An experimental–numerical study on shape memory behavior of PU/PCL/ZnO ternary blend

Abstract

Shape memory polymer composites have attracted significant attention due to novel properties and great applications. In this article, we focus on the fabrication and simulation of polyurethane/polycaprolactone nanocomposites. The polyurethane/polycaprolactone blends containing ZnO nanoparticles (5 to 30 wt%) are fabricated using a solution mixing and casting method. It is found that significant improvement of polyurethane/polycaprolactone composites in Young’s modulus is achieved by incorporating 20 wt% of ZnO nanoparticles; also, the results of the shape recovery ratio reveal that adding an optimum amount of ZnO (the reinforcement) can increase the shape recovery ratio (for 20 wt% of ZnO). These results could most likely be explained by the fact that some particles restrict the hard segment–soft segment interactions and provide more mobility to polycaprolactone components, while the other nanoparticles can act as the nucleating agent for polycaprolactone chains. A generalized Maxwell model is then used to examine the shape memory behavior of shape memory polymer composites. The dynamic mechanical thermal analysis results are utilized to define the model coefficients and the simulation is carried out to determine the shape recovery ratio. Simulation of this shape recovery ratio for shape memory polymer composites reveals that the numerical results are in good agreement with those of the experimental data.

Keywords

Polyurethane/polycaprolactone nanocomposite shape memory finite element method

Introduction

Shape memory polymers (SMPs) have the property to store the energy of applied shape under different stimuli like temperature, light, electric field, magnetic ﬁeld, pH, specific ions, or enzyme and quickly regain their original shapes (Guo et al., 2015; Liu et al., 2009; Mazaheri et al., 2016). Temperature is expressed as the most usual stimulus, and thermal-sensitive SMPs have been extensively investigated because they have a great potential for application in industry (Leng et al., 2008; Tobushi et al., 1998).

SMPs have some advantages like low manufacturing cost, low toxicity, low density, potential biocompatibility, and biodegradability (Leng et al., 2009). However, the practical applications of SMPs are hindered due to their relatively poor elastic properties (Peponi et al., 2014), and various methods are developed by researchers to overcome this defect. Jamalimehr et al. (2017) numerically studied the SMP beams reinforced with corrugated polymeric structures. Their results showed a significant increase in load capacity of the reinforced beam against a decrease in shape fixity. In a different way, the addition of nanoparticles was employed to improve the mechanical properties of SMPs. Chen et al. (2016) added thermally reduced graphite oxide (TrGO) in epoxy-based shape memory polymer (ESMP). They concluded that Young’s modulus and tensile strength of reinforced epoxy were increased by 41%–71% and 44%–64%, respectively, compared with pristine ESMP. Taherzadeh et al. (2016) used three-dimensional (3D) finite element modeling to study the characteristic thermomechanical behavior of SMPs reinforced with different volume fractions and aspect ratios of graphene nanoplatelets (GNPs). Their results revealed a significant enhance in the elastic modulus with addition of GNPs.

Among SMPs, polyurethane (PU) and its nanocomposites were studied by many researchers (Cai et al., 2013; Gu et al., 2016; Zhang et al., 2015). Thermoplastic polyurethane (TPU), as a popular SMP material, is shown to be a suitable material for medical equipment because of its flexibility, durability, and biological stability (Simmons et al., 2006). At room temperature, the soft and hard segments of TPU are incompatible, which is responsible for the shape memory properties (Lin and Chen, 1998). Poly(ε-caprolactone) (PCL) as the crystallizable component is the second part of the blend (Minakov and Schick, 1999) which is very useful due to its advantages, for example, biocompatibility, biodegradability, and elasticity (Middleton and Tipton, 2000; Sun et al., 2006). Polyurethane/polycaprolactone (PU/PCL) blends, as thermally activated SMPs, have been studied recently (Li et al., 2009; Ansari et al., 2018) and showed that nanoparticles can be used to improve the elastic properties in order to be applied in practical applications. Gu et al. (2018) used Fe₃O₄ nanoparticles in PU matrix to obtain dual-responsive shape memory effects (SMEs) around the body temperature. Their results highlighted a favorable improvement in mechanical properties with incorporation of 6 wt% nanoparticles. In addition, the nanofillers can be effective on shape memory actuation. Lu et al. (2016b) showed that electrically induced shape recovery and actuation efficiency of SMP enhanced using conductive carbon nanotubes (CNTs) and nafion/silica nanofibers. Also, Lu et al. (2016a) indicated that hydrogen bonding of PU macromolecules and mass migration of CuCl₂ particles are the mechanisms behind the chemo-responsive SME in PU SMP.

More specifically, zinc oxide (ZnO) was employed as the modulus reinforcement because of its specific physical and chemical properties that provide practical applications in the biomedical field such as bio-imaging and drug delivery (Xiong, 2013). In addition, dispersion of ZnO particles is easier compared to other particles (such as aluminum oxide (Al₂O₃) or titanium dioxide (TiO₂)) (Zheng et al., 2005). Augustine et al. (2014) fabricated various morphologies of PCL fibers by including different concentrations of ZnO nanoparticles for novel tissue engineering scaffolding. The prepared samples improved the cell proliferation and also increased the tensile modulus by adding ZnO content over 1 wt%. Li et al. (2009) fabricated PU-based coatings reinforced by ZnO nanoparticles to improve both the mechanical strength and the antibacterial properties of the samples. As a result, the ZnO-doped PU films showed excellent antibacterial activity and, on the other hand, the addition of ZnO nanoparticles improved the tensile strength and Young’s modulus significantly.

