M. Abdolah Zadeh,
A. M. Grande,
S. van der Zwaag and
S. J. Garcia*
Novel Aerospace Materials Group, Faculty of Aerospace Engineering, Delft University of Technology, Kluyverweg 1, 2629 HS, Delft, The Netherlands. E-mail: s.j.garciaespallargas@tudelft.nl
First published on 20th September 2016
In the present work we show the effect of the crosslinking degree on the mechanical and healing behaviour of a healable thermoset dual-network polymer. A hyphenated rheological test (i.e. simultaneous rheology and FTIR) was used to follow the effect of the curing process on the mechanical behaviour in relation to the underlying chemical reactions. The effect of curing on the bulk properties and the polymer interfacial healing was studied using gap closure kinetics and a fracture mechanical test. The increased crosslinking density at longer curing times led to a more temperature-stable polymer network with significantly higher mechanical properties (elastic modulus and strength at break). It was found that the damage closure kinetics decrease with the curing degree but the ultimate interfacial healing efficiency does not. The results here reported highlight the effect of the crosslinking density on the kinetics of damage closure with a low impact on the maximum interfacial healing efficiency as long as the amount of reversible bonds remains constant.
As a result of the dynamic nature of the covalent disulphide (S–S) bonds, disulphide-based polymers exhibit good mechanical properties as well as efficient healing performance using a wide range of stimuli. Various triggering agents such as temperature,9–11 shear forces,2 reduction reactions12,13 or UV-irradiation14,15 can lead to selective scission of S–S bonds and on-demand flow of the polymeric networks. The selective opening of disulphide bridges can be facilitated and accelerated by the presence of nucleophilic reagents such as phosphine16 and thiol groups.17,18 In this later case the S–S bridge scission occurs via thiol-di/oligo-sulphide exchange reaction. Nevertheless, in has been reported that unfavourable oxidation of thiol groups can reduce the overall healing kinetics.11,18 Additionally, the nature of the groups directly connected to the disulphide bridges also affects the kinetics of the disulphide bridge opening.19 Furthermore, lower bond strengths and therefore higher bond interchange rates can be achieved by replacing disulphide bonds with tetrasulphides.6–8
Recently we introduced a healable hybrid thermoset dual network with dynamic di/tetra-sulphide bridges.11 The thermos-reversible di/tetrasulphide bridges enabled thermo-mechanically induced flow of the crosslinked networks while the irreversible organic and inorganic crosslinks preserved the mechanical integrity of the polymer during the healing process. This dual network showed remarkable macroscale damage closure capabilities (gap closure kinetics and final network restoration) at an optimal healing temperature of 70 °C.11
A part from the reversible moiety used, the healing performance of polymers is strongly affected by the polymer chemistry and architecture, e.g. content of the reversible bonds, crosslinking density, chain stiffness and intra-molecular interactions.20–23 Moreover, the overall properties of the healed interface (e.g. mechanical, barrier) highly depend on the newly formed polymer architecture at the healed site (i.e. scar). A good understanding and evaluation of the healed zone and its long-term performance is of paramount importance for the ultimate industrial implementation of this new class of materials and is therefore attracting increasing attention.
Standard tensile tests are usually employed to evaluate the healing performance of polymers.24,25 Although, the tensile procedure fails to distinguish the different processes taking place at the interface and can lead to overestimated values in terms of healing efficiency its value as a fast evaluation and screening tool is un-doubtful.26 However, methods sensitive to discontinuities, such as fracture mechanics,27 can potentially lead to a better understanding of the phenomena involved during the healing process and can provide more realistic values of the obtained healing degrees.28,29 In the case of self-healing polymers with elasto-plastic behaviour (such as the dual network presented in our previous work11) accurate quantification of the fracture toughness is feasible using a fracture mechanics protocol based on the J-integral evaluation.30
In this work we analyse the effect of curing time on the time-resolved behaviour of a healable hybrid dual network polymer and its impact on mechanical, viscoelastic and healing properties. Rheological measurements were performed to evaluate the time/frequency dependent properties of the polymer through the application of the well-known time–temperature superposition (TTS) principle.31 The effect of curing time was followed by monitoring the evolution of the dynamic shear moduli at different temperatures. Furthermore, to investigate the effect of curing time on the polymer bulk properties and its interfacial healing performance, flow and fracture tests were carried out on the polymer cured for 2 and 48 h. It was found that long curing times lead to a more stable polymeric network with improved mechanical properties and reduced flow kinetics. However, the maximum interfacial healing efficiency was found to be independent of the curing degree. Such behaviour is attributed to a completion of the inorganic crosslinking of the alkoxysilanes leading to an increase in the network stiffness but not affecting the availability of the reversible S–S bridges responsible for efficient interfacial healing.
