Introduction

Shape-memory polymers (SMPs) are a class of materials that can change their shape when an external stimulus is applied. These polymers can be processed to have a permanent shape by conventional techniques, and then deformed and fixed in a new or temporary shape that can remain stable until the stimulus is applied [1]. This shape changing is called the shape-memory effect (SME) and it is usually driven by heat. However, it can also be driven by light, a magnetic field, or an electrical current. These smart materials have attracted a lot of interest in recent years due to their wide range of applications including self-deployable structures for aerospace applications, biomedical devices, or smart fiber and fabrics [2–4].

The most common SMPs are thermoplastics because they are extremely ductile and easy to process [5, 6], and, therefore, suitable for a wide variety of potential applications. There are also some examples of SMP composites with improved mechanical performance [7, 8]. However, in recent years, a number of studies have reported shape-memory thermosets with excellent mechanical and shape-memory properties, such as epoxy-based or acrylate-based SMPs [9, 10].

Epoxy resins are widely used in many applications (coatings, adhesives, or matrices in composites) because of their chemical resistance, thermal stability, and good mechanical properties [11]. However, their brittle behavior and low elongation at break greatly limit their use as SMPs. Many studies have focused on overcoming this limitation by selecting suitable crosslinking agents which enhance local mobility and deformability of the networks [12–15]. Although the toughness of epoxies is greatly improved in such networks, the mechanical properties in terms of tensile strength may remain a limitation. So new kinds of shape-memory epoxy polymer need to be found that combine high values of tensile strength and maximum deformation above and below the transition temperature for more mechanically demanding applications such as thermomechanical actuators.

Our research group recently reported the use of a low molecular weight hyperbranched polymer as crosslinking agent for epoxy-based SMPs [9]. The results revealed that the presence of hyperbranched polymer led to good thermal, mechanical, and shape-memory properties. These results indicate that other hyperbranched polymers should be used if shape-memory epoxy polymers are to be further improved. To this end, in this study, two hyperbranched polymers with different molecular weights have been used as crosslinking agents to determine the effect of the hyperbranched structure and molecular weight on the thermomechanical and shape-memory properties of epoxy-based SMPs.

Materials and Experimental Methods

Materials

Diglycidyl ether of bisphenol A (Araldite GY 240, Huntsman) with a weight per epoxy equivalent of 182 g/mol was used as the base epoxy resin (Fig. 1a). An aliphatic diamine (Jeffamine D400, Huntsman) with M w = 430 g/mol (Fig. 1b) and two different commercial hyperbranched poly(ethyleneimine)s Lupasol® FG with M w = 800 g/mol and Lupasol® PR8515 with M w = 2000 g/mol (BASF) (Fig. 1c) were used as crosslinking agents.

Fig. 1
figure 1

a Chemical structure of DGEBA (n = 0.08); b chemical structure of Jeffamine D400 (x = 6.1); c chemical structure of Lupasol® FG

Full size image

Various formulations with different contents of diglycidyl ether of bisphenol A (DG), Jeffamine-D400 (D400), and Lupasol® FG (LP FG) or Lupasol® PR8515 (LP PR) were prepared (Table 1) as crosslinking agents. The proportions of reactants in each formulation were selected so that the molecular structure consisted of stoichiometric DG-D400 and DG-LP networks. The sample nomenclature X-Y-Z used in Table 1 indicates the weight fraction of stoichiometric DG-D400 (X) and DG-LP (Y) networks and the type of Lupasol® used (Z).

Table 1 Composition of the formulations studied
Full size table

Thermomechanical Characterization

The thermomechanical properties were measured using a DMA Q800 (TA Instruments) equipped with a 3-point bending clamp. Prismatic rectangular samples of ca. 30 mm × 5.5 mm × 2.5 mm were analyzed at 1 Hz, 0.1 % strain, and a heating rate of 3 °C/min. The glass transition temperature was determined from the peak in tan δ. The values of storage modulus E′ were evaluated below and above glass transition.