Numerical modeling is an appropriate tool in fabricating SMP structures through the prediction of SMP properties. Two main approaches have been applied to model the shape memory behavior of SMPs.

The first approach, phase transition, considers the material as a mixture of active and frozen phases and the shape recovery results from conversion of the active to frozen phase and the change in volume fraction of both phases (Nguyen, 2013). A theoretical framework was first suggested by Liu et al. (2006). Complementarily, Baghani et al. (2012b) introduced a constitutive model for SMPs in the small strain regime. Based on this model, they developed a 3D constitutive model under time-dependent multiaxial thermomechanical loading conditions (Baghani et al., 2014a; Baghani et al., 2014b; Baghani and Naghdabadi, 2012). In another work, Baghani et al. (2012a) studied the shape memory behavior of helical springs under axial force, both analytically and numerically.

The other approach describes the SME through the thermo-viscoelastic behavior and was reported by some researchers (see, for example, Lin and Chen, 1999; Xiao et al., 2013). The initial model of this approach used simple viscoelastic elements (Lin and Chen, 1999). Later on, more complex models were developed to explain the stress–strain behavior (Srivastava et al., 2010; Xiao et al., 2013). The suggested model by Diani et al. (2012) used the time–temperature superposition principle and a generalized Maxwell model. A useful advantage of this model is obtaining all the material parameters from standard tests involving viscoelastic results (e.g. dynamic mechanical thermal analysis (DMTA)). Since the model is based on viscoelastic properties, the effect of the temperature and deformation rate are considered simultaneously and shape memory parts can be analyzed in the large deformation regime.

Although the inclusion of nanoparticles in SMPs is expected to improve their mechanical properties, some negative effects may be induced to the composite, like the reduction in shape recovery properties due to a disruption of the switching phase functionality. In conclusion, the mechanical behavior and shape recovery properties should be optimized simultaneously.

The fabrication of TPU/PCL nanocomposites using ZnO and the simulation of its shape recovery behavior would be useful to design biomedical devices like stents. In this article, according to the previous investigation of Ajili et al. (2009), we investigate the ratio of 70/30 for PU/PCL for its practical application for the shape memory stent in the body temperature range. This work will provide a new approach to control the mechanical and shape memory properties in a polymer blend nanocomposite.

The first part of the article presents the used material and the procedure of sample preparation in the experimental section. Then, the applied measurements on samples are introduced, and in the following section, the results of the experimental tests are investigated in detail. The obtained results are applied to the numerical study of the thermomechanical behavior of the sample. Finally, we present a summary and draw conclusions.

Experimental section

Materials

Thermoplastic PU/PCL ester–based TPU (LPR2102-85 AE) was provided from Coim Co. with a density of 1.16 g cm⁻³. The hard segments of TPU are based on $4^{¯}, 4'$ -methylenediphenyl diisocyanate (MDI) and 1,4-butanediol (BDO) and soft segments of PCL with a M_n of 2000 g mol⁻¹. PCL was purchased from Sigma-Aldrich Co. with a M_n of 42,500 g mol⁻¹, a density of 1.145 g cm⁻³, and a melting point of 60°C.

ZnO nanoparticles with an average size of 50 nm, supplied by Daypetronic Co., were used as the modulus reinforcement. The dimethylformamide (DMF) was delivered from Sigma-Aldrich Co. The PU and PCL were dried in vacuum, respectively, at 100°C and 60°C for 24 h prior to the usage.

Sample preparation

Step 1. The blend samples were prepared using a solvent casting technique. Due to the discrepancy of the melting temperature of the PU and PCL components, the procedure of blending was performed by first dissolving a certain amount of PU granules in DMF by magnetic stirring. Then, 30 wt% PCL was poured into the solution on a magnetic stirring in a water bath of 60°C.

Step 2. To prepare the composite samples, the desired fraction of nanoparticles (5, 10, 20, and 30 wt%) was mixed with the polymer solution and sonicated for 20 min at 0°C. High-power ultrasonic was used with the aid of a magnetic stirrer to guarantee the good dispersion and distribution of nanoparticles in the matrix. To avoid a temperature rise within the ultrasound process, the solution was kept in an ice bath where the magnetic stirring prevented the sedimentation of nanoparticles.

The final solutions of both steps were mixed separately for about 4 h to evaporate the solvent, and finally, a vacuum oven was used for 24 h at 45°C to remove the excess DMF of the prepared samples.

Measurements

Fourier transform infrared spectroscopy

Fourier transform infrared spectroscopy (FTIR) measurements were applied at the room temperature in the transmission mode to characterize the existing bonds. The thin films for FTIR measurements were prepared using a hot press process.

Scanning electron microscopy

Scanning electron microscopy (SEM) analysis was performed using a Philips XL30 scanning electron microscope. Samples were fractured under liquid nitrogen and the cross section was coated with gold using a sputter coater.

Differential scanning calorimetry

The thermal properties of samples were investigated by differential scanning calorimetry (DSC). DSC has been performed under nitrogen flow. The samples were heated from −60°C up to +200°C, then cooled down to −60°C again. Each step was performed at a rate of 10°C min⁻¹.

