Photopolymerized Triazole-Based Glassy Polymer Networks with Superior Tensile Toughness

Abstract

Photopolymerization is a ubiquitous, indispensable technique widely applied in applications from coatings, inks, and adhesives to thermosetting restorative materials for medical implants, and the fabrication of complex macro-scale, microscale, and nanoscale 3D architectures via additive manufacturing. However, due to the brittleness inherent in the dominant acrylate-based photopolymerized networks, a significant need exists for higher performance resin/oligomer formulations to create tough, defect-free, mechanically ductile, thermally and chemically resistant, high modulus network polymers with rapid photocuring kinetics. This study presents densely cross-linked triazole-based glassy photopolymers capable of achieving preeminent toughness of ≈70 MJ m⁻³ and 200% strain at ambient temperature, comparable to conventional tough thermoplastics. Formed either via photoinitiated copper(I)-catalyzed cycloaddition of monomers containing azide and alkyne groups (CuAAC) or via photoinitiated thiol-ene reactions from monomers containing triazole rings, these triazole-containing thermosets completely recover their original dimensions and mechanical behavior after repeated deformations of 50% strain in the glassy state over multiple thermal recovery–strain cycles.

1. Introduction

The spatiotemporal control associated with photopolymerization reactions enables rapid in situ formation of glassy polymer networks polymerized from initially low viscosity resins containing multifunctional monomers and oligomers that are particularly useful in additive manufacturing and other emerging applications.^[1,2] Currently, conventional thermosetting photopolymers used in industrial applications are most often prepared from either multifunctional (meth)acrylate-based photoinitiated radical chain-growth polymerizations or epoxy-based cationic polymerizations triggered via photoacid generators.^[3–5] These reactions exhibit rapid polymerization kinetics in conjunction with the formation of polymers with both mechanical stiffness and strength. However, the maximum conversion achieved in these networks is often limited by vitrification at the cure temperature,^[6–8] and the formation of heterogeneous network structures^[9] also results in impaired material properties. Also, these photo­polymerized materials typically form highly brittle polymers,^[3] leading to premature failure and limiting their applications, particularly as compared with tough thermoplastic materials that are more robust but generally not readily formed by reasonable photopolymerization processes. Therefore, on-going challenges remain in developing photopolymerized networks that will exhibit behavior including not only high stiffness and strength but also superior thermoplastic­like ductile behavior and toughness.

Tough polymers such as linear thermoplastic polycarbonates^[10,11] have high-energy absorbing capacities, generally measured by the area under the engineering stress–strain curve. These materials often exhibit ductile behavior and yield significantly prior to fracture in response to a large applied load.^[12] By contrast, when cross-linked thermosetting polymers such as those formed via typical photopolymerizations are strained, eventual breaking of covalent bonds leads to the formation of nanosized cracks and ultimately failure at a relatively low elongation to break.^[13,14] For a ductile thermoplastic material, multiple crazes appear in a comparatively large volume of the polymer, or shear deformation occurs without crazing,^[15,16] and this is macroscopically evident after the yield point by a reduction in stress followed by the specimen necking at constant stress during the cold-drawing region, as the necked section is extended.^[17] The material’s ductility is dependent on complex factors including temperature relative to the glass transition (T_g) or melting temperatures, testing rate, degree of crystallinity, chain alignment, chain length, chain branching and cross-linking, chain packing, intrinsic chain rigidity, and other intermolecular forces, among others.^{[3,12,16,18–20]}

Several strategies for toughening glassy cross-linked polymers have been developed to improve tensile and/or fracture toughness as well as impact strength. Particulate fillers including inorganic microparticle or nanoparticles,^[21,22] microfibres^[23] and carbon nanotubes^[24] and platelets^[25,26] have been added to act as crack deflectors or to absorb energy by debonding from the matrix. Low molecular weight linear elastomers,^[27] thermoplastics,^[28] or block polymers^[29,30] have been added to epoxy or acrylate-based thermosets to enhance yielding in the matrix or induce cavitation/debonding of the particle. Similarly, both core/shell particles containing an elastomeric core coated with a glassy shell layer^[31,32] and “spheres on spheres” (raspberry-like) structures containing elastomeric spheres attached to a glassy core^[33] were also employed to toughen epoxy-based polymer networks. Regardless of an increase in toughness associated with these additives, the viscosity increase, reduction in modulus from rubber-based additives, coloration of carbon nanotubes, and light scattering or opacity from inorganic or polymeric particles have limited the practical applications of additives as toughening agents.^[3]

