Shape memory epoxy vitrimers based on DGEBA crosslinked with dicarboxylic acids and their blends with citric acid

Abstract

Thermosetting polymers were synthesized from a commercial epoxy resin (diglycidyl ether of bisphenol A, DGEBA) and tricarboxylic (citric, CA) and different dicarboxylic (sebacic, SA and glutaric, GA) acids. Crosslinking of DGEBA–SA and DGEBA–GA systems was achieved using excess epoxy, which was homopolymerized after all the acid groups were consumed. It was found that the properties of the material depend on the diacid length and on the excess epoxy, and with the proper formulation, vitrimers with T_g values ranging from 51 °C to 62 °C, and a high rate of stress relaxation (less than 1 h at 160 °C to achieve 63% of relaxation) could be obtained. Notably, using a mixture of tri-functional CA with SA allowed a reduction in the epoxy excess while maintaining a high T_g value and faster stress relaxation. Three of the formulations were selected and their shape memory performance was studied. Good shape fixity and shape recovery ratios (>99%) were obtained, which indicate an overall good shape memory performance. These properties can be used to create different permanent and temporary shapes on a thermosetting polymer obtained from widely available and affordable raw materials.

---

Introduction

Shape memory epoxies and their nanocomposites have been receiving increasing interest due to their excellent mechanical and thermal properties, very good shape fixing and recovery and their versatility to position the glass transition temperature (T_g) in a convenient temperature range.^1–29 In a typical case, a temporary shape is produced at temperatures above the T_g by exerting a force. This shape is fixed by cooling below the T_g while keeping the force. The initial shape is recovered by removing the force and heating again above the T_g. Alternatively, keeping the temporary shape during the second heating stage enables the material to behave as a force actuator. Several practical applications have been devised for both types of responses. In recent years, new formulations have been developed, including blends and nanocomposites that are different in nature, which expand the potential applications of shape memory epoxies. Examples are the possibility to produce remote activation by alternating magnetic fields, electrical currents or light irradiation,^6,13–16,18 and materials with multiple shape memory behavior,^{14,16,17,21,26} or materials that exhibit the desired combination of high strength and elongation.^5,9,23–25

Leibler and co-workers recently introduced the concept of vitrimers as a new class of cross-linked polymeric materials, which exhibit an interchange of covalent bonds above the temperature T_v (located above T_g).^30–32 The importance of this new type of cross-linked polymer is that its behaviour above T_v makes self-healing and recycling possible, and the relaxation of residual stresses. Shape memory materials based on vitrimers enable the production of complex temporary shapes outside a mould. This is simply achieved by generating the permanent shape at a suitable temperature above T_v for the necessary period of time to erase the stress.^30,31,33 This permanent shape is fixed by cooling below T_v and can be used as the starting point to generate temporary shapes in the temperature range located between the T_g and T_v. The use of vitrimers can also provide a flexible strategy to achieve a multishape memory effect. Layers of premade epoxy vitrimers with different thermal transitions self-assemble through covalent bonds by hot-pressing at a temperature above T_v. This provides an easy way to obtain spatial control of the shape memory of 3D structures.³⁴

Although many different chemistries have been developed to generate vitrimers,³⁵ the focus of this study is placed on the chemistry explored by Leibler and co-workers.^30–32 Specifically, they showed that epoxies containing 2-hydroxyester groups produced by the epoxy–carboxylic acid reaction (Fig. 1a), exhibit a fast interchange of covalent bonds (Fig. 1b) above a characteristic temperature (T_v), in the presence of Zn salts. Although the fast transesterification of 2-hydroxyester bonds produced by epoxy–carboxylic acid has been reported in the literature,^36,37 the group of Leibler was the first to use this chemistry to develop the concept of vitrimers. Vitrimers based on this chemistry are receiving increasing attention in the literature with the use of different formulations and catalysts.^34,38–43 Our specific aim is to develop formulations based on DGEBA (diglycidyl ether of bisphenol A), which is the most used epoxy monomer for practical applications and is commercially available, and cheap poly(carboxylic acids).

Fig. 1 (a) 2-Hydroxyester formation through the reaction between an epoxy and a carboxylic acid groups and (b) transesterification reaction between two 2-hydroxyester links.