For the TTS analysis storage (G′) and loss (G′′) shear moduli were measured as a function of temperature, frequency and time. A shear strain amplitude of 0.5% was employed to stay in the linear viscoelastic range of the hybrid polymer. The frequency sweep scans were performed in the range of 0.1–10 Hz at different temperatures from 25 °C to 70 °C with a temperature step of ΔT = 5 °C. Storage (G′) and loss (G′′) modulus master curves were then generated applying the time temperature superposition principle (TTS).
The effect of curing time was studied by measuring the dynamic shear modulus at a frequency of 1 Hz at seven different temperatures in the range of 50–110 °C with a step of 10 °C. The evolution of the dynamic shear modulus was followed for at least 2 hours at each of the testing temperatures. Simultaneous rheological and FTIR data were collected at every 10 minutes during the experiment. The time-resolved evolution of Si–O–Si bonds was followed by monitoring the changes in peak intensity of its characteristic resonance (υSi–o–Si = 1036 cm−1).
![]() | (1) |
![]() | (2) |
The fitting approach of the shift factors led to an activation energy (Ea) significantly higher for the 48 h-cured polymer (31.7 kcal mol−1) than for the 2 h-cured one (19.1 kcal mol−1), yet in general agreement with values reported in literature for polysulfide containing rubbers.4
As Fig. 1 shows, 48 h curing time led to a polymer network with higher elastic (G′) and loss (G′′) moduli in all the frequency range compared to 2 h curing. Such a trend further suggests an increase in the crosslinking density with the curing time and implies a higher chain mobility potential for low cured samples (2 h curing time). As will be presented later on, these aspects having a direct impact on the network mobility play a critical role on the healing process.
In order to follow the evolution of the crosslinking processes and their effect on mechanical properties a hyphenated experimental procedure combining rheology and FTIR measurements was performed. From the mechanical point of view, time sweep scans illustrated a gradual increase of the storage modulus (G′) depending on the sample temperature during a curing time of 2 h (Fig. 2a). In particular, higher temperatures resulted in a more rapid growth of the storage modulus (G′) over the course of the measurements (e.g. 110 °C). As Fig. 2a shows, below 60 °C, there is no change in the storage modulus for the longest tested time (2 h). However, at temperatures higher than 60 °C, the storage modulus increased linearly without reaching a plateau on the time scale of 2 h. When defining the storage modulus growth rate as the slope of the different temperature curves in Fig. 2a, an exponential dependence on the curing temperature was obtained as shown in Fig. 2b. The observed G′ evolution suggests the formation of new bonds stiffening the polymeric network.
![]() | ||
Fig. 2 Rheological time-sweep scans at 1 Hz and different temperatures (a) and the corresponding growth rate of the elastic modulus (G′) with temperature (b). |
Fig. 3 shows the FTIR spectra of the polymer cured for 2 h and further cured at 100 °C for 120 minutes during a time sweep rheological scan. The FTIR spectra shows several characteristic bands in the spectral range of 4000–500 cm−1; e.g. υ = 3450 cm−1 corresponding to O–H and N–H stretching, υ = 2968, 2925 and 2880 cm−1 associated with C–H stretching, υ = 1608 and 1450 cm−1 assigned to the aromatic ring stretching, υ = 1340 cm−1 associated with C–N stretching, υ = 1036 cm−1 corresponding to Si–O–Si stretching and υ = 550 cm−1 assigned to C–S stretching.33 To get a better insight into the nature of chemical reactions proceeding during the second curing step, the FTIR spectra were studied in the whole spectral range. As Fig. 3b shows the peaks associated with the organic network (e.g. υCC–C = 1450 cm−1, υC–N = 1340 cm−1 and υC–S = 550 cm−1) exhibited no significant variations during the measurement. However, the characteristic peaks of the organically modified silicone alkoxides (OMSAs) i.e. υSi–OH = 956 cm−1, υSi–OC2H5 = 1075 cm−1, υSi–OCH3 = 1100 cm−1 and υSi–O–Si = 1036 cm−1 varied considerably over 2 hours at 100 °C (Fig. 3c). While the intensity of the characteristic resonances of the SiOCH3, SiOC2H5 and SiOH decreased, the one of the Si–O–Si significantly increased during the time sweep rheological scan (Fig. 3c). The variation of the aforementioned peak intensities can be explained by the following condensation reactions:34
![]() | (3) |
As eqn (3) illustrates, the condensation reactions of the alkoxy silanes (i.e. SiOCH3 and SiOC2H5) and silanol groups (SiOH) result in formation of Si–O–Si bonds, justifying the descending trend in the peak intensities of the former groups and the ascending trend of Si–O–Si resonance in Fig. 3.
