Introduction

Stimuli-responsive polymers (SRPs) are smart materials which can show noticeable changes in their properties by environmental stimulus variations [1]. Shape memory polymers (SMPs) are an interesting class of SRPs, which have attracted great concern due to their specific properties and potential applications in various fields from biomedicine to aerospace [2]. SMPs can be deformed and fixed into a stable predefined temporary shape and have the capability to recover their original permanent shape under an external stimulus such as temperature [3], light [4], pH [5], moisture [6], chemicals [7], electrical or magnetic field changes [8, 9]. While, the shape memory effect can be activated electronically, magnetically, or electromagnetically, the most common shape memory trigger is thermal activation. Thermo-responsive SMPs are stimuli-sensitive materials whose shape is modulated by heat [7]. In general, thermally induced SMPs are considered to consist of molecular switches and net-points [10, 11]. The net-points can be either physically or chemically cross-linked structures. The latter usually shows higher mechanical strengths due to stronger interactions between the polymer chains. The switching segments determine the switching temperature, which needs to be exceeded to induce the temporary shape and recover the permanent shape when cycled through their thermal transition temperature (Ttrans) [12]. Ttrans typically may be either glass transition temperature (Tg) or melting point temperature (Tm). However, Tm is preferred due to a sharper transition and better shape recovery [13]. From biomedical point of view, combination of shape memory effect and the good biocompatibility of polymers has led to SMPs which is mostly interested for biomedical applications such as drug delivery, biosensors and biomedical devices, and implant materials [2, 14, 15]. Examples of such devices include stents [16, 17], heart valves, and septal defect occluder [15, 18]. Improvement of the degradable SMPs such as implant materials due to their minimally invasive surgery as well as no need for second surgery to remove implant materials is highly regarded [19, 20]. Various types of biodegradable SMPs based on poly(DL-lactide) (PLA) and poly(ε-caprolactone) (PCL) have been prepared for the development of smart implant devices [21]. The SMPs consist of PCL are of particular interest in biomedical applications due to their biocompatibility, biodegradability, elasticity, sharp and tunable Ttrans, high shape recoverability, high shape fixity ratios and good mechanical properties [22,23,24]. Thus, PCL-based SMPs have been widely studied and prepared as both physically and chemically crosslinked networks. Some methods have been reported to modify the Tm of PCL to the body temperature [25], which adjusting the average molecular weight (Mn) of PCL chains and designing the branch structure for PCL are the most popular methods. In addition to biodegradability, thermally induced SMPs indicated the potential for use as intelligent materials in controlled drug delivery systems [15, 17, 21, 26]. Neffe et al. prepared a crosslinked SMPs by UV-curing of oligo[(ε-caprolactone)-co-glycolide]-dimethacrylates. Then, Ethacridine lactate (EL) and Enoxacin (EN) were incorporated into the polymer matrix. The influence of incorporation of drugs on the shape memory functionality, biodegradation of the SMP network on drug diffusion, as well as the programming process and shape recovery on drug release were studied [12, 15, 17]. Nagahama et al. developed a new branched architecture SMPs based on chemically crosslinking of branched oligo(ε-caprolactone) precursors with hexamethylene diisocyanate (HDI) in the presence of theophylline. The resulted polymer network, exhibited good sustained drug releasing in 35 days with remarkable temperature sensitivity and shape recovery in a narrow temperature range from 37 to 39 °C [15, 17, 27]. Also, the SMP network of star shape amorphous copolyester urethane was prepared by Wiscke et al. The resulting polymer demonstrated a controlled release of the loaded drug without losing its SMP properties in 80 days [17, 21, 28]. Recently, Serrano et al. reported a new class of crosslinked shape memory elastomers with capability of eluting drugs based on citric acid and dioles [17, 29].

We are interested in preparing new biocompatible and biodegradable PCL-based SMPs with switching temperature around body temperature. We have also developed drug-eluting SMPs to develop controlled release devices for minimally invasive surgery systems. Although, a few multifunctional polymers with shape-memory effect, biodegradability and controlled drug release have been developed, our research strategy for the development of such a multifunctional polymer system was based on used methods of the incorporation of drug molecule to affect the shape memory functionality and drug release profile. Hence, in our previous work [30], hydroxypropyl cellulose (HPC), a nontoxic, biodegradable, and biocompatible cellulose derivative, was used as a multifunctional initiator for grafting PCL in different reaction conditions. The grafted samples were crosslinked by HDI and subsequently the process was optimized by the response surface methodology (RSM). HPC was essential to reduce the PCL degree of crystalinity, therefore, a reduced Tm for PCL could be observed for prepolymers (HPC-g-PCL) and chemically crosslinked polymeric networks. The synthesized polymer networks based on PCL and HPC (X(HPC-g-PCL)) had adjustable Ttrans ranging from 32 to 53 °C.

