the PCL-diester (3) containing composites are much higher than expected if all non-reacted components were extracted from the thermosetting network. Nevertheless, there are no significant differences for extending the extraction time from one day to three days. Apart from small amounts of residual hydroxyl end groups which are able to be integrated into the epoxy network, this observation leads to the assumption that PCL-diester (3) is also able to be integrated to a small proportion into the thermoset for example by transesterification reaction. Transesterification in thermosetting epoxy resins has already been observed by Chen et al., Leibler et al., and Yang et al.^45–49 But also the entrapment of small PCL-diester (3) amounts into the epoxy network cannot be excluded totally, although increasing extraction time does not increase the amount of extractables. However, the fraction of PCL-diester (3) which is integrated into the network structure by exchange reaction is relatively low compared to the reactive PCL (2) species.

Fig. 2 Gel yield of cationically polymerized epoxide (1)/PCL polymers in dependence on the epoxide content. The black curve shows the polymers with PCL-diol (2) and the gray curve illustrates the polymers containing PCL-diester (3).

Based on different polymerization mechanisms, the samples containing PCL-diol (2) and PCL-diester (3) have different appearances as shown in Fig. 3. All PCL-diol (2) compounded polymers appear transparent whereas the ones with PCL-diester (3) occur opaque down to a content of 20 wt% in the epoxy resin. This opacity results from crystallization which is evidenced by the results of cyclic DSC measurements listed in Table 2. The first heating was carried out after storage for one month in order to obtain a crystallization as complete as possible. The degree of crystallinity was determined by the melting enthalpy according to the literature.⁵⁰ Only opaque specimens exhibit a distinct endothermic signal indicating crystallinity. This concerns samples with a minimum of 20 wt% of PCL-diester (3). The degree of crystallinity as well as the melting temperature decreased with reduced PCL-diester (3) content due to higher dilution of PCL and smaller crystallites in the epoxy resin as observed by polarization microscopy (discussed later). The second heating step is unaffected from a long crystallization process and gives information about crystallinity which is formed during cooling with 5 K min⁻¹ in the DSC cooling run. The sample containing 20 wt% PCL-diester (3) crystallized too slow within the cooling process caused by limited chain segment mobility due to high network density of the matrix. Thus, there is no melting signal observed under given conditions. A fast cooling process leads both to a decrease in the degree of crystallinity and in the melting temperature compared to the first heating step. For this, the formation of crystallinity is to some extent hindered by faster solidification of the PCL phase in a less ordered state and furthermore the melting temperature is reduced due to occurrence of smaller crystallites. The composites with PCL-diol (2) exhibit no crystallinity according to DSC measurements because of the integration of this reactive polyester into the epoxy network as shown by extraction experiments so that phase separation is limited. Crystallization as one driving force of phase separation is obviously higher for PCL-diester (3) compared to PCL-diol (2) since there is no covalent bonding or strong solvation by the resin. Furthermore, the examination of the glass transition temperature T_g by the loss factor of dynamic mechanical analysis (DMA) emphasizes the stronger phase separation of PCL-diester (3) composites compared to the PCL-diol (2) ones as depicted in Fig. 4. It was observed that the glass transition temperature increases linearly with rising epoxide (1) content due to formation of higher network density and a reduction of network widening by both kind of poly(ε-caprolactone). The distortion of the epoxy network is considerably enhanced by PCL-diol (2) compared to its non-reactive counterpart (3). This is caused by chain transfer reaction (AM) of the alcohol functions of polyester (2) with the epoxy resin (1), as observed by DSC, leading to lower network density due to direct integration into the three dimensional network and therefore to a greater miscibility and only partial phase separation. In case of PCL-diester (3), the influence on the epoxy network formation is not as strong as for PCL-diol (2) because the end group blocked polyester (3) is not able to react with the epoxy resin so that the ability of segregation is enhanced. For this reason, the epoxy resin (1) only reacts by the activated chain end mechanism (ACE) forming a more segregated structure with higher T_g values for PCL-diester (3) compared to the PCL-diol (2) containing polymers.

Fig. 3 Occurrence of epoxide (1)/PCL composites with 10, 20, 30, 40, and 50 wt% of PCL. The upper samples are composites containing PCL-diol (2) whereas the lower composites contain the non-reactive PCL-diester (3).

