Two-way shape memory property and its structural origin of cross-linked poly(ε-caprolactone)

Abstract

In this work, two-way shape memory (TWSM) properties and the corresponding structural origin of cross-linked poly(ε-caprolactone) (cPCL) with different gel contents obtained by adopting different weight percentages of benzoyl peroxide (BPO) were investigated. The effects of gel contents on melting temperatures, crystallization temperatures and crystallinity of the cPCL materials were studied with differential scanning calorimetry. The TWSM properties of the samples were determined by dynamic mechanical analysis. It was found that gel content was the key factor determining the TWSM behavior, and higher gel content would result in the so-called robust TWSM effect for cPCL samples (cPCLx, x denoting the weight percentage of BPO). Compared with cPCL10, cPCL5 had larger elongation and lower recovery capabilities due to its lower gel content. However, the sample with much lower gel content (cPCL3) displayed almost no TWSM behavior, implying that an appropriate gel content was responsible for the TWSM characteristic. The crystalline structure of cPCLx subsequently changed when subjected to external stress. As the samples were cooled down under constant stress, higher stress would lead to a more oriented crystalline structure. Furthermore, concurrent wide-angle and small-angle X-ray scattering investigations revealed the structural evolution occurring during the TWSM process, indicating that the crystal orientation along the stretching direction took place simultaneously with the elongation during the cooling process.

---

1 Introduction

Shape memory polymers (SMPs) are materials that can change their bulk shape in response to a vast array of external stimuli such as heat, light and electric field.^1–19 The initial and final shapes of SMPs can be programmed and reprogrammed multiple times. As a consequence, SMPs may find wide application in temperature sensors,²⁰ minimally invasive surgery,²¹ stents²² and other fields.²³ However, it must be admitted that their application is limited by the intrinsic irreversibility of shape memory transformations, namely that the shape memory process of samples can occur repeatedly, but external stress for shape deformation and its removal is necessary.^4,6,24 This is called a one-way shape memory (OWSM) process, or a not fully reversible process. Full reversibility is required for many applications, such as in actuators and artificial muscles.²⁵ Consequently, the development of polymers featuring a fully reversible process is the goal of recent researches, in which the material is able to vary between distinct shapes with only the need for a temperature change. This is the so-called two-way shape memory (TWSM) process.

Recently, several groups have pioneered researches to investigate the properties of TWSM materials.^26–37 An important advance is the discovery by Mather et al.²⁶ that a cross-linked semi-crystalline polymer network has the capability of exhibiting the TWSM effect under constant stress. It has been proved that TWSM behavior could be achieved^26–29,34 when a constant tensile stress is applied to a cross-linked semi-crystalline material that is repeatedly heated above the melting temperature (T_m) and cooled below the crystallization temperature (T_c). This is because a significant elongation would take place upon cooling below T_c and the reverse upon heating above T_m. Such processes are called crystallization-induced elongation (CIE) upon cooling and melting-induced contraction (MIC) upon heating, respectively. Westbrook et al.³⁵ have developed a one-dimensional constitutive model to accurately describe the complex thermomechanical TWSM effect, in which the strain actuation mechanism during the cooling process can be ascribed to two aspects: the first one is due to the decrease in the modulus of entropic elasticity, and the second one is due to the formation of the stretching-induced crystallization (SIC) phase. Westbrook et al.³⁶ have also prepared an SMP composite that demonstrates TWSM effects in response to changes in temperature without the requirement of a constant load. Ge et al.³⁷ have presented a theoretical model for an SMP composite actuator to investigate its thermo-mechanical behavior and the mechanisms during the actuation cycles. It was found that higher gel content was needed for cross-linked poly(ethylene-co-vinyl acetate) to exhibit a TWSM effect.²⁷ Previous work by Pandini et al.²⁹ has also shown that cross-linked poly(ε-caprolactone) with different gel contents, prepared with various macromolecular architectures starting from linear, three- and four-arm star poly(ε-caprolactone) functionalized with methacrylate end-groups, presents both OWSM and TWSM properties. It seems that some material parameters, such as gel content and crystalline transition, are important requirements for semi-crystalline polymers to show TWSM properties. Most previous reports paid attention to the synthesis of TWSM polymers and factors affecting the TWSM performance. However, a systematic understanding of the correlation among gel content, TWSM behavior and structure evolution is still needed.

As a low-toxicity, fully bio-/hydrolytic degradable and biocompatible polymer, poly(ε-caprolactone) (PCL) has potential applications in the biomedical field.³⁸ The shape memory effect of PCL is possibly realized at around the human body temperature by lowering the shape memory transition temperature (T_trans). The T_trans corresponding to the melting temperature of PCL can be tailored through changing the gel content. Therefore, cross-linked poly(ε-caprolactone) (cPCL) was selected to study the TWSM property in the present work. The cPCL samples were prepared by simple radical-initiated cross-linking of PCL. The gel contents of cPCL samples could be controlled through using different weight percentages of benzoyl peroxide (BPO). The TWSM behavior could be tuned easily by altering the gel content and external stress. More importantly, the instantaneous crystallization and segmental orientation behaviors of material during the cooling process in the TWSM cycle were studied. Based on the experimental results, a possible mechanism is proposed in an effort to understand the structural origin of TWSM behavior.