In the present study, we have developed SMPs comprising PCL and HPC with excellent shape fixing capability and mechanical properties. They also have demonstrated good shape recovery properties and their Ttrans are slightly above the body temperature (39–41 °C). Naproxen is a propionic acid derivative and a non-steroidal anti-inflammatory drug (NSAID) with anti-inflammatory, antipyretic and analgesic activities and as a model drug was incorporated into the polymer matrix by two different methods. In situ drug incorporation method allows the crosslinking of HPC-g-PCL films in the presence of drug. In the swelling method the drug loads after the crosslinking step. So, the main objective of this article is to investigate the thermal and mechanical properties and shape memory behavior of cross-linked HPC-g-PCL polymers before and after drug incorporation. Furthermore, drug release behaviors from the drug-loaded films were studied. The chemical structure of the HPC-g-PCL polymers and networks were characterized using 1H NMR spectroscopy, differential scanning calorimetry (DSC), X-ray diffraction (XRD), tensile test, and shape memory programming processes.

Experimental section

Materials

Hydroxypropyl cellulose (HPC) was purchased from Aldrich (Milwaukee, WI, USA; Mn = 100,000 gmol−1 according to the manufacturer) and dried under vacuum at room temperature for 24 h before use. The molar substitution of propoxy groups (MSHPC) was approximated 1.25 using 1H NMR spectrum of HPC [31]. ε-Caprolactone (CL) was purchased from Merck (Germany) and dried over freshly powdered CaH2 for 24 h and then purified by twice distillation process under reduced pressure. Tin (II) bis (2-ethylhexanoate) (Sn(Oct)2, 96%) was purchased from Alfa Acer. Hexamethylene diisocyanate (HDI) was provided from Fluka and used without any further purification. Naproxen was purchased from Pharmacia and recrystallized in ethanol and distilled water for three times. Chloroform and toluene were dried by heating over sodium with benzophenone as indicator and freshly distillated before use. Tetrahydrofuran and n-hexane were used as received.

Synthesis procedure

Synthesis of HPC-g-PCL polymers

The synthesis procedure of the HPC-g-PCL prepolymers were detailed in our previous article (Scheme S1, Supporting Information) [30]. The preparation method of the HPC-g-PCL-98 (with 98%wt. of CL) is described as a representative case. Thus, HPC and CL at 2:98 weight ratio were added into a dry flask and the mixture was slowly stirred at room temperature for 24 h until HPC was completely swelled by CL. Then, the swelled mixture was placed in an oil bath at 130 °C under argon atmosphere. Then 1.25% (w/w) of Sn(Oct)2 to CL was dissolved in dry toluene (0.05 g/mL) and added into the mixture. The resulted mixture was stirred for 24 h at the same temperature. In order to remove PCL homopolymer as the main by-product, the obtained polymer was dissolved in THF at a concentration of 5% (w/v). It was precipitated subsequently in equal volume of cold n-hexane. The purification procedure was repeated three times and the pure grafted polymer was dried overnight to yield branched HPC-g-PCL-98 prepolymer [31].

Preparation of X(HPC-g-PCL) networks

Shape memory films were previously prepared in our laboratory by crosslinking a series of HPC-g-PCL prepolymers having different chain lengths of PCL graft with different quantities of HDI through two different routes [30]. In this work, SMP films of PCL with thickness about 300 μm were prepared by dissolving 1 g of prepolymer in 15 mL of anhydrous CHCl3. As a homogenous viscose solution was obtained, 15% (v/w) HDI was added as a cross-linker. The mixture was molded into a Teflon dish and dried at room temperature for 24 h. Then, the casted film was cured in an oven at 80 °C for another 24 h (Route II in our previous work) [30] to obtain X(HPC-g-PCL)-N networks (N is related to the weight content of CL in samples).

Preparation of drug- loaded X(HPC-g-PCL) films

As a model drug, naproxen was incorporated into the polymer network by two different methods.

In situ method of drug-loading was conducted by dissolving two different amounts of naproxen (3.0 and 10.0%wt.) in a chloroform based solution of prepolymer. The solutions were stirred for 30 min. The resulted drug/polymer mixtures were subjected to crosslinking to obtain X(HPC-g-PCL)-N/P% (In situ) (N and P are related to the weight content of CL and weight percent of dissolving drug in samples).

Swelling method of drug-loading was directed by immersing the dry polymer networks in a 100-fold excess (v/w) of a saturated drug solution in chloroform for 24 h with subsequent drying the drug-loaded films at reduced pressure to obtain X(HPC-g-PCL)-N/P% (Swelling). The amount of incorporated naproxen was quantified by acidic methanolysis of the polymer network. Then, the concentration of extracted naproxen was determined by UV spectroscopy as described by Neff et.al. [12].

Characterization techniques

The swelling ratio (Q) and gel content (G) of X(HPC-g-PCL) networks were determined to evaluate the cross-linking degree of grafted polymer networks [32, 33]. For this purpose, samples were cut into small slices and weighted carefully (mo). After swelling in chloroform for 48 h at room temperature, the swollen extracted specimens were weighted again (ms). Extracted samples were dried at ambient temperature until constant weight was achieved and the final weight (md) was recorded. Q and G were calculated according to the following equations:

$$ {\displaystyle \begin{array}{c}Q=1+\frac{\uprho_1}{\rho_2}\left(\frac{m_s}{m_d}-1\right)\\ {}G=\left(\frac{m_d}{m_o}\right)\end{array}} $$

Where ρ1 and ρ2 are the swelling medium and polymer specific density, respectively. ρ2 was measured using a pycnometer [33].

