Investigations on Green Blends Comprising Biodegradable Polymer and Ionic Liquid

The green blends of an ionic liquid, 1-ethyl-3-propylimidazolium bis(trifluoromethanesulfonyl)imide {[EPrI][TFSI]}, and a biodegradable polymer, poly(3-hydroxybutyrate) (PHB), were investigated in this study. The influence of an ionic liquid on the crystallization behaviors of a biodegradable polymer was explored. In the blends, the presence of [EPrI][TFSI] decreased the Tg and Tm of PHB. Incorporating [EPrI][TFSI] in the blends reduced the degree of crystallinity of PHB, inferring that the [EPrI][TFSI] weakened the crystallization of PHB. It further showed retarded isothermal and non-isothermal crystallization for PHB with the presence of [EPrI][TFSI]. The smaller K and 1/t0.5 estimated by the Avrami equation for the blends indicated that [EPrI][TFSI] weakened the isothermal crystallization of PHB with exhibiting the slower crystallization rate. The present study also discussed non-isothermal crystallization of the blends. We found that the Mo model, which is generally used to discuss the non-isothermal crystallization, adequately described the non-isothermal behaviors of the [EPrI][TFSI]/PHB blends. By increasing the [EPrI][TFSI] content, the rate-related parameter F(T) systematically increased, inferring a decreased crystallization rate of PHB with the addition of [EPrI][TFSI] in the blends. The FTIR results suggested an ion–dipole interaction between [EPrI][TFSI] and PHB. This proposes the occurrence of possible complexation between [EPrI][TFSI] and PHB.


Introduction
Conventional petrochemical-based plastics have caused the environmental pollution for our Earth in the last several decades because of their lower degree of biodegradation [1,2]. Recently, much attention has been paid to the biodegradable polymers with the rise of concern about the environmental protection [3][4][5][6][7]. The biodegradable polymers can be mainly classified into the biosynthetic biodegradable polymers and chemosynthetic biodegradable polymers [8,9]. Among them, bacterially-synthesized poly(3-hydroxybutyrate) (PHB) is probably one of the most extensively studied biodegradable polymers. PHB has attracted much interest because of its biocompatibility and biodegradability. Therefore, PHB can be applied for both ecological and biomedical end-uses [10].
In general, PHB presents higher brittleness and a narrow processing window. The brittleness of PHB is mainly associated with its degree of crystallization, and the narrow processing window is due to the poor thermal stability between the processing temperature and melting temperature of PHB. The brittleness and narrow processing window of PHB have been the problems to extend the application of PHB in various fields [11,12]. Blends and copolymerization are often adopted to overcome the shortcomings of PHB [13,14]. It has been found that PHB is miscible with the small molecular additive, diglycidyl ether of bisphenol A (DGEBA) [15]. For the polymer blends of PHB, [EPrI] [TFSI], was provided by Department of Chemical and Materials Engineering, National Yulin University of Science and Technology, Yunlin, Taiwan, Prof. Tzi-Yi Wu's laboratory in collaboration with us. [EPrI][TFSI] was synthesized from 1-ethyl-3-propylimidazolium bromide [27]. The detailed preparation procedure of [EPrI][TFSI] is shown in the literature [27]. The [EPrI][TFSI]/PHB blending samples were prepared by the solution-casting method using chloroform as solvent. The film-casting procedure was performed by evaporating the solvent at 45 • C, followed by vacuum drying at 60 • C for at least 24 h. After evaporating the solvent completely, blending films showing different compositions were obtained.

Instruments and Experiments
Differential scanning calorimetry (DSC) (Perkin-Elmer DSC-8500, Perkin Elmer, Waltham, MA, USA) was utilized to study the thermal and crystallization behaviors of the blends. For T g measurement, the blended samples were heated at a rate of 20 • C/min, and T g values were noted as the onset of transition. Samples for studying the isothermal crystallization were first heated above the melting temperature of the PHB (~190 • C) and then rapidly cooled to various crystallization temperatures (T c ) to crystallize. Exothermal curves for heat flow as a function of time were recorded to analyze the isothermal crystallization. On the other hand, for observing non-isothermal crystallization, all specimens were first annealed at 190 • C for 3 min and then cooled at different cooling rates to enable investigating non-isothermal crystallization.