Mechanical Characterization

The mechanical properties at room temperature were evaluated using an electromechanical universal testing machine (Zwick 1445) with specially designed grips. The specimens were sized in accordance with ASTMD638 requirements and had a Type IV dog-bone shape. The experiments were performed at a crosshead speed of 1 mm/min.

The mechanical properties at \(T_{\text{g}}^{{E^{\prime}}}\) were measured using a DMA Q800 equipped with a film-tension clamp in force-controlled mode. Prismatic rectangular samples of ca. 20 mm × 2.5 mm × 0.5 mm were analyzed at \(T_{\text{g}}^{{E^{\prime}}}\) at a force rate of 1 N/min. In both cases, the stress–strain relationship was obtained using the specifications established in ASTMD638.

Shape-Memory Characterization

The shape-memory properties were measured using a DMA Q800 with a force-controlled mode and equipped with a film-tension clamp. Prismatic rectangular samples of ca. 20 mm × 2.5 mm × 0.5 mm were used in shape-memory performances. The method used to create a temporary shape and trigger the SME was a thermomechanical procedure called programming (Fig. 2). This programming consisted of four steps. (1) First the sample was heated to programming temperature T prog and deformed to a prescribed value of maximum stress σ m at 1 MPa/min. In this stage, the deformation of the sample was ε m. (2) The next step was to cool the sample to below transition temperature T low (~25 °C) so as to fix the temporary shape. (3) After fixation, the stress was released at the same stress rate of 1 MPa/min. Once the sample had been unloaded, the deformation of the sample was ε u. (4) The SME was triggered by heating the sample to a recovery temperature T recovery (T g + 10 °C) at a heating rate of 3 °C/min. The amount of non-recoverable deformation at the end of programming was ε p.

Fig. 2
figure 2

Scheme of the thermomechanical programming

Full size image

The programming temperature was chosen as the onset of the glass transition temperature \(T_{\text{g}}^{{E^{\prime}}}\) of each sample. According to Yakacki et al., [16] a peak in the deformability of shape-memory acrylate-based polymers can be obtained at a temperature coinciding with the onset of glass transition temperature. Feldkamp and Rousseau [17] reported the same phenomenon in epoxy-based systems.

The most significant parameters for quantifying shape-memory properties are the shape-recovery ratio (R r) and the shape-fixity ratio (R f). The shape-recovery ratio (Eq. 1) quantifies the ability of the SMP to recover its original shape and was calculated as the total deformation recovered with respect to the maximum deformation reached during the programming. The shape-fixity ratio, calculated from Eq. (2), quantifies the ability of the SMP to fix the temporary shape. It was calculated as the deformation after the stress was released with respect to the maximum deformation.

$$R_{\text{r}} \left( \% \right) = \frac{{\varepsilon_{\text{m}} - \varepsilon_{\text{p}} }}{{\varepsilon_{\text{m}} }} \times 100,$$
(1)
$$R_{\text{f}} \left( \% \right) = \frac{{\varepsilon_{\text{u}} }}{{\varepsilon_{\text{m}} }} \times 100.$$
(2)

Another parameter of interest in evaluating shape-memory ability is the shape-recovery rate (V r). This quantifies the velocity at which the permanent shape is recovered and it is calculated according to Eq. 3:

$$V_{\text{r}} \left( {{\raise0.7ex\hbox{$\% $} \!\mathord{\left/ {\vphantom {\% { \hbox{min} }}}\right.\kern-0pt} \!\lower0.7ex\hbox{${ \hbox{min} }$}}} \right) = \frac{{\left( {\frac{{\varepsilon_{{{\text{rec}},15\% }} - \varepsilon_{{{\text{rec}},85\% }} }}{{\varepsilon_{{{\text{rec}},15\% }} }}} \right) \times 100}}{{\varDelta t_{15 - 85\% } }},$$
(3)

where ε rec,15 % is the deformation corresponding to a shape recovery of 15 %, ε rec,85 % is the deformation corresponding to a shape recovery of 85 %, and ∆t 15–85 % is the time interval between these two points. The difference between deformations corresponding to a shape recovery of 15 and 85 %, ε rec,15 %–ε rec,85 %, was calculated with respect to ε rec,15 % in order to avoid the influence of the maximum deformation on V r.