FT-IR spectra were recorded on a Brucker PS-15 spectrometer (Bruker Optics, Ettlingen, Germany) at wave numbers ranging from 400 to 4000 cm−1. 1H NMR spectra of the HPC-g-PCL polymers were recorded using a 1H NMR spectrometer (400 MHz Brucker SP-400 Avance spectrometernsile Bruker Biospin, Rheinstetten, Germany) using chloroform as solvent and tetramethylsilane as the internal standard. The enthalpy change (ΔHm), melting temperature (Tm) and degree of crystallinity (Xc) of the PCL prepolymer and their corresponding cross-linked films were measured by differential scanning calorimetry (type DSC 822 from Mettler-Toledo, Switzerland) under nitrogen purge from 0 to 100 °C at a heating rate of 10 °C/min in three cycles (first heating, first cooling and second heating) [30]. The crystalline structure of the polymer films was analyzed using X-ray diffraction (XRD) patterns by a Bruker AXS model D8 advanced diffractometer for Cu Kα radiation (λ = 1.54187 Å) at 40 kV and 35 mA with Bragg angle ranging from 3 to 70°. The microstructures of the crosslinked samples, before and after drug incorporating were examined on a VEGA\\TESCAN-XMU scanning electron microscope (SEM). The samples were sputter-coated with gold prior to SEM observation and the examination voltage and magnification level were 10.0 kV and 10,000×, respectively. Mechanical properties of the polymer networks (dumbbell shape specimen, according to ISO 527-2 (type 5B)) was determined in an elongation rate of 5 mm/min at 22, 37 and 60 °C using a tensile test machine (type Z010 from Zwick/Roell, Germany) equipped with 10KN load cell [27, 34,35,36]. The shape-memory behaviors of the X(HPC-g-PCL) networks were investigated by cyclic tensile thermo-mechanical tests in a tensile testing machine (type Z010 from Zwick/Roell, Germany) [19, 36]. To examine the shape memory effect, the samples were heated to 65 °C for 10 min and stretched from permanent shape to temporary shape with an elongation rate of 5 mm/min, and hold for 10 min (εm). The aforementioned test was conducted for at least three specimens and was repeated three times. This temporary shape was fixed by cooling the sample to 20 ± 2 °C with the same εm. Subsequently, after unloading the samples and measuring their length (εu), the temperature increased up to Tm and kept again for 10 min to recover the original shape (εp). Finally, the shape fixity (Rf) and the shape recovery (Rr) were calculated according to the following equations [13, 27, 32, 37]:

$$ {\displaystyle \begin{array}{c}{R}_f=\frac{\varepsilon_u\left(\mathrm{N}\right)}{\varepsilon_m}\\ {}{R}_r=\frac{\varepsilon_u\ \left(\mathrm{N}\right)-{\varepsilon}_p\ \left(\mathrm{N}\right)\ }{\varepsilon_u\ (N)-{\varepsilon}_p\ \left(N-1\right)}\end{array}} $$

All drug loaded films were cut dumbbell shape according to ISO 527-2 (type 5B) and immersed into sterilized PBS buffer (pH 7.4, I = 0.14 phosphate buffered solution (PBS)) in a vial (10 mL) with a dialyzed cap (MWCO = 10,000). Subsequently, the vial immersed into a bigger one containing 80 mL of the same release media and incubated at 37 °C. At each selected time point, 0.5 mL of PBS was collected for analysis and a fresh portion of PBS was added to continue the drug release experiment [27]. The concentration of released drug was measured with a UV-vis spectrometer at 229 nm.

Results and discussion

Characterization of SMP networks and drug- loaded films

In our previous work [30], a series of HPC-g-PCL-N prepolymers with different CL/HPC ratios and Sn(Oct)2 catalyst amounts were prepared and cross-linked using various amounts of HDI to obtain X(HPC-g-PCL)-N networks. The detailed chemical structure of HPC-g-PCL prepolymers and networks is shown in Scheme S1 (Supporting Information). Among the prepared networks, the X(HPC-g-PCL)-90, X(HPC-g-PCL)-93 and X(HPC-g-PCL)-95 samples were selected for this study due to their high Tc (crystallization temperature) and Xc (degree of crystallinity) which are necessary to obtain good mechanical properties and SMP behavior. Furthermore, the X(HPC-g-PCL)-98 sample was prepared and studied due to its highest degree of crystallinity for comparison with aforementioned samples.

Figure S1 (Supporting Information) shows the FT-IR spectra of the prepared HPC-g-PCL-98 and X(HPC-g-PCL)-98 samples. The characteristic absorption peaks for the HPC-g-PCL-98 prepolymers were observed in the wave number region of ester carbonyl group stretching modes at 1726 cm−1. Also, –OH group stretching mode at 3438 cm−1 attributed to the existence of terminal -OH of PCL. From the FT-IR spectrum of X(HPC-g-PCL), the crosslinking was confirmed by the presence of the new absorption bands corresponding to the N-H bending and stretching vibration of the urethane bonds appeared in 1580 and 3320 cm−1, respectively. Also, the absorption peak of C=O in X(HPC-g-PCL) was slightly stronger and wider than the HPC-g-PCL due to the presence of carbamate and ester peaks [32]. Appearance of a new bond in 1620 cm−1 can be attributed to urea linkage which produced by a side reaction of H2O with isocyanate groups, during the crosslinking procedure. It indicated successful crosslinking of prepolymers through terminal hydroxyl groups of PCL grafts and atmospheric H2O by HDI. 1H NMR spectrum of HPC-g-PCL-98 is shown in Fig. S2 (Supporting Information). As reported in previous research [30] the characteristic peaks of the HPC-g-PCL appeared as peaks a-e (Fig. S2, Supporting Information), approved the well synthesis of HPC-g-PCL-98.