Scanning electron microscopy (SEM) (Hitachi S3000, Hitachi, Tokyo, Japan) was carried out to confirm the phase morphology with greater magnification. Blended films for SEM observation were solution-casted to be thick enough so that the fracture surface of the thickness (cross-section) could be conveniently examined. Before SEM observation, the fractured blend samples were coated with gold by vapor deposition using vacuum sputtering.
Fourier-transform infrared spectrometer (FTIR) (Perkin-Elmer Frontier TM , Perkin Elmer, Waltham, MA, USA) was used to explore the interactions among the constituents. It recorded all spectra at a resolution of 4 cm −1 and an accumulation of 64 scans in the range of 400~4000 cm −1 . The films for measurements were prepared by dropping the blended solution onto KBr pellets, followed by vacuum drying at 60 • C for at least 24 h to remove residual solvent. The film thickness was controlled so that the IR measurements could be properly processed.  Figure 1, SEM reveals a homogeneous morphology rather than the morphology with phase separation for the blends.

The thermal behaviors of [EPrI]
[TFSI]/PHB blends were also investigated by DSC. Blending samples were first melted at the temperature just above PHB's Tm and then quenched for DSC measurements. Figure 2 shows the sequential DSC scans after melting/quenching each blending sample. As shown in Figure 2, all thermograms exhibited only one single Tg (arrow-marked) for each of the [EPrI][TFSI]/PHB blends with various compositions. In addition, the Tgs and Tms of [EPrI][TFSI]/PHB blends are relatively lower than that of neat PHB, as shown in Figure 2. DSC results demonstrated that the [EPrI][TFSI] in the blends influenced the thermal properties of PHB. It might further suggest that [EPrI][TFSI] moderately weakened the intramolecular interactions among PHB chains and decreased the chain cohesion of PHB as a molecular diluent. This phenomenon would influence the physical properties of PHB in the blends. Similar situations have been found in the literature reporting for other polymer-diluent pairs [30]. The peaks in the range of 25 to 50 °C in Figure 2 are the cold crystallization peaks of PHB in the blends. The degree of crystallinity (Xc) was also evaluated by the melting enthalpy (ΔHm) measured by DSC. For each of the compositions in the blends, the value of ΔHm was calculated by the area of its melting transition shown in Figure 2. It should note that the ΔHm values were normalized by the content of PHB in the blends. The Xc of neat PHB and the [EPrI][TFSI]/PHB blends were further estimated by the following equation: where Δ is the melting enthalpy for 100% crystalline PHB. The Δ value of neat PHB is 146.51 J/g [31]. Table 1 Figure 2 shows the sequential DSC scans after melting/quenching each blending sample. As shown in Figure

The thermal behaviors of [EPrI]
[TFSI]/PHB blends were also investigated by DSC. Blending samples were first melted at the temperature just above PHB's Tm and then quenched for DSC measurements. Figure 2 shows the sequential DSC scans after melting/quenching each blending sample. As shown in Figure   In addition, the Tgs and Tms of [EPrI][TFSI]/PHB blends are relatively lower than that of neat PHB, as shown in Figure 2. DSC results demonstrated that the [EPrI][TFSI] in the blends influenced the thermal properties of PHB. It might further suggest that [EPrI][TFSI] moderately weakened the intramolecular interactions among PHB chains and decreased the chain cohesion of PHB as a molecular diluent. This phenomenon would influence the physical properties of PHB in the blends. Similar situations have been found in the literature reporting for other polymer-diluent pairs [30]. The peaks in the range of 25 to 50 °C in Figure 2 are the cold crystallization peaks of PHB in the blends. The degree of crystallinity (Xc) was also evaluated by the melting enthalpy (ΔHm) measured by DSC. For each of the compositions in the blends, the value of ΔHm was calculated by the area of its melting transition shown in Figure 2. It should note that the ΔHm values were normalized by the content of PHB in the blends. The Xc of neat PHB and the [EPrI][TFSI]/PHB blends were further estimated by the following equation: where Δ is the melting enthalpy for 100% crystalline PHB. The Δ value of neat PHB is 146.51 J/g [31]. Table 1 [30]. The peaks in the range of 25 to 50 • C in Figure 2 are the cold crystallization peaks of PHB in the blends. The degree of crystallinity (X c ) was also evaluated by the melting enthalpy (∆H m ) measured by DSC. For each of the compositions in the blends, the value of ∆H m was calculated by the area of its melting transition shown in Figure 2. It should note that the ∆H m values were normalized by the content of PHB in the blends. The X c of neat PHB and the [EPrI][TFSI]/PHB blends were further estimated by the following equation: where ∆H 0 m is the melting enthalpy for 100% crystalline PHB. The ∆H 0 m value of neat PHB is 146.51 J/g [31]. Table 1 [TFSI] in PHB would reduce and weaken the crystallization of PHB. To further explore the influence of [EPrI][TFSI] on the crystallization behaviors of PHB, discussions on the isothermal crystallization and non-isothermal crystallization will be performed in following sections.  