A polymer blend approach to tailor the ferroelectric responses in P(VDF–TrFE) based copolymers

Polymer 54 (2013) 2373e2381

Contents lists available at SciVerse ScienceDirect

Polymer journal homepage: www.elsevier.com/locate/polymer

A polymer blend approach to tailor the ferroelectric responses in P(VDFeTrFE) based copolymers Xiang-Zhong Chen a, b, Xinyu Li a, Xiao-Shi Qian a, Shan Wu a, Sheng-Guo Lu a, Hai-Ming Gu a, Minren Lin a, Qun-Dong Shen b, Q.M. Zhang a, * a

Materials Research Institute and Department of Electrical Engineering, The Pennsylvania State University, University Park, PA 16802, USA Department of Polymer Science & Engineering and Key Laboratory of Mesoscopic Chemistry of MOE, School of Chemistry & Chemical Engineering, Nanjing University, Nanjing 210093, China b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 10 December 2012 Received in revised form 8 February 2013 Accepted 18 February 2013 Available online 27 February 2013

The electroactive properties of PVDF-based ferroelectric polymers can be tailored by blending. In order to investigate the tunability of electrocaloric effect (ECE) and ferroelectric responses, blends of ferroelectric relaxor poly(vinylidene difluorideetrifluoroethyleneechlorofluoroethylene) (P(VDF-TrFE-CFE)) terpolymer and normal ferroelectrics poly(vinylidene difluorideecoetrifluoroethylene) (P(VDFeTrFE)) copolymer are studied. At low copolymer content (<15 wt%), the coupling between the relaxor terpolymer and the nano-phase copolymer converts the copolymer into relaxor and causes an increase in the crystallinity compared with neat terpolymer. As a result, the blends exhibit an enhanced relaxor polarization response and a significant increase in the electrocaloric effect compared with those in the neat terpolymer. At high copolymer content, the blends exhibit mixed structures of the two components. By varying composition, the dielectric and ferroelectric properties of blends can be tuned in the range between the copolymer and terpolymer. This blend system provides a model system to study how random defects influence the polarization response in the normal ferroelectric copolymer, and to understand the relationship between the polarization response and ECE in the blends. The results demonstrate the promise of nanocomposite approaches in tailoring and enhancing ECE and ferroelectric properties in the ferroelectric polymers. Ó 2013 Elsevier Ltd. All rights reserved.

Keywords: P(VDFeTrFE) Ferroelectric Electrocaloric effect

1. Introduction Poly(vinylidene difluoride) (PVDF) based polymers have attracted a great deal of attention since they are multifunctional electroactive polymers with superior mechanical properties and chemical inertness. Their potential to be integrated into organic ferroelectric memory devices [1e3], high permittivity dielectrics [4e9], electrostrictive actuators [10e13], and high energy density capacitors [14e21] has been widely explored in the last few decades. Recently, the PVDF-based copolymers have shown promising applications in solid-state cooling devices due to their large electrocaloric effect (ECE) [22e25]. The ECE is the change in temperature and entropy of dielectric materials caused by the electric field induced change in dipolar states [26e28]. Hence, high polarizations induced by and large changes of polarization with electric field are necessary to realize large ECE. These considerations indicate that it

* Corresponding author. Tel.: þ1 814 863 8994. E-mail address: [email protected] (Q.M. Zhang). 0032-3861/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.polymer.2013.02.041

is advantageous to utilize ferroelectric materials such as P(VDFe TrFE) to achieve large ECE. Among various crystalline phases, ferroelectric b phase is most likely to form in P(VDFeTrFE) due to the steric hindrance of bulky TrFE group [29,30]. b phase is a highly ordered state, in which polymer chains adopt all-trans (TTTT) conformation, where all the fluorine atoms are on one side of the chain, forming dipoles perpendicular to the chain direction. When temperature goes up, the ferroelectric b phase will undergo a phase transition and turn into paraelectric phase, which consists of a random sequence of trans-gauche (TG) bonds, such as TGTG0 , and T3GT3G0 isomers [30]. Correspondingly, the dipoles in the crystallites also change from an ordered state to a disordered state. When temperature is just above the ferroelectriceparaelectric (FEePE) transition, the dipoles still can be aligned by external electric field. In other words, the dipole ordered state can be switched on by electric field just above the FEe PE transition. Upon removal of the electric field, the ordered dipoles will again become disordered along with absorbing entropy from the cool load, which causes a decrease in temperature [22,31]. It is reported that in P(VDFeTrFE) 55/45, a temperature change of