Table 2 Melting enthalpy Δ_mH and melting temperature T_m examined by DSC for samples containing epoxide (1) and PCL-diester (3). The degree of crystallinity X_c was determined according to the literature.⁵⁰ The first heating step by DSC is characteristic for the slow crystallization process of the samples which were all subject of the same conditions whereas the second heating step gives information about the crystallization behavior during a fast cooling process

Sample	1. Heating step			2. Heating step
	Tm [°C]	ΔmH [J g^−1]	Xc [%]	Tm [°C]	ΔmH [J g^−1]	Xc [%]
100/0	—	—	—	—	—	—
90/10	—	—	—	—	—	—
80/20	46	2	2	—	—	—
70/30	52	8	6	45	6	5
60/40	52	12	9	45	8	6
50/50	53	29	22	47	19	14

---

Fig. 4 Dependence of the glass transition temperature of epoxide (1) composites each with PCL-diol (2) (black) and PCL-diester (3) (gray) on the epoxy content.

Mechanistic study on phase separation behaviour by X-ray scattering

Structuring of thermosets by both self-assembly and by reaction-induced phase separation (RIPS) have been shown as feasible concepts.^{12,30–37,44} To get insight into phase separation mechanism as well as information about the nanostructure, small-angle X-ray scattering (SAXS) was performed.

SAXS measurements were carried out at a temperature of 22 °C. The samples are the pure epoxy resin (1), mixtures of epoxide (1) each with PCL-diol (2) and PCL-diester (3) and in each case with 10 and 50 wt% content. All samples were measured in the uncured (liquid) state as well as in the cured (solid) state. Furthermore, uncured samples containing 10 wt% polyester were measured in the crystalline state and in the molten one. This was not possible for samples containing 50 wt% PCL due to fast crystallization at the given temperature. Fig. 5 illustrates the scattering intensity in dependence on the scattering vector q. Unfortunately, a quantitative SAXS analysis of the scattering profiles respective form factors could not be applied because the signals and shoulders appeared very broad. Thus, only a qualitative analysis could be carried out to get information about phase separation mechanism and mean domain distances. The pure uncured epoxy resin (1) shows no correlation peak as expected for completely amorphous and homogeneous resins. Compared to this, the uncured mixtures with 10 and 50 wt% polyester show a signal with a scattering maximum at around 0.4 nm⁻¹ which corresponds to an average domain distance of 16 nm. As shown by opacity and DSC measurements of crystalline samples, there are additionally macroscopic domains. After melting of the 10 wt% containing, uncured polyester composites, the samples turn transparent and the correlation peak at around 0.4 nm⁻¹ disappeared. This indicates homogeneity and complete miscibility of the resin and PCL in the molten state. Furthermore, self-assembly as phase separation mechanism can be excluded for cured samples as the samples pass a temperature profile during polymerization which exceeds the melting point of PCL before the curing reaction takes place.

Fig. 5 SAXS pattern of uncured (black curves: A–E and I and J) and cured (gray curves: F–H and K and L) mixtures of epoxy resin (1) each with PCL-diol (2) and PCL-diester (3) at a polyester content of 10 and 50 wt%. Graphic (1): (A) 10 wt% PCL-diol in crystallized state, (B) 10 wt% of PCL-diester in crystallized state, (C) 10 wt% PCL-diol in molten state, (D) 10 wt% PCL-diester in molten state, (E) pure epoxy resin, (F) cured epoxy resin, (G) 10 wt% PCL-diol (cured), (H) 10 wt% PCL-diester (cured). (Graphic 2): (I) 50 wt% PCL-diol (cured), (J) 50 wt% PCL-diester (cured), (K) 50 wt% PCL-diol (cured), and (L) 50 wt% PCL-diester (cured). The scattering vector is q = (4π/λ) sin (θ/2) with the wavelength λ = 0.134 nm and the scattering angle θ. For a better visualization, the curves are staggered. Both graphics (1) and (2) show the same arbitrary units.

