Epoxy Polymer Networks with Improved Thermal and Mechanical

Oct 6, 2015 - The polymer isomers are those with the same PRS content and different PRS chemical conversion. At the same PRS loading, compact tension ...
13 downloads 10 Views 7MB Size
Download PDF
Article pubs.acs.org/Macromolecules

Epoxy Polymer Networks with Improved Thermal and Mechanical Properties via Controlled Dispersion of Reactive Toughening Agents Majid Sharifi, Changwoon Jang, Cameron F. Abrams, and Giuseppe R. Palmese* Department of Chemical and Biological Engineering, Drexel University, 3141 Chestnut Street, Philadelphia, Pennsylvania 19104, United States ABSTRACT: Highly cross-linked polymers are extensively used in the formulation of protective coatings, adhesives, and composite materials. The inherent densely crosslinked structure makes these materials strong but brittle limiting their use. This investigation reveals that rearranging the molecular structure of these polymers can significantly improve toughness without degrading and sometimes even enhancing other essential thermal and mechanical properties. Epoxy polymers are synthesized such that they have the same overall chemical composition but different molecular arrangements. These systems are referred to as polymer network isomers. The structural variations are introduced by using a processing technique termed “partially reacted substructures (PRS)”. The PRS were made from the stoichiometric reactions of poly(propylene oxide)diamine and diglycidyl ether of bisphenol A (DGEBA) up to specified extents of reaction (0, 60, 70, and 80%). The resulting still reactive PRS were mixed and cured with stoichiometric blends of DGEBA and diethyltoluenediamine at specified PRS contents (10, 15, and 20 wt %). The polymer isomers are those with the same PRS content and different PRS chemical conversion. At the same PRS loading, compact tension results indicated a remarkable improvement in fracture toughness (G1c) with increasing PRS conversion. SEM micrographs indicated the presence of voids on the fracture surfaces of the modified samples. Cavitation-induced shear deformation was found to be responsible for increasing fracture toughness in PRS-modified systems. Small-angle X-ray scattering results revealed the existence of heterogeneous nanodomains with domain sizes that were directly proportional to PRS conversion. Additionally, quasi-static tension tests showed that mechanical properties such as Young’s modulus and strength were almost invariant for a given set of network isomers. It was also observed that the glass transition temperature (Tg) of the modified systems was higher than those of the corresponding control systems, thus resulting in materials with both improved fracture toughness and Tg. In summary, polymer network isomers with improved thermal and mechanical properties are obtained by rearranging the network molecular structures via the addition of a single processing step. These findings can be directly and economically applied to current industrial coating, adhesive, and composite formulations.

■

of coarse-grained molecular simulations.15,16 The simulation studies suggested that upon tensile extension a void-opening or surface-drawing mechanism in highly cross-linked polymers enhances the material’s ductility without affecting tensile strength. On the basis of this, Sharifi et al.17 synthesized materials with molecular-scale defects, known as “protovoids”, in the cross-linked structure of epoxy polymers, using a processing technique called “reactive encapsulation of inert solvent followed by drying and annealing (the RESD method)”.17 In that study, it was shown that at fixed monomer composition polymer network topology could be tuned to optimize mechanical performance. However, the use of solvent and the drying and annealing stages of the RESD method limit its applicability to small parts with thin cross sections. It was necessary to find a method for controlling network topology that does not rely on the use of solvent so that it can be widely applied in industrial settings. The work presented

INTRODUCTION Thermosets are an important class of polymers used extensively in coatings, adhesives, and as matrices for structural composites. Their inherently dense cross-linked molecular structures make them strong but brittle. Consequently, studies have focused on finding ways to toughen cross-linked polymers. Among these studies, it has been demonstrated that mechanical toughness can be significantly improved by using diblock copolymers,1 triblock copolymers,2 interpenetrating networks,3 core/shell particles,4 rubber particles,5 elastomers,6 and inclusion of a thermoplastic phase.7,8 However, the presence of soft inclusions always reduces the strength, stiffness, and glass transition temperature. It has also been shown that uniform distribution of hard nanoparticles such as titania, alumia,9 silica,10 graphene platelets,11 carbon nanotubes,12 fullerene,13 and POSS14 can overcome the disadvantages of using traditional soft tougheners by improving modulus and toughness without sacrificing Tg. However, the incorporation of such fillers makes processing sometimes intractable and in many cases reduces strength. This work is motivated by the recently proposed and novel approach to toughening thermosets shown by means of a series © XXXX American Chemical Society

Received: March 31, 2015 Revised: September 14, 2015

A

DOI: 10.1021/acs.macromol.5b00677 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules

Figure 1. Molecular structure of the epoxy resin and the curing agents used in this study. (a) EPON 825 (DGEBA), n = 0.07, (b) Jeffamine D-4000 (PPODA), x = 68, and (c) EPIKURE W (DETDA) (“n” and “x” are number of the repeating units for DGEBA and PPODA, respectively).