DMTA

The thermo-viscoelastic responses of samples were measured with a DMTA with sample dimensions of 15 mm × 0.6 mm × 5 mm. The dynamic temperature sweep was performed in the temperature range of −70°C to +100°C. All tests were conducted at 0.1% strain, a heating/cooling rate of 5°C min⁻¹, and a frequency of 1 Hz.

Shape recovery

Shape recovery is an important characteristic of SMPs. The behavior of fabricated materials under bending loads helps us to use these smart materials in practical applications. To evaluate the performance of pure PU/PCL blend and its nanocomposites at various temperatures, a shape recovery test upon bending was carried out based on the procedure mentioned by Lan et al. (2009).

The schematics of the shape recovery test together with experimental measurements of shape recovery are presented in Figure 1. It includes three steps: (1) the specimen is soaked in a water bath for 5 min at T_h, where T_h is defined as T_ms + 10°C (T_ms is the melting temperature of the soft segment). (2) The specimen is bent around a mandrel with a radius of 4 mm to the primary angle θ₀ (storage angle) and after that the deformed sample is cooled in a water bath at T_c. In this step, the specimen is placed under an external restriction for 5 min. (3) The load is finally removed and the specimen is immersed into another water bath with given temperatures so the angle will be changed to θ_r (recovery angle).

At the first step, the applied heating puts the sample in the rubbery state. The mechanical load is then applied and kept fixed on the sample to freeze the elastic energy in the crystallized components. The temporary shape is then achieved for the specimen after unloading. Finally, immersing the specimen in water at the desired temperature and recovering of the specimen shape would be measured using a conveyor.

The shape recovery ratio (R) can then be obtained according to equation (1)

R = \frac{θ_{0} - θ_{r}}{θ_{0}} \times 100 %

(1)

where $θ_{r}$ is the residual angle in the recovery state of bending process and $θ_{0}$ shows the storage angle of the sample in the first step as shown in Figure 1 (left).

Results and discussion

Material results

Infrared spectroscopy

Infrared (IR) spectroscopy can be used to identify the functional groups of materials in order to analyze hydrogen bonding between the hard and soft segments and also to investigate the effect of ZnO nanoparticles on phase separation between hard and soft segments.

Among various bands in the IR spectrum, the change of band intensities in the N-H (3100–3500 cm⁻¹) and the C=O (1650–1750 cm⁻¹) regions is commonly observed (Bistričić et al., 2010). Carbonyl groups are known to exist in PU, both in the hard segment (i.e. urethane) and soft segment (i.e. ester), and PCL also has O=C_ester in its structure. Each of these carbonyls can form hydrogen bonding structures.

An important peak in this region, observed at 3115 cm⁻¹, is related to the bending vibration of N-H group in the hard segments. The intensity ratio of N-H over the C=C peak (stretching vibration of C=C in benzene ring of urethane (Lamba et al., 1998) is considered to eliminate the effects of the concentration and thickness differences in the samples. Table 1 reports the relative intensity of absorbance at 3115 cm⁻¹ to that at 1596 cm⁻¹ (N/C). According to the results, decreasing the amount of N/C for PU/PCL/10 and PU/PCL/20 compared to PU/PCL presents the increment of hard segment–hard segment reactions; therefore, the morphology undergoes more phase separations (Wang and Wei, 2005).

Table 1.

FTIR data for PU/PCL and nanocomposites.

Samples	PU/PCL	PU/PCL/5	PU/PCL/10	PU/PCL/20	PU/PCL/30
DPS	0.4	0.41	0.44	0.47	0.42
N/C	0.38	0.34	0.29	0.23	0.31

FTIR: Fourier transform infrared spectroscopy; PU/PCL: polyurethane/polycaprolactone; DPS: degree of phase separation.

Figure 2(a) depicts IR analysis of PU/PCL and its nanocomposites. Figure 2(b) shows that the intensity of a small peak related to free N-H group at 3440 cm⁻¹ becomes weaker by increasing the amount of nanoparticles, which reveals that NH groups have less interaction with soft segment with increasing ZnO content (Khosravi et al., 2014). Therefore, the hydrogen bondings of N-H…O=C_ester are restricted by more incorporation of ZnO particles, particularly in PU/PCL/10 and PU/PCL/20.

For a better investigation of the effects of ZnO nanoparticles on the phase separation, the amount of C=O band is examined according to equation (2). In IR spectrum, the observed peaks at 1722 and 1703 cm⁻¹ are associated with the free and hydrogen-bonded carbonyl stretching vibration peaks, respectively. The degree of phase separation (DPS) can be calculated from the following equation (Jung et al., 2012)

DPS = \frac{A (C_{bonded})}{[(A (C_{bonded}) + A (C_{free}))]}

(2)

in which $A (C_{bonded})$ and $A (C_{free})$ are the absorbance of the hydrogen-bonded and free carbonyl peaks, respectively.

As can be observed from Table 1, the highest amount of DPS is attributed to the PU/PCL/20; the data show that the increase in weight percent of the nanoparticles enhances the phase separation except for the PU/PCL/30. In this sample, the ascending trend stops and the phase separation declines.