In order to use the resulting vitrimer for shape memory applications, it is desirable to develop a network with a T_g located above room temperature (RT) in order to employ cooling to RT to fix the temporary shape. If there are no extra requirements for the use of the material, a T_g located somewhere in the range of 50–80 °C would have the advantage of minimizing the heat needed to produce the temporary shape. This also decreases the temperature gap between the high and the low (RT) temperature of the thermal cycle, which is a factor that enables an increase in the fixity ratio by decreasing the shrinkage of the material during the cooling step. Several fatty acids have been reported in the literature to cure DGEBA in order to produce vitrimers. However, all of them present T_g values significantly lower than 40 °C. The use of a commercial blend of di- and tri-carboxylic fatty acids (Pripol 1040) led to a material with a T_g close to RT.^30–32,41 Among the family of typical fatty diacids, only adipic acid (C6) yielded a material with a T_g close to 40 °C when it is employed in a stoichiometric ratio with respect to epoxy groups.⁴⁰ Sebacic acid (C10) has also been employed to cure DGEBA (1 : 1 stoichiometry) to produce epoxy vitrimers for shape memory applications.^34,39 The glass transition temperature of the cured material is about 30 °C which requires a temperature of 0 °C to set the temporary shape.³⁴ Another problem that has not been paid enough attention is the use of a fatty diacid (A2) in stoichiometric proportion with DGEBA (B2), which would ideally produce a linear polymer through the epoxy–acid reaction. Transesterification reactions lead to gelation and formation of a polymer network.³⁶ However, a significant sol fraction is present at the end of polymerization which is not desirable for practical applications.³⁶ An increase in the crosslinking density, and hence in the T_g value can be achieved by employing an excess of epoxy groups and their homopolymerization subsequently to the epoxy–acid reaction. Tertiary amines are efficient catalysts/initiators to produce both reactions in succession.^9,37

Another possible crosslinker of DGEBA is citric acid, a tricarboxylic acid present in lemon, lime and orange juice, which is nowadays produced from sugars through a microbiological process.⁴⁴ Citric acid has already been employed as an aqueous solution to crosslink epoxidized soybean oil (ESO),^38,42 and an epoxidized sucrose soyate.⁴⁵ The 2-hydroxyesters generated by the reaction of citric acid with ESO showed an acceptable transesterification rate in the absence of any catalyst.^38,42 Citric acid is not soluble in DGEBA but homogeneous solutions can be prepared in the presence of a fatty acid, as is reported in this study.

In this work, we present two different strategies to synthesize shape memory epoxy vitrimers based on DGEBA and polycarboxylic acids, with the T_g located at temperatures high enough to enable the temporary shape to be set at 20 °C. The first strategy consisted of dissolving citric acid in melted sebacic acid (in a 50 : 50 ratio of carboxylic groups), followed by the addition of DGEBA (10% over stoichiometry) to enable the formation of a homogeneous solution that was thermally cured. An imidazole (1-methylimidazole, 1MI) was added with the aim of catalyzing the transesterification reactions, which are necessary to relax the stress when converting the temporary shape into a permanent shape. 1MI also proved to be a good catalyst for the epoxy–acid reaction. The second strategy was based on the preparation of DGEBA–sebacic acid and DGEBA–glutaric acid blends with an excess of epoxy groups. This produced two sequential reactions in the presence of a tertiary amine (1MI in our case): the first reaction was the epoxy–acid reaction and the second the homopolymerization of the excess epoxy groups.⁶ By selecting an adequate epoxy excess it was possible to tune both the T_g and the rate of the transesterification reactions to desired values. The shape memory properties of the resulting vitrimers are reported with particular emphasis on the possibility of transforming temporary shapes into permanent shapes.

Experimental

Materials

A diglycidyl ether of bisphenol A (DGEBA) based diepoxy monomer (D.E.R. 332 from Dow Chemical; epoxide equivalent weight EEW = 174 g eq.⁻¹), citric acid (CA) monohydrate (C₆H₈O₇·H₂O; >98%; acid equivalent weight AEW = 70 g eq.⁻¹), glutaric acid (GA, C₅H₈O₄; >98%; AEW = 94 g eq.⁻¹) and 1-methylimidazole (1MI; 99%) were purchased from Sigma-Aldrich. Sebacic acid (SA; C₁₀H₁₈O₄; >98%; AEW = 101 g eq.⁻¹) was kindly supplied by Castor Oil (Buenos Aires, Argentina). All reagents were used as received without any further purification, and their chemical structures are shown in Fig. 2.