Figure 2. (a) FTIR spectra of epoxy−amine reactions of PRS (DGEBA + PPODA) at 80 °C. (b) Reaction fractional conversion based on epoxide consumption. substructures (PRS) with controlled extent of polymerizations, (2) making blends of blank epoxy (DGEBA + DETDA) with specific contents of the reactive PRS, and (3) curing the blends. PRS was produced by adding/mixing stoichiometric blends of PPODA and DGEBA at 80 °C. The chemical conversion of these mixtures was monitored using a Nexus 870 FTIR spectrometer (Thermo Nicolet Corp.) in the near-IR region. Figure 2a shows representative FTIR spectra taken over a period of time for this reaction. The peak at 4530 cm−1 corresponds to the stretching and bending of oxirane bands, and the peak at 4930 cm−1 corresponds to the NH2 combination bands.19,20 The decrease of epoxy and amine concentrations signifies the presence of epoxy−amine reactions. Figure 2b shows the fractional conversion of epoxy with respect to time obtained using the epoxy peak area. Additionally, the approximate gel point was measured using a steady shear viscosity measurement during curing reactions using a TA rheometer apparatus AR2000. Gelation is known to happen at the point where steady shear viscosity rises rapidly to infinity.21 A combined viscosity−conversion measurement on DGEBA−PPODA systems at 80 °C showed that the approximate gel point for these systems is taking place at around 84% oxirane conversion. Cure was stopped at 60, 70, and 80% conversion as shown in Figure 2b by cooling to room temperature. The resulting pregelled PRS samples were labeled 60, 70, and 80% conversion PRS, respectively. Specific amounts of the each of the resulting PRS were then added to stoichiometric blends of DGEBA and DETDA to make solutions of 10, 15, and 20 wt % PRS. The cured samples were called PRS-modified samples and were labeled according to their PRS chemical conversion. Additionally, “one-pot” batches were prepared by mixing the three components (DGEBA, DETDA, and PPODA) such that compositions were kept the same as the corresponding PRS-modified samples at 10, 15, and 20% PRS loading. The resulting cured samples were called PRS-control samples (labeled “0% conversion PRS”). We refer to the “one-pot” 0% conversion PRS samples as control systems. Samples with the same PRS content for any PRS chemical conversion were termed polymer network isomers because they possess the same overall chemical composition but differ in molecular arrangement of

herein provides a method to synthesize polymer networks that possess the same overall composition but with varying molecular arrangement. These systems are referred to as “polymer network isomers”. It is important to note that concept of polymer network isomers, in this study, is different than the definitions used elsewhere.18 This methodology, termed the “partially reacted substructure (PRS)” method, involves subgel partial curing of a relatively low-Tg diamine/epoxy system followed by its blending and complete curing in a high-Tg diamine/epoxy system while maintaining overall stoichiometry. Polymer networks isomers are therefore prepared by varying the degree of cure for the low-Tg diamine/epoxy system (PRS) before it is coreacted with the high-Tg monomers. The base isomer for comparison is prepared by mixing all reactants without prereactionin this case the degree of cure of the PRS is 0%. The PRS method overcomes the limitation of RESD in that it does not involve sacrificial solvent, thereby allowing fabrication of samples of any size and geometry and enabling measurement of plane-strain fracture toughness using linear elastic fracture mechanics (LEFM). We demonstrate significant toughness improvements using the PRS method without sacrificing strength and with increased Tg. We further demonstrate that the degree of toughness enhancement is correlated to the PRS degree of cure.