Using the C–H stretching band (υC–H = 2969 cm−1) as the internal standard, the amount of SiOCH3, SiOC2H5 and SiOH groups as well as Si–O–Si links at given times and temperatures were calculated using their normalized peak intensities based on eqn (4) and plotted in Fig. 4:
![]() | (4) |
Fig. 4 shows that at 50 and 60 °C the normalized peak intensities of the relevant groups did not vary significantly with time. Nevertheless at temperatures beyond 60 °C, the amount of SiOCH3, SiOC2H5 and SiOH groups decreased linearly over the course of the measurements. Higher temperatures led to more rapid decay of the aforementioned groups. Such an effect was more pronounced at temperatures equal or higher than 100 °C due to a major solvent evaporation during the condensation of silanol and alkoxy silane groups (eqn (3)) and therefore condensation reactions. As a result of parallel phenomena taking place in the polymer during the post curing (e.g. chemical reactions and solvent evaporation), the process could not be modelled with a single Arrhenius process.
The observed decrease in the content of the alkoxy silane and silanol groups is associated with an increase in the content of the Si–O–Si bridges and therefore the crosslinking density of the inorganic network. In agreement with the rheological time sweep scans, the growth rate of the Si–O–Si links increased exponentially as a function of the testing temperature, manifesting the direct correlation between the content of newly formed irreversible bonds and the enhanced mechanical properties.
Fig. 5 shows the evolution of the storage modulus and the content of the Si–O–Si links in a 2 h-cured polymer during a post-curing step of 48 hours at 70 °C. As can be seen the storage modulus (G′) of the polymer increased exponentially over 48 hours, reaching a plateau at the end of the measurement. The amount of the Si–O–Si links calculated using the peak intensity criteria (eqn (4)) followed the same trend, further confirming the direct correlation between increased crosslinking density due to formation of the irreversible Si–O–Si bridges and the improved mechanical properties demonstrated by higher storage modulus values (G′). The obtained results further illustrate stabilization of the mechanical properties in the 48 h-cured hybrid sol–gel polymer thus not being affected by further temperature treatments.
![]() | ||
Fig. 5 Evolution of G′ (a) and content of Si–O–Si links (b) in a 2 h-cured hybrid sol–gel polymer during a rheological time-sweep scan performed at 70 °C. |
Raman analysis (Fig. 6) also revealed the presence of free thiol groups in both 2 h and 48 h cured samples by presence of the characteristic resonance of S–H bonds (υS–H = 2570 cm−1).33 Furthermore, the characteristic resonance of S–S stretching (υS–S = 510 cm−1)33 was chosen for the identification and qualitative quantification of S–S bonds in the two sets of samples. Using the C–C stretching band (υC–C = 1186 cm−1) as the internal standard, the amount of S–S and S–H bonds at room temperature were calculated using the peak area ratio. While the amount of S–H groups was 3.8% lower after 48 h curing than after 2 h, the content of the S–S bridges was about 1.5% higher, suggesting the temperature triggered oxidation of free thiol groups to S–S in air.11
Fig. 7 shows that longer curing times (48 h) led to a yield stress of about three times higher than that of short curing times (2 h). Moreover a pronounced strain hardening behaviour (slope of the stress–strain curve after yielding) was observed in the high crosslinked polymer leading to three times higher strength at break. On the other hand, the mildly crosslinked polymer showed an extended plastic region (from ∼2.5% strain up to the break point). All the observed aspects indicate a reduced flow tendency and a less ductile behaviour when the crosslinking is higher.35
Fig. 8 shows the effect of the crosslinking density (2 h vs. 48 h at 70 °C) on the gap closure kinetics. The results clearly show reduced flow kinetics of the highly crosslinked polymer compared to the less crosslinked one. While the partially crosslinked sample is able to close a gap with an average width of 500 μm in less than 5 min, the fully crosslinked one (48 h) took more than 20 min to cover and close a gap of the same dimension. These results are well in-line with previous mechanical and rheological measurements, where higher Ea was found for the highly crosslinked polymer. The lower gap closure kinetics of the highly crosslinked polymer can be attributed to the higher crosslinking density achieved by formation of new Si–O–Si bridges, as demonstrated in the FTIR studies. It is therefore clear that higher crosslinking reduces the gap closure kinetics.