Fig. 2 Chemical structure of the reagents. (a) DGEBA, n = 0.028; (b) citric acid; (c) sebacic acid; (d) glutaric acid; and (e) 1-methylimidazole.

Synthesis of epoxy–acid vitrimers

The formulations used to obtain the networks are detailed in Table 1 (see Table S1 in the ESI† for further details). The experimental procedure carried out in each case is described below.

Table 1 Formulations used to synthesize the epoxy–acid vitrimers. Base: 1.0 acid milliequivalents

Label	DGEBA, meq.	CA·H2O, meq.	SA, meq.	GA, meq.	1MI, mmol	R
CS11	1.10	0.50	0.50	0	0.055	1.1
S12	1.20	0	1.00	0	0.060	1.2
S15	1.50	0	1.00	0	0.075	1.5
S20	2.00	0	1.00	0	0.100	2.0
G15	1.50	0	0	1.00	0.075	1.5

---

DGEBA/CA–SA/1MI. CA·H₂O and SA powder were mixed in a glass vial and heated in a convection oven at 140 °C. Once the SA melted, the vial was manually agitated several times until complete dissolution of CA, and the oven temperature was lowered to 125 °C. At this time, the required amount of DGEBA was placed in the oven. After 10 min, the pre-heated DGEBA was added to the CA–SA solution, and the mixture was manually mixed to obtain a clear solution. The DGEBA/CA–SA blend was allowed to react at 125 °C for 4 minutes before 1MI was added, and subsequently the clear, pale yellow mixture was casted in a disposable aluminium mould and cured for 3 h at 120 °C and 6 h at 160 °C.

DGEBA/SA/1MI and DGEBA/GA/1MI. DGEBA and SA (or GA) were heated separately in aluminium disposable beakers placed in an oven at 140 °C (105 °C), until the diacid melted completely. The oven temperature was subsequently lowered to 130 °C (100 °C), and DGEBA and the diacid were mixed together with a glass rod. The blend was allowed to react for 4 minutes, and then 1MI was added and manually mixed. Finally, the system was cured for 3 h at 120 °C and 6 h at 160 °C.

Methods

DMTA. Dynamic-mechanical thermal analysis (DMTA) was performed in the dynamic mode, employing a TA Instruments Q800 Dynamic Mechanical Analyzer with a three-point bending geometry, at a frequency of 1 Hz. Specimens of rectangular cross-sections (1.5 mm × 5 mm) were tested, with a 20 mm span. Tests were carried out at a heating rate of 2 °C min⁻¹, from 20 °C to 180 °C, following the evolution of the storage modulus (E′) and loss modulus (E′′). Alpha relaxation temperatures (T_α) were defined as the maxima of the tan δ (=G′′/G′) peaks.

DSC. Dynamic scanning calorimetry (DSC) tests were performed on a Shimadzu DSC 50 calorimeter under a nitrogen atmosphere, at a heating rate of 10 °C min⁻¹, from room temperature to 170 °C. Glass transition temperature (T_g) values were measured at the onset of the transition.

Stress relaxation. Shear stress relaxation was determined with an AntonPaar Physica MCR 301 rheometer, using specimens with a rectangular cross section (1.5 mm × 5 mm) and a distance of 18 mm between the clamps. Isothermal tests were carried out at different temperatures in the range of 120 °C to 160 °C. When the sample reached the desired temperature, the torsion deformation γ = 5%, which corresponds to a deflection angle (θ) of about 30° was applied and the stress relaxation, which is expressed as the relaxation modulus, G(t), was recorded.

Shape memory. Shape memory performance was studied with an AntonPaar Physica MCR 301 rheometer, using specimens with the same dimensions as that used for stress relaxation experiments. In these tests, the materials' behaviour during several cycles of temporary shape fixation and shape recovery was analyzed. Each cycle was comprised of the following steps:

- A constant γ = 5% was applied at 100 °C and maintained upon cooling to 20 °C.

- After equilibrating at 20 °C for 5 min, the specimen was released by setting torque to τ = 0, and the evolution of γ was measured for the following 10 min.

- Temperature was increased, keeping τ = 0, to recover the permanent shape, and the evolution of γ was measured for 10 min after reaching T = 100 °C.