Figure 3 shows the DSC traces for neat PHB and the [EPrI][TFSI]/PHB blends. Blending samples were heated above the melting temperature of the PHB (~190 • C) and held for 3 min to erase any thermal history. Subsequently, the samples were rapidly cooled to various crystallization temperatures (T c = 65, 68, 70, 73 • C) and then maintained at T c to crystallize isothermally. on the crystallization behaviors of PHB, discussions on the isothermal crystallization and nonisothermal crystallization will be performed in following sections.  Figure 3 shows the DSC traces for neat PHB and the [EPrI][TFSI]/PHB blends. Blending samples were heated above the melting temperature of the PHB (~190 °C) and held for 3 min to erase any thermal history. Subsequently, the samples were rapidly cooled to various crystallization temperatures (Tc = 65, 68, 70, 73 °C) and then maintained at Tc to crystallize isothermally. The well-known Avrami equation [32,33] was used to analyze the isothermal crystallization kinetics of the [EPrI][TFSI]/PHB blends. The Avrami equation is given as follows:

Isothermal Crystallization Kinetics
where Xt is the relative degree of crystallinity at time t, the exponent n is a constant with a value depending on the mechanism of crystallization, and the parameter k is a rate constant. In general, the larger k value implies the crystallization process with a faster crystallization rate, and the smaller k The well-known Avrami equation [32,33] was used to analyze the isothermal crystallization kinetics of the [EPrI][TFSI]/PHB blends. The Avrami equation is given as follows: where X t is the relative degree of crystallinity at time t, the exponent n is a constant with a value depending on the mechanism of crystallization, and the parameter k is a rate constant. In general, the larger k value implies the crystallization process with a faster crystallization rate, and the smaller k value means the crystallization showing a slower crystallization rate. The log-log representation of the Avrami equation is also presented as the following equation: Figure 4 shows the log-log plots of Avrami Table 2 shows relative parameters estimated by the fitting results. Meanwhile, the crystallization half-time (t 0.5 ), which is defined as the time at which the extent of crystallization is 50%, can also be determined from the estimated kinetic parameters according to the following form:   Figure 5 shows the DSC thermograms for neat PHB and the [EPrI][TFSI]/PHB blends at different cooling rates of 2.5, 5, 7.5 and 12.5 • C/min. The peak temperature of non-isothermal crystallization (T p ) and the heat of non-isothermal crystallization (∆H n,c ) were analyzed. We summarized the main results of Figure 5 in Figure 6. Figure 6a shows that the T p shifted to a lower temperature as the [EPrI][TFSI] content was increased, regardless of changes in the cooling rate. Figure 6b shows a similar tendency for the values of ∆H n,c (heat of non-isothermal crystallization), which decreased as the [EPrI][TFSI] content was increased. These two phenomena preliminary showed that [EPrI][TFSI] might retard and decline the non-isothermal crystallization of crystalline PHB. The non-isothermal crystallization kinetics could be described by different models including the modified Avrami equation [36], Ozawa analysis [37], and Mo method [38]. The Avrami equation modified by Jexiorny [36] was firstly used to analyze the non-isothermal crystallization kinetics of the [EPrI][TFSI]/PHB blends. The modified Avrami equation suggests that the non-isothermal crystallization with a fixed cooling rate comprises a series of infinite small isothermal crystallization steps. The kinetics of the overall non-isothermal crystallization rate is expressed as the following equation: where the exponent n is a constant depending on the type of nucleation and crystal growth dimension. Also, Z t is the crystallization rate constant related to the nucleation and growth parameters.