After polymerization, both the pure epoxy resin (1) and the copolymer with 10 wt% PCL-diol (2) exhibit no correlation peak and therefore no heterogeneity. For the latter, this means no segregation during polymerization and complete miscibility in the cured state. In case of the polymer alloy containing 10 wt% PCL-diester (3), a distinct broad shoulder appeared in the range of 0.2–0.3 nm⁻¹ which indicates a heterogeneous morphology with average domain distances of around 21–31 nm. Obviously, already small amounts of non-reactive PCL-diester (3) suffice to form a heterogeneous nanostructure resulting from segregation of unbound PCL. The blends containing 50 wt% polyester exhibit very broad signals evidencing also a heterogeneous morphology. Furthermore, the copolymer with 50 wt% PCL-diol (2) shows a shoulder at 0.4–0.9 nm⁻¹ which corresponds to average domain distances of around 7–16 nm. This copolymer reveals nanostructures despite the reaction between the PCL-diol (2) and the epoxy resin (1) because of a high amount of non-reacted polyester as shown by gel yield measurements. These unbound polyester should be able to segregate preferentially leading to nanostructures. On the other hand, the polymer alloy containing 50 wt% PCL-diester (3) reveals a shoulder at 0.4–0.7 nm⁻¹ but also a strong increase of the signal at 0.2–0.3 nm⁻¹ indicating average domain distances of 9–31 nm. In case of the polymer alloys, the domain distances decrease with increasing amount of PCL-diester (3) as expected for the presence of larger domains and more material able to segregate.

PCL is distributed as fine crystallites in the uncured epoxy resin for all examined concentrations. After melting the crystalline PCL phase, a homogeneous morphology results without any nanostructure. This subsequently turns into heterogeneous morphology during polymerization. In conclusion, the heterogeneous morphology of the polymers results from reaction-induced phase separation (RIPS), especially reaction-induced microphase separation (RIMS), leading to a partially ordered or short range ordered morphology. A preservation of a structure preformed in the liquid system during polymerization can be excluded.

Additional to SAXS measurements, X-ray powder diffraction was performed in a diffraction angle range of 10–30° (2θ) (Fig. 6) (WAXS). Only the polymer alloy containing 50 wt% PCL reveals sharp signals at 21.4°, 21.9°, and 23.7° which are typical for crystalline PCL as described in the literature.⁵¹ This shows that the PCL in the epoxy matrix crystallizes in the same structure as pure PCL. The absence of distinct diffraction peaks in the diagrams shows that the domains detected in part of the samples by SAXS are amorphous. This is in agreement with the DSC results of the respective samples. Furthermore, the amorphous halo of the epoxy resin shifts slightly to higher diffraction angles when 10 wt% PCL-diol (2) or PCL-diester (3) are part of the polymer due to disturbance of the three-dimensional network. This is more pronounced when the polyester content increases as seen in the X-ray diffraction pattern (Fig. 6) which is in agreement with the observation in literature for shape memory polymers based on an epoxy resin and more than 70 wt% PCL.⁵¹ Additionally, the amorphous halo broadens in case of 50 wt% PCL confirming a network change of the epoxy matrix.

Fig. 6 X-ray diffraction pattern of the pure epoxy resin (1), the copolymers containing 10 wt% (2) and 50 wt% PCL (4), and the polymer alloys containing 10 wt% (3) and 50 wt% PCL (5). The intensities are given in arbitrary units (a.u.) and the diffraction profiles are shifted along the y-axis for a better visualization.

Morphological investigation

The thermally initiated cationic polymerization of epoxy resin (1) with PCL follows the reaction-induced phase separation (RIPS) mechanism as shown by SAXS. To identify the structure of the polymeric samples in more detail, scanning transmission electron microscopy (STEM) was carried out. Phosphotungstic acid was used as negative stain for contrasting the samples according to the literature.²² This means, there is a strong preference for staining the epoxy resin (bright areas in the STEM image) in contrast to PCL (dark positions in the STEM image) as was measured by energy dispersive X-ray spectroscopy (EDX). The STEM images of copolymers and polymer alloys each with 10 and 50 wt% PCL are shown in Fig. 7. The transparent looking copolymer containing 10 wt% PCL-diol (2) (Fig. 7: image A) appears homogeneous as expected for well miscible samples resulting in a uniform contrast due to weak or homogeneous staining by phosphotungstic acid (the stripes are cutting artefacts due to different sample thickness). The homogeneous morphology resulted due to a very high degree of integrated PCL by AM mechanism as shown by a gel content of 98% (Fig. 2) wherefore segregation seems to be strongly inhibited. A content of 50 wt% PCL-diol (2) leads to a wormlike nanostructured epoxy resin with a domain size of around 10–30 nm (Fig. 7: image B). As described before, also polymers with high amounts of PCL-diol (2) lead to segregation which might preferentially the unbound PCL. The polymer alloys consisting of 10 and 50 wt% PCL-diester (3) reveal sphere like nanostructures of PCL with 10–20 nm (Fig. 7: image C) and 10–60 nm (Fig. 7: image D) domain sizes, respectively. With increasing PCL amount these structures grow only in size but do not change their spherical occurrence. The results observed by STEM are in good agreement with the SAXS measurements as the average nanodomain distances in the epoxy resin are increasing in the following way: 50 wt% PCL-diol (2) (ca. 11 nm) < 50 wt% PCL-diester (3) (ca. 11–28 nm) < 10 wt% PCL-diester (3) (ca. 22–33 nm).