As can be observed from Figure 2(c), increasing the ZnO amount decreases the intensity of the free carbonyl peak gradually and the hydrogen-bonded carbonyl intensity increases to some extent. As mentioned, the ratio of carbonyl peaks approximately becomes equal for PU/PCL/20. This reveals more hydrogen bonding between the urethane carbonyl groups and the urethane N-H groups in the hard segments. This behavior could come from the dispersion of ZnO nanoparticles in the soft segments through hydrogen bonding between ZnO hydroxyl groups and ester groups. Hence, the number of carbonyl sites on the soft segments to form a hydrogen bond with N-H group decreases, so it is more likely for N-H groups to interact with carbonyl groups of urethane segments. In other words, ZnO nanoparticles create some restrictions for hard segments to interact with PCL chains and the phase separation enhances accordingly (Sadeghi et al., 2013).

SEM results

The level of particle dispersion and distribution in PU/PCL blend is one of the crucial points in investigating the nanocomposite behavior. The surface morphology of the nanocomposites was investigated by SEM. Figure 3 gives the images of PU/PCL blend and its nanocomposites. At low nanoparticle content level (5 and 10 wt%), ZnO nanoparticles were well dispersed in the blend, as it can be observed from Figure 3(a) and (b). As shown in Figure 3(c) and (d), for higher content levels, some small clusters composed of several particles are observed in 20 and 30 wt%, showing that the agglomeration content has increased. The strong van der Waals interactions between the nanoparticles in the higher concentration can lead to the formation of a percolated network, which hinders the chain mobility of the soft segment. It is therefore expected that the dependent properties of viscoelastic behavior would change significantly. For PU/PCL/30, the existence of pores or defects could come from the following reasons: as the amount of nanoparticle increases, the viscosity of nanocomposite enhances which makes solvent evaporation harder and also some air bubble can be formed as defects. Besides, addition of nanoparticle in higher amounts lowers their dispersion because of the particle tendency to have more filler–filler interactions; therefore, some voids can be observed due to the weak polymer–particle interactions. This will consequently deteriorate the mechanical properties (Fonseca et al., 2013).

Investigation of the interactions

ZnO nanoparticles may interact more with soft segments according to IR results; hence, they create some physical confinement on PU chains due to the lower possibility of reactions between the hard and soft segments which cause reduction in the hard-phase circumscriptions on freedom of the soft chains, so the PCL chains gain more mobility to locate in the crystal domains.

Addition of higher concentration of filler can affect the value of crystallinity, but as the solution mixing was used for the sample preparation, the solvent evaporation may provide longer times for PCL chains to have higher mobility, in order to orient and crystallize. PCL is known as a fast crystallizing polymer, so it is expected that PCL will achieve higher crystallinity even in the presence of the particles.

The influence of particle addition on the soft segment crystallinity was investigated by DSC. According to equation (3), there is a direct relationship between the degree of crystallinity and the heat of fusion

χ_{C} (%) = \frac{Δ H_{f}}{w_{P C L} \times Δ H_{100}}

(3)

where $χ_{C} (%)$ is the degree of crystallinity and $Δ H_{100}$ and $w_{PCL}$ are the heat of fusion of 100% crystalline PCL and the weight fraction of the PCL component, respectively. Table 2 reports the heat of fusion $(Δ H_{f})$ results for PU/PCL and its nanocomposite samples after the first thermal scan. The maximum and minimum values of the melting crystallites, respectively, are related to PU/PCL/20 and PU/PCL/30.

Table 2.

Heat of fusion for PU/PCL and its nanocomposites.

Sample	PU/PCL	PU/PCL/5	PU/PCL/10	PU/PCL/20	PU/PCL/30
ΔH_f (j/g)	25.91	23.13	24.68	27.44	19.66

PU/PCL: polyurethane/polycaprolactone.

However, the crystallinity of PCL soft segments decreased after reactions of the soft chains with nanoparticles except the PU/PCL/20. As the heat of fusion of the fully crystalline PCL was taken to be 32.4 cal g⁻¹ (Minakov and Schick, 1999), the crystallinity of PCL chains was dramatically reduced with incorporation of the hard segment of PUs. For this reason, if the nanoparticles can confine the hard segment motions, the PCLs can pack together in the crystalline domains. However, the changes in soft segments crystallinity could be attributed to the following reasons: (1) the particles can create physical bonds with the polymer chains so the crystallinity of PU/PCL/5, PU/PCL/10, and PU/PCL/30 decreased; (2) particles can help the PCL chains to form crystallized segments and act as nucleating agent for soft chains as the crystallinity increased for the PU/PCL/20 (Meng et al., 2007; Rezanejad and Kokabi, 2007).

Thermo-mechanical results

DMTA

The viscoelastic properties of samples were evaluated by DMTA. The storage modulus versus temperature for PU/PCL blend and its nanocomposites (5, 10, 20, and 30 wt% of ZnO) are presented in Figure 4 (experimental). Different behaviors are observed for the blends which need to be detailed.

The viscoelastic properties for filled blend are severely dependent on the polymer–filler interface, especially since a higher amount of interface results in a higher storage and loss modulus.

It can be observed from Figure 4 that PU/PCL/5 and PU/PCL/10 show mostly a higher modulus than pure SMP. This can be attributed to the increase in the polymer–filler interface. The addition of 20 wt% ZnO particles caused an increase in the storage modulus ( $E'$ ) over the whole temperature range. This highest influence on the storage modulus could be explained by the fact that polymer–filler interaction is dominant or that fillers can act as the nucleating agent (Lin and Chen, 1998) in order to increase the crystallites in the soft domains. Thus, a higher amount of polymer–filler interactions show an increase in the storage modulus due to the creation of stronger interfacial adhesions, which can mobilize more polymer chains of the hard segments.