A separate approach that has been used is to modify the polymer structures to enhance toughness. For example, incorporating addition fragmentation chain transfer agents in methacrylate-based cross-linked polymers^[34,35] and incorporating thiol-ene or yne-based step-growth polymers in (meth)acrylate-based cross-linked polymers,^[36,37] respectively, reduced the brittleness of these materials. Urethane-based glassy copolymer networks have also improved toughness and stiffness^[38] where the urethane groups caused hydrogen bonding interactions between the polymer strands, but the cross-link density of these polymers was very low, as evidenced by the low gel fraction.^[38] In a glassy cross-linked polymer formed via ring-opening metathesis polymerization (ROMP) of dicyclopentadienes, a substantial increase in toughness with >100% elongation at break was also achieved by tuning the cross-linking density using 2nd generation Grubbs’ catalysts.^[39,40] However, due to the lack of temporal control of the catalyst used in ROMP polymerization,^[39,40] a more facile and efficient approach for developing readily polymerized thermosetting polymers with high toughness is desired particularly for applications such as additive manufacturing.

Photoresponsive “click” chemistries offer spatially and temporally controlled reactions that allow efficient, clean, and orthogonal reactions.^[41,42] In particular, the copper(I)-catalyzed azide-alkyne cycloaddition (CuAAC)^[43–45] reaction has been utilized in forming step-growth bulk photopolymer networks where the CuAAC photopolymerization generates triazole adducts as a product that are thermally, chemically, and mechanically stable.^[46,47] This attribute of forming triazole moieties, especially in a densely cross-linked network, substantially enhances the mechanical stiffness and strength of these materials during the polymerization, which is a necessary element for many industrial photopolymers. Other key merits associated with this network as opposed to conventional chain-growth polymerizations include the formation of homogeneous network structures and readily tunable backbone functionalities achieved in step-growth polymerization. The broad ability to tailor these networks enables systematic modification of resin viscosity, polymerization kinetics, gel-point conversion, vitrification, polarity, and the ultimate mechanical performance.

Here, glassy cross-linked triazole­based photopolymers are developed with dramatically enhanced toughness, elongation to break, and tensile strength based on the in situ formation of triazoles via the photoinitiated CuAAC polymerization. We hypothesize that noncovalent interactions of the rigid triazole rings and sufficient freely rotatable bonds between network junctions are the cause for the enhanced mechanical performance of these glassy photopolymers. Various resin formulations utilizing different photopolymerization methodologies were systematically designed to support this hypothesis. Furthermore, the mechanical properties of the triazole-based network polymers were investigated via dynamic mechanical thermal analysis (DMTA), and the effect of strain rate and physical ageing on the tensile stress–strain behavior as well as shape memory recovery behavior were examined.

2. Results and Discussion

Several monomer formulations were selected for tensile testing and thermal analysis (Figure 1)—these utilized different reactions, including chain-growth radical photopolymerization and either CuAAC or thiol-ene photopolymerization. Specifically, the triazole-based CuAAC network (T1) and the triazole-based thiol­ene network (T2) were compared with three structurally relevant control networks (C1, C2, C3), as shown in Figure 1B. The first of these comparisons was a triazole-based CuAAC control system (C1), synthesized from a diazide monomer, and designed to have limited freely rotatable bonds between the two triazole units formed during polymerization, as opposed to the nonsterically hindered triazole­based CuAAC system (T1). The second comparative network was a rubbery thiol-ene control system (C2) composed of an analogous network structure to the tough triazole-based CuAAC system (T1) except that thioether cross-links essentially replace the triazole cross-links in an otherwise identical polymer network structure. The last comparison was a bis(glycidylether) bisphenol-A dimethacrylate/triethylene glycol dimethacrylate (BisGMA/TEGDMA) network (C3) formed using free radical chain growth polymerization, which contrasts with the step-growth mechanism used to form all of the other networks. This collection of comparative networks provides an opportunity to assess relatively independently the influence of several molecular characteristics of the polymer network structure and their effects on the mechanical behavior of the triazole­based networks.