Fig. 7 STEM images of cured epoxide (1)/PCL copolymers and polymer alloys: (A) 10 wt% PCL-diol (2), (B) 50 wt% PCL-diol (2), (C) 10 wt% PCL-diester (3), and (D) 50 wt% PCL-diester (3). The samples are stained with phosphotungstic acid. Scale bar of image (A) is 200 nm and the others have a scale bar of 100 nm.

Polarization microscopy (PM) was performed to identify crystallinity and crystal size due to birefringence. Furthermore, PM measurements served as elucidation of the macroscopic morphology of epoxy (1)/PCL-diester (3) composites and complemented the observations by simple visual appearance of the samples (opacity), by WAXS, as well as results by cyclic DSC measurements (melting behavior). Fig. 8 illustrates the PM images in 20 times magnification. The pure epoxy resin (1) (Fig. 8: image A) and the other transparent looking samples show no birefringence as expected for samples without crystallinity. In case of the sample with 20 wt% PCL-diester (3), the crystals are too small to be observable by PM, but crystallinity was detected by DSC as mentioned before. Large spherulites (Maltese cross) were observed for the polymer alloy containing 50 wt% PCL-diester (3) due to strong segregation processes (Fig. 8: image D). The size of spherulites has a maximum of around 100 μm. The crystal size diminishes drastically by decreasing the content of unreactive PCL (3) (Fig. 8: image B and C). In summary, PCL-diester (3) based polymer alloys exhibit macroscopic demixing in the form of spherulites of increasing size with increasing PCL amount starting with about 30 wt%. In contrast, PCL-diol (2) based copolymers are free of crystallinity in the applied concentration range because of well miscibility in the cured state. These observations support the suggestion of a stronger ability for phase separation of the non-reactive PCL (3) within the epoxide matrix showing large crystal structures next to nanodomains.

Fig. 8 Polarization microscopy images of the pure epoxy resin (1) (A) and its polymer alloys with PCL-diester (3) containing 30 wt% PCL (B), 40 wt% PCL (C), and 50 wt% PCL (D). The images were recorded in 20 times magnification.

At last, scanning electron microscopy (SEM) was carried out showing differences in fracture surfaces as depicted in Fig. 9. The SEM micrographs show the fracture surfaces of composites consisting of epoxide (1) each with PCL-diol (2) and PCL-diester (3). The fracture surface of the pure epoxy resin (1) (not shown here) and its composites with 10 wt% of both the reactive and the non-reactive PCL look homogeneous and exhibit no significant differences (Fig. 9A and D). In case of PCL-diol (2) blended polymers, the surface roughness only slightly increases by enhancing the polyester content from 10 wt% to 50 wt% (Fig. 9A–C). On the other hand, polymer alloys containing PCL-diester (3) reveal a strong increase in surface roughness by enhancing the polyester content (Fig. 9D and E). The higher the content of PCL-diester (3) in the respective epoxy based polymer alloy the more heterogeneous is the formed morphology. This is accompanied by an increase in crystallinity and phase separation leading to the observed raising roughness of the fracture surfaces.

Fig. 9 SEM images of epoxide (1)/PCL blends fracture surfaces. The samples are broken at cryo conditions. The copolymers of epoxy resin (1) and PCL-diol (2) are illustrated in (A) with 10 wt% PCL, in (B) with 30 wt% PCL, and in (C) with 50 wt% PCL. The images of the polymer alloys consisting of epoxide (1) and PCL-diester (3) are shown in (D) with 10 wt% PCL, in (E) with 30 wt% PCL, and in (F) with 50 wt%.