In non-isothermal crystallization, time (t) can be related to temperature (T) as the following relationship: where Φ is the cooling rate. It can obtain the values of n and Z t from the slope and intercept of the straight regime from plots of log[−ln(1 − X t )]-v.s.-log(t). Figure 7 where the exponent n′ is a constant depending on the type of nucleation and crystal growth dimension. Also, Zt is the crystallization rate constant related to the nucleation and growth parameters. In non-isothermal crystallization, time (t) can be related to temperature (T) as the following relationship: where Φ is the cooling rate.       The Ozawa equation was also used to discuss the non-isothermal crystallization kinetics in the blends. Ozawa [37] has extended the Avrami equation to further investigate the non-isothermal crystallization process. It was considered that the non-isothermal crystallization process should consist of a large number of infinitesimal isothermal crystallization steps. The Ozawa equation for studying the non-isothermal crystallization process is expressed as the following equation: where X T is the relative crystallinity at a temperature T, K(T) is the cooling function for the overall crystallization rate, and Φ is the cooling rate and m is the Ozawa exponent. The Ozawa exponent depends on the dimension of the crystal growth. One could rearrange the Ozawa equation by taking logarithms on both sides to show the following form: The Ozawa equation was also used to discuss the non-isothermal crystallization kinetics in the blends. Ozawa [37] has extended the Avrami equation to further investigate the non-isothermal crystallization process. It was considered that the non-isothermal crystallization process should consist of a large number of infinitesimal isothermal crystallization steps. The Ozawa equation for studying the non-isothermal crystallization process is expressed as the following equation: where XT is the relative crystallinity at a temperature T, K(T) is the cooling function for the overall crystallization rate, and Φ is the cooling rate and m is the Ozawa exponent. The Ozawa exponent depends on the dimension of the crystal growth. One could rearrange the Ozawa equation by taking logarithms on both sides to show the following form:   Mo and coworkers [38] have also derived a model to study the non-isothermal crystallization. We further applied the Mo model to investigate the non-isothermal crystallization kinetics of the blends comprising [EPrI][TFSI] and PHB. The equation is given below: where logF(T) = [K(T)/k] 1/m and the Mo index "a" is the ratio between the Avrami exponent (n) and Ozawa exponent (m). The F(T) refers to the value of the cooling rate required to reach a defined degree of crystallinity at a certain temperature in the unit crystallization time. It has been shown that the F(T) could be related to the rate of non-isothermal crystallization [41], and a higher value of F(T) could be associated with a retarded crystallization with lower crystallization rate. The parameters a and F(T) can be determined by the slope and intercept of the plot of logΦ versus log(t) at defined relative crystallinity. The plot between logΦ and log(t) gives a straight-line relationship for the blends of [EPrI][TFSI]/PHB as shown in Figure 9, suggesting that the Mo model can properly describe the non-isothermal crystallization kinetics of [EPrI][TFSI]/PHB blends.  Mo and coworkers [38] have also derived a model to study the non-isothermal crystallization. We further applied the Mo model to investigate the non-isothermal crystallization kinetics of the blends comprising [EPrI][TFSI] and PHB. The equation is given below: where logF(T) = [K(T)/k] 1/m and the Mo index "a" is the ratio between the Avrami exponent (n) and Ozawa exponent (m). The F(T) refers to the value of the cooling rate required to reach a defined degree of crystallinity at a certain temperature in the unit crystallization time. It has been shown that the F(T) could be related to the rate of non-isothermal crystallization [41], and a higher value of F(T) could be associated with a retarded crystallization with lower crystallization rate. The parameters a and F(T) can be determined by the slope and intercept of the plot of logΦ versus log(t) at defined relative crystallinity. The plot between logΦ and log(t) gives a