Figure 1.

A) Schematic illustration of the triazole-based networks before and after deformation and the ductile behavior of the networks—the shaded section represents the thermally recoverable deformation region for the network. The changes in mechanical properties before physical ageing, after ageing, and after deageing are also shown. B) Structures of the five different monomeric systems (T1, T2, C1, C2, C3), the copper catalyst, and DMPA as a radical-generating photoinitiator. The CuAAC systems (T1, C1) were composed of a molar ratio per functional group of 1:1:0.01:0.02 (alkyne:a zide:CuCl₂[PMDETA]:DMPA), the thiol-ene resins (T2, C2) were composed of a molar ratio per functional group of 1:1:0.01 (thiol:ene:DMPA), and the methacrylate resin (C3) was composed of a weight ratio of 70:30:1 (BisGMA:TEGDMA:DMPA). Each mixture was irradiated at room temperature using 10 mW cm⁻² of UV light (λ_max = 366 nm) for 5 min followed by postcure at 70 °C for 24 h. C) The two photopolymerization chemistries that were utilized to form the triazole-containing networks (T1, C1, T2)—in the first case (T1, C1), the triazoles are formed directly during the photopolymerization by photoinitiated CuAAC polymerization, whereas in the second case (T2), a thiol-ene photopolymerization reaction is used to form the network from triazole-containing monomers.

Each mixture was photocured at ambient temperature (additionally, see Figure S1 of the Supporting Information for in situ kinetics at 60 °C) followed by thermal postcuring at 70 °C for 24 h to eliminate the potential for additional reaction during thermal cycling. With the exception of the BisGMA/TEGDMA system (which suffers from topological restrictions to “full cure”^[48]), this approach ensured that near-quantitative conversion was achieved in each system in Figure 1B (see Figure S2, Supporting Information). As a result of the postcuring treatment, and with the exception of the rubbery thiol-ene polymer (C2) with T_g of 17 °C, the remaining four systems (T1, T2, C1, C3) had a similar room temperature storage modulus ranging from 1.5 to 2.2 GPa and a T_g (as measured at the tan q maximum) well above ambient between 48 and 150 °C when scanned by DMTA through two subsequent heating cycles ranging from 0 to 200 °C, as shown in Figure 2A,B. The minimal changes in T_g of each system between the two thermal cycles confirm that functional group conversions up to the topological limit (i.e., to the point where any remaining functional groups are topologically trapped) were thus successfully achieved as a result of the postcuring.^[47,48] Noticeably, the C3 network presented a much broader glass transition region due to heterogeneity of the networks polymerized via chain-growth while the step-growth networks (T1, T2, C1, C2) all formed relatively homogeneous networks, as indicated by the narrow glass transition region (Figure 2B).

Figure 2.

A,B) Thermomechanical properties of five different photopolymer systems as measured by DMTA for the five different photopolymer systems. The near overlap of the data for the first (solid line) and second DMTA scans (dotted line) indicates that nearly complete cure (or to the topological limit) was achieved after photopolymerization and the subsequent thermal postcure at 70 °C for 24 h. C) Tensile testing of five different photopolymer systems (for replicates, see Figures S3–S7, Supporting Information). Each specimen was heated above T_g for 1 min followed by 3 min of cooling at ambient temperature prior to mechanical testing. D) Polarized optical microscopy images of 100% strained triazole-based CuAAC polymer specimens (T1) viewed through crossed-polarizers at 0°, 45°, 90°, and 135° relative to the tensile strain direction as shown by the arrows.

Surprisingly, tensile stress–strain experiments at ambient temperature (Figure 2C) show that the triazole-based CuAAC (T1) and the triazole-based thiol-ene networks (T2) have strain-to-break values of more than 180%, despite each exhibiting glassy moduli of ≈2 GPa (Figure 2A) and glass transition temperatures 25–40 °C above ambient temperature (Figure 2B). In addition, analysis of the stress–strain data in Figure 2C reveals that the triazole-based CuAAC network (T1) with flexible and relatively long network strands (as given by the average number of rotatable bonds between network junctions, as given in Table 1) has a very high tensile toughness of 72 MJ m⁻³ energy (as measured by the area under the stress–strain curve up to the maximum extension), and this value compares favorably with other tough but uncross-linked polymers such as bisphenol-A polycarbonate, which has been reported^[10] to have a toughness of 50 MJ m⁻³. The triazole-based thiol-ene network (T2) also exhibited high toughness (44 MJ m⁻³). To shed further light on this unusual tensile behavior, the mechanical behavior of the three other control networks (C1, C2, C3), despite having numerous structural features in common with the triazole-containing networks (T1, T2), all have strain-to-break and/or toughness values that are at least an order of magnitude lower than the T1 and T2 networks (Figure 2C and Table 1).