However, the mechanical properties decreased with the addition of 30 wt% nanoparticles which may be due to the formation of a filler–filler network at higher amounts of 20 wt%; hence, the slippage of the soft segment chains on the filler network decreases the storage modulus and the high amount of filler interactions becomes the stress concentration points and thus the modulus drops remarkably.

Next, the shape memory behavior of the samples was simulated based on the 3D constitutive model suggested by Diani et al. (2012). This numerical model uses the basic elements of the classic generalized Maxwell model within the range of thermo-viscoelastic finite strains. The model includes a spring (with coefficient of $E_{\infty}$ ) as the equilibrium element and some non-equilibrium elements including a series of springs and dampers (Maxwell elements). To satisfy the time–temperature superposition, the relaxation time of the non-equilibrium elements is considered as a function of the temperature (equation (4))

τ_{i} (T) = τ_{0}^{i} α_{T} (T)

(4)

in which, $τ_{i} (T)$ is the relaxation time of the element i at temperature T, $τ_{0}^{i}$ is the element relaxation time at the reference temperature and $α_{T} (T)$ stands for the shift factor that follows the Williams–Landel–Ferry (WLF) equation $(\log (a_{T}) = C_{1} (T - T_{ref}) / C_{2} + (T - T_{ref}))$ (Diani et al., 2012), in which $C_{1}$ and $C_{2}$ are material parameters. It is worth noting that the coefficients of the WLF equation (C₁ and C₂) are obtained from the master curve of the DMTA results.

The storage modulus under the sinusoidal strain $(ε = ε_{0} \sin ω t)$ is

E' (ω) = E_{\infty} + \sum_{i = 1}^{n} \frac{E_{i} τ_{i}^{2} ω^{2}}{1 + τ_{i}^{2} ω^{2}}

(5)

in which $ω$ is the angular frequency, $E_{i}$ (the moduli of the element i), $τ_{i}$ , and $E_{\infty}$ (infinite moduli) coefficients can be obtained from fitting equation (5) on the experimental curves of DMTA.

In other words, the parameters of the constitutive model are measured from fitting Prony series on the relaxation moduli results. A series of 12 terms are assumed here; therefore, 12 pairs of $E_{i}$ , $τ_{i}$ , and $E_{\infty}$ are calculated for each sample. In order to decrease the number of variables and also to make the fitting easier, the geometric sequence is applied to relaxation times $τ_{i} s$ and only the sentence with the highest amount $(τ_{1})$ and its magnitude ratio are considered in the curve fitting procedure (the number of parameters to fit was reduced to 15).

Figure 4 shows the fitting results of equation (5) on the reference curves of the storage modulus ( $E'$ ). The fitting was carried out using the least mean square method. The root mean square error (RMSE) was 1.75 in the worst case (for PU/PCL/20) and 0.08 in the best case (for PU/PCL/30). The obtained parameters of each sample are presented in Table 3.

Table 3.

Model parameters for PU/PCL and its nanocomposites.

	PU/PCL		PU/PCL/5		PU/PCL/10		PU/PCL/20		PU/PCL/30
i	$E_{i} (MPa)$	$E_{i} (MPa)$	$E_{i} (MPa)$	$E_{i} (MPa)$	$E_{i} (MPa)$	$τ_{i} (s)$	$τ_{i} (s)$	$τ_{i} (s)$	$τ_{i} (s)$	$τ_{i} (s)$
∞	0.1	–	0.1	–	3	–	2	–	3.1	–
1	7.294	4.576	0.238	2.942	0.875	19,290.2	925.5	108,218.8	40,684.50	67,392.6
2	10.6	36.819	14.381	99.333	1.517	713.03	26.48	3695.9	571.80	3701.9
3	8.107	19.913	33.098	32.293	3.387	26.36	7.577E−01	126.221	8.036	203.35
4	7.306	19.523	19.990	63.715	4.522	9.742E−01	2.168E−02	4.311	1.129E−01	11.17
5	9.745	25.353	24.119	106.241	6.578	3.601E−02	6.204E−04	1.472E−01	1.587E−03	6.136E−01
6	13.967	36.080	29.641	176.023	8.960	1.331E−03	1.775E−05	5.028E−03	2.231E−05	3.370E−02
7	22.806	55.665	37.747	277.899	8.200	4.920E−05	5.079E−07	1.717E−04	3.136E−07	1.851E−03
8	41.174	83.923	54.334	273.402	15.767	1.819E−06	1.453E−08	5.864E−06	4.407E−09	1.017E−04
9	64.481	114.462	74.405	181.255	21.650	6.722E−08	4.159E−10	2.003E−07	6.194E−11	5.586E−06
10	80.843	106.608	99.314	111.656	25.906	2.485E−09	1.190E−11	6.840E−09	8.705E−13	3.068E−07
11	45.595	137.755	72.402	572.844	19.726	9.185E−11	3.405E−13	2.336E−10	1.223E−14	1.686E−08
12	226.985	502.841	133.628	714.729	24.036	3.395E−12	9.742E−15	7.977E−12	1.719E−16	9.259E−10

PU/PCL: polyurethane/polycaprolactone.