In order to evaluate the effect of drug loading by different methods on polymer properties, naproxen as a well water-soluble drug at different loading levels was incorporated by in situ and swelling methods before and after crosslinking of HPC-g-PCL prepolymers, respectively. However, in spite of four samples prepared by the in situ method named: X(HPC-g-PCL)-95/3%, X(HPC-g-PCL)-95/10%, X(HPC-g-PCL)-98/3% and X(HPC-g-PCL)-98/10%, the samples prepared by the swelling method were unavoidably non-identical in terms of the drug content. The obtained results showed that naproxen loading in matrix of X(HPC-g-PCL)-95 and X(HPC-g-PCL)-98 polymers in order reached to 7.8 wt% and 12 wt% so they were named as X(HPC-g-PCL)-95/7.5% and X(HPC-g-PCL)-98/12%, respectively [12].

The values of gel content (G) and degree of swelling (Q) of the various X(HPC-g-PCL) and drug loaded-X(HPC-g-PCL) samples along with the calculated values of Mn for grafted PCL in corresponding precursor polymers are listed in Table 1. The average length of the PCL segments (L) was estimated by 1H NMR signals at δ = 3.64 and 4.06 ppm, corresponding to the initial (IHe) and terminal (IHe′) methylene protons of the CL units, respectively (Fig. S2, Supporting Information). Based on this assignment, (L) could be calculated according to L = IHe/IHe′ [27].

Table 1 Gel content and degree of swelling of X(HPC-g-PCL) networks without and with drug loading
Full size table

As shown in Table 1, Q and G strongly depend on the molecular weight of PCL segment and content of loaded drug. Due to the high efficiency of curing through the crosslinking step, all systems present high values of G ranging from 92% to 99%. According to Table 1, as the chain length of PCL grafts decreases the number of grafting PCL chains increases, which led to an increase of the cross-link density. Also, by increasing PCL chain length the free volume in the polymer network became larger and its capacity to swell by more amount of solvent increased [33]. Similar finding was obtained for drug-loaded samples. As it is concluded from Table 1, in situ incorporation of drug into the polymer matrix, especially for high concentration of drug, are likely to disturb the crosslinking reaction. The resulting lower crosslinking densities could decrease G and increase Q compared to unloading drug state. Therefore, minimum values of G are belonged to drug loaded X(HPC-g-PCL)-95 with high amount of drug (10%). The obtained results for Q can be explained in the same way. Q has various data ranging from 10.2 for X(HPC-g-PCL)-95 to 28.18 for the drug loaded X(HPC-g-PCL)-98 sample due to the increasing of PCL chain length from 2451 to 5242. In this case, experimental data confirmed that drug loading reduces crosslink density and therefore led to greater swelling capability [32].

The thermal properties of synthesized prepolymers and networks were characterized using the DSC analysis. The curves obtained from the second DSC heating run for HPC-g-PCL prepolymers and X(HPC-g-PCL) films are shown in Fig. 1a and b, respectively. Their main thermal characteristic is summarized in Table S1 (Supporting Information). As it can be easily concluded from Table S1 (Supporting Information), decreasing of CL level leads to lower Tm and Xc. This result can be explained based on two reasons; the first is the amorphous nature of HPC that inhibited the crystal formation of grafted PCL chains via the incorporation HPC as a backbone. The second reason is the enhancement of overall number of active hydroxyl groups on the initiator by increasing the concentration of HPC. This can be resulted the lower length of PCL grafts as side chains. A notable decrease in Tm and Xc are observed upon network formation indicating that it has a strong influence on the degree of crystallinity via more limitation in mobility of PCL chains [27, 30, 32]. Also, in situ drug loaded networks show higher Tm, Tc and Xc than unloaded networks (Table 2) due to the interference of in situ drug loading by crosslinking reaction. The interaction of the naproxen carboxylic group with polycaprolactone hydroxyl groups in competition with diisocyanate molecules prevents chains chemically crosslinking reactions, which leads to lower crosslinking density and increases the degree of crystallinity of the network. However, in drug loaded networks by the swelling method, there is a significant reduction in these parameters for X(HPC-g-PCL)-95/7.8% and X(HPC-g-PCL)-98/12% networks, which evidently verify the networks morphological changes due to the drug incorporating (Fig. 1c and d). For further explanation, it can be attributed to the hydrogen bonds which resulted from the carboxylic group of naproxen with some of the hydroxyl groups of PCL which have been not participated in the crosslinking reaction. Consequently, this interaction slows down the polymeric chain movements, and ultimately reduces the degree of crystallinity of the polymeric network. On the other hand, naproxen as an impurity and a small molecule can place between polymeric chains, and inhibit polymer molecular movements. So, it reduces the degree of the crystallinity of PCL chains.