■

EXPERIMENTAL SECTION

Materials. Diglycidyl ether of bisphenol A (DGEBA, EPON 825, Miller-Stephensen), n = 0.07, was used as the epoxy resin. In addition, poly(propylene oxide)diamine (PPODA, Jeffamine D4000, Huntsman), x = 68, and diethyltoluenediamine (DETDA, EPIKURE W, Miller-Stephensen) were used as the curing agents. The molecular structures of these materials are shown in Figure 1. Synthesis Procedure. The formulation of the polymer isomers involved three basic steps: (1) synthesizing the partially reacted B

DOI: 10.1021/acs.macromol.5b00677 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules network building blocks. For additional comparison, PPODA-free samples (blank samples) were also prepared by curing stoichiometric blends of DGEBA and DETDA. All blends were well mixed and degassed with a centrifugal mixer. They were then oven-cured at 80 °C for 24 h and postcured at 160 °C for 4 h. Molecular Dynamics Simulations. MD systems of mixtures of DGEBA, PPODA, and DETDA were modeled using the Generalized Amber Force Field (GAFF)22,23 and simulated using Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS).24 Crosslinked systems were generated using the capture-radius approach previously utilized.25,26 Tg’s were measured with a temperature ramp from 700 to 145 K at a cooling rate of 15 K/2 ns. All observables were measured as averages of three replicas.

prereaction of some of the components, increased the fracture toughness of the material by well over 100%. This can be clearly observed by comparing the 0% conversion PRS to the corresponding 80% conversion PRS isomers. G1c values seem to go through a maximum as the concentration of PRS in the high Tg network increases. In fact, G1c values are lower for the materials with 20% PRS compared to materials with 10% and 15% PRS. It is not clear to us why this is the case. However, it could be a result of the significantly reduced modulus at higher PRS content. This behavior should be investigated further. Fracture Interface Nanostructure. Figure 4 contains representative SEM micrographs of the series of samples containing 10% PRS that show significant differences in fracture surface morphology. Comparing the fracture surfaces of blank and control samples in Figures 4a and 4b, respectively, shows that the surface morphologies are rather similar and vary only in roughness. However, comparing the surface morphology of the polymer network isomers shown in Figures 4b, 4c, 4d, and 4e reveals the presence of cavities on the fracture surfaces of modified samples shown in Figures 4c, 4d, and 4e and the absence of cavities on the surface of the control shown in Figure 4b. The cavities on the fracture surfaces may explain why the modified systems possess much higher fracture toughness. It should be noted that no indication of macroscopic phase separation existed, since all blank, control, and modified samples were transparent to visible light as shown by the inset pictures on the top right-hand corner of each micrograph in Figure 4. The average size of the cavities on the fracture surfaces was measured using the NIH image processing toolbox “ImageJ”. Table 1 contains estimates of void diameters for modified samples. The average void diameter for the relatively circular features is roughly 30−45 nm. It was observed that the average void diameter of the 80% modified systems is slightly higher than the other modified systems even though the error bars are overlapping. Small-angle X-ray scattering was used to characterize the polymer samples prior to mechanical deformation. A SAXS/ WAXS apparatus (Rigaku S-MAX 3000) was utilized for this purpose. Specimens were cut into thin disk-shaped geometries and polished. Each sample was irradiated with an X-ray having a wavelength of 1.54 Å (CuKα), and the scattered beam was collected with a 2D X-ray detector. Figure 5 shows scattered intensity (integrated over the azimuthal angle, χ) versus reciprocal space q-vector. Figure 5a represents the scattered intensity of the control samples at different PRS (PPODA) contents, whereas Figure 5b plots the identical quantities of the polymer network isomers at 20% PRS loading. A scattering peak is observed at q = 0.049 Å−1 in Figure 5a, suggesting that the use of PPODA results in the formation of heterogeneous domains. According to Bragg’s law,27,28 this q-value corresponds to a d-spacing of about 13 nm. Increasing PRS (PPODA) content enhances the scattered intensity by increasing the concentration of the scatterers, which appears as an increase in the peak area. However, the peak position did not change with increasing PPODA content, which suggests that the size of the heterogeneities in control samples does not change with increasing PPODA content. The measured size of the control sample heterogeneous domains is about 13 nm and is larger than the size of a single PPODA monomer. So, we expect this scattering to be from a few linked monomers. On the other

■

RESULTS AND DISCUSSION Fracture Toughness. After the blank, control, and modified samples were cured, they were cut into a compact tension specimen geometry. Compact tension testing was conducted under quasi-static tensile loading with the crosshead speed of 1 mm/min following ASTM D5045. The values of the critical stress intensity factor, K1c, and the strain energy release rate, G1c, were measured according to the theory of linear elastic fracture mechanics (LEFM). For each sample we ensured the size of sample thickness, crack length, and uncracked ligament to be greater than the LEFM size criteria, 2.5(KQ/σy)2. The G1c values are shown in Figure 3. A