2374

X.-Z. Chen et al. / Polymer 54 (2013) 2373e2381

w12  C around its phase transition temperature 67  C has been achieved at 120 MV/m [23]. By tuning the copolymer compositions, this phase transition temperature can be adjusted between 65  C and 130  C [29,30]. However, further lowering the phase transition temperature to below 65  C by adjusting the P(VDFeTrFE) copolymer composition is not possible. Meanwhile, the transition in normal ferroelectrics is sharp. Thus large ECE can be achieved only in a small temperature range around phase transition temperature. Introducing defects in P(VDFeTrFE) copolymer, either by electron irradiation or copolymerizing with a third monomer with bulky group like chlorofluoroethylene (CFE), provides promising solutions to this problem [10,12,32e35]. When the defect content reaches 7%w9%, the normal ferroelectric polymers will be converted to ferroelectric relaxor, in which randomly distributed nano-polar regions are embedded in a non-polar matrix. When an external electric field is applied, the random dipoles in the relaxor ferroelectrics will reorient along the electric field direction, causing a structure change in the polymer along with a temperature change as compensation for the entropy change. Compared with the normal ferroelectrics, the relaxor exhibits a broad diffused dielectric peak at ca. 25  C, and also can provide large ECE over a broad temperature range around room temperature. A most evident advantage of polymeric materials is their tunability. Apart from the chemical processes such as copolymerization or irradiation, some physical methods, e.g., changing processing conditions, are much easier to tune the properties of the polymers. For example, stretching P(VDFeTrFEeCFE) terpolymer can lead to different electrocaloric responses [36]; casting terpolymer films at different temperatures causes variations in structures and dielectric responses [37]; even different annealing temperatures result in different electromechanical properties [38]. Besides, adding nano-particles or other polymers into the matrix polymer to make composites is also an easy way to tailor the properties or endow the composites new properties [16e18]. In this paper, we investigate structure, dielectric properties, electrocaloric effect and ferroelectric properties of P(VDFeTrFEe CFE) terpolymer/P(VDFeTrFE) copolymer blends. Experimental results show that terpolymer and copolymer do not co-crystallize, as evident by the separate melting peaks for the two polymers, but the presence of the copolymer increases the crystallinity of the blends compared with the neat terpolymer. At low copolymer content, the random defects in the terpolymer influence the ferroelectric response in the copolymer, presumably through interfaces between the two polymers, and consequently, the whole blends exhibit relaxor ferroelectric response. These combined effects of increased crystallinity and interfacial couplings lead to an enhanced polarization and a larger (>30%) increase in ECE in the blends compared with that in the neat terpolymer. At high copolymer content, the two phases become more distinguished, and each component tends to keep their own structures, leading to intermediate electric properties between pure terpolymer and copolymer. The blend system provides a model system to study how the random defects in the terpolymer influence the polarization response in the copolymer, and also provide a direct figure to understand the relationship between the polarization response and ECE in the blends. The results demonstrate the promise of composite approaches in tailoring and enhancing ECE and ferroelectric properties in PVDF-based ferroelectric polymers. 2. Experimental P(VDFeTrFEeCFE) terpolymer (62.5/29/8.5 mol%) was synthesized using a suspension polymerization process by Piezotech (France) [39]. P(VDFeTrFE) 55/45 mol% were supplied by Solvay.