Results and Discussion

Thermomechanical Properties

The thermomechanical properties of formulations with Lupasol® FG and Lupasol® PR8515 are listed in Table 2. The glass transition temperatures of these materials can be controlled by varying the chemical composition of the formulations. T gs show values between 60 and 117 °C with LP FG and 56 and 112 °C with LP PR. Formulations with LP PR have slightly lower T gs than formulations with LP FG. The higher molecular weight of Lupasol® PR8515 results in longer branches than Lupasol® FG, and these longer branches improve the molecular motion of the network in comparison with formulations with the lower molecular weight counterpart. Accordingly, these materials can be used in shape-memory applications that require a broad range of transition temperatures.

Table 2 Thermomechanical data obtained by DMA: glass transition temperature (T g), onset temperature of the glass transition temperature (\(T_{\text{g}}^{{E^{\prime}}}\)), storage modulus in glassy, and rubbery regions (E′ g and E′ r respectively)
Full size table

A higher content of Lupasol® means that the crosslinking density is also higher (Table 1) because of the higher number of reactive amine groups per unit mass and the presence of internal branching points. This decreases the average chain length between netpoints and restricts the molecular mobility, which leads to an increase in T g. As can be observed in Table 2, an increase in Lupasol® also lowers and broadens the curve of tan δ. The shape of the tan δ curve can be correlated with its network structure: the higher and narrower the shape of the curve of tan δ, the more homogeneous and mobile the network structure is [18]. The presence of Lupasol® leads to a more heterogeneous network structure due to the inherently disperse nature of the hyperbranched poly(ethyleneimine) and the presence of different types of crosslinks from the pre-existing tertiary amines and the reacted primary and secondary amines. The longer branches of LP PR, which decrease the transition temperature in comparison with formulations with LP FG, increase the heterogeneity and dispersity of the networks, which have lower and wider curves of tan δ (Table 2).

All formulations show very high values of storage modulus in the glassy region around 3 GPa, which is higher than most of the shape-memory epoxy resins found in the literature [13–15]. The storage modulus in the rubbery region E′ r increases from 16 MPa to 52 MPa as the LP FG content increases and from 15 to 58 MPa as the LP PR content increases. According to the theory of rubber elasticity, E′ r is roughly proportional to the crosslinking density and the rubbery modulus (Table 2) is expected to increase with the crosslinking density. Given the hyperbranched structure of Lupasol®, which has a considerable number of internal branching points and very short ethylene segments within, unlike the long, and flexible aliphatic structure of Jeffamine D400, mobility can be expected to be restricted and deformability lower when the Lupasol® content is increased in the cured thermosets.

Mechanical Properties

Most of the articles about SMPs found in the literature focus on mechanical properties at high temperature so that deformations can be high during the programing of the temporary shape. However, the characterization of the mechanical properties at room temperature is important for applications such as thermomechanical actuators, which must perform recovery stress or work against an external load. Figure 3 shows the values of stress at break and strain at break of formulations with Lupasol® FG (Fig. 3a) and with Lupasol® PR8515 (Fig. 3b) evaluated at T room .