2 Experimental

2.1 Materials

PCL (M_n = 45 000 g mol⁻¹) and BPO (98%) were purchased from Sigma-Aldrich and J&K Chemical, respectively. Tetrahydrofuran (THF) was of analytical grade. All chemicals were used as received without any further purification.

2.2 Sample preparation

PCL and BPO were mixed in a Thermo Scientific Haake Minilab II mixer (Germany) at 65 °C for 10 min and the resulting uncured materials were processed by compression molding at 130 °C and 10 MPa for 20 min to get cross-linked samples (ESI, Fig. S1†). The BPO content varied from 3 wt% to 10 wt% based on the weight of PCL. All the cross-linked samples were named cPCLx (x = 3, 5 and 10, x denoting the weight percentage of BPO). For thermal analysis and shape memory characterization, rectangular strips with a size of 15 mm in length, 4 mm in width and 0.15 mm in thickness were cut from the compression-molded plaques.

2.3 Determination of gel content

Specimens (initial mass, m₀) were cut out from the cPCL samples, and put into a large volume of THF at room temperature for about 24 h to extract the non-cross-linked linear PCL component. In order to determine the residual mass after extraction (m_d), the extracted specimens were dried in a vacuum oven at 40 °C until constant weight was obtained. The gel content (G) was calculated according to the following equation:³⁰where G represents the weight fraction of cPCL component in the sample. The reported value was the average calculated from at least three samples that were prepared with the same composition and under the same processing conditions.

2.4 Thermo-mechanical characterization

The thermal behavior of samples was determined using a differential scanning calorimeter (DSC Q2000, TA Instruments), which was calibrated with indium before measurements. Each sample was analyzed under a thermal program including a first heating run from 0 °C to 80 °C at 10 °C min⁻¹, followed by a cooling run to −30 °C at 10 °C min⁻¹ and by a second heating run to 80 °C at 10 °C min⁻¹. The melting temperature (T_m), crystallization temperature (T_c) and the heat of fusion (ΔH_m) were evaluated from the second heating run. The crystallinity of samples was evaluated according to the most commonly used equation:where ΔH_m is the measured melting enthalpy from the second heating run, and ΔH⁰_m is the melting enthalpy of a 100% crystalline PCL polymer (134.9 J g⁻¹).²⁹

Dynamic-mechanical behavior was investigated by means of dynamic mechanical analysis (DMA; Q800, TA Instruments) of the rectangular strips. The tests were performed in tension mode from −100 °C to 90 °C at 3.0 °C min⁻¹. The frequency and the strain were 1 Hz and 0.1%, respectively. In order to avoid severe temperature gradient, the low rate of 3.0 °C min⁻¹ was selected for the heating and cooling runs of DMA tests.

2.5 TWSM testing

The TWSM properties of samples were investigated by the aforementioned DMA machine in a controlled force mode. The TWSM cycle contained a deformation step and a subsequent recovery step under constant non-zero stress, in which the samples were deformed at 70 °C under a certain stress, and cooled to −10 °C at 3 °C min⁻¹ while holding the applied stress constant. Then the samples were heated to 70 °C at 3 °C min⁻¹ under the same constant stress. This cycle was repeated three times to evaluate the cyclic reproducibility. A schematic diagram of the TWSM cycles is shown in Fig. 1.

Fig. 1 Schematic diagram of the temperature and stress as a function of time during the TWSM process. The heating and cooling rate is 3 °C min⁻¹.

2.6 Structural evolution

The crystal structure at the end of the cooling step during the TWSM cycle was investigated by ex situ wide-angle and small-angle X-ray scattering (WAXS and SAXS). The tests were carried out on cPCL5 and cPCL10 samples that had been previously heated to 70 °C, deformed at various levels of stress (in the range between 0.15 and 0.6 MPa) and cooled to −10 °C under fixed stress. These samples were kept in a refrigerator (at about −15 °C) prior to WAXS and SAXS experiments.

The structural evolution of samples in the TWSM cycle was investigated by means of in situ WAXS. The tests were carried out on cPCL10 sample with a tensile hot stage (Linkam TST 350). The sample was stretched at 70 °C to a strain of 100%, and then cooled to 0 °C at 3 °C min⁻¹ under fixed stress.

WAXS and SAXS were carried out on beamlines BL14B1 and BL16B1, respectively, in the Shanghai Synchrotron Radiation Facility (SSRF). The wavelength of the radiation source was 1.24 Å. The WAXS patterns were collected by a MAR 225 detector with a resolution of 3072 × 3072 pixels (pixel size = 73 × 73 μm²). The SAXS patterns were collected by a MAR 165 detector with a resolution of 2048 × 2048 pixels (pixel size = 79 × 79 μm²). The image acquisition time was 6 s for WAXS and 60 s for SAXS. The sample-to-detector distance was 386.0 mm for WAXS and 5220.8 mm for SAXS. All the scattering patterns were corrected for background scattering, air scattering and beam fluctuations.