P(VDFeTrFEeCFE) terpolymer and P(VDFeTrFE) copolymer powders were dissolved in N,N-dimethylformamide at room temperature, respectively. Then the two solutions were mixed by proper ratios for different blend compositions and then cast on cleaned glass plates and dried at 70  C for 24 h. Afterwards, the films were peeled off from the glass plates and further annealed at 110  C for 24 h. The thickness of final films is 8e10 mm. FTIR spectra were obtained using a Bruker V70 infrared spectrometer equipped with a diamond ATR accessory. To calculate the variation of infrared absorption intensities for polymer chains of all trans (Tm>3, absorbance peak at 1285 cm1), T3GT3G0 (504 cm1), and TGTG0 (613 cm1) conformation with respect to copolymer content, first, the fraction Fi of each chain conformation in every sample was calculated:

Fi ¼

Ai ; AI þ AII þ AIII

where i ¼ I, II, III, and AI, AII, AIII are the absorption area of the chain conformations with all trans (Tm>3), T3GT3G0 , and TGTG0 , respectively. After this procedure, the total fraction of all the chain conformations in each sample is normalized to 1, thus the difference brought by the experiments can be eliminated. The fractions of a certain chain conformation from different samples are then compared with that of terpolymer to get the final data. Melting points and melting heat of the pure terpolymer and blends were measured by differential scanning calorimetry (DSC) (TA Q100). The heating rate is 10  C/min. X-ray diffraction (XRD) was performed using a PANalytical X’Pert Pro MPD diffractometer equipped a copper target with an average wavelength of 1.542  A. Dynamic mechanical analysis (DMA) was performed using a TA RSA-G2 solids analyzer at 1 Hz with heating rate of 3  C/min under tension mode. Gold electrodes were sputtered on both surfaces of the polymer films for electric characterization. The dielectric properties as a function of temperature were characterized using a precision LCR meter (HP 4284A) equipped with a temperature chamber (Delta 9023). The data were recorded every 3  C. The dielectric properties as a function of frequency were recorded by an HP 4294A Precision Impedance Analyzer from 40 Hz to 10 MHz. Polarizationeelectric field (PeE) loops were measured using a modified SawyereTower circuit at a frequency of 10 Hz at room temperature. The adiabatic temperature change DT was deduced from DT ¼ Qad/CE, where CE is specific heat which can be measured by modulated DSC (TA Q100), and Qad is the heat ejected or absorbed by the film upon the application/removal of the applied field during the adiabatic process. Since the heat changes in the adiabatic process Qad of the samples will exchange with the ambient through the isothermal process and finally achieve balance, the heat measured in the isothermal process Qiso is actually equal to heat changes in the adiabatic process Qad. Therefore, the adiabatic temperature change DT can be deduced from DT ¼ Qad/CE ¼ Qiso/CE. Qiso is measured using a high sensitivity heat flux sensor in a temperature controlled chamber [36,40]. 3. Results and discussion 3.1. Structure study of terpolymer and copolymer blends FTIR is employed to detect local conformational structure changes of PVDF-based copolymers. The normalized FTIR absorption spectra of the terpolymer, copolymer and their blends are shown in Fig. 1(a). The terpolymer shows mixed chain conformations consisting of trans and gauche bonds, i.e. TGTG0 (613 cm1), T3GT3G0 (504 cm1) and Tm>3 (1285 cm1) conformations [12,41].

X.-Z. Chen et al. / Polymer 54 (2013) 2373e2381

2375

Fig. 2. X-ray diffraction patterns of the terpolymer, copolymer, and their blends with different composition.

Fig. 1. (a) FTIR spectra for terpolymer, copolymer and their blends. (b) Variation of infrared absorption intensities for polymer chains of TTTT (1285 cm1), TGTG0 (613 cm1) and T3GT3G0 (504 cm1) conformation with respect to copolymer content. The lines are drawn to guide eyes.

With addition of copolymer, crystal bands at 1285 cm1 and 848 cm1 corresponding to all trans (Tm>3) conformation gradually grow up, while those bands representing TGTG0 conformation slowly diminish. The ratio of representative absorption bands’ intensities of blends to those of terpolymer are presented in Fig. 1(b) to show the changes in portions of chain conformation with respect to the copolymer content. It can be seen that at low copolymer content (<20%), the increase of Tm>3 conformation and decrease of TGTG0 conformation is not obvious, which means the copolymer embedded in the blends adopts a mixed structure consisting of three conformations just as terpolymer does. As copolymer content becomes higher than 20%, a quick increase/decrease is observed, which indicates the influence of copolymer becomes larger. The T3GT3G0 conformation only increases a little in the whole copolymer content range. To further understand the