To evaluate the effect of crosslinking density on the interfacial healing performance of the hybrid sol–gel polymers a fracture mechanics based test protocol for ductile polymers was employed. Fracture experiments were performed on both 2 and 4 h-cured polymers.
The load–displacement curves for the virgin DENT samples and the ones healed at 70 °C for different times (10 min, 30 min, 1 h, 2 h, 4 h and 12 h) are presented in Fig. 9. As Fig. 9 shows, the 2 h-cured polymer showed lower mechanical properties than the 48 h-cured one. However, for both polymers, lower mechanical properties were exhibited by the healed DENT specimens compared to the virgin specimens. Additionally, a preliminary healing time dependent behaviour was detected for the two crosslinking degrees (2 h and 48 h).
![]() | ||
Fig. 9 Force–displacement plots of the virgin and healed DENT specimens for the 2 h-cured (a) and the 48 h-cured (b) hybrid sol–gel polymer. |
To estimate the material resistance to crack propagation, the critical J-integral value (JIC) was selected as reference parameter.28 Critical fracture energy values, JIC, for each sample were calculated according to the following equation:
![]() | (5) |
The healing efficiency was calculated based on the following eqn:
![]() | (6) |
The interfacial healing efficiency of the 2 h and 48 h-cured samples measured by fracture mechanics analysis are presented in Fig. 11. As Fig. 11 shows, the healing efficiency of the mildly crosslinked polymer (2 h) continuously increased with the healing time from about 20% reaching a plateau of 60% after 2 hours. The highly crosslinked polymer (48 h) exhibited a remarkably high recovery of the fracture property (around 45%) already at short healing times. Interestingly, for short healing times the healing efficiency of the fully crosslinked polymer was higher than that of the mildly crosslinked one. The surprisingly higher healing efficiency of the highly crosslinked polymer at short times can be explained by the higher interfacial wetting facilitated by the formation of a smoother surface during the fracture process due to the more brittle character of this polymer. The effect of surface wetting is also reported in literature20,28 demonstrating its prominent influence on the kinetics of the healing process in polymeric materials. Interestingly, despite the differences in healing degrees at short healing times, both polymers showed a similar degree of healing at long healing times (t > 600 min) suggesting an equivalent extent of chain bridging at the fracture site.
![]() | ||
Fig. 11 Healing efficiencies of the 2 h and 48 h cured hybrid sol–gel polymers calculated based on JIC. |
Based on the obtained results, it is possible to state that longer curing times of the hybrid sol–gel polymer increases the mechanical properties due to an increase in the crosslinking density. This affects the gap closure kinetics but does not affect the final overall healing behaviour. These results suggest that high crosslinking times do not directly affect the dynamics of the reversible bonds, although it clearly has an influence on the global viscoelastic behaviour of the material and on the morphology of the fracture surfaces thereby affecting the short term healing efficiency. It becomes thus clear that, while maintaining the same amount of reversible bonds, both fracture surface and the viscoelastic behaviour have a direct impact on the healing process (closure kinetics and final interface strength) since they both affect the capability of the material to flow and to promote an efficient contact between the fracture surfaces.
This journal is © The Royal Society of Chemistry 2016 |