Figure 3. Strain energy release rate, G1c.

clear increase in fracture energy was observed for the PRSmodified systems. Assuming full conversion of curing reactions after postcure, the difference between control and modified samples at fixed PRS content (same overall composition) is attributed only to the degree of PRS chemical conversion. The G1c results show that the network isomerization state substantially influences fracture behavior. Also, by looking at the results at the same PRS loading, it was observed that the degree of toughening is directly proportional to the chemical conversion of the embedded PRS. The same trend was observed in the three batches with different PRS contents. The observed enhancement in toughness is significant because toughening was achieved not by incorporating the second phase (soft matter or nanoparticle) but by means of rearranging the network structure. The results indicate that adding an additional processing step, i.e., the C

DOI: 10.1021/acs.macromol.5b00677 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules

Figure 4. SEM images of the fracture surfaces of the polymer network isomers. Surface morphology of the (a) blank samples, (b) control sample (0% conversion PRS), (c) PRS-modified sample (60% conversion PRS), (d) PRS-modified sample (70% conversion PRS), and (e) PRS-modified sample (80% conversion PRS). The control and modified samples are at 10% PRS content. The scale bars are 1 μm.

Table 2. d-Spacings of the Heterogeneous Domains for Each of the Samples in This Study

Table 1. Void Diameters (nm) of the Structural Features (Cavities) Measured by ImageJ

d-spacings

void diameter (nm) PRS content (%)

PRS conv 0%

PRS conv 60%

PRS conv 70%

PRS conv 80%

PRS content (%)

PRS conv 0%

PRS conv 60%

PRS conv 70%

PRS conv 80%

10 15 20

N/A N/A N/A

39.6 ± 11.9 32.8 ± 8.9 39.5 ± 12.4

40.1 ± 11.1 30.9 ± 5.1 41.5 ± 13.6

44.8 ± 12.4 44.9 ± 15.3 44.3 ± 13.6

0 10 15 20 30 40

N/A 13.1 13.4 13.0 12.6 11.3

N/A 19.6 18.0 18.4

N/A 23.4 21.5 21.2

N/A 23.4 25.4 25.0

hand, prereacting PPODA molecules shifts the peak to smaller q-values as shown in Figure 5b, suggesting that the heterogeneous domains become larger with increasing PRS conversion. The calculated d-spacing values for each of the samples are listed in Table 2.

Table 2 shows that the concentration of PRS has little effect on the size of nanodomains. However, the size of the structural

Figure 5. Scattered X-ray intensity vs q: (a) control samples; (b) polymer isomers at 20% PRS content. D

DOI: 10.1021/acs.macromol.5b00677 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules features is directly proportional to PRS conversion at fixed composition. A general schematic of molecular network structure that illustrates this behavior is given in Figure 6. This figure shows that PPODA containing domains are smaller in control structures and become larger as determined by prereaction in modified structures.

bound parts of the polymer network, interface debonding commonly found in other toughening systems methods is not occurring. This is clearly shown by the uniquely ragged edges of the voids on the fracture surfaces. SEM micrographs clearly show cavitation and enhanced plastic shear yielding. The extent of void growth can be inferred by comparing the d-spacing obtained by SAXS given in Table 2 to the void sized measured using image analysis in Table 1. The cold-fractured surface morphology of a PRS-modified sample broken in liquid nitrogen was compared to the surface morphology of the same material fractured at room temperature (RT). The corresponding micrographs are shown in Figure 7. Qualitatively, it appears that void formation is reduced at cold temperatures. The quantitative measurement of void surface coverage and void size were 0.3% and 37.4 ± 8.9 nm, respectively, for the cold-fractured system, whereas these measurements were significantly higher, roughly 1.7% surface coverage and 45.7 ± 15.7 nm, for the RT-fractured systems. The significant difference in void size and surface coverage between RT-fractured and cold-fractured surfaces provides further evidence that the cavitation-aided shear yielding is the dominant micromechanism in toughening of these polymer networks. However, in colder environments, or similarly at higher strain rates, this mechanism is moderated, and therefore voids are smaller and less observable. Nevertheless, cavitation are still visible in cold fractures systems which suggests that this toughening mechanism is uncharacteristically activated even in extremely cold environments or analogously at high strain rates. This behavior could enable a new generation of epoxy systems used for adhesives and composites that endure extreme dynamic environments. Even though the toughening mechanism is similar to that observed in traditional rubber and particle toughened systems, there are some important differences.36 Compared to traditional liquid rubber modifiers the PRS technique allows for direct control of domain size via reaction at a very small scale. The domains that develop are on the size of a 10−25 nm. In comparison to small particle (core−shell) and block copolymer systems, the PRS domains are on the low end of the size scale but more importantly are directly linked to rest of the network. In fact, we argue that the PRS is part of the network. Because of this, fracture micrographs do not show the characteristically spherical inclusions of the other systems but rather more asymmetric shapes with ragged edges that could in part be responsible for improving fracture toughness. Furthermore,