Table 1.

Tensile testing and DMTA mechanical behavior for five different photopolymer systems from Figure 2. Errors listed are the SD obtained from multiple measurements.

	T1	T2	C1	C2	C3
Tg [°C] from tan δ maximum	67	48	85	16	150
DMTA tensile storage modulus @ T=25 °C [GPa]	2.2	2.1	1.6	0.006	1.5
DMTA tensile storage modulus @ Tg + 30 °C [MPa]	7.5	8.7	5.2	3.8	40
Network strand densitya) [mmol g−1]	0.82	0.99	0.54	0.48	3.54
Average molecular weight per strand, Mca) [g mol−1]	1220	1010	1860	2090	280
Average number of rotatable network strand bonds between junctions	30	16.9	6	34	10.5
Tensile yield stress [MPa]	38 ± 2	16 ± 2	36 ± 3	2 ± 0.2b)	49 ± 14
Tensile Young’s modulus [MPa]	1000 ± 100	520 ± 60	650 ± 50	0.02 ±0.002	1500 ± 50
Tensile toughness [MJ m−3]	72 ± 5	44 ± 7	3.3 ± 0.7	1 ±0.2	1.4 ± 0.9
Tensile elongation at break [%]	200 ± 10	210 ± 10	13 ± 2	100 ± 10	5 ± 2

^a)Calculated from the affine theory of rubber elasticity;^[49]

^b)For the C2 network yielding is not observed and so the stress to break is listed.

The high mechanical deformation withstood by the triazole-containing T1 and T2 polymers might suggest that they undergo irrecoverable viscous flow at high extension, but this postulate is disproved by the polarization microscopy image in Figure 2D of the T1 specimen when strained to 100% deformation and then released—bright images are observed when the specimen is viewed through crossed-polarizers at 45° and 135° relative to the strain direction indicating residual strain due to chain orientation because anisotropic minerals affect the polarization of light passing through them, while dark images are found at 0° and 90°.

The DMTA behavior (Figure 2A) reveals that the storage moduli in the rubbery region of the tough T1 and T2 triazole networks were extremely high (see Table 1), suggesting that despite their ability to undergo high deformation, these materials are indeed highly cross-linked networks. To quantify this conclusion, if one neglects loops and dangling chain ends, the classical affine theory of rubber elasticity^[49] allows the network strand density (v) to be calculated from the rubbery modulus (E) from the expression

$$E\text{ = 3}vdRT$$ (1)

where d is the mass density (assumed here for simplicity to be 1000 kg m⁻³), R is the gas constant, and T is the absolute temperature. The network strand densities calculated from these data are 0.82 mmol g⁻¹ for T1 and 0.99 mmol g⁻¹ for T2, as given in Table 1. Additionally, with the approximation that both the triazole-based CuAAC (T1) and triazole-based thiol-ene polymers (T2) are polymerized to full conversion, the theoretical strand densities were also calculated on the basis of the stoichiometry involved in the network structure.^[50] From this analysis, the theoretical values of T1 and T2 are 1.44 and 3.09 mmol g⁻¹, respectively, which are two to three times greater than that estimated from the rubbery modulus (0.82 and 0.99 mmol g⁻¹). This discrepancy may suggest that the networks have topological defects such as loops and dangling ends; however, it should be noted that if the phantom theory of rubber elasticity^[51,52] is used to estimate the cross-link density, then the theoretical values for the T1 and T2 system (1.44 and 3.09 mmol g⁻¹) are much closer to than that obtained from the rubbery modulus (2.45 and 2.39 mmol g⁻¹). In either case, it is clear that both networks are relatively highly cross-linked, as also indicated from the average molecular weights between cross-links, M_c (see Table 1).