straight-line relationship for the blends of [EPrI][TFSI]/PHB as shown in Figure 9, suggesting that the Mo model can properly describe the non-isothermal crystallization kinetics of [EPrI][TFSI]/PHB blends. Mo and coworkers [38] have also derived a model to study the non-isothermal crystallization. We further applied the Mo model to investigate the non-isothermal crystallization kinetics of the blends comprising [EPrI][TFSI] and PHB. The equation is given below: where logF(T) = [K(T)/k] 1/m and the Mo index "a" is the ratio between the Avrami exponent (n) and Ozawa exponent (m). The F(T) refers to the value of the cooling rate required to reach a defined degree of crystallinity at a certain temperature in the unit crystallization time. It has been shown that the F(T) could be related to the rate of non-isothermal crystallization [41], and a higher value of F(T) could be associated with a retarded crystallization with lower crystallization rate. The parameters a and F(T) can be determined by the slope and intercept of the plot of logΦ versus log(t) at defined relative crystallinity. The plot between logΦ and log(t) gives a straight-line relationship for the blends of [EPrI][TFSI]/PHB as shown in Figure 9, suggesting that the Mo model can properly describe the non-isothermal crystallization kinetics of [EPrI][TFSI]/PHB blends.  We further tabulated the parameters estimated by the Mo model in Table 3. At the same extent of crystallinity, one can find that the F(T) value increases systematically with increasing the amount of [EPrI][TFSI] in the blends. The relevant analyses demonstrated that the presence of [EPrI][TFSI] in the blends decreased the non-isothermal crystallization rate of PHB. The [EPrI][TFSI] in the blends would cause a reduction in proceeding non-isothermal crystallization of PHB. We suggest that the efficient dilution effect exerted by the [EPrI][TFSI] in the blends retarded the non-isothermal crystallization of PHB. That is, [EPrI][TFSI] diluted the polymer chains of PHB and significant enlarged the inter-chain distance of PHB by the efficient dilution. The corporative diffusion and aggregation of PHB for proceeding crystallization were diminished, and the crystallization of PHB was then retarded and declined. By the above results, it suggests that the [EPrI][TFSI] significantly influenced the crystallization behaviors of PHB in the blends, and it retarded the isothermal and non-isothermal crystallization of PHB by the dilution effect.  Figure 10. Figure 10a shows the IR spectra of carbonyl stretching region (1800-1660 cm −1 ) for [EPrI][TFSI]/PHB blends. The absorption peak of neat PHB was found at 1740 cm −1 . As the [EPrI][TFSI] content was increased, the peak shifted to 1735 cm −1 , suggesting the presence of interactions between PHB and [EPrI] [TFSI]. Samples were kept at the molten amorphous state for measurements to avoid the complexity of crystallization. Figure 10b shows the IR C-H stretching vibration band (3225-3050 cm −1 ) of the imidazolium cation ring in [EPrI][TFSI]. The C-H stretching vibration band from 3225 to 3050 cm −1 for the imidazolium cation ring was assigned in the literature [42]. For the neat [EPrI][TFSI], a shoulder at 3168 cm −1 and three peaks at 3149, 3116, and 3093 cm −1 can be shown in Figure 10b. A detail discussion of the IR results of imidazolium cation ring was performed by spectral deconvolution.  Figure 11 shows deconvoluted IR spectra for the C-H stretching vibrational band of imidazolium cation ring with different compositions in the blends. It displays that for the [EPrI][TFSI]/PHB blends the peak at 3093 cm −1 split into two peaks as at the 3087 and 3097 cm −1 . The smaller inserted diagrams are the plots with expanded x-axis (from 3125-3050 cm −1 ) for relevant compositions. A similar situation has also been found in the blends of poly(vinylidene fluoride-cohexafluoropropylene) (PVDF-HFP) and an ionic liquid containing an imidazolium ring cation [42]. According to the literature [42], this feature of IR spectra could result from two forms of the conformation: (i) IL cation complexed with the polymer chain; and (ii) uncomplexed IL cation, and the absorption band at lower wavenumber (3087 cm −1 ) and higher wavenumber (3097 cm −1 ) could be assigned to the complexed and uncomplexed forms of the conformation, respectively.