Figure 6. Probable structural formation of the polymer network isomers: (upper) control systems; (lower) PRS-modified systems.

Fracture Mechanism. In highly cross-linked epoxy polymers, plastic shear deformation is considered the main energy absorbing process.29,30 In PRS modified systems, similar to rubber toughened epoxy systems, plastic shear yielding is aided by cavitation (void formation) in the heterogeneous soft microdomains that show up as cavities on fracture surfaces.29,31,32 The size of the observed cavities on fracture surfaces of the PRS-modified samples was significantly smaller than the size of the cavities in rubber-toughened epoxy systems. According to the SAXS results in Figure 5, greater conversion of the prereacted substructures resulted in larger heterogeneous domains. The cavitation induced shear yielding is higher when the heterogeneous domains are larger as in the 80% PRSmodified systems. The enhanced shear yielding mechanism improves G1c values as shown in Figure 3. The size analysis of the cavities from the SEM micrographs supports this argument as an increasing trend of size with PRS chemical conversion was observed (Table 1). In summary, cavitation-induced shear yielding was higher in the PRS-modified systems than the controls. This observation agrees with the notion of a sizedependent cavitation criterion proposed by Lazzeri et al.33 and Bucknall et al.34,35 Since the soft PRS domains are covalently

Figure 7. Room-temperature fractured (a) vs cold fractured (b) surface morphology of an 80% PRS-modified sample at 10% PRS content. The scale bars are 1 μm. E

DOI: 10.1021/acs.macromol.5b00677 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules

Glass Transition Temperature. The glass transition temperature for the polymer networks being investigated was measured using a TA-Q800 dynamic mechanical analyzer in single cantilever mode. DMA specimens were heated at a ramp rate of 2 °C/min under cyclic loading with a frequency of 1 Hz and amplitude of 15 μm. Tg values were obtained from the peak position of the tan δ curves. It is interesting to note that the Tg values were not the same for the system isomers. In fact, the PRS-modified systems revealed higher Tg values compared to the corresponding control systems. Figure 9a shows the tan δ curves of the blank, control, and modified samples at 10% PRS content. It was observed that the Tg values of the control systems follow the linear trend as predicted by Fox.37 However, the Tg values of the PRS-modified systems deviate from the linear fit (Fox model) beyond 60% PRS conversions. These striking results are shown in Figure 9b where many formulations show Tg well above the Fox model predictions. One explanation for the elevated Tg in the modified systems compared to the corresponding control systems could be the dispersion of PPODA molecules within the networks. When relatively dilute, each PPODA monomer is likely surrounded by DEDTA and DEGBA, but when aggregated into PRS regions as determined by prereaction, each PPODA monomer is likely surrounded by other PPODA monomers. As PPODA has a flexible backbone, it likely enhances the mobility of its neighbors in the network. Therefore, at the same overall composition, diluted PPODA enhances the mobility of more DEDTA than does aggregated PPODA, thereby degrading Tg in the former case much more than in the latter relative to pure DEDTA/DEGBA networks. This is consistent with the observation that the control systems in which PPODA is dilute show lower glass transition temperatures than the corresponding modified systems. Atomistic Molecular Dynamic Simulation. The structural formation of control and PRS-modified systems was modeled using atomistic molecular dynamic (MD) simulations. Mixtures of the two diamines, PPODA and DETDA, were used together with DGEBA to compare the two blending protocols at constant overall DGEBA/DETDA/PPODA 2:1 epoxy:amine stoichiometry. The control involves curing DGEBA with both

particle dispersion techniques are not required during processing. Network Cross-Link Density. Figure 8 shows DMA storage plots for 0% (blank), 10% PRS, and 20% PRS control

Figure 8. Plots of storage modulus vs temperature for the blank, control, and PRS-modified (60% conversion PRS) epoxy thermosets at 0, 10, and 20% PRS content.

samples (0% PRS conversion). As expected, the rubbery modulus plateau and average cross-link density are significantly reduced by increasing PRS content. Our intention was to explore and compare the average cross-link density of the polymer network isomers. So, also included in Figure 8 are the storage modulus plots for 60% conversion PRS corresponding to 10 and 20% PRS content. Comparing the rubbery modulus plateau of control and 60% conversion PRS systems at fixed PRS contents in Figure 8 shows that PRS chemical conversion does not affect rubbery modulus and average cross-link density, suggesting that average cross-link density is not affected by the way that PPODA is dispersed in the network.