Fig. 1
figure 1

DSC thermograms of HPC-g-PCL-90, −93, −95 and − 98 prepolymers (a), X(HPC-g-PCL)-90, −93, −95 and − 98 films (b), unloaded and drug loaded X(HPC-g-PCL)-95 films (c) and unloaded and drug loaded X(HPC-g-PCL)-98 films (d)

Full size image
Table 2 Melting and crystallization temperatures, and degree of crystallinity of drug loaded X(HPC-g-PCL) films
Full size table

Figure 2a-d shows X-ray diffraction (XRD) patterns of the HPC-g-PCL prepolymers and their corresponding drug loaded and unloaded cross-linked films. As shown in Fig. 2a, the main diffraction peaks located at 2θ = 21.56 and 23.84° in agreement with those reported for PCL in the literature [30]. As seen in Table 2, the values of Xc, obtained from the XRD analysis are in well agreement with DSC results. These results suggest that the partially crystalline phase in the HPC-g-PCL originated from PCL side chains. Since the crystallinity strongly depends on PCL chains, the crystallinity of PCL side chains increases by increasing of their chain length. In accordance with the results of the DSC analysis, the intensity of crystalline diffraction peaks of PCL also decreased in crosslinked films (Fig. 2b). It can be interpreted based on the more amorphous nature of crosslinked films than their corresponding grafted copolymers.

Fig. 2
figure 2

XRD patterns of HPC-g-PCL-90, −93, −95 and − 98 prepolymers (a), X(HPC-g-PCL)-90, −93, −95 and − 98 films (b), unloaded and drug loaded X(HPC-g-PCL)-95 films (c) and unloaded and drug loaded X(HPC-g-PCL)-98 films (d)

Full size image

The degree of crystallinity of the crosslinked films is also affected by drug loading (Fig. 2c-d). According to Tables 1 and 2, as a consequence of the drug concentration enhancement, crosslink density is decreased and it caused significant increment in the -PCL crystallinity. In spite of drug loading via the in situ method, naproxen incorporation by the swelling method decrease the degree of crystallinity in the both X(HPC-g-PCL)-95 and X(HPC-g-PCL)-98 films. To explain the reason of this feature, as mentioned in previous section, it is suggested that the drug incorporation induced some changes in the network morphology and it decreased the PCL crystallinity. However, the results show the degree of crystallinity of the X(HPC-g-PCL)-95 is more affected by drug loading in compared to X(HPC-g-PCL)-98. The reason for this behavior can be probably interpreted that PCL chain length in X(HPC-g-PCL)-95 is shorter than X(HPC-g-PCL)-98. So, expect in the equal weights of both X(HPC-g-PCL)-95 and X(HPC-g-PCL)-98 networks, X(HPC-g-PCL)-95 have more hydroxyl groups, which can be led to the more chain-drug interactions. Consequently, X(HPC-g-PCL)-95 experiences more changes in its morphology than X(HPC-g-PCL)-95, and these more morphological changes will cause to more the degree of crystallinity reduction for X(HPC-g-PCL)-95 compared to X(HPC-g-PCL)-98.

The SEM micrographs represented in Fig. 3 show the cross-sectional view of the fracture surfaces of samples including the HPC-g-PCL-95 and HPC-g-PCL-98 prepolymers, their related crosslinked networks, and X(HPC-g-PCL)-98/3% and X(HPC-g-PCL)-98/12% drug loaded films via in situ and swelling methods, respectively. The SEM images verify the change of the HPC-g-PCL morphology after crosslinking and drug incorporation processes. A smooth ductile morphology was observed in both HPC-g-PCL-95 and 98 samples (Fig. 3a, b), whilst the brittle structure was seen for X(HPC-g-PCL)-95 and 98 (Fig. 3c, d) films. The morphological studies also supported the DSC and XRD results. As seen in Fig. 3e, the arrows marked presence of naproxen into the X(HPC-g-PCL)-98/3% (In situ). This sample shows smoother ductile fracture surfaces than the respected unloaded crosslinked network due to decrease of the crosslink density via the in situ naproxen loading method. On the other hand, drug loading by swelling (X(HPC-g-PCL)-98/12%) show more brittle fracture surfaces than the respected unloaded X(HPC-g-PCL)- 98 network (Fig. 3f).