In order to characterize the shape-memory behaviour of the networks, shape fixity (R_f) and shape recovery (R_r) ratios were calculated from these tests. A detailed explanation of the calculations is given in the ESI.†

Additionally, tests with higher deformation values (γ = 10, 15, 20, 25, 30 and 40%, which correspond to approximately θ = 60, 90, 120, 150, 180 and 240°, respectively) and one single cycle were also performed to evaluate the same parameters.

Results and discussion

Thermal and dynamic-mechanical properties

The dynamic-mechanical properties of the epoxy–acid networks are shown in Fig. 3. Curves of E′ vs. T display the typical shape for thermosetting polymers, with a high modulus at low temperatures, which corresponds to the glassy state, followed by an α relaxation and rubber region characterized by a lower E′ value. All formulations have high flexural storage moduli in the glassy state (1.5–3.0 GPa), which are characteristic of epoxy networks.^46,47 The plateau in the rubbery state, with the E′ value increasing with T also indicates the presence of covalent crosslinks, which demonstrate that the epoxy excess effectively underwent homopolymerization. Significant differences in T_α values were obtained in the DMA tests, which range from 45 °C for S12 to 78 °C for G15. The variations of T_α can be explained by differences in the concentration of phenyl groups, hydroxyl moieties, crosslink points and aliphatic long chains (particularly from SA). It is known that an increment in the former three factors leads to an increase in T_α whereas the fourth factor has the opposite effect.⁴⁶ An analysis of the SA cured networks shows that T_α increases from 45 °C for R = 1.2 to 66 °C for R = 1.5, and to 73 °C for R = 2.0. It can be seen (Table S2 in ESI†) that the concentration of phenyl groups increases, and that of SA aliphatic chains decreases in the same way, whereas the crosslinking density is an increasing function of R, since in these materials all the crosslinking points can only be created by the homopolymerization of the epoxy excess. The effect of shortening the aliphatic chain length can be inferred from a comparison of the S15 and G15 networks, which provides clear evidence that the presence of long SA chains enhances the local mobility and lowers the T_α value significantly. Finally, it was also evidenced that the addition of CA increased T_α up to 73 °C for an R value as low as 1.1, due mainly to the higher crosslinking density and –OH concentration. It is also worth mentioning that CS11 presents a broader transition, with the tan δ peak showing a shoulder at lower temperatures, which probably originates from heterogeneity at the molecular level due to the use of a mixture of acids as the crosslinker.

Fig. 3 (a) Storage modulus (E′) and (b) damping factor (tan δ) of the vitrimers as a function of temperature.

The DSC curves (Fig. 4) show a very similar trend for T_g values, ranging from 30 °C for S12 to 62 °C for G15 and are 15–20 °C lower than the T_α values measured by DMTA. S15, S20 and CS11 showed intermediate values of 51 °C, 54 °C and 53 °C, respectively, which confirm the DMTA analysis results. With the exception of S12, these T_g values enable temporary shape setting at RT.

Fig. 4 DSC thermograms of the vitrimers (exothermic heat flow points upwards).

Stress relaxation

The stress relaxation behaviour of the vitrimers was assessed to evaluate the transesterification reaction rate, which is responsible for reshuffling the network topology and setting a new permanent shape.^{30–32,38–42,48} The stress relaxation properties measured at 160 °C for each formulation are shown in Fig. 5. A comparison of the uncatalyzed DGEBA/CA–SA network (R = 1.0) with the CS11 vitrimer shows the strong effect of the 1MI catalyst, which enhances the transesterification rate by around three orders of magnitude and enables the permanent re-shaping of the vitrimer in about 1 h.

Fig. 5 Stress relaxation of the vitrimers at 160 °C. The stress relaxation test carried out on a stoichiometric DGEBA/SA–CA network without 1MI is shown here for comparison purposes.

Besides the use of a transesterification catalyst, there are other network parameters that can affect the relaxation rate, and they were also analyzed. Firstly, when the DGEBA–SA–1MI systems are compared the results show that stress relaxation proceeds slower for higher values of R. This is a direct consequence of the lower ratio of 2-hydroxyester groups to non-reversible ether links (generated by homopolymerization) for higher R values. It should be noted that the concentration of 1MI increases with R (Table S2†). However experimental evidence shows that the effect of the 2-hydroxyester concentration is much stronger than that of a slight increase in catalyst concentration.