The high elongation to break and toughness of these glassy, highly cross-linked triazole systems (T1, T2) suggest efficient intermolecular interactions between functional groups embedded in the network strands, which are able to resist deformation up to the yield stress and then yield and enable cold-drawing, leading to the high toughness values achieved. As discussed above, the triazole unit is associated with the desirable mechanical characteristics in two (T1 and T2) of the three triazole networks. Also, the absence of a triazole ring in the thiol-ene C2 network (which otherwise has a nearly chemically identical, analogous structure to the tough T1 polymer as shown in Figure 1A) leads to a rubbery polymer with a much lower strain to break (≈100% versus 200% for T1), maximum stress, and toughness (see Table 1). Furthermore, the free radical chain growth polymerized C3 also lacks a triazole ring (see Figure 1A) and only has a small number of bonds between the cross-links (see Table 1), leading to low extensibility and toughness (see Figure 2C and Table 1). However, the triazole-based CuAAC control system (C1) has a similar overall cross-link density and triazole density as compared to the T1 system, but the triazole structures in the C1 network are expected to have limited ability to rotate and participate in intermolecular interactions. While the glass transition temperature of the C1 network is 20–40 °C higher than the T1 and T2 network, which limits mobility of networks at ambient temperature, the length of the network strands between junction (i.e., cross-link) points is also much lower. Thus, in contrast to the triazole-containing T1 and T2 networks, the triazole-based C1 network exhibited brittle failure with significantly lower elongation to break (13%). As noted above, the strand density of the C1 network, as calculated from the affine theory of rubber elasticity,^[49] was lower than the other two triazole-based T1 and T2 systems (Table 1) so that the cross-link density per se cannot be the main factor affecting the extensibility and toughness. Thus, the brittle behavior of the triazole-based C1 network is believed to arise from the limited triazole and chain mobility, which may limit the stacking efficiency of triazole rings (e.g., π–π bonding) and hydrogen bonding potential of the triazoles, and would restrict the network strand extensibility. In fact, a fully polymerized triazole-based C1 network consists of only six rotatable bonds between two ideal junction points, which is approximately only one fifth the number of rotatable bonds available in the T1 system (see Table 1). Unsurprisingly, the chain growth polymerized BisGMA/TEGDMA system (C3) also resulted in brittle behavior, as shown in Figure 2C, due to the limited chain mobility associated with the presence of only two rotatable bonds between the trifunctional junction points. These results suggest that it is specifically the presence of the triazole moie-ties along with sufficient freely rotatable bonds between network junctions which are essential elements to the enhanced stiffness and toughness of these highly cross-linked networks.

To address further the superior ductility and toughness of these highly cross-linked triazole-based networks, programmed thermal recovery–strain tests were performed on the T1 polymer (monomer structures shown in Figure 1B). The T1 polymer was first strained to 50% at ambient temperature in the glassy state (resulting in yielding and cold drawing), released from the stress and was then heated above T_g to 80 °C for 1 min for thermal recovery, followed by cooling to ambient for 3 min prior to subsequent tensile testing. Figure 3A presents the overlay of three subsequent stress–strain–thermal recover cycles of the T1 system. In none of the stress–strain tests was fracture observed. Although the stress–strain curve exhibits typical ductile behavior such as yielding, strain softening, and cold drawing, nearly perfect recovery of the initial shape and mechanical properties was achieved after each of two subsequent thermal recovery–strain cycles. This “shape memory–recovery” behavior on the cold-drawn T1 polymer shows that the deformation occurring as a result of 50% strain of this glassy material is thermally reversible, which is extremely surprising and nearly unprecedented for a photopolymerized, thermosetting material with high cross-link density. Additionally, the thermal recovery–strain tests to 100 and 150% strain were performed on the T1 polymer (Figures S9 and S10, Supporting Information)—for specimens deformed to 100% strain, half of these failed on the second deformation cycle while the others broke only on the third strain cycle, whereas all specimens failed on the second deformation cycle at 150% strain. Despite this, the shape of these materials was recovered completely after each heating cycle in which the samples did not break.

Figure 3.