For the finite element numerical model, the viscoelastic properties require a set of corresponding shear relaxation moduli $(G_{i})$ , relaxation times $(τ_{i})$ , and bulk moduli $(K_{i})$ . Assuming the bulk modulus to be constant, $G_{i}$ values can be calculated from the $E_{i}$ values with $G = E / 2 (1 + ν)$ and the $k_{i} s$ can be set to zero. In addition, the coefficients and the reference temperature of the time–temperature superposition function (WLF) should be imported. Since these coefficients are all relative, some reference elastic properties are determined. The hyperelastic properties of the material are considered as neo-Hookean with $C_{10}$ and $D_{1}$ coefficients defined, respectively, by $G_{0} / 2$ and $2 / K_{0}$ . $G_{0}$ and $K_{0}$ are calculated from equation (6). In these equations, $E_{0} = E_{\infty} + \sum_{i = 1}^{n} E_{i}$ and $ν$ are assumed to be 0.4

G_{0} = \frac{E_{0}}{2 (1 + ν)}, K_{0} = \frac{E_{0}}{3 (1 - 2 ν)}

(6)

Shape recovery predictions versus experimental results

The shape recovery ratio is simulated using finite element method according to the experimental sample size (3 × 9 × 50 mm³). For these simulations, dynamic explicit simulations in ABAQUS/Standard are employed. Samples are meshed with 32 × 6 × 4 = 768 C3D8H elements (an eight-node linear brick, hybrid, constant pressure). Figure 5 shows the strain recovery of the specimen at different temperatures for the recovery steps. The maximum logarithmic principal strain of SMP is shown in Figure 5 which is 29% at 30°C, and by increasing the temperature, this strain is released. The experimental and simulation results of shape recovery are presented in Figure 6.

The sample is programmed according to the shape recovery section described previously in section “Shape recovery.” In simulation and experiments, the temperature of 60°C was selected as the T_h temperature (as predefined field with uniform distribution applied across the sample), since all the crystals of PCL almost melt at this temperature. In a second step, the sample was bent around the mandrel and immediately cooled to 0°C (T_m−40°C) to form the crystals of PCLs and fix the temporary shape. After unloading, a partial recovery occurred (storage). Subsequently, the sample was immersed at different temperatures (according to the horizontal axis of Figure 6).

Figure 6 shows the comparison between the simulation results and the experimental data for the shape recovery of the samples as a function of the heating temperature. The finite element simulation results show a good agreement with the experimental data except for PU/PCL/30, in which the complete shape recovery could not be achieved due to the significant confinement of the soft segmental motion caused by the high volume fraction of ZnO particles.

According to Figure 6, no recovery occurs below a temperature of about 25°C. However, the shape recovery increases sharply when the temperature exceeds about 40°C. The sample will reverse its temporary shape fully to its initial shape. Complete shape recovery is not achieved for PU/PCL/5 and is slightly lower in PU/PCL/10 than the pure blend, which may be attributed to restriction of the segmental motions through the interactions with ZnO particles.

As it can be observed, PU/PCL/20 achieves the highest shape recovery at temperatures higher than 40°C. Since the degree of crystallinity of the soft segments can drastically change the shape memory properties, adding 20 wt% of particles showed to obtain the desirable amount of PCLs’ crystallinity with more phase separation, so the chains can easily deform and orientate along the direction of strain. Another physical reason of the shape recovery behavior of PU/PCL/20 can be related to the function of ZnO particles as cross-linking agents. They will help the switching phase to recover faster to its original shape. This behavior is also confirmed in the work by Park et al. (2008).

For 30 wt% of nanoparticles, the composite behavior has changed significantly. This could be attributed to the formation of a percolated network of nanoparticles. It seems that the particle–particle interactions become dominant in the nanocomposite, so these interactions hinder the chain mobility of the soft segment and therefore no shape recovery effect is observed because the particle network can interfere in the shape memory properties of the composites.

In conclusion, addition of nanoparticles could induce two different behaviors to the blend: (1) the particles will restrict hard segment chains to form bonding with soft segment that affects the mobility of PCLs and (2) particles will act as the nucleating agents for PCLs to satisfy the favorable crystallinity and store the energy to behave as the switching component.

Summary and conclusion

In this study, PU/PCL nanocomposites were successfully prepared by solution mixing. DMTA analysis showed that adding ZnO particles increased the elastic modulus and changed the shape recovery simultaneously. The results revealed the effect of nanoparticles on the shape recovery properties of PU/PCL blend. The maximum of the shape recovery was achieved for PU/PCL/20 in comparison with other filler contents. This could be due to (1) the occupation of carbonyl groups in soft segments by ZnO hydroxyl groups, there was an obstacle for hard segments to interact with soft segments so more freedom was achieved for PCL component and (2) incorporating particles as the nucleating agent of PCL chains and increasing the crystalline domains of the soft segment. The occurrence of these two phenomena can offer an optimized point to improve both the modulus and shape recovery properties. In complement, the complete shape recovery was not obtained for PU/PCL/30 due to the remarkable change in the sample behavior caused by the formation of filler–filler network. The generalized Maxwell model was applied to predict the SMPCs’ behavior. The model was first calibrated by the DMTA results of our experimental data and then the simulation was conducted on the samples to validate the shape recovery ratios. It was concluded that simulation results were in good agreement with the experimental test of the shape memory behavior.