The comparison of G15 and S15 shows that the latter takes about 45% more time to achieve the same level of relaxation (G/G₀ = 0.37). One of the reasons for this can be the higher concentrations of 1MI and 2-hydroxyester groups in G15, however a complementary explanation for the differences between relaxation rates could be found in the distance between adjacent reacting groups (2-hydroxyesters). Since the length of the GA aliphatic chain is about one half of that of SA, the 2-hydroxyester linkages in the G15 network are much closer to each other, and hence less hindered to react. These two effects add up, but they cannot be distinguished and quantified individually from these experiments.

The CS11 vitrimers showed the fastest relaxation rate of the series at 160 °C, as a result of the combined effects of a higher concentration of 2-hydroxyester moieties and the proximity between them. The contribution of –OH concentration is quite clear here, since CS11 has a lower 1MI concentration than S15 and G15. It is important to stress that CA provides not only three –COOH groups (yielding three 2-hydroxyesters) within less than 6 carbon atoms of distance, but also an additional hydroxyl group. It is known that –OH concentration plays a substantial role in the transesterification reaction,^31,38 and hence it can be assumed that these additional –OH groups contribute to increase the reaction rate. The higher functionality of CA allows a highly crosslinked material to be obtained without the need for a high value of R, and keeps a convenient ratio of 2-hydroxyester groups to ether links generated through the homopolymerization reaction.

DMTA, DSC and stress relaxation tests at 160 °C showed that the CS11, S15 and G15 networks have adequate combinations of properties (i.e. fast relaxation rates and T_g values above 50 °C). Hence, further characterization of the transesterification reactions was carried out for these vitrimers by measuring the stress relaxation at different temperatures (Fig. 6), and obtaining the activation energy (E_a) at two different relaxation values. E_a values were calculated from the relaxation time–temperature relation for each vitrimer, using an Arrhenius-type equation (eqn (1)):

ln τ = E_a/(RT) − ln A (1)

where, τ is the time needed to attain a given stress relaxation value, A is a pre-exponential factor and R is the gas constant.

Fig. 6 G/G₀ vs. time at different temperatures for the CS11 (a), S15 (b) and G15 (c) vitrimers.

Eqn (1) provides an excellent fitting of the experimental data (ESI, Fig. S2†). Activation energies can be used to predict the relaxation stress rate within a temperature range, and also indicate the sensitivity of the transesterification reaction rate to a change in temperature. The E_a values are shown in Table 2, and range from about 90 kJ mol⁻¹ for S15 to around 120 kJ mol⁻¹ for G15, with CS11 showing an intermediate value of ∼107 kJ mol⁻¹. The E_a values for CS11 and G15 are somewhat higher than those reported for other catalyzed vitrimers³² and those of uncatalyzed ESO–CA networks,³⁸ and significantly lower than that reported for polylactide vitrimers with stannous octoate as the transesterification catalyst.⁴³ Additionally, the T_v values of CS11, S15 and G15 were determined to be 105 °C, 107 °C and 117 °C, respectively.

Table 2 Activation energy (E_a) for the transesterification reactions and topology freezing temperature (T_v), calculated using eqn (1) from the tests shown in Fig. 6

	Ea (G/G0 = 0.6), kJ mol^−1	Ea (G/G0 = 0.37), kJ mol^−1	Ea (average), kJ mol^−1	Tv, °C
CS11	107.7	106.0	106.9	105
S15	91.2	87.5	89.3	107
G15	115.1	122.6	118.8	117

---

Shape memory

Fig. 7 shows the strain- and temperature–time curves obtained in the shape memory experiments for the CS11, S15 and G15 formulations, and Table 3 summarizes the fixity ratio (R_f) and recovery ratio (R_r) for each individual cycle. In general, fixity ratios above 99% were measured for all the vitrimers, which is a very good result. The recovery ratios were also very high, which demonstrate the ability of these polymers to go back to their permanent shape.

Fig. 7 Deformation (γ) and temperature vs. time curves for the 5-cycle shape memory tests for CS11, S15 and G15, with γ = 5%.

Table 3 R_f and R_r for CS11, S15 and G15 vitrimers, for γ = 5% and N = 1 to 5

	N	1	2	3	4	5
CS11	Rf (%)	100.0	100.0	100.0	100.0	100.0
	Rr (%)	98.4	99.2	99.3	99.3	99.4
S15	Rf (%)	99.6	99.6	99.7	99.6	99.5
	Rr (%)	99.0	99.7	98.7	100.0	99.8
G15	Rf (%)	99.8	99.8	99.8	99.8	99.8
	Rr (%)	99.5	99.7	99.8	99.7	99.8

---

It is worth mentioning the repeatability of the results, which show no significant variation for different cycles. Differences among the formulations are also very low.