A) Thermal recovery–strain of the triazole-based CuAAC polymer (T1) as measured in the MTS instrument. Each of the three dogbone specimens was first deformed at 0.75 mm min⁻¹ to 50% strain before the stress was released, and the specimen was heated on a hot plate at 80 °C (above the T_g) for 1 min followed by 3 min of cooling at ambient prior to the next tensile testing cycle (for thermal recovery–strain measurements at 100 and 150% strain, see Figures S9 and S10, Supporting Information). B) Microscopy of profile of the shoulder and neck of a nominally 0.25 mm thick T1 dogbone specimen before deformation, after straining to 150% in the MTS instrument, and after stress release and heating to 80 °C. (for images of the shoulder and neck using 50 and 100% strain, see Figure S11, Supporting Information). C) A shape recovery image of a dogbone specimen strained to 150% in the MTS instrument. D) Multiple shape memory-recovery behavior of the T1 polymer. The specimen was deformed at glassy state followed by subsequent thermal recovery cycle.

The thickness and width of the necking shoulder of the cold-drawn T1 specimen before and after being deformed to 150% strain in the glassy state are shown in Figure 3B,C, and these images reveal that the necked specimen had a 50% reduction in thickness and 30% reduction in width. The formation of a necking shoulder also appeared on a 50% strained specimen, and a fully developed neck was observed in a 100% strained specimen (see Figure S11, Supporting Information). Despite the large tensile strain, this fully necked T1 specimen also completely recovered to its original dimensions upon rapid heating slightly above T_g (see Figure 3B,C; Figure S11, Supporting Information). Interestingly, the behavior observed here is significantly different from that normally used with either conventional thermoplastic or thermoset shape memory approaches.^[52,53] In those approaches, the deformations are generally imposed on the specimen at a temperature above the T_g or T_m, and the shape was fixed by cooling the sample below this temperature in the deformed state. The original shape is recovered (usually only partially) by heating above the T_g or T_m, depending on the nature of the material.^[53–55] Contrarily, in the behavior observed here, significant deformation is applied below the T_g and, after the deformation is complete, the shape is recovered by heating above the T_g. In addition to the recovery behavior for the T1 network deformed under tensile strain (Figure 3B,C), a rectangular bar (31 × 3 × 0.25 mm³) composed of the T1 network rapidly deformed into a random coil structure while in the glassy state at ambient temperature also demonstrated complete recovery of its original shape upon heating above T_g over multiple cycles (Figure 3D).

The triazole-based CuAAC polymer (T1) also exhibits time dependent mechanical behavior. As presented in Figure 4A, the Young’s modulus in the glassy state of the network was increased from 780 to 1300 MPa with varying strain rate from 1 to 100% min⁻¹ (Table S1, Supporting Information), as expected of a viscoelastic polymer.^[56] Additionally, the yield stress was effectively increased from 30 to 50 MPa over the same strain rate range. Despite this sharp increase in yield stress, the T1 polymer exhibited fairly high toughness and elongation to break over this entire strain rate range (see details in Table S1, Supporting Information). The effect of strain rate on yielding is often explained by the Eyring theory,^[57,58] which relates the uniaxial yield stress (σ) to strain rate $\left(\dot{\epsilon }\right)$ by the expression

$$\sigma =\frac{2\Delta H}{v_{\text{ flow }}}-\frac{2RT}{v_{\text{ flow }}}\ln \left(\frac{{\dot{\epsilon }}_{\text{o}}}{\dot{\epsilon }}\right)$$ (2)

where v_flow is the activation volume (or volume of the flow unit), ΔH is the activation energy, R is the gas constant, T is the absolute temperature, τ is the shear yield stress, and ${\dot{\epsilon }}_{\text{o}}$ is a constant $\left({\dot{\epsilon }}_{\text{o}}\gg \dot{\epsilon }\right)$. From this equation, the average activation volume of the T1 system was found to be ≈1.7 nm³ (see Figure S12, Supporting Information), which is similar to the volume of one network strand between two centers of alkyne molecules covalently linked with one azide molecule, estimated to be ≈1.2 nm³. Furthermore, the activation volume obtained for the T1 polymer is comparable to that found for other glassy polymers (i.e., 1.5 nm³ for anhydride cross-linked epoxy networks,^[59] 2 nm³ for amine-cross-linked epoxy networks,^[58] and 1.9 nm³ for polycarbonates^[60]).