Conclusions

The cationic polymerization of a cycloaliphatic epoxy resin with a reactive as well as a non-reactive poly(ε-caprolactone) was examined. In case of a heterogeneous polymer, the domains are formed by the reaction-induced (micro-)phase separation mechanism as observed by SAXS. Furthermore, the DSC study on reaction behavior point out that the PCL-diol reacts according to the activated monomer mechanism (AM) forming a copolymer whereas the non-reactive PCL-diester does not react with the epoxy resin resulting in a polymer alloy. This was additionally supported by gel content determinations. As evidenced in the literature,^41,42 we could confirm that the AM mechanism proceeded faster than the ACE mechanism. This kinetic difference is responsible for a high degree of PCL-diol integration into the three dimensional network and moreover for a distinct miscibility of these two components in the cured state as indicated by a drastic decrease in glass transition temperature, by the absence of crystallinity, and smooth fracture surfaces. STEM images and SAXS measurements of this copolymer show a homogeneous morphology for small contents of PCL but exhibit a wormlike segregated nanostructure for a high content of 50 wt%. To simply modify the morphology of such epoxy based polymers, manipulation of the reaction mechanisms, especially the activated monomer mechanism (AM), is beneficial to force the polyester into a separated phase by RIPS for huge domains and by RIMS in case of small domains. The exclusion of the AM mechanism by esterification of the reactive end groups of PCL already lead to heterogeneity for small polyester contents. STEM images reveal sphere like nanodomains up to 50 wt% PCL-diester content. Due to a stronger segregation during polymerization and the absence of the chain transfer reaction, the polymer alloys exhibit significantly higher glass transitions temperatures than their copolymer counterparts. Furthermore, the polymer alloys show crystallinity up to an epoxy resin content of 80 wt% as consequence of demixing during the RIPS mechanism, respectively. Further work will focus on structure–property relations respective mechanical and adhesion behaviour for both polymer types.

Acknowledgements

Financial support by Deutsche Forschungsgemeinschaft DFG, grant number HA 2420/14-1 is gratefully acknowledged.