Figure 9. (a) Tan δ curves of the blank, control, and modified samples at 10% PRS content. (b) Glass transition temperature of the control and modified samples at different PRS loading. The dashed line represents the Fox model. F

DOI: 10.1021/acs.macromol.5b00677 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules DETDA and PPODA, and the modified involves first partially curing DGEBA with PPODA and then adding DETDA and DGEBA to continue to a fully cured sample. In this study, the 15 wt % PRS at 0 and 80% partial cure was modeled, and they were labeled “control” and “modified”, respectively. The glass transition temperatures from control and modified systems were obtained from a specific volume versus temperature curve by averaging over three independent replicas per system as shown in Figure 10. Predicted Tg was 410 K for

Figure 11. Root-mean-squared fluctuation in the center-of-mass of DETDA rings vs temperature for the control (0% conversion PRS) and modified systems (80% converson PRS) at 15% PRS content.

used to show that topology-based toughening does not sacrifice mechanical strength or material stiffness. To test this theory on the network isomers considered in this study, quasi-static mechanical testing was performed in uniaxial extension mode using the servo-hydraulic INSTRON apparatus. At least eight dog-bone-shaped tensile coupons were produced, and the stress−strain curves were collected from at least six samples in each batch. The results are presented in Table 3. It was found that increasing PRS content generally increases failure strain and reduces Young’s modulus and tensile strength. However, the data show that these properties do not vary appreciably within sets of isomers. This is particularly clear at 10% PRS content.

Figure 10. Specific volume vs temperature for the control (0% conversion PRS) and modified (80% conversion PRS) systems at 15% PRS content. Glass transition temperature is around 410 K for the control and 425 K for the modified systems.

the control systems and 425 K for the modified systems. This trend agrees with the results of thermomechanical analysis (TMA) experiments conducted at 2 °C/min using a TA Instruments 2940 TMA. The experimentally obtained Tg for the compositions of interest was 420 K for the control and 437 K for the modified system. The main factors that contribute to the discrepancies between the simulation and experiment are size and time scale. The simulated samples were composed of 35 000 atoms, which is much smaller than the size of experimental samples. Additionally, time scales were substantially different from simulation and experiment; the simulation temperature ramp rate was 15 K/2 ns compared to 2 K/min for experiments. The root-mean-squared fluctuation (RMSF) of the DETDA center of mass in the cross-linked systems was calculated as a function of temperature (Figure 11). The RMSFs clearly show that the DETDA rings fluctuate more in the control sample than in the PRS samples at all temperatures. This results in lower molecular flexibility in the PRS-modified samples and consequently higher Tg’s relative to the control samples. The higher molecular flexibility and lower Tg of control samples stems from the dilution of PPODA molecules. In the control samples, PPODA is well-dispersed, leading to a higher likelihood of DETDA/PPODA nonbonded contacts in the control vs the PRS samples. Quasi-Static Mechanical Properties. So far, it has been shown that the modified samples are tougher and have more elevated glass transition temperatures than the corresponding control samples. Earlier, coarse-grained molecular simulations,15,16 atomistic simulations,25 and experiments17 were

■

CONCLUSION A novel processing technique for toughening thermosets was introduced through which partially reacted substructures of a low-Tg thermoset at different extents of polymerization were embedded into a high-Tg majority-phase thermoset. It was shown that the PRS-modified samples are tougher than the corresponding controls, and the toughness is directly proportional to the chemical conversion of the embedded PRS. The 80% conversion PRS-modified samples were almost twice as tough as the corresponding control samples. SEM micrographs from fracture surfaces of the modified samples showed voids whereas none were seen on the fracture surfaces of the control samples. This implies that the principal energy dissipating mechanism during fracture in the modified structures is cavitation with associated plastic shear yielding. SAXS results represented the existence of heterogeneous domains of about 12−25 nm. The domain size increased with increasing of PRS conversion, indicating that the prereaction step in PRS technique led the PPODAs to be more aggregated. Also observed was that the glass transition temperatures of the modified systems were higher than those of the control systems. In addition, quasi-static mechanical extension did not show substantial differences in mechanical strength and modulus between system isomers. Overall, polymer network isomers with improved thermal and mechanical properties were developed by rearranging the molecular structures through an G