The 2D SAXS radially integrated intensities I(q) (q = 4π sin θ/λ) were obtained by integration in the azimuthal angular range of 0° ≤Φ ≤ 360°. The long period (L) was calculated with the Bragg equation:

where q_max corresponds to the peak position in the scattering curves. The meridional intensities were obtained by integration in the azimuthal angular range of 45° ≤Φ ≤ 135° (meridional direction is along the stretching direction).³⁹

3 Results and discussion

3.1 Materials structure and properties

Three cPCL samples designated as cPCL3, cPCL5 and cPCL10 were obtained in this study, with the numbers reflecting the weight percentages of BPO (Table 1). For comparison purposes, a reference non-cross-linked PCL (cPCL0) is included. It is obvious that the gel content of cPCL samples increases monotonically with BPO content. The marked change of gel content is from cPCL3 (64.0%) to cPCL5 (85.2%), whereas the increasing rate of gel content is slower as the BPO further increases.

Table 1 Measured gel contents and thermal characterization of the various cPCL samples

Sample	BPO (wt%)	Gel content (%)	Melting temperature Tm (°C)	Crystallization temperature Tc (°C)	Crystallinity χc (%)
cPCL0	0	0	55.7	32.5	56.0
cPCL3	3	64.0	52.3	28.9	46.9
cPCL5	5	85.2	49.7	25.1	43.2
cPCL10	10	95.7	41.6	13.1	32.5

---

The melting temperature (T_m), crystallization temperature (T_c) and crystallinity (χ_c) of the samples were measured, as shown in Fig. 2a and b, and the results are summarized in Table 1. The crystallization and melting temperatures of the cPCL samples are much reduced with respect to the typical values of a linear PCL, which may be attributed to the cross-linked structure, and the reduction becomes more important for higher gel content. The same is the case for crystallinity. The higher the gel content, the lower the crystallinity. So the crystallization is inhibited by the cross-linked structure. As a result, it is possible to vary the shape memory transition temperature through changing the BPO content in the initial PCL–BPO mixtures.

Fig. 2 DSC thermograms of the cPCL samples, a: the second heating curves; b: the cooling curves. c: storage modulus (E′) vs. temperature for the cPCL samples.

Although the DSC results provide a quick glance at the thermal properties of the polymers, the shape memory behaviors of polymers are closely related to their thermo-mechanical properties. DMA tests were carried out for all the samples and the curves are shown in Fig. 2c. A rather pronounced glass transition (T_g) at −40 °C is observed in the DMA curves. The uncross-linked polymer (cPCL0) exhibits flow behavior after the melting transition. Comparison among all the curves in Fig. 2c suggests that the cross-linked structure significantly impacts the modulus above melting transitions, for which all the cPCL samples reach modulus plateaus. It is easily seen that the plateau modulus increases with gel content. On the one hand, the modulus plateau of cPCL3 is remarkably less flat, indicating partial suppression of flow due to its much lower gel content and very poor cross-linked network. In contrast, the modulus plateau of cPCL10 is more flat and large, suggesting almost complete suppression of flow due to its higher gel content and more perfect cross-linked network.

3.2 Evidence of two-way shape memory property

The TWSM behaviors of the cPCL samples were investigated under fixed stress. In a typical TWSM process, the polymer at zero strain was heated to 70 °C and a stress was then applied, leading to a pre-stretching of the sample (εⁱ_H). During the cooling process, an evident elongation occurred, resulting in a low-temperature strain (ε_L). By reheating to 70 °C, the sample reversed the elongation, leading to another strain (ε^f_H). The TWSM property was quantitatively characterized in terms of two parameters,²⁶ namely: the actuation magnitude representing the elongation taking place during cooling, and the recovery magnitude representing the percent of the elongation recovered during heating process. These parameters were defined as:

Actuation magnitude R_act =(ε_L − εⁱ_H) × 100% (4)

The TWSM property of cPCL3 was firstly investigated. Three consecutive shape memory cycles were performed under identical experimental conditions and the curves are presented in Fig. 3a. The values of R_act and R_rec for each cycle are summarized in Table 2. A continuous creep-like strain increase is observed during the cooling step in Fig. 3a instead of a distinct crystallization-induced elongation (CIE) event. Regardless of the identical experimental conditions, there is a significantly marked upper shifting of strain from one cycle to the next due to the much lower gel content of cPCL3. Moreover, R_rec is rather minimal in all three cycles. Overall, cPCL3 has almost no TWSM property.

Fig. 3 3D TWSM cycles of the cPCL samples; a: cPCL3; b: cPCL5; c: cPCL10; the black, red and blue lines represent cycle 1, cycle 2 and cycle 3, respectively. In view of the DMA range, a stress of 0.03 MPa was applied to cPCL3, which was much smaller than that applied to cPCL5 and cPCL10 (0.3 MPa).