Tests with higher deformation levels were also performed, which are shown in Fig. 8. The fixity and recovery ratios calculated for these tests are summarized in Table 4. The tests show that the R_f and R_r values are very high regardless of the deformation applied to the material. An × indicates material failure with the corresponding applied strain at 100 °C. Decreasing this temperature could enable the extension of the deformation range without failure.⁵ Overall, these materials show excellent fixity and recovery abilities when compared to other epoxy shape memory polymers.^{1–5,7–13,17,18,24–29} It should be recalled, however, that shape memory experiments were carried out in torsion mode, hence the deformation values cannot be directly compared with other experiments where only uniaxial tensile or compressive stresses are applied.

Fig. 8 Shape memory tests with different deformation values: (a) CS11; (b) S15; and (c) G15. (d) G15 sample with an applied deformation of 30%, which corresponds to a deflection angle of more than 180°.

Table 4 R_f and R_r for the CS11, S15 and G15 vitrimers for different values of γ and N = 1. An × symbol means that the specimen broke when the deformation was applied. Shape memory tests with γ = 25% were not carried out on S15 and G15, since tests with γ = 30% could be performed

	γ (%)	5	10	15	20	25	30	40
CS11	Rf (%)	100.0	100.0	100.0	100.0	100.0	×	—
	Rr (%)	98.4	98.5	99.6	99.2	99.0	×	—
S15	Rf (%)	99.6	99.5	99.6	99.6	—	99.7	×
	Rr (%)	99.0	99.0	99.0	99.3	—	100.0	×
G15	Rf (%)	99.8	99.9	99.9	99.9	—	99.9	×
	Rr (%)	99.5	99.4	98.7	98.4	—	98.8	×

---

Temporary and permanent shape changes

The combination of temporary and permanent shape changes for the CS11 vitrimer were evaluated both qualitatively and quantitatively. For the qualitative test (Fig. 9; a movie clip showing this experiment is also available as ESI†), the temporary shape was set by heating the sample at 90–100 °C using hot water and manually twisting it before cooling to RT. Afterwards, the permanent shape #1 was recovered by submerging the sample in hot water again. To set the permanent shape #2, the sample was twisted in the same way as described for the temporary shape setting, but it was subsequently subjected with clamps to preserve the shape, and placed in a convection oven at 160 °C for 1 h. After the thermal treatment, it was no longer possible to recover the permanent shape #1 by heating the sample above the T_g. Instead, a new temporary shape could be obtained by manually applying a stress before the specimen cooled below the T_g. Finally, a new immersion in hot water led to the permanent shape #2 again.

Fig. 9 Qualitative assessment of shape memory and permanent shape changing for CS11 vitrimer.

The quantitative demonstration of both temporary and permanent shape changes is shown in Fig. 10. In the first step, a temporary shape with a deformation γ = 5% was programmed at 100 °C, and then the specimen was cooled to 20 °C. This shape was retained upon releasing the specimen (γ = 0), however when temperature was raised above the T_g, the permanent shape #1 (corresponding to γ = 0%) was recovered. The next step involved the modification of the permanent shape. For this purpose, a deformation of 5% was fixed for 1 h at a temperature of 160 °C, and then the sample was cooled again maintaining γ = 5%. The deformation corresponding to the permanent shape #2 was measured as 4.7% when the sample was heated to 100 °C without applying any torque (τ). The discrepancy between the new permanent γ and the fixed value of 5% is due to incomplete stress relaxation for the temperature and time used, and is in agreement with a stress relaxation of 91% at 1 h for the same temperature (Fig. 5 and 6). Finally, a new temporary shape (γ = 0%) was fixed at 100 °C. This shape was maintained at 20 °C when the sample was released, and when the temperature surpassed the T_g, the sample went back to the permanent shape #2.

Fig. 10 Combined shape memory and stress relaxation experiment for the CS11 vitrimer. The scheme at the left illustrates that the rate of transesterifications becomes negligible below T_v.