Footnotes

Acknowledgements

The authors would like to thank Dr Nadereh Golshan Ebrahimi, associate professor of Polymer Engineering at Tarbiat Modares University, for her supportive help.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Center for International Scientific Studies & Collaboration (CISSC).

ORCID iD

M Baghani

References

Ajili

Ebrahimi

Soleimani

(2009) Polyurethane/polycaprolactane blend with shape memory effect as a proposed material for cardiovascular implants. Acta Biomaterialia 5(5): 1519–1530.

Ansari

Golzar

Baghani

et al . (2018) Shape memory characterization of poly(ε-caprolactone) (PCL)/polyurethane (PU) in combined torsion-tension loading with potential applications in cardiovascular stent. Polymer Testing 68: 424–432.

Augustine

Malik

Singhal

et al . (2014) Electrospun polycaprolactone/ZnO nanocomposite membranes as biomaterials with antibacterial and cell adhesion properties. Journal of Polymer Research 21(3): 1–17.

Baghani

Naghdabadi

(2012) A thermodynamically-consistent 3D constitutive model for shape memory polymers. International Journal of Plasticity 35: 13–30.

Baghani

Arghavani

Naghdabadi

(2014a) A finite deformation constitutive model for shape memory polymers based on Hencky strain. Mechanics of Materials 73: 1–10.

Baghani

Mohammadi

Naghdabadi

(2014b) An analytical solution for shape-memory-polymer Euler-Bernoulli beams under bending. International Journal of Mechanical Sciences 84: 84–90.

Baghani

Naghdabadi

Arghavani

(2012a) A semi-analytical study on helical springs made of shape memory polymer. Smart Materials and Structures 21(4): 045014.

Baghani

Naghdabadi

Arghavani

et al . (2012b) A constitutive model for shape memory polymers with application to torsion of prismatic bars. Journal of Intelligent Material Systems and Structures 23(2): 107–116.

Bistričić

Baranović

Leskovac

et al . (2010) Hydrogen bonding and mechanical properties of thin films of polyether-based polyurethane-silica nanocomposites. European Polymer Journal 46(10): 1975–1987.

10.

Cai

Jiang

J-S

Zheng

et al . (2013) Synthesis and properties of magnetic sensitive shape memory Fe₃O₄/poly(ε-caprolactone)-polyurethane nanocomposites. Journal of Applied Polymer Science 127(1): 49–56.

11.

Chen

Liu

et al . (2016) Nanocomposites of epoxy-based shape memory polymer and thermally reduced graphite oxide: mechanical, thermal and shape memory characterizations. Composites Part B: Engineering 91: 75–82.

12.

Diani

Gilormini

Frédy

et al . (2012) Predicting thermal shape memory of crosslinked polymer networks from linear viscoelasticity. International Journal of Solids and Structures 49(5): 793–799.

13.

Fonseca

Abreu

Gonçalves

FAMM

et al . (2013) Shape memory polyurethanes reinforced with carbon nanotubes. Composite Structures 99: 105–111.

14.

Chang

Jin

(2018) A dual-induced self-expandable stent based on biodegradable shape memory polyurethane nanocomposites (PCLAU/Fe3O4) triggered around body temperature. Journal of Applied Polymer Science 135(3): 1–10.

15.

S-Y

Jin

S-P

Gao

X-F

et al . (2016) Polylactide-based polyurethane shape memory nanocomposites (Fe3O4/PLAUs) with fast magnetic responsiveness. Smart Materials and Structures 25(5): 55036.

16.

Guo

Wang

Tong

et al . (2015) Shape memory and thermo-mechanical properties of shape memory polymer/carbon fiber composites. Composites Part A: Applied Science and Manufacturing 76: 162–171.

17.

Jamalimehr

Baghani

Baniassadi

et al . (2017) Thermomechanical behavior of shape memory polymer beams reinforced by corrugated polymeric sections. Meccanica 52(8): 1947–1962.

18.

Jung

Kim

Hayashi

et al . (2012) Fabrication of transparent, tough, and conductive shape-memory polyurethane films by incorporating a small amount of high-quality graphene. Macromolecular Rapid Communications 33(8): 628–634.

19.

Khosravi

Sadeghi

Banadkohi

et al . (2014) Polyurethane-silica nanocomposite membranes for separation of propane/methane and ethane/methane. Industrial & Engineering Chemistry Research 53(5): 2011–2021.

20.

Lamba

NMK

Woodhouse

Coope

(1998) The chemistry of polyurethane copolymers. In: Bamford CH (Ed) Polyurethanes in Biomedical Applications. Boca Raton, FL: CRC Press, 205–241.

21.

Lan

Liu

et al . (2009) Fiber reinforced shape-memory polymer composite and its application in a deployable hinge. Smart Materials and Structures 18(2): 24002.

22.

Leng

Liu

et al . (2008) Conductive nanoparticles in electro activated shape memory polymer sensor and actuator. In: Proceedings of SPIE 6931 (nanosensors and microsensors for bio-systems 2008). Available at: https://www.spiedigitallibrary.org/conference-proceedings-of-spie/6931/693109/Conductive-nanoparticles-in-electro-activated-shape-memory-polymer-sensor-and/10.1117/12.775743.short?SSO=1

23.