Table 2 TWSM behavior of the cPCL samples in three cycles

Sample	Actuation magnitude Ract (%)			Recovery magnitude Rrec (%)
	Cycle 1	Cycle 2	Cycle 3	Cycle 1	Cycle 2	Cycle 3
cPCL3	26.9	12.7	6.7	3.3	9.9	47.0
cPCL5	65.1	49.1	46.1	65.1	88.4	95.9
cPCL10	14.5	13.1	13.2	86.3	97.3	97.1

---

In contrast, distinct CIE, melting-induced contraction (MIC) phenomenon and no noticeable displacement of strain for both cPCL5 and cPCL10 are observed in all three cycles (Fig. 3b and c). cPCL5 and cPCL10 show highly repeatable TWSM effect due to the low plastic flow of the non-cross-linked PCL that is almost completely suppressed by the surrounding cross-linked network. Nevertheless, cPCL5 and cPCL10 exhibit different elongation and recovery capabilities in the TWSM process (Table 2). The lower the gel content, the higher the pre-stretching elongation (εⁱ_H) and the cooling-induced elongation (ε_L − εⁱ_H). However, the recovery ability is better for samples with higher gel content. As a consequence, it is found that higher gel content is a requisite for cross-linked semi-crystalline polymers to exhibit TWSM property. It has been reported that the overall strain in the TWSM process consists of the reversible strain change and the irreversible strain change (the upshift of the strain range between cycles).⁴⁰ For cPCL3 with much lower gel content, the irreversible strain is much larger than the reversible strain in the TWSM process. But for cPCL5 and cPCL10, the reversible strain is much larger.

What is more, an apparent improvement in R_rec for all the samples is observed as the number of cycles increases, in accordance with previous reports.^27,40 This is largely due to the fact that the irreversible strain would slow down upon consecutive cycling, whereas the reversible strain remains relatively unchanged upon cycling.

3.3 Effects of gel content on the two-way shape memory property

Here, the impact of gel content on TWSM property is further investigated. For samples deformed under fixed stress, the instantaneous deformation speeds V_d (% °C⁻¹) and recovery speeds V_r (% °C⁻¹) are calculated using:where and are the temperature derivative of strain directly obtained from the deformation and strain recovery data, respectively.

The CIE and MIC events are clearly illustrated by plotting the strain derivatives (i.e., instantaneous deformation rate and recovery rate) against the temperature. Fig. 4 shows the deformation behavior and strain recovery behavior of cPCL samples in cycle 1 under fixed stress. Upon cooling from 70 °C, a pronounced increase of V_d is visible in a certain temperature range for both cPCL5 and cPCL10, corresponding to the CIE events. V_d and onset temperature of CIE for cPCL5 are higher than those of cPCL10, suggesting a larger CIE level, since cPCL5 has a higher crystallization temperature due to its lower gel content. By reheating the samples to a temperature below T_m, V_r of the two samples show almost no change. Then a remarkable growth of V_r of both cPCL5 and cPCL10 appears, which are named MIC events.

Fig. 4 (a) Deformation behaviors and (b) strain recovery behaviors of cPCL samples in cycle 1 under fixed stress.

Ex situ WAXS and SAXS have been applied to provide a complementary study of the effects of gel content on the TWSM property (Fig. 5). In the unstretched state, the WAXS patterns of both cPCL5 and cPCL10 show two evident sharp Debye–Scherrer reflections, as typically found for isotropic crystalline orientation. These rings correspond to the two characteristic reflections of the PCL orthorhombic crystal form, located at 2θ = 17.1° and 19.0° and related respectively to the (110) and (200) planes,⁴¹ as shown in Fig. 6a. For all samples, no change in angular position along 2θ is found, suggesting that gel content and stress have no influence on the orthorhombic structure of the crystal phase. The samples treated with a load of 0.3 MPa present discontinuous Debye–Scherrer rings, indicating the presence of preferred orientation of crystallites.²⁹ The crystalline phase of elongated samples displays an oriented structure, which could be explained similarly to what has been observed in PCL/PVC blends,⁴² in which a row nucleation under strain produced linear nuclei parallel to the strain direction, and then folded-chain lamellae could grow epitaxially from the row nuclei with their chain axis parallel to the strain direction.

Fig. 5 WAXS and SAXS patterns of cPCL5 and cPCL10 samples as unstretched and subjected to cooling under fixed stress. The stretching direction is vertical.

Fig. 6 (a) The 1D intensity profiles of cPCL5 and cPCL10 samples; (b) the corresponding azimuthal scanning profiles of the (110) plane reflection; (c) 1D SAXS profiles along the meridian; the samples were unstretched and subjected to cooling under fixed stress. a1,b1,c1: unstretched cPCL5; a2,b2,c2: unstretched cPCL10; a3,b3,c3: cPCL5 subjected to cooling under a load of 0.3 MPa; a4,b4,c4: cPCL10 subjected to cooling under a load of 0.3 MPa.

For the purpose of providing a precise evaluation of the preferred orientation, the azimuthal scanning profiles for (110) plane reflection of all samples are presented in Fig. 6b. Table 3 shows the values of peak width at half height and long period of all samples under different conditions. It is well known that a lower value of peak width at half height is associated with a more oriented structure and a practically constant signal represents the absence of a preferred orientation. The value of peak width at half height of cPCL5 is smaller, suggesting a higher degree of orientation. This is due to the larger strain of cPCL5 that results from its lower gel content. The long period of cPCL5 is smaller under zero stress, as the lower gel content will lead to higher crystallinity and thus to smaller long period. Furthermore, the long period of cPCL5 increases a lot under a load of 0.3 MPa, while that of cPCL10 changes little. This is also ascribed to the larger strain of cPCL5 that gives rise to a great increase of long period.