Fig. 3
figure 3

SEM images of HPC-g-PCL-95 (a), HPC-g-PCL-98 prepolymers (b), X(HPC-g-PCL)-95 (c), X(HPC-g-PCL)-98 films(d), drug loaded X(HPC-g-PCL)-98/3% (In situ) (e) and X(HPC-g-PCL)-98/12% (swelling) (f)

Full size image

Table S1 (Supporting Information) shows that Tm of X(HPC-g-PCL) films could be reduced to around body temperature by using of short PCL chains. Also, based on the DSC and XRD analyses, it could be concluded that X(HPC-g-PCL)-95 and X(HPC-g-PCL)-98 films may have suitable mechanical properties and the crystalline structure. Therefore, these films were chosen to undergo further analysis of the thermo-mechanical properties and shape memory effect. The mechanical properties of samples were evaluated from different points of view such as the effect of PCL graft chain length at different temperatures (22 °C, 37 °C and 65 °C), the influence of the drug loading method and the drug content of films. Young’s modulus (E), elongation at break (εb %) and tensile strength (σm) of cross linked samples, obtained from tensile stress–strain curves, are summarized in Tables 2, 3, and 4. Figures 4a-b compare the mechanical properties of X(HPC-g-PCL)-95 and X(HPC-g-PCL)-98 samples at different temperatures (22 °C, 37 °C and 65 °C). The mechanical properties of the X(HPC-g-PCL)-98 film is significantly higher than the previously reported PCL-based SMP (Fig. 4b) [23, 27]. It is essential to recognize that thermo-mechanical properties affected by two important factors; the crystalline domains of the PCL chains and the crosslink density of network. The former acts as physical crosslinks and its effect is observed at lower temperature (22 °C), where both samples are at below their respective Tm and are partially crystalline. But, the latter has the more effect at high temperature (65 °C), because both samples are above their respective Tm and are completely amorphous. At 37 °C the samples show a partially crystalline nature which is an intermediate mode between the samples at 22 °C and the ones at 65 °C. Therefore, the E values are significantly controlled by crystallinity at low temperatures, but it is influenced by the crosslink density at higher temperatures. As shown in Table 3, by increasing of temperature, samples become much softer (lower E, and (σm)) and their mechanical properties decrease [23, 38]. According to the Tables 1 and 2, the longer PCL segment chain lengths, leads to the lower crosslink density and higher crystallinity. Hence, as observed in Table 3, at 22 °C the maximum and minimum values of E referred to X(HPC-g-PCL)-98 and X(HPC-g-PCL)-95 samples, respectively, while this order was completely inversed at 65 °C [21]. In addition, the mechanical properties of polymer networks containing drugs were determined by tensile test at 22 °C. Figures 4c-d depict the tensile curves in the unloaded and drug loaded states by two different loading methods of naproxen. According to Table 4, drug incorporation has no significant effect on σm and εb compared to the unloaded networks. On the other word, due to no change in the PCL chain length, both unloaded and drug loaded networks have almost the same σm and εb values. These results achieve while significant changes have been seen in Young’s modulus after drug incorporation. While drug content reaches to 3% through the in situ method, a significant enhancement in Young’s modulus can be seen for X(HPC-g-PCL)-98 and X(HPC-g-PCL)-95 samples (increasing from 376 to 528 MPa and from 143 to 326 MPa, respectivly). Further increase in naproxen content to 10% resulted less enhancement of E compared to their respective 3% drug loading samples. It should be related to the drug aggregation that cannot really affect well on the increasing the modulus. As expected, both X(HPC-g-PCL)-98 and X(HPC-g-PCL)-95 loaded networks by the swelling method exhibited a partially drop of E amount. This behavior could be resulted from changing morphology of network due to the drug incorporation.

Table 3 Mechanical properties of X(HPC-g-PCL) films, measured at 22, 37 and 65 °C
Full size table
Table 4 The effect of drug content, loading method and release of the drug on thermo-mechanical properties of X(HPC-g-PCL) films, measured at 22 °C
Full size table
Fig. 4
figure 4

Stress-strain curves of X(HPC-g-PCL)-95 films at 22, 37 and 65 °C (a), X(HPC-g-PCL)-98 films at 22, 37 and 65 °C (b), X(HPC-g-PCL)-95 drug loaded and unloaded films at 22 °C (c), X(HPC-g-PCL)-98 drug loaded and unloaded films at 22 °C (d)

Full size image

According to Table 4, it is well seen that the entire drug loaded networks experienced a significantly reduction in their mechanical properties after the drug releasing procedure. This reduction in the mechanical properties can be related to the creation of some defects in polymer matrix during the drug releasing procedure. Briefly, it can be noted that morphological and not structural changes (crosslink density) during the swelling method lead to significant mechanical changes for this type of films after drug release than the one loaded via the in situ method. Although the incorporation of drug into the polymer matrix decreases some mechanical properties, it could not distinctly influence the application of the materials in the biomedical field [39].