DOI: 10.1021/acs.macromol.5b00677 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules Table 3. Mechanical Properties: Young’s Modulus, Tensile Strength, and Failure Strain Values tensile strength, σy (MPa)

Young’s modulus, E (GPa)

failure strain, εr (%)

PRS content (%)

PRS conv 0%

PRS conv 70%

PRS conv 80%

PRS conv 0%

PRS conv 70%

PRS conv 80%

PRS conv 0%

PRS conv 70%

PRS conv 80%

0 10 15 20

2.0 ± 0.1 2.1 ± 0.1 1.8 ± 0.1

2.7 ± 0.4 2.2 ± 0.1 2.1 ± 0.2 1.6 ± 0.1

2.1 ± 0.1 1.7 ± 0.2 1.5 ± 0.1

49.0 ± 4.0 51.4 ± 6.9 44.9 ± 1.5

64.0 ± 14.7 46.0 ± 3.8 52.8 ± 2.6 38.9 ± 2.0

46.0 ± 2.4 46.5 ± 2.6 38.6 ± 2.8

4.4 ± 0.7 5.5 ± 0.7 6.3 ± 1.5

5.3 ± 0.8 3.6 ± 0.7 4.4 ± 1.1 5.0 ± 2.0

4.0 ± 0.4 5.0 ± 0.7 6.0 ± 2.0

(8) Kinloch, A.; Yuen, M.; Jenkins, S. J. Mater. Sci. 1994, 29 (14), 3781−3790. (9) Wetzel, B.; Rosso, P.; Haupert, F.; Friedrich, K. Eng. Fract. Mech. 2006, 73 (16), 2375−2398. (10) Rosso, P.; Ye, L.; Friedrich, K.; Sprenger, S. J. Appl. Polym. Sci. 2006, 100 (3), 1849−1855. (11) Park, Y. T.; Qian, Y.; Chan, C.; Suh, T.; Nejhad, M. G.; Macosko, C. W.; Stein, A. Adv. Funct. Mater. 2014, 25, 575. (12) Gojny, F. H.; Wichmann, M. H. G.; Fiedler, B.; Schulte, K. Compos. Sci. Technol. 2005, 65 (15−16), 2300−2313. (13) Rafiee, M. A.; Yavari, F.; Rafiee, J.; Koratkar, N. J. Nanopart. Res. 2011, 13 (2), 733−737. (14) Kim, G. M.; Qin, H.; Fang, X.; Sun, F. C.; Mather, P. T. J. Polym. Sci., Part B: Polym. Phys. 2003, 41 (24), 3299−3313. (15) Mukherji, D.; Abrams, C. F. Phys. Rev. E 2008, 78 (5), 050801. (16) Mukherji, D.; Abrams, C. F. Phys. Rev. E 2009, 79 (6), 061802. (17) Sharifi, M.; Jang, C. W.; Abrams, C. F.; Palmese, G. R. J. Mater. Chem. A 2014, 2 (38), 16071−16082. (18) Groeninckx, G.; Vanneste, M.; Everaert, V. In Polymer Blends Handbook; Utracki, L. A., Ed.; Kluwer Academic Publishers: Boston, MA, 2002. (19) Mijovic, J.; Andjelic, S. Macromolecules 1995, 28 (8), 2787− 2796. (20) Raman, V. I.; Palmese, G. R. Macromolecules 2005, 38 (16), 6923−6930. (21) Sperling, L. H. Introduction to Physical Polymer Science; John Wiley & Sons: New York, 2015. (22) Wang, J.; Wang, W.; Kollman, P. A.; Case, D. A. J. Mol. Graphics Modell. 2006, 25 (2), 247−260. (23) Wang, J.; Wolf, R. M.; Caldwell, J. W.; Kollman, P. A.; Case, D. A. J. Comput. Chem. 2004, 25 (9), 1157−1174. (24) Plimpton, S. J. Comput. Phys. 1995, 117 (1), 1−19. (25) Jang, C.; Sharifi, M.; Palmese, G. R.; Abrams, C. F. Polymer 2014, 55 (16), 3859−3868. (26) Jang, C.; Sirk, T. W.; Andzelm, J. W.; Abrams, C. F. Macromol. Theory Simul. 2015, 24, 260. (27) Cowley, J. M. Diffraction Physics; Elsevier: Amsterdam, 1995. (28) Bragg, W. L. In The Diffraction of Short Electromagnetic Waves by a Crystal, Proceedings of the Cambridge Philosophical Society; 1913; p 4. (29) Kinloch, A.; Shaw, S.; Tod, D.; Hunston, D. Polymer 1983, 24 (10), 1341−1354. (30) Kinloch, A. Mechanics and mechanisms of fracture of thermosetting epoxy polymers. In Epoxy Resins and Composites I; Springer: Berlin, 1985; pp 45−67. (31) Kinloch, A.; Shaw, S.; Hunston, D. Polymer 1983, 24 (10), 1355−1363. (32) Liu, J.; Sue, H.-J.; Thompson, Z. J.; Bates, F. S.; Dettloff, M.; Jacob, G.; Verghese, N.; Pham, H. Macromolecules 2008, 41 (20), 7616−7624. (33) Lazzeri, A.; Bucknall, C. B. J. Mater. Sci. 1993, 28 (24), 6799− 808. (34) Bucknall, C. B.; Paul, D. R. Polymer 2013, 54 (1), 320−329. (35) Bucknall, C. B.; Heather, P. S.; Lazzeri, A. J. Mater. Sci. 1989, 24 (6), 2255−61. (36) Bagheri, R.; Marouf, B. T.; Pearson, R. A. Polym. Rev. (Philadelphia, PA, U. S.) 2009, 49 (3), 201−225. (37) Fox, T. G., Jr.; Flory, P. J. J. Appl. Phys. 1950, 21 (6), 581−591.