Fig. 3
figure 3

Stress at break and strain at break of formulations with Lupasol® FG (a) and Lupasol® PR8515 (b) evaluated at T room

Full size image

Both the stress at break and the deformation at break increase from 42 to 55 MPa and from 3.5 % to nearly 5.0 % with LP FG and from 44 to 52 MPa and from 3.7 to 4.8 % with LP PR. The observed tendency for stress at break values clearly shows that strength increases almost linearly with increasing crosslinking density and T g in both formulations with LP FG and LP PR. The strain at break does not seem to be affected differently by crosslinking density in formulations with LP FG. Two distinctive regions are observed: from 90-10-FG to 60-20-FG, with values of strain at break around 3.6 %, and from 50-50-FG to 30-70-FG, with values of strain at break around 5 %. In formulations with LP PR, the strain at break increases linearly from 3.6 % in formulation 90-10-PR to 5.0 % in formulation 50-50-PR. From this point, the strain at break shows values around 5.0 % in the other formulations. These results suggest that the increase in the strain at break seems to be possible up to a certain value when the increasing crosslinking density limits the deformability of the network.

Properties below the T g depend on a combination of factors such as cohesive forces and the presence of local mobility. Strain and stress at break increase with increasing T g and depend mainly on the difference between the test temperature and T g. It should be taken into account that when the Lupasol® content is increased, the DG content also increases (Table 1), which increases the number of hydroxyl groups from the epoxy-amine condensation and contributes to the cohesion of the network structure. The internal flexible structure of Lupasol®, which has limited local mobility below the glass transition temperature, may contribute to the observed behavior.

Figure 4 shows the values of stress at break and strain at break of formulations with LP FG (Fig. 4a) and with LP PR (Fig. 4b) at \(T_{\text{g}}^{{E^{\prime}}}\). All formulations show very high values of stress at break, above 10 MPa in all cases and up to 15 MPa in formulation 30-70-FG. Several studies have reported shape-memory epoxy networks with very high deformations at break at \(T_{\text{g}}^{{E^{\prime}}}\). However, the values of stress at break at \(T_{\text{g}}^{{E^{\prime}}}\) in shape-memory epoxy networks are usually lower than those reported in this paper [14, 17, 19].

Fig. 4
figure 4

Stress at break and strain at break of formulations with Lupasol® FG (a) and Lupasol® PR8515 (b) evaluated at \(T_{\text{g}}^{{E^{\prime}}}\)

Full size image

The values of stress at break and strain at break have a more rigid structure as Lupasol® content increases, which means that strain at break decreases from 58 to 17 % in formulations with LP FG and from 50 to 13 % in formulations with LP PR. In general, an increase in the crosslinking density results in a less mobile and ductile network. On the macroscopic scale this leads to a lower failure strain and a higher stress at break. These results can be explained by the internal hyperbranched structure of Lupasol® which has short ethylene segments within and is less deformable than the long, flexible Jeffamine D400 structure.

Overall, for both temperatures, formulations with LP PR have values of stress at break and strain at break that are slightly lower than formulations with LP FG. These differences may be caused by the higher crosslinking density of formulations with LP PR and the presence of possible molecular interactions due to the longer branches of LP PR.

Shape-Memory Properties

To make a comparative study with the same level of load, every sample was stretched during the programming to a prescribed maximum stress (σ m) that was 75 % of the stress at break (σ m = 0.75σ b).

Figure 5 shows the shape-recovery ratio and the shape-fixity ratio of formulations with Lupasol® FG (Fig. 5a) and with Lupasol® PR8515 (Fig. 5b). The average shape-recovery ratio and shape-fixity ratio over three cycles were calculated when programming under these conditions (σ m and \(T_{\text{g}}^{{E^{\prime}}}\)). All formulations show shape-recovery ratio values above 93 % and shape-fixity ratio values around 97 %, which demonstrates the excellent shape-memory properties of materials with both Lupasol® FG and Lupasol® PR8515.