The thermally induced shape-memory behavior of X(HPC-g-PCL)-N films is shown in Fig. 5a-d . To the best of our knowledge, shape recovery is mostly affected by cross-linking density while shape fixity is influenced by crystallinity of the switching segments in networks. The shape-recovery (Rr), and shape fixity (Rf) of unloaded and drug loaded X(HPC-g-PCL)-95 and X(HPC-g-PCL)-98 films were measured by thermo-mechanical tensile experiments and the results are listed in Tables 5 and 6. During the test, the samples were first elongated at 65 °C to a certain strain (εm) of 50% (for the X(HPC-g-PCL)-95 film) (Fig. 5a) or 100% (for the X(HPC-g-PCL)-98 film) (Fig. 5b), by applying a constant elongation rate of 5 mm/min. Temporary shape was fixed by cooling the samples to 22 °C at the same constant strain, then unloaded strain (εu) was correspondingly recorded. Subsequently, shape recovery was determined upon heating the samples up to their corresponding Tm and εp was recorded. This completed thermo-mechanical cycle (N = 1), and three additional cycles were carried out using the same method [19, 37, 40, 41]. The shape-memory behavior for X(HPC-g-PCL) networks was clearly observed. By considering almost fully crosslinking of all unloaded HPC-g-PCL polymers (Table 1), both X(HPC-g-PCL)-98 and X(HPC-g-PCL)-95 samples exhibited excellent shape recovery (Rr) values even after 3 cycles, which were reliably higher than 95%. [32, 42]. The Rr values are relatively low in the first cycle, but exceeded to 99% after 3 cycles. Although, molecular weight and crystallinity of PCL segments have only a negligible effect on Rr, their effects on Rf are quite obvious. The Rf value was increased by increasing of molecular weight of the PCL side chain. Increasing the crystallinity of the PCL side chain causes the material to be able to hold inner stress to preserve temporary shape. Hence, as seen in Table 5, the X(HPC-g-PCL)-98 sample with the high degree of crystallinity content of PCL side chain exhibits higher value of shape fixity than X(HPC-g-PCL)-95 sample [19, 32].

Fig. 5
figure 5

Force–strain curves obtained from thermo-mechanical tests with initial deformation at room temperature for unloded X(HPC-g-PCL)-95 films (a), unloaded X(HPC-g-PCL)-98 films (b), drug loaded X(HPC-g-PCL)-95 and X(HPC-g-PCL)-98 films after drug releasing (c), Shape recovery ratio vs. temperature for unloaded X(HPC-g-PCL)-95 and X(HPC-g-PCL)-98 films in water (d)

Full size image
Table 5 Shape fixity, shape recovery, and transition temperature of X(HPC-g-PCL) films at different thermo-mechanical test cycles
Full size table
Table 6 Shape fixity, shape recovery, and transition temperature of drug loaded X(HPC-g-PCL) films after drug releasing
Full size table

Shape memory behavior of X(HPC-g-PCL)-95 and X(HPC-g-PCL)-98 films was also investigated after drug releasing. For this purpose, all samples were elongated up to 50% of their original length under the same condition which previously described for unloaded samples. The results are summarized in Table 6. As seen in Fig. 5c, Rf values have been significantly decreased by reducing the crystallinity of PCL side chains. Low crystallinity causes the PCL chains not to be able to preserve the internal stress to maintain the temporary shape [32].

As seen in Tables 2, 3%w/w naproxen loading into the polymer matrix through the in situ method did not affect significantly on the crystallinity of the X(HPC-g-PCL)-98 network. Therefore, in situ drug-loaded (after drug releasing) networks as well as unloaded polymer ones show comparable shape fixity (Table 5). As expected, the swelling incorporation method allows further morphological changes and subsequent reduction of networks crystallinity. Thereby, the swelling incorporation of naproxen reduces Rf as was observed in Table 5. For instance, the differences of shape fixity for unloaded X(HPC-g-PCL)-98 and loaded X(HPC-g-PCL)-98 after drug releasing was significant. Rf in X(HPC-g-PCL)-98 drug loaded by swelling experienced a decrease of 23% after drug releasing)Rf = 72%). The results showed drug releasing from X(HPC-g-PCL)-95, which loaded by both drug incorporation methods, led to evidently reduction of Rf.

As mentioned before, the in situ drug loading method reduces the networks crosslinking density and both X(HPC-g-PCL)-95 and X(HPC-g-PCL)-98 samples exhibited lower Rr compared to unloaded films. However, both X(HPC-g-PCL)-95 and X(HPC-g-PCL)-98 samples, which drug loaded by the swelling method, kept their shape recoverability in comparison with the corresponding unloaded networks. These results confirmed that the drug loading by the swelling method have no influence on the networks crosslinking density.

Strain recovery curves of the X(HPC-g-PCL)-95 and X(HPC-g-PCL)-98 samples as a function of temperature was conducted in water according to the description follow and the results are shown in Fig. 5d. First, the samples were elongated up to 200% of their original length in 65 °C hot water. Then cooled down below their melting point (fixing the temporary shape) and, eventually, it placed into a 20 °C water bath and warmed in a rate of 1 °C per minute (till 39 °C). As seen in Fig. 5d, both the measured curves are S-shaped, and Ttrans for X(HPC-g-PCL)-95 and X(HPC-g-PC)-98 are around 36 and 41 °C, respectively. The Ttrans of both networks are significantly lower than their Tm, and are in a good agreement with the melting onset point on the corresponding DSC thermograms.

Shape memory behavior of X(HPC-g-PCL) is demonstrated in Fig. 6a-c. X(HPC-g-PCL) in its permanent shape (6a) was heated at 55 °C and kept for 2 min. The sample was deformed into its temporary shape and cooled into a 0 °C ice water bath for freezing the stress and fixing into the temporary shape for 1 min (6b) [42]. Eventually, it placed into a 20 °C water bath and warmed in a rate of 1 °C per minute. When the water bath temperature reached to 39 °C the deformed sample come back to its permanent shape just in 10s (6c).