Table 3 The peak width at half height and long period (L) for cPCL5 and cPCL10 samples unstretched and subjected to cooling under fixed stress

Sample	cPCL5		cPCL10
Stress (MPa)	0	0.3	0	0.3
Peak width at half height (°)	—	38.5	—	49.3
Long period (nm)	16.5	17.1	17.1	17.3

---

3.4 Effect of stress on the two-way shape memory property

The effects of external stresses on the TWSM property were explored by applying stresses ranging from 0.15 MPa to 0.6 MPa. The curves are shown in Fig. 7 and the results are reported in Tables 4 and 5. As can be seen, higher stresses lead to larger elongational effects. Compared with cPCL10, the cPCL5 samples show a similar trend, but with larger R_act and smaller R_rec at the same stress. The applied stress for cPCL5 cannot be 0.6 MPa, as the resulting strain goes beyond the range of DMA. Depending on the applied stress and the gel content, the elongations of all the samples range from a few percent up to about 100%. For a given stress, higher R_act and lower R_rec are found for the samples with lower gel content. For example, under a stress of 0.45 MPa, the R_act and R_rec values in cycle 1 of cPCL10 are about 25% and 88%, whereas those of cPCL5 are around 97% and 58%. In addition, the difference of R_act between cPCL5 and cPCL10 becomes more pronounced at higher stress, which can be ascribed to the much greater extensibility of the less cross-linked structure.

Fig. 7 TWSM responses of cPCL10 sample under different stresses.

Table 4 Results of the TWSM characterization of cPCL10 samples tested at various levels of stress

cPCL10	Actuation magnitude Ract (%)			Recovery magnitude Rrec (%)			TCIE (°C)
Applied stress (MPa)	Cycle 1	Cycle 2	Cycle 3	Cycle 1	Cycle 2	Cycle 3	Cycle 2
0.15	6.7	6.1	6.1	90.8	99.8	100.3	13.0
0.3	14.5	13.1	13.2	86.3	97.3	97.1	16.7
0.45	25.6	24.1	23.8	87.9	94.6	96.6	18.9
0.6	44.8	45.6	44.2	91.6	93.3	97.0	21.1

---

Table 5 Results of the TWSM characterization of cPCL5 samples tested at various levels of stress

cPCL5	Actuation magnitude Ract (%)			Recovery magnitude Rrec (%)			TCIE (°C)
Applied stress (MPa)	Cycle 1	Cycle 2	Cycle 3	Cycle 1	Cycle 2	Cycle 3	Cycle 2
0.15	14.3	10.1	9.9	62.1	93.3	97.1	29.9
0.3	65.1	49.1	46.1	65.1	88.4	95.9	33.8
0.45	96.9	77.3	79.1	57.3	92.6	94.6	40.6

---

A further effect of stress is to increase the temperature at which the elongational effects take place. This temperature is indicated by T_CIE, representing the inflection point of the curve. T_CIE of cPCL5 is much larger than that of cPCL10, since cPCL5 without strain has a higher crystallization temperature. Besides, higher stress will lead to higher T_CIE, which is caused by the increased crystallization temperature when chains crystallize under stretching. Moreover, the increase of T_CIE for cPCL5 is more pronounced than that for cPCL10 at the same stress, as the strain increment of cPCL5 is much greater. For instance, when the applied stress is increased from 0.15 MPa to 0.45 MPa, T_CIE of cPCL5 increases by about 11 °C, whereas T_CIE of cPCL10 only increases by about 6 °C.

Fig. 8 presents the WAXS and SAXS patterns of cPCL10 samples subjected to cooling under different fixed stresses. Tests have been carried out on the cPCL10 samples that were prepared by heating to 70 °C and applying various stresses (0.15, 0.3, 0.45 and 0.6 MPa) before cooling under fixed stress. The results achieved for these samples were compared with those for an unstretched one.

Fig. 8 WAXS and SAXS patterns of cPCL10 samples as unstretched and subjected to cooling under different fixed stresses. The stretching direction is vertical.

As has been discussed in Section 3.3, the unstretched cPCL10 shows isotropic crystalline orientation. The samples treated with different stresses present discontinuous Debye–Scherrer rings and the discontinuity of the rings becomes more evident as the stress increases, indicating that much more preferentially oriented crystallites are formed. The orthorhombic structure of the crystal phase remains unchanged with an increase of stress, as no change in angular position along 2θ is found (Fig. 9a).

Fig. 9 (a) The 1D WAXS intensity profiles of cPCL10; (b) the corresponding azimuthal scanning profiles for the (110) plane reflection of cPCL10; (c) 1D SAXS profiles along the meridian; (d) evaluation of the peak width at half height and long periods (L) of cPCL10. The long periods are calculated from the SAXS scattering along the meridian direction. The samples were unstretched and subjected to cooling under different fixed stresses.

In order to provide a precise evaluation of the preferred orientation, the azimuthal scanning profiles for the (110) plane reflection of cPCL10 samples cooled under different fixed stresses are presented in Fig. 9b. The values of peak width at half height and long periods are plotted in Fig. 9d. It is noticeable that higher stress leads to lower values of peak width at half height and larger values of long period, namely that higher stress generates a more oriented crystalline structure.