Conclusions

The results presented herein show the possibility of synthesizing thermosetting vitrimers from affordable and commercially available raw materials, such as DGEBA, inexpensive poly(carboxylic acids), and an imidazole as a catalyst for both the epoxy–acid reaction and the homopolymerization of the epoxy excess. Interestingly, the imidazole was an excellent catalyst for the transesterification reaction, which speeded up the re-shuffling of the polymeric networks and enabled permanent re-shaping of the material within a reasonable timeframe. The shape memory properties of these vitrimers were evaluated through shape fixity and recovery ratios, and overall excellent results were obtained. Regarding the differences between formulations, the network synthesized with glutaric acid has better properties than that obtained using sebacic acid, though the latter is more environmentally friendly due to the renewable origin of sebacic acid. Moreover, the use of citric acid (dissolved in sebacic acid) is very advantageous, since it leads to highly crosslinked networks, decreases the need for an epoxy excess and accelerates the stress relaxation.

Acknowledgements

Authors would like to express their gratitude to the National Research Council (CONICET), the National Agency for the Promotion of Science and Technology (ANPCyT) and the National University of Mar del Plata (UNMdP) for the funding. FIA would also like to thank Fundación Bunge y Born (Argentina) for the financial support.

Notes and references

1. T. Xie and I. A. Rousseau, Polymer, 2009, 50, 1852–1856.
2. I. A. Rousseau and T. Xie, J. Mater. Chem., 2010, 20, 3431.
3. D. M. Feldkamp and I. A. Rousseau, Macromol. Mater. Eng., 2010, 295, 726–734.
4. Y. Liu, C. Han, H. Tan and X. Du, Mater. Sci. Eng., A, 2010, 527, 2510–2514.
5. A. B. Leonardi, L. A. Fasce, I. A. Zucchi, C. E. Hoppe, E. R. Soulé, C. J. Pérez and R. J. J. Williams, Eur. Polym. J., 2011, 47, 362–369.
6. J. Puig, C. E. Hoppe, L. A. Fasce, C. J. Pérez, Y. Piñeiro-Redondo, M. Bañobre-López, M. A. López-Quintela, J. Rivas and R. J. J. Williams, J. Phys. Chem. C, 2012, 116, 13421–13428.
7. S. Pandini, F. Bignotti, F. Baldi and S. Passera, J. Intell. Mater. Syst. Struct., 2013, 24, 1583–1597.
8. Y. Dong, Q.-Q. Ni, L. Li and Y. Fu, Mater. Lett., 2014, 132, 206–209.
9. N. Zheng, G. Fang, Z. Cao, Q. Zhao and T. Xie, Polym. Chem., 2015, 6, 3046–3053.
10. M. Fan, X. Li, J. Zhang and J. Cheng, J. Therm. Anal. Calorim., 2015, 119, 537–546.
11. Y. Dong, Q.-Q. Ni and Y. Fu, Composites, Part A, 2015, 72, 1–10.
12. D. Santiago, X. Fernández-Francos, F. Ferrando and S. De la Flor, J. Polym. Sci., Part B: Polym. Phys., 2015, 53, 924–933.
13. W. Wang, D. Liu, Y. Liu, J. Leng and D. Bhattacharyya, Compos. Sci. Technol., 2015, 106, 20–24.
14. Y. Wu, J. Hu, C. Zhang, J. Han, Y. Wang and B. Kumar, J. Mater. Chem. A, 2015, 3, 97–100.
15. X. Liu, H. Li, Q. Zeng, Y. Zhang, H. Kang, H. Duan, Y. Guo and H. Liu, J. Mater. Chem. A, 2015, 3, 11641–11649.
16. L. Yu, Q. Wang, J. Sun, C. Li, C. Zou, Z. He, Z. Wang, L. Zhou, L. Zhang and H. Yang, J. Mater. Chem. A, 2015, 3, 13953–13961.
17. H. Guo, Y. Li, J. Zheng, J. Gan, L. Liang, K. Wu and M. Lu, RSC Adv., 2015, 5, 67247–67257.