Fig. 6
figure 6

Thermally induced shape-memory behavior of X(HPC-g-PCL) SMP films (with 4.8 cm length and 0.1 mm thickness) (a) Permanent shape (b) temporary shape fixed at o °C (c) the recovered shape after 10 s immersion in water at 39 °C

Full size image

In vitro naproxen release behavior

Drug-loaded SMP films were prepared using naproxen as a conventional model drug. Figure 7 shows cumulative profiles of naproxen release from X(HPC-g-PCL)-95 and X(HPC-g-PCL)-98 films which loaded by two different methods. Figure 7a-c exhibit drug release profiles of X(HPC-g-PCL)-95/3%, (In situ) and X(HPC-g-PCL)-98/3% (In situ) (a); X(HPC-g-PCL)-95/10% (In situ) and X(HPC-g-PCL)-98/10% (In situ) (b); and X(HPC-g-PCL)-95/7.8% (Swelling) and X(HPC-g-PCL)-98/12% (Swelling) (c) films. As seen in Fig. 7, the release curves are basically identical for all samples. Results show that the initial release of the drug depends on the amount of the loaded drug, so higher amount of loaded drug leads to greater initial release. In both (a) and (b) films, regardless of the amount of loaded drug, it is clearly seen that the initial release of the drug from X(HPC-g-PCL)-95 is more than the X(HPC-g-PCL)-98. It can be attributed to the higher degree of entanglement in X(HPC-g-PCL)-98 chains, due to its longer chain length, which resulted to more trapping drug and more slowly rate of releasing [27, 29, 38, 40,41,42,43,44]. Drug release profile of (c) films showed two importance differences than the in situ method: (i) a burst initial release was seen for both network due to dominate surface adsorption of swelling compared to the in situ method; (ii) the initial release of the drug from X(HPC-g-PCL)-95 is less than the X(HPC-g-PCL)-98. This result can be related to more swelling degree for X(HPC-g-PCL)-98 than X(HPC-g-PCL)-95 which led to more drug incorporated into X(HPC-g-PCL)-98 compared to X(HPC-g-PCL)-95. On the other side, inherently PCL is a hydrophobic polymer, so it is reasonable that PCL chains experiences no swelling in water as a drug releasing medium. Hence, in this study, the drug releasing rate is independent from network swelling degree, but related to the amount of drug loading and the chain entanglements.

Fig. 7
figure 7

In vitro release profile of naproxen from X(HPC-g-PCL)-95/3%, (In situ), X(HPC-g-PCL)-98/3% (In situ) (a), X(HPC-g-PCL)-95/10% (In situ), X(HPC-g-PCL)-98/10% (In situ) (b), and X(HPC-g-PCL)-95/7.8% (Swelling) and X(HPC-g-PCL)-98/12% (Swelling) films in PBS at 37 °C

Full size image

Below a temperature of 37 °C, no significant UV absorption of naproxen was obtained after the burst initial release for both in situ and swelling drug loaded networks. It is a strong reason for the thermo-sensitivity of the X(HPC-g-PCL) polymer for the controlled drug release.

Conclusion

In summary, a series of PCL-based network polymers X(HPC-g-PCL) were prepared via ring opening polymerization of CL by HPC and then were chemically cross-linked the PCL side chains in excellent yields. Obtained networks were characterized in terms of their gel content, degree of swelling, melting temperature, degree of crystallinity, and morphology. The degree of crystallinity and the melting temperature of resulted cross-linked polymers were affected by altering the molecular weight of PCL side chains. The results of DSC and XRD analyses showed that the crystalline structure of the X(HPC-g-PCL) sample originated from PCL side chains and related to CL/HPC ratio in the reaction feed. Among synthesized samples, X(HPC-g-PCL)-95 and − 98 films satisfy thermo-mechanical properties with combining shape memory effect. The results showed that the mechanical properties significantly depend on the polycaprolactone chains length. On the other hand, the amount of loaded drug and the type of the loading method caused significant effects on mechanical properties. The X(HPC-g-PCL)-98 sample with higher molecular weight of PCL showed high shape recovery behavior by a narrower recovery temperature ranging from 39 to 41 °C as well as higher shape fixity than the X(HPC-g-PCL)-95 sample. As a result, by tailoring the graft copolymer architecture precisely, shape-memory switching temperatures could be successfully adjusted near body temperature while retaining a sharp transition in a narrow temperature range. X(HPC-g-PCL)-95 and − 98 samples as drug-loaded shape memory polymers were successfully prepared by introducing naproxen as a model drug through two different drug loading methods (loading by swelling and in situ methods). Drug loading by the swelling method is typically introduced a non-controllable loading levels with burst initial release depending on the drugs physicochemical properties. Besides, the in situ loading method related to controllable drug payload and resulted in lower burst and higher subsequent release. Despite the initial release which depends on the initial amount of the incorporated drug, network capacity and chain entanglements, both drug loading methods showed the same release profiles. X(HPC-g-PCL)-95 and − 98 samples’ thermo-mechanical properties and shape memory effect decreased after the sustained drug release over 70 days, but their ability to use in medical applications is debatable. Therefore, HPC-g-PCL networks could be introduced as SMPs with Ttrans relatively higher than the body temperature (39–41 °C), excellent mechanical properties, shape fixing potential, shape recovery and drug release capability to make a promising system for biomedical applications.