Leng

Liu

et al . (2009) Shape-memory polymers—a class of novel smart materials. MRS Bulletin 34(11): 848–855.

24.

Hong

et al . (2009) Effects of ZnO nanoparticles on the mechanical and antibacterial properties of polyurethane coatings. Progress in Organic Coatings 64(4): 504–509.

25.

Lin

Chen

(1998) Study on shape-memory behavior of polyether-based polyurethanes. I. influence of the hard-segment content. Journal of Applied Polymer Science 69(8): 1563–1574.

26.

Lin

Chen

(1999) Shape-memorized crosslinked ester-type polyurethane and its mechanical viscoelastic model. Journal of Applied Polymer Science 73: 1305–1319.

27.

Liu

Gall

Dunn

et al . (2006) Thermomechanics of shape memory polymers: uniaxial experiments and constitutive modeling. International Journal of Plasticity 22(2): 279–313.

28.

Liu

Lan

et al . (2009) Review of electro-active shape-memory polymer composite. Composites Science and Technology 69(13): 2064–2068.

29.

Huang

et al . (2016a) Quantitative separation of the influence of copper (II) chloride mass migration on the chemo-responsive shape memory effect in polyurethane shape memory polymer. Smart Materials and Structures 25(10): 105003.

30.

Huang

et al . (2016b) Synergistic effects of carboxylic acid-functionalized carbon nanotube and nafion/silica nanofiber on electrical actuation efficiency of shape memory polymer nanocomposite. Composites Part B: Engineering 25(10): 146–151.

31.

Mazaheri

Baghani

Naghdabadi

(2016) Inhomogeneous and homogeneous swelling behavior of temperature-sensitive poly-(N-isopropylacrylamide) hydrogels. Journal of Intelligent Material Systems and Structures 27(3): 324–336.

32.

Meng

Zhu

(2007) Shape-memory polyurethane/multiwalled carbon nanotube fibers. Journal of Applied Polymer Science 106(2): 837–848.

33.

Middleton

Tipton

(2000) Synthetic biodegradable polymers as orthopedic devices. Biomaterials 21(23): 2335–2346.

34.

Minakov

Schick

(1999) Advanced AC calorimetry of polycaprolactone in melting region. Thermochimica Acta 330(1–2): 109–119.

35.

Nguyen

(2013) Modeling shape-memory behavior of polymers. Polymer Reviews 53(1): 130–152.

36.

Park

Chung

Lee

et al . (2008) Shape memory effects of polyurethane block copolymers cross-linked by celite. Fibers and Polymers 9(6): 661–666.

37.

Peponi

Navarro-Baena

Kenny

(2014) Shape memory polymers: properties, synthesis and applications. In: Montemor

(Ed) Smart Polymers and Their Applications. Cambridge: Woodhead Publishing Limited, 204–236.

38.

Rezanejad

Kokabi

(2007) Shape memory and mechanical properties of cross-linked polyethylene/clay nanocomposites. European Polymer Journal 43: 2856–2865.

39.

Sadeghi

Mehdi Talakesh

Ghalei

et al . (2013) Preparation, characterization and gas permeation properties of a polycaprolactone based polyurethane-silica nanocomposite membrane. Journal of Membrane Science 427: 21–29.

40.

Simmons

Hyvarinen

Poole-Warren

(2006) The effect of sterilisation on a poly(dimethylsiloxane)/poly(hexamethylene oxide) mixed macrodiol-based polyurethane elastomer. Biomaterials 27(25): 4484–4497.

41.

Srivastava

Chester

Anand

(2010) Thermally actuated shape-memory polymers: experiments, theory, and numerical simulations. Journal of the Mechanics and Physics of Solids 58(8): 1100–1124.

42.

Sun

Mei

Song

et al . (2006) The in vivo degradation, absorption and excretion of PCL-based implant. Biomaterials 27(9): 1735–1740.

43.

Taherzadeh

Baghani

Baniassadi

et al . (2016) Modeling and homogenization of shape memory polymer nanocomposites. Composites Part B: Engineering 91: 36–43.

44.

Tobushi

Hashimoto

Ito

et al . (1998) Shape fixity and shape recovery in a film of shape memory polymer of polyurethane series. Journal of Intelligent Material Systems and Structures 9: 127–136.

45.

Wang

L-F

Wei

Y-H

(2005) Effect of soft segment length on properties of fluorinated polyurethanes. Colloids and Surfaces B: Biointerfaces 41(4): 249–255.

46.

Xiao

Choi

Lakhera

et al . (2013) Modeling the glass transition of amorphous networks for shape-memory behavior. Journal of the Mechanics and Physics of Solids 61(7): 1612–1635.

47.

Xiong

H-M

(2013) ZnO nanoparticles applied to bioimaging and drug delivery. Advanced Materials 25(37): 5329–5335.

48.

Zhang

Ogunmekan

Ibekwe

et al . (2015) Healing of shape memory polyurethane fiber-reinforced syntactic foam subjected to tensile stress. Journal of Intelligent Material Systems and Structures 27: 1792–1801.

49.

Zheng

Ozisik

Siegel

(2005) Disruption of self-assembly and altered mechanical behavior in polyurethane/zinc oxide nanocomposites. Polymer 46(24): 10873–10882.