Figure 4.

A) Tensile testing of the triazole-based CuAAC polymer (T1) at varying apparent strain rates (for replicates, see Table S1, Supporting Information). It should be noted that the strain rate is not the true strain rate but the apparent strain rate due to the fact that the strain varies in different sections of the dogbone specimen after the yield point. B) Physical ageing of the triazole-based CuAAC polymer (T1) conducted at ambient temperature via tensile testing (for replicates, see Table S2, Supporting Information).

Surprisingly, the T1 system exhibited a time-dependent ageing process on the mechanical properties when specimens were stored at ambient temperature well below the T_g (Figure 4B), and this behavior is presumably associated with physical ageing. Unlike the process of chemical ageing, physical ageing occurs due to segment or chain relaxation from a nonthermodynamic equilibrium state to a more densified state resulting in a free volume reduction;^[61] however, this reduction in free volume and any resulting changes in material properties of physically aged samples are generally erased upon heating above T_g.^[61] Physical ageing experiments were conducted by ageing samples at ambient temperature, following an initial heating to 80 °C (>T_g) in an oven for a mini mum of 1 h to erase any thermal history. Significant changes in mechanical performance as indicated by changes in the Young’s modulus, yield stress, and the elongation at break were evident during 16 h of ageing (Figure 4B; see details in Table S1, Supporting Information). However, this physical ageing was thermally reversible by heating above the T_g; indeed after 32 h of physical ageing, specimens heated to 80 °C and rapidly cooled had identical mechanical performance to the nonaged samples, confirming that the changes observed were entirely due to physical ageing.

3. Conclusions

We have demonstrated unique ductile behavior of photopolymerized, densely cross-linked triazole-containing networks with high tensile toughness achieved in the glassy state. More detailed studies of one of these triazole-bearing glassy networks (T1), using thermal recovery–strain tests, showed a complete restoration of physical dimensions and material properties (e.g., yield stress, Young’s modulus, elongation to break, and energy absorption upon deformation) over multiple deformation cycles. These triazole networks exhibited typical time dependent viscoelastic and yielding behaviors. However, rapid physical ageing was observed on storage at ambient temperature, resulting in a more brittle material along with substantially increased yield stress. This physical ageing was reversible by thermal treatment and recovered the desirable toughness of the photopolymerized thermoset.

4. Experimental Section

Materials:

Detailed synthetic procedures of monomers along with nuclear magnetic resonance (NMR) results are presented in the Supporting Information. The monomers used were two thiol monomers, namely, tris[2-(3-mercaptopropionyloxy)ethyl]isocyanurate and bis (6-mercaptohexyl) (1,3-phenylenebis(propane-2,2-diyl))dicarbamate, two azide monomers namely bis(6-azidohexyl) (1,3-phenylenebis(propane-2,2-diyl))dicarbamate and 2,2-bis(azidomethyl)propane-1,3-diyl bis (hexylcarbamate), an yne monomer 1-(prop-2-yn-1-yloxy)-2,2-bis((prop-2-yn-1-yloxy)methyl)butane, two allyl monomers tetraallyl-1,1′-(((((1,3-phenylenebis(propane-2,2-diyl))bis(azanediyl))bis(carbonyl))bis(oxy)) bis(hexane-6,1-diyl))bis(1H-1,2,3-triazole-4,5-dicarboxylate) and 1-(allyloxy)-2,2-bis((allyloxy)methyl)butane, and a 70/30 w/w comonomer mixture of bis(glycidylether) bisphenol-A dimethacrylate and triethylene glycol dimethacrylate. The structures of these monomers are given in Figure 1B, and further details are given in the Supporting Information.