Notes and references

1. J.-P. Pascault and R. J. J. Williams, Epoxy Polymers: New Materials and Innovations, Wiley-VCH, Weinheim, 2010.
2. C. B. Bucknall and T. Yoshii, Br. Polym. J., 1978, 10, 53.
3. J. Ma, M.-S. Mo, X.-S. Du, P. Rosso, K. Friedrich and H.-C. Kuan, Polymer, 2008, 49, 3510.
4. A. Hartwig, A. Lühring and J. Trautmann, Macromol. Mater. Eng., 2009, 294, 363.
5. L. Könczöl, W. Döll, U. Buchholz and R. Mülhaupt, J. Appl. Polym. Sci., 1994, 54, 815.
6. J. Liu, H.-S. Sue, Z. J. Thompson, F. S. Bates, M. Dettloff, G. Jacob, N. Verghese and H. Pham, Macromolecules, 2008, 41, 7616.
7. J. M. Dean, R. B. Grubbs, W. Saad, R. F. Cook and F. S. Bates, J. Polym. Sci., Part A: Polym. Chem., 2003, 41, 2444.
8. J. Liu, Z. J. Thompson, H.-J. Sue, F. S. Bates, M. A. Hillmyer, M. Dettloff, G. Jacob, N. Verghese and H. Pham, Macromolecules, 2010, 43, 7238.
9. R. Yu and S. Zheng, Macromolecules, 2011, 44, 8546.
10. S. Wu, Q. Guo, M. Kraska, B. Stühn and Y.-W. Mai, Macromolecules, 2013, 46, 8190.
11. S. Wu, Q. Guo, S. Peng, N. Hameed, M. Kraska, B. Stühn and Y.-W. Mai, Macromolecules, 2012, 45, 3829.
12. W. Fan, L. Wang and S. Zheng, Macromolecules, 2010, 43, 10600.
13. X. Yang, F. Yi, Z. Xin and S. Zheng, Polymer, 2009, 50, 4089.
14. J. M. Dean, N. E. Verghese, H. Q. Pham and F. S. Bates, Macromolecules, 2003, 36, 9267.
15. Y. S. Thio, J. Wu and F. S. Bates, Macromolecules, 2006, 39, 7187.
16. J. Mijovic, M. Shen, J. W. Sy and I. Mondragon, Macromolecules, 2000, 33, 5235.
17. L. Boogh, B. Pettersson and J.-A. E. Manson, Polymer, 1999, 40, 2249.
18. M. Sangermano, G. Malucelli, R. Bongiovanni, A. Priola, A. Harden and N. Rehnberg, Polym. Eng. Sci., 2003, 43, 1460.
19. M. Sangermano, A. Priola, G. Malucelli, R. Bongiovanni, A. Quaglia, B. Voit and A. Ziemer, Macromol. Mater. Eng., 2004, 289, 442.
20. H. Lützen, P. Bitomsky, K. Rezwan and A. Hartwig, Eur. Polym. J., 2013, 49(1), 167.
21. H. Lützen, T. M. Gesing and A. Hartwig, React. Funct. Polym., 2013, 73, 1038.
22. H. Lützen and A. Hartwig, Macromol. Mater. Eng., 2013, 289, 1275.
23. K. Koschek, Composites, Part A, 2015, 69, 21.
24. A. Arnebold, O. Schorsch, J. Beckmann and A. Hartwig, React. Funct. Polym., 2013, 73, 1625.
25. M. Cetinkaya, S. Xiao, B. Markert, W. Stacklies and F. Gräter, Biophys. J., 2011, 100, 1298.
26. A. R. Studart, Adv. Mater., 2012, 24, 5024.
27. A. Walther, I. Bjurhager, J.-M. Malho, J. Pere, J. Ruokolainen, L. A. Berglund and O. Ikkala, Nano Lett., 2010, 10, 2742.
28. Y. Oaki and H. Imai, Angew. Chem., Int. Ed., 2005, 44, 6571.
29. F.-Z. Cui and J. Ge, J. Tissue Eng. Regener. Med., 2007, 1, 185.
30. F. Meng, Z. Xu and S. Zheng, Macromolecules, 2008, 41, 1411.
31. F. Meng, S. Zheng, W. Zhang, H. Li and Q. Liang, Macromolecules, 2006, 39, 711.
32. F. Meng, S. Zheng, H. Li, Q. Liang and T. Liu, Macromolecules, 2006, 39, 5072.
33. P. M. Lipic, F. S. Bates and M. A. Hillmyer, J. Am. Chem. Soc., 1998, 120, 8963.
34. M. A. Hillmyer, P. M. Lipic, D. A. Hajduk, K. Almdal and F. S. Bates, J. Am. Chem. Soc., 1997, 119, 2749.
35. H. Garate, S. Goyanes and N. B. D'Accorso, Macromolecules, 2014, 47, 7416.
36. S. Goossens, B. Goderis and G. Groeninckx, Macromolecules, 2006, 39, 2953.
37. G. V. Poel, S. Goossens, B. Goderis and G. Groeninckx, Polymer, 2005, 46, 10758.
38. B. Dillman and J. L. P. Jessop, J. Polym. Sci., Part A: Polym. Chem., 2013, 51, 2058.
39. S. Penczek, P. Kubisa and R. Szymanski, Makromol. Chem., Macromol. Symp., 1986, 3, 203.
40. Y. Yagci and W. Schnabel, Angew. Makromol. Chem., 1999, 270, 38.
41. T. Biedron, R. Szymanski, P. Kubisa and S. Penczek, Makromol. Chem., Macromol. Symp., 1990, 32, 155.
42. A. Hartwig and M. Sebald, Eur. Polym. J., 2003, 39, 1975.
43. T. Endo and H. Uno, J. Polym. Sci., Polym. Lett. Ed., 1985, 23, 359–363.
44. I. A. Zucchi and W. F. Schroeder, Polymer, 2015, 56, 300.
45. J.-L. Chen, H.-M. Huang, M.-S. Li and F.-C. Chang, J. Appl. Polym. Sci., 1999, 71, 75.
46. D. Montarnal, M. Capelot, F. Tournilhac and L. Leibler, Science, 2011, 334, 965.
47. M. Capelot, D. Montarnal, F. Tournilhac and L. Leibler, J. Am. Chem. Soc., 2012, 134, 7664.
48. M. Capelot, M. M. Unterlass, F. Tournilhac and L. Leibler, ACS Macro Lett., 2012, 1, 789.
49. Y. Yang, Z. Pei, X. Zhang, L. Tao, Y. Wie and Y. Ji, Chem. Sci., 2014, 5, 3486.
50. C. J. Ong and F. P. Price, J. Polym. Sci., Part A: Polym. Chem., 1978, 63, 45.
51. H. Lützen, T. M. Gesing, B. K. Kim and A. Hartwig, Polymer, 2012, 51, 6089.

---