3.5 Structural evolution of sample cooled under fixed stress

In situ WAXS has been used to study the instantaneous crystallization and orientation behavior of materials. The strain vs. temperature curve and the WAXS patterns of cPCL10 cooled under fixed stress are shown in Fig. 10. The corresponding 1D WAXS intensity profiles and azimuthal scanning profiles for (110) plane reflection of cPCL10 are presented in Fig. 11. Firstly, the strain increases very slowly when cooled from 70 °C to 36 °C (a–c). Secondly, upon continuing cooling (36 °C–24 °C, (c–f)), the strain increases rapidly. Fig. 11a shows that the rapid elongation process is accompanied by crystallization, which is the so-called CIE event. It should be noted that, due to the difference of instruments, the level and the starting temperature of the CIE event on a tensile hot stage are rather different from that tested by DMA. Finally, the strain is almost unchanged below 24 °C.

Fig. 10 The strain vs. temperature curve and the WAXS patterns of cPCL10 sample cooled under fixed stress; a: 70 °C; b: 40 °C; c: 36 °C; d: 32 °C; e: 28 °C; f: 24 °C. The stretching direction is vertical.

Fig. 11 (a) The corresponding 1D WAXS intensity profiles and (b) azimuthal scanning profiles for (110) plane reflection of cPCL10 sample cooled under fixed stress.

In Table 6, it is very apparent that the values of peak width at half height increase firstly (34 °C–28 °C), and then decrease slightly (28 °C–24 °C). This phenomenon could be explained as follows. Before crystallization (70 °C–36 °C), there is no visible orientation of the PCL chain in the amorphous state as a practically constant signal is found in Fig. 11b, although the applied strain leads to a reorientation of the cross-linked network. When the sample starts to crystallize under fixed stress, two behaviors occur, including crystallization and a quite large increase of strain (CIE). Upon cooling from 34 °C to 28 °C, the values of peak width at half height increase greatly, indicating that many crystals, with their molecular chains having some deviation from the stretching direction, are formed. At this region, the crystallization dominates in the process, while the increase of strain has little influence on the orientation behavior. At temperatures between 28 °C and 24 °C, the values of peak width at half height decrease slightly, representing a higher degree of orientation. This may be because the increase of strain dominates, leading to a more oriented crystal structure along the stretching direction. Accordingly, it is safe to draw the conclusion that the orientation of crystals is a process that takes place simultaneously with the deformation one.

Table 6 Results of the peak width at half height of cPCL10 sample tested under fixed stress

Temperature (°C)	70	40	36	34	32	30	28	26	24
Peak width at half height (°)	—	—	—	24.4	28.7	32.0	32.8	31.8	31.5

---

3.6 Structural origin of two-way shape memory property

From the above discussion, it can be concluded that TWSM property of materials depends strongly on the congregated structures (gel content). The structural characteristics of cPCL samples are much more complicated. Understanding the structural origin of the TWSM property is only possible in the larger context of complex morphology and molecular dynamics. The structure of cPCL samples is considered as a combination of cross-linked PCL network and free linear PCL molecular chains. A schematic sketch of the structural evolution of samples with different gel contents in the TWSM cycle is shown in Fig. 12.

Fig. 12 2D TWSM cycles of cPCL samples. (A1) Samples with high gel content; (B1) samples with low gel content. The sample (a1 and b1) is heated above T_m (a2 and b2), deformed at a given stress (a3 and b3), and then cooled to low temperature under fixed stress undergoing an entropy elongation (a4 and b4) and a crystallization-driven elongation (a5 and b5), and finally recovers to a certain strain (a3′ and b3′) when heated above T_m. Schematic representation of the structural evolution and the corresponding photographs during the TWSM cycle of the cPCL samples: (A2) samples with high gel content; (B2) samples with low gel content.

The structure of samples with high gel content can be seen as a combination of more perfect cross-linked PCL network and fewer free linear PCL molecular chains. Firstly, the sample is in a random orientation state and has no preferentially oriented crystallites, as shown in Fig. 12a1. When the sample is heated to 70 °C, all crystallites are molten (Fig. 12a2). Then the sample is stretched at 70 °C under fixed stress (Fig. 12a2 and a3). At this time, the cross-linked sample is deformed in the rubbery state and it is the cross-linked PCL network that sustains the most stress. Afterwards it is cooled to −10 °C under constant stress (Fig. 12a3–a5). Before crystallization, the modulus of entropic elasticity decreases with the temperature,³⁵ leading to a slow increase of strain (Fig. 12a3 and a4). When part of the cross-linked PCL network starts to crystallize (Fig. 12a4 and a5), the ability of the polymer to carry the stress reduces. This is because that there is less of a cross-linked PCL network available and the newly formed crystalline phase in a stress-free state is unable to carry the stress either. This is in accordance with previous reports.^43–45 The only way that the sample can sustain the constant stress is to stretch, so a deviation from the first slow increase of strain takes place suddenly. With further deformation of the sample, the newly formed crystalline phase also deforms. This crystallization and elongation process will be repeated until the crystallites are totally formed (Fig. 12a5). Just as has been shown in Fig. 8, the PCL crystallites are preferentially oriented parallel to the direction of strain. Upon reheating the sample, the first slight increase of strain originates from thermal expansion and then the recovery process takes place at temperatures above T_m (Fig. 12a5–a3′). It must be admitted that the sample can almost recover to its pre-stretched state and a number of TWSM processes with better repeatability could be realized by cooling and heating the sample cyclically under the same fixed stress.