additional step in the processing of these polymers. This further supports the idea that the network isomerization state can be tuned to produce materials with improved toughness and glass transition temperature without sacrificing essential mechanical properties such as strength and stiffness. The results on the epoxy−amine system investigated herein can be directly and economically applied to current industrial coating, adhesive, and composite formulations.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail [email protected] (G.R.P.). Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Research was sponsored by the Army Research Laboratory and was accomplished under Cooperative Agreement W911NF-122-0022. The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of the Army Research Laboratory or the U.S. Government. The U.S. Government is authorized to reproduce and distribute reprints for Government purposes notwithstanding any copyright notation herein. The authors are grateful to the Huntsman Corporation for the generous donation of the Jeffamine reagents for research purposes.

■

ABBREVIATIONS PRS, partially reacted substructures; DGEBA, diglycidyl ether of bisphenol-A; DETDA, diethyltoluenediamine; PPODA, poly(propylene oxide)diamine; MD, molecular dynamics; LAMMPS, Large-scale Atomic/Molecular Massively Parallel Simulator; GAFF, Generalized Amber Force Field; RESD, reactive encapsulation of solvent/drying; RMSF, root-mean square fluctuation; SEM, scanning electron microscopy; SAXS/ WAXS, small/wide-angle X-ray scattering; DMA, dynamic mechanical analysis; FTIR, Fourier transform infrared; LEFM, linear elastic fracture mechanics; ASTM, American Society for Testing and Materials; NIH, National Institutes of Health.

■

REFERENCES

(1) Thompson, Z. J.; Hillmyer, M. A.; Liu, J.; Sue, H.-J.; Dettloff, M.; Bates, F. S. Macromolecules 2009, 42 (7), 2333−2335. (2) Meng, F.; Zheng, S.; Zhang, W.; Li, H.; Liang, Q. Macromolecules 2006, 39 (2), 711−719. (3) Shaker, Z. G.; Browne, R. M.; Stretz, H. A.; Cassidy, P. E.; Blanda, M. T. J. Appl. Polym. Sci. 2002, 84 (12), 2283−2286. (4) Sue, H.-J. J. Mater. Sci. 1992, 27 (11), 3098−3107. (5) Pearson, R. A.; Yee, A. F. J. Mater. Sci. 1991, 26 (14), 3828−3844. (6) Yee, A. F.; Pearson, R. A. J. Mater. Sci. 1986, 21 (7), 2462−2474. (7) Pearson, R. A.; Yee, A. F. Polymer 1993, 34 (17), 3658−3670. H

DOI: 10.1021/acs.macromol.5b00677 Macromolecules XXXX, XXX, XXX−XXX