Sample Preparation:

CuAAC networks (T1, C1) were prepared from a stoichiometric mixture of azide and alkyne groups with 1 mol% CuCl₂[PMDETA] and 2 mole% 2,2-dimethoxy-2-phenylacetophenone (DMPA) per functionality as a photoredox pair. Thiol-ene networks (T2, C2) were prepared from a stoichiometric mixture of thiol and ene groups with 1 mol% DMPA per functionality as a photoinitiator. Since these monomer mixtures are not completely miscible, methanol was used to homogenize the mixture, and this was later removed in vacuo, as verified by ¹H-NMR using a Bruker Avance-III 400 MHz spectrometer which showed the solvent remaining was less than 1 wt% prior to any polymerization. BisGMA/TEGDMA networks (C3) were prepared from a 70/30 w/w comonomer mixture of BisGMA and TEGDMA with 1 wt% DMPA as a photoinitiator. For mechanical testing, two consecutive cycles of DMTA scans up to 150 °C were performed on each postcured sample and this confirmed negligible solvent plasticization of the glass transition. Thin polymer specimens (0.25 mm) for DMTA and tensile testing were photopolymerized using 10 mW cm⁻² of UV irradiation (λ_max = 366 nm) for 5 min at ambient temperature followed by postcuring at 70 °C for 24 h. The functional group conversion was monitored via Fourier transform near infrared spectra (Nicolet 8700, Fisher Scientific) prior to the polymerization and after postcuring to ensure near-complete conversion.

¹H-NMR and ¹³C-NMR Experiments:

The experiments were performed in deuterated chloroform (CDCl₃) to determine the purity of the synthesized molecules using a Varian Mercury Plus 400 MHz NMR spectrometer. The number of transients for ¹H and ¹³C are 32 and 512, respectively, and a relaxation time of 1 s was used for the integrated intensity determination of ¹H NMR spectra.

Real-Time Polymerization Kinetics:

The kinetics (presented in the Supporting Information) were analyzed using a Fourier transform infrared spectrometer (Nicolet 8700, Fisher Scientific) in transmission mode, combined with a heating stage set at 60 °C. Irradiation of samples placed between NaCl plates was performed using a light guide connected to a mercury lamp (Acticure 4000, EXFO) with a 365 nm bandgap filter and an irradiance of 10 mW cm⁻². The thiol, ene, and azide peaks of the monomers were monitored in the absorption range between 2602 and 2525 cm⁻¹, 956 and 920 cm⁻¹, and 2250 and 2000 cm⁻¹, respectively, with 4 scans s⁻¹ and at 2 cm⁻¹ resolution.

DMTA Measurements:

These measurements were performed with a TA Instruments Q800 DMA in tension mode at a frequency of 1 Hz, using a heating rate of 3 °C min⁻¹ to a maximum temperature of 150 °C, to yield the storage (E”) and loss moduli (E’) and the tan q (equal to the ratio E”/E’ and relating the energy dissipation relative to the energy stored in the material). The glass transition temperature, T_g, was taken to be the temperature at the peak of the tan q curve. The DMTA specimens were ≈0.25 mm in thickness, 5 mm in width, and 7 mm in length between grips. Each specimen (n = 3) was thermally cycled twice.

Tensile Testing:

It was performed using an MTS Exceed E42 universal testing machine with a 500N load cell to give the engineering stress–strain curve, the Young’s modulus (determined from the initial linear elastic region of the stress–strain curve), the yield stress (the stress at the maximum), the elongation to break, and the toughness (as measured from the area under the stress–strain curve). Dogbone samples were cut or molded (for brittle specimens) similar to the ASTM dogbone die D638-V (see details listed at the end of the Supporting Information), with a 3.15 mm width and 0.25 mm thickness, however the gage length was ≈15 mm rather than the 7.62 mm specified by ASTM D638-V^[62]—according to Saint Venant’s principle,^[63,64] this change should have no significant effect on the data accuracy, as the effect of stress intensity due to the grip and the dogbone shape would be minimized by this change. The specimens were clamped in the grip areas and tested under uniaxial tensile loading at a crosshead speed of 0.75 mm min⁻¹ or otherwise specified such as the cases for 0.15, 7.5, and 15 mm min⁻¹. The stress was calculated from the applied force over the original cross-sectional area of the gage section, while the strain was determined from the ratio of the crosshead displacement and the gage length.

Polarization Microscopy:

It was conducted on a Nikon Optical Microscope using a 4× objective. The specimen was first strained to 100% using a MTS Exceed E42 universal testing machine, and the stress was released. The specimen was then placed between cross-polarized lenses at different angles (0°, 45°, 90°, and 135°) to observe birefringence. Seeing light at angles other than 0° and 90° is the indication of birefringence and anisotropy in the material.