18. A. B. Leonardi, J. Puig, J. Antonacci, G. F. Arenas, I. A. Zucchi, C. E. Hoppe, L. Reven, L. Zhu, V. Toader and R. J. J. Williams, Eur. Polym. J., 2015, 71, 451–460.
19. T. Tanpitaksit, eXPRESS Polym. Lett., 2015, 9, 824–837.
20. T. Tsujimoto, eXPRESS Polym. Lett., 2015, 9, 757–763.
21. Y. Dong, H. Xia, Y. Zhu, Q.-Q. Ni and Y. Fu, Compos. Sci. Technol., 2015, 120, 17–25.
22. J. Karger-Kocsis and S. Kéki, in Multifunctionality of Polymer Composites, Elsevier, 2015, pp. 822–841.
23. Y. Liu, J. Zhao, L. Zhao, W. Li, H. Zhang, X. Yu and Z. Zhang, ACS Appl. Mater. Interfaces, 2016, 8, 311–320.
24. S. Ponyrko, R. K. Donato and L. Matějka, Polym. Chem., 2016, 7, 560–572.
25. A. Arnebold and A. Hartwig, Polymer, 2016, 83, 40–49.
26. X. Kuang, G. Liu, X. Dong and D. Wang, Polymer, 2016, 84, 1–9.
27. X. Wu, X. Yang, Y. Zhang and W. Huang, J. Mater. Sci., 2016, 51, 3231–3240.
28. L. Chen, W. Li, Y. Liu and J. Leng, Composites, Part B, 2016, 91, 75–82.
29. C. Li, J. Dai, X. Liu, Y. Jiang, S. Ma and J. Zhu, Macromol. Chem. Phys., 2016, 217, 1439–1447.
30. D. Montarnal, M. Capelot, F. Tournilhac and L. Leibler, Science, 2011, 334, 965–968.
31. M. Capelot, D. Montarnal, F. Tournilhac and L. Leibler, J. Am. Chem. Soc., 2012, 134, 7664–7667.
32. M. Capelot, M. M. Unterlass, F. Tournilhac and L. Leibler, ACS Macro Lett., 2012, 789–792.
33. Q. Zhao, W. Zou, Y. Luo and T. Xie, Sci. Adv., 2016, 2, e1501297.
34. Z. Pei, Y. Yang, Q. Chen, Y. Wei and Y. Ji, Adv. Mater., 2016, 28, 156–160.
35. W. Denissen, J. M. Winne and F. E. Du Prez, Chem. Sci., 2016, 7, 30–38.
36. K. Dušek and L. Matějka, in Rubber-Modified Thermoset Resins, ed. C. K. Riew and J. K. Gillham, American Chemical Society, Washington, D.C., 1984, vol. 208, pp. 15–26.
37. C. E. Hoppe, M. J. Galante, P. A. Oyanguren and R. J. J. Williams, Macromol. Mater. Eng., 2005, 290, 456–462.
38. F. I. Altuna, V. Pettarin and R. J. J. Williams, Green Chem., 2013, 15, 3360–3366.
39. Z. Pei, Y. Yang, Q. Chen, E. M. Terentjev, Y. Wei and Y. Ji, Nat. Mater., 2013, 13, 36–41.
40. Y. Yang, Z. Pei, X. Zhang, L. Tao, Y. Wei and Y. Ji, Chem. Sci., 2014, 5, 3486.
41. K. Yu, P. Taynton, W. Zhang, M. L. Dunn and H. J. Qi, RSC Adv., 2014, 4, 10108.
42. F. I. Altuna, J. Antonacci, G. F. Arenas, V. Pettarin, C. E. Hoppe and R. J. J. Williams, Mater. Res. Express, 2016, 3, 45003.
43. J. P. Brutman, P. A. Delgado and M. A. Hillmyer, ACS Macro Lett., 2014, 3, 607–610.
44. M. Sauer, D. Porro, D. Mattanovich and P. Branduardi, Trends Biotechnol., 2008, 26, 100–108.
45. S. Ma and D. C. Webster, Macromolecules, 2015, 48, 7127–7137.
46. J.-P. Pascault, H. Sautereau, J. Verdu and R. J. J. Williams, Thermosetting Polymers, Marcel Dekker, New York, 1st edn, 2002.
47. J.-P. Pascault and R. J. J. Williams, in Epoxy Polymers, ed. J.-P. Pascault and R. J. J. Williams, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, 2010, pp. 1–12.
48. W. Denissen, G. Rivero, R. Nicolaÿ, L. Leibler, J. M. Winne and F. E. Du Prez, Adv. Funct. Mater., 2015, 25, 2451–2457.

---

Footnote

† Electronic supplementary information (ESI) available: Complete vitrimers formulations, details on the calculation of R_f and R_r, network parameters, and calculation of activation energy and T_v. See DOI: 10.1039/c6ra18010h

---