Although the samples with high gel content have TWSM property, they would exhibit different elongation and recovery capabilities in the TWSM process due to the difference of gel content. Compared with cPCL10, cPCL5 with lower gel content has a slightly less perfect cross-linked PCL network and more free linear PCL molecular chains, resulting in a larger strain (both pre-stretching elongation and cooling-induced elongation) and lower R_rec under the same stress. This can be mostly ascribed to the easier uncoiling of less cross-linked structures and the slippage of free linear PCL molecular chains. In addition, the crystallinity and T_CIE of cPCL5 are larger than those of cPCL10 due to the lower gel content of cPCL5.

The structure of samples with much lower gel content can also be seen as a combination of a very poor cross-linked PCL network and a lot of free linear PCL molecular chains (Fig. 12b1). The deformation process above T_c during the cooling process results from both the deformational behavior of cross-linked sample in the rubbery state and tremendous slippage of free linear PCL molecular chains (Fig. 12b3 and b4). However, the CIE event is extremely unobvious when the sample begins to crystallize (Fig. 12b4 and b5). The quickly completed crystallization and the higher crystallinity lead to the fact that the stress is not large enough to quickly stretch the already formed stiff crystalline sample. On the other hand, due to the high plastic flow of the free linear PCL molecular chains, only a very minor recovery takes place at temperatures above T_m (Fig. 12b5–b3′); that is, the sample cannot recover to the pre-stretched state.

4 Conclusions

In this work, the TWSM property of cPCL samples with different gel contents and its structural origins have been studied. The cPCL samples were synthesized via radical-initiated cross-linking of PCL polymers. Using different weight fraction of BPO, it was possible to obtain materials with different gel contents, melting and crystallization temperatures, and crystallinity. The results showed that gel content had a significant influence on the TWSM properties of samples. Higher gel content was required for a much stronger TWSM effect. When heated and cooled in a certain temperature region, the samples with higher gel content showed a TWSM effect under fixed stress, which was ascribed to the significant elongation that took place on cooling and the large contraction that happened on heating. The elongational process contained two contributions: one caused by a rubber elasticity process prior to crystallization and the other due to a crystallization-induced process. Furthermore, it was very apparent that the samples exhibited different elongation and recovery capabilities in the TWSM process due to the difference of gel content. For example, cPCL5 had larger elongation and lower recovery magnitude than cPCL10. However, for cPCL3 with much lower gel content, the plentiful plastic flow of free linear PCL molecular chains during the deformation process and the quick crystallization meant that it had nearly no TWSM effect. WAXS and SAXS results showed that the oriented crystal structure was closely related to the values of gel content and the applied stress. Under the same constant stress, the lower the gel content, the higher the degree of orientation. Higher stress would result in a more oriented crystalline structure. Moreover, the orientation of crystals was a process that took place simultaneously with the elongational process.

Acknowledgements

We would like to thank the generous financial support of the following grants: National Natural Sciences Foundation of China (no. 51173195, 51203170), R&D Program of the Ministry of Science and Technology (no. 2013BAE02B02).