Fig. 5
figure 5

Shape-recovery ratio (R r) and shape-fixity ratio (R f) of formulations with Lupasol® FG and (a) and with Lupasol® PR8515 (b)

Full size image

The shape-recovery ratio decreases slightly as the Lupasol® content increases in both groups of formulations: from 96.3 % to 93.4 % with LP FG and from 96.5 to 92.9 % with LP PR. The maximum is for formulations 80-20-FG and 80-20-PR with R r = 97.4 % and R r = 97.5 %, respectively. The shape-fixity ratio is high with both types of Lupasol® and ranges from 98 to 94 % as the Lupasol® content increases.

These results can be correlated with thermomechanical properties and network structure. Formulations with a higher Lupasol® content have broader network relaxation because of the more heterogeneous network structure (Table 2) so it is more difficult to release the stored imposed stresses (step 1 in Fig. 2) during the recovery process. This decreases the shape-recovery ratio R r. The local deformation of the internal structure of Lupasol® in the network, which has short ethylene segments within, has a lower contribution to conformational, entropic deformation than longer D400 chains and can therefore relax immediately after stress release. A higher amount of spontaneous elastic recovery may be expected when formulations with a higher Lupasol® content are unloaded. In this case, shape-fixity ratios R f are lower.

As mentioned above, LP PR has longer branches than LP FG. This means that formulations with LP PR have a more heterogeneous structure than formulations with LP FG and have slightly lower values of R r and R f.

Figure 6 shows the shape-recovery velocity V r of formulations with Lupasol® FG (Fig. 6a) and Lupasol® PR8515 (Fig. 6b). It is observed that V r decreases as the Lupasol® content increases. Formulations with a higher Lupasol® content have broader network relaxation because of the more heterogeneous network structure which hinders the recovery process. Thus, it takes longer to release the stored internal stresses during the recovery process in materials with a higher Lupasol® content. Formulations with LP PR have lower values of V r due to the more heterogeneous networks than formulations with LP FG (see “Thermomechanical properties” section).

Fig. 6
figure 6

Shape-recovery velocity (V r) of formulations with Lupasol® FG (a) and with Lupasol® PR8515 (b)

Full size image

Conclusions

A series of shape-memory epoxy polymers was synthesized using an aliphatic amine and two commercial hyperbranched poly(ethyleneimine)s with different molecular weights as crosslinking agents. The influence of the structure of the hyperbranched polymers on the thermal, mechanical, and shape-memory properties was analyzed and discussed.

The thermomechanical measurements showed that an increase in the Lupasol® content led to more heterogeneous networks with higher crosslinking densities and higher T gs. Formulations with Lupasol® PR8515 had lower T gs but wider transitions than formulations with Lupasol® FG.

The tensile tests at room temperature showed that an increase in the Lupasol® content resulted in an increase in the stress at break and deformation at break, showing values around 55 MPa and 5 % in both groups of formulations. At the programming temperature T E’g , the stress at break showed excellent values above 10 MPa in all formulations and up to 15 MPa and deformations at break of almost 60 %. The higher crosslinking density and the presence of possible molecular interactions of formulations with Lupasol® PR8515 resulted in slightly lower values of stress at break and strain at break than formulations with Lupasol® FG.

Shape-memory performances showed high values of shape-recovery, shape-fixity ratios, and shape-recovery velocity. The shape-memory properties were reduced when the Lupasol® content was increased because it restricted chain-conformational changes during deformation and molecular dynamics, which hindered the recovery and fixity processes and decreased the recovery velocity. Formulations with Lupasol® PR8515 presented slightly lower shape-memory properties because they were more heterogeneously structured than formulations with Lupasol® FG.

The results obtained show that hyperbranched polymers can be used as crosslinking agents for epoxy-based SMPs. The hyperbranched polymer content and molecular weight should be controlled. While materials with low molecular weight hyperbranched polymers showed slightly better mechanical and shape-memory properties, materials with higher molecular weight hyperbranched polymers had lower transition temperatures. Other mechanical properties can be analyzed in future work to determine the effect of both types of hyperbranched poly(ethyleneimine)s.