We further estimated relative ratios between the absorption intensity of 3087 cm −1 (I3087) to 3097 cm −1 (I3097). The values of (I3087/I3097) are presented in Figure 12. Figure 12 demonstrates the plot of (I3087/I3097)-v.s.-PHB content in the blends. The value of (I3087/I3097) increased when the PHB content was increased. In other words, the value of (I3087/I3097) increased when the [ [TFSI] and the carbonyl group of PHB could be proposed. Similar situation has also been reported for the interactions between a cation and a carbonyl-containing polymer in their blends [43]. Kuo et al. [43] have studied the blends comprising lithium salt (LiClO4) and poly(vinylpyrrolidone) (PVP). They have proposed that the complexation between LiClO4 and PVP should be resulted from the ion-dipole interactions between the cation of Li + and the carbonyl group of PVP. In addition, according to the FTIR results in our blending system, it also suggests that the excess loading of [EPrI][TFSI] would not benefit the complexation. The illustration presenting the complexation between [EPrI][TFSI] and PHB is shown in Figure 13.  Figure 11 shows deconvoluted IR spectra for the C-H stretching vibrational band of imidazolium cation ring with different compositions in the blends. It displays that for the [EPrI][TFSI]/PHB blends the peak at 3093 cm −1 split into two peaks as at the 3087 and 3097 cm −1 . The smaller inserted diagrams are the plots with expanded x-axis (from 3125-3050 cm −1 ) for relevant compositions. A similar situation has also been found in the blends of poly(vinylidene fluoride-co-hexafluoropropylene) (PVDF-HFP) and an ionic liquid containing an imidazolium ring cation [42]. According to the literature [42], this feature of IR spectra could result from two forms of the conformation: (i) IL cation complexed with the polymer chain; and (ii) uncomplexed IL cation, and the absorption band at lower wavenumber (3087 cm −1 ) and higher wavenumber (3097 cm −1 ) could be assigned to the complexed and uncomplexed forms of the conformation, respectively.
We further estimated relative ratios between the absorption intensity of 3087 cm −1 (I 3087 ) to 3097 cm −1 (I 3097 ). The values of (I 3087 /I 3097 ) are presented in Figure 12. Figure 12 demonstrates the plot of (I 3087 /I 3097 )-v.s.-PHB content in the blends. The value of (I 3087 /I 3097 ) increased when the PHB content was increased. In other words, the value of (I 3087 /I 3097 ) increased when the [ [TFSI] and the carbonyl group of PHB could be proposed. Similar situation has also been reported for the interactions between a cation and a carbonyl-containing polymer in their blends [43]. Kuo et al. [43] have studied the blends comprising lithium salt (LiClO 4 ) and poly(vinylpyrrolidone) (PVP). They have proposed that the complexation between LiClO 4 and PVP should be resulted from the ion-dipole interactions between the cation of Li + and the carbonyl group of PVP. In addition, according to the FTIR results in our blending system, it also suggests that the excess loading of [EPrI][TFSI] would not benefit the complexation. The illustration presenting the complexation between [EPrI][TFSI] and PHB is shown in Figure 13.

Conclusions
In this study, we discussed blends comprising ionic liquid and biodegradable polymer, and investigated the influence of ionic liquid on the crystallization behaviors of the biodegradable polymer.  [TFSI] in the blends decreased the non-isothermal crystallization rate of PHB. The polymer chains of PHB could be diluted by the [EPrI][TFSI] to enlarge the inter-chain distance of PHB. The corporative diffusion and aggregation of PHB for proceeding crystallization were diminished, and the crystallization of PHB was retarded and declined. The FTIR results suggested the occurrence of the ion-dipole interaction between [EPrI][TFSI] and PHB.