References

1. A. Lendlein and S. Kelch, Angew. Chem., Int. Ed., 2002, 41, 2034.
2. M. Behl and A. Lendlein, Mater. Today, 2007, 10, 20.
3. C. Liu, H. Qin and P. T. Mather, J. Mater. Chem., 2007, 17, 1543.
4. J. S. Leng, X. Lan, Y. J. Liu and S. Y. Du, Prog. Mater. Sci., 2011, 56, 1077.
5. T. Xie, Polymer, 2011, 52, 4985.
6. J. L. Hu, Y. Zhu, H. H. Huang and J. Lu, Prog. Polym. Sci., 2012, 37, 1720.
7. M. Heuchel, T. Sauter, K. Kratz and A. Lendlein, J. Polym. Sci., Part B: Polym. Phys., 2013, 51, 621.
8. J. Zhao, M. Chen, X. Y. Wang, X. D. Zhao, Z. W. Wang, Z. M. Dang, L. Ma, G. H. Hu and F. H. Chen, ACS Appl. Mater. Interfaces, 2013, 5, 5550.
9. X. Y. Wang, J. Zhao, M. Chen, L. Ma, X. D. Zhao, Z. M. Dang and Z. W. Wang, J. Phys. Chem. B, 2013, 117, 1467.
10. A. H. Torbati, H. B. Nejad, M. Ponce, J. P. Sutton and P. T. Mather, Soft Matter, 2014, 10, 3112.
11. E. Huitron-Rattinger, K. Ishida, A. Romo-Uribe and P. T. Mather, Polymer, 2013, 54, 3350.
12. Q. Ge, X. F. Luo, C. B. Iversen, P. T. Mather, M. L. Dunn and H. J. Qi, Soft Matter, 2013, 9, 2212.
13. E. Kurahashi, H. Sugimoto, E. Nakanishi, K. Nagata and K. Inomata, Soft Matter, 2012, 8, 496.
14. G. X. Fei, G. Li, L. S. Wu and H. S. Xia, Soft Matter, 2012, 8, 5123.
15. G. X. Fei, C. Tuinea-Bobe, D. X. Li, G. Li, B. Whiteside, P. Coates and H. S. Xia, RSC Adv., 2013, 3, 24132.
16. H. J. Zhang and Y. Zhao, ACS Appl. Mater. Interfaces, 2013, 5, 13069.
17. Y. Zhu, J. L. Hu, K. W. Yeung, H. J. Fan and Y. Q. Liu, Chin. J. Polym. Sci., 2006, 24, 173.
18. X. F. Luo and P. T. Mather, Soft Matter, 2010, 6, 2146.
19. H. F. Lee and H. H. Yu, Soft Matter, 2011, 7, 3801.
20. J. Kunzelman, T. Chung, P. T. Mather and C. Weder, J. Mater. Chem., 2008, 18, 1082.
21. A. Lendlein and R. Langer, Science, 2002, 296, 1673.
22. L. A. Xue, S. Y. Dai and Z. Li, Biomaterials, 2010, 31, 8132.
23. T. Xie and I. A. Rousseau, Polymer, 2009, 50, 1852.
24. L. S. Zhang, Y. H. Jiang, Z. Xiong, X. Q. Liu, H. N. Na, R. Y. Zhang and J. Zhu, J. Mater. Chem. A, 2013, 1, 3263.
25. D. L. Thomsen, P. Keller, J. Naciri, R. Pink, H. Jeon, D. Shenoy and B. R. Ratna, Macromolecules, 2001, 34, 5868.
26. T. Chung, A. Rorno-Uribe and P. T. Mather, Macromolecules, 2008, 41, 184.
27. J. J. Li, W. R. Rodgers and T. Xie, Polymer, 2011, 52, 5320.
28. S. Pandini, S. Passera, M. Messori, K. Paderni, M. Toselli, A. Gianoncelli, E. Bontempi and T. Ricco, Polymer, 2012, 53, 1915.
29. S. Pandini, F. Baldi, K. Paderni, M. Messori, M. Toselli, F. Pilati, A. Gianoncelli, M. Brisotto, E. Bontempi and T. Ricco, Polymer, 2013, 54, 4253.
30. M. Messori, M. Degli Esposti, K. Paderni, S. Pandini, S. Passera, T. Ricco and M. Toselli, J. Mater. Sci., 2013, 48, 424.
31. J. Zotzmann, M. Behl, D. Hofmann and A. Lendlein, Adv. Mater., 2010, 22, 3424.
32. J. Zotzmann, M. Behl, Y. K. Feng and A. Lendlein, Adv. Funct. Mater., 2010, 20, 3583.
33. S. a. T. Jing Zhou, S. M. Brosnan, Q. Li, J.-M. Y. Carrillo, O. G. Dmytro Nykypanchuk, V. S. Ashby, A. V. Dobrynin and S. S. Sheiko, Macromolecules, 2014, 47, 1768.
34. J. M. Raquez, S. Vanderstappen, F. Meyer, P. Verge, M. Alexandre, J. M. Thomassin, C. Jerome and P. Dubois, Chem.–Eur. J., 2011, 17, 10135.
35. K. K. Westbrook, V. Parakh, T. Chung, P. T. Mather, L. C. Wan, M. L. Dunn and H. J. Qi, J. Eng. Mater. Technol., 2010, 132, 041010.
36. K. K. Westbrook, P. T. Mather, V. Parakh, M. L. Dunn, Q. Ge, B. M. Lee and H. J. Qi, Smart Mater. Struct., 2011, 20, 065010.
37. Q. Ge, K. K. Westbrook, P. T. Mather, M. L. Dunn and H. J. Qi, Smart Mater. Struct., 2013, 22, 055009.
38. A. Wurm, E. Zhuravlev, K. Eckstein, D. Jehnichen, D. Pospiech, R. Androsch, B. Wunderlich and C. Schick, Macromolecules, 2012, 45, 3816.
39. Z. J. Xiong, G. M. Liu, X. Q. Zhang, T. Wen, S. De Vos, C. Joziasse and D. J. Wang, Polymer, 2013, 54, 964.
40. T. Xie, J. Li and Q. Zhao, Macromolecules, 2014, 47, 1085.
41. T. Kamal, T. J. Shin and S. Y. Park, Macromolecules, 2012, 45, 8752.
42. Y. Zhao, D. Keroack and R. Prud'homme, Macromolecules, 1999, 32, 1218.
43. I. J. Rao and K. R. Rajagopal, Z. Angew. Math. Phys., 2002, 53, 365.
44. G. Barot and I. J. Rao, Z. Angew. Math. Phys., 2006, 57, 652.
45. G. Barot, I. J. Rao and K. R. Rajagopal, Int. J. Eng. Sci., 2008, 46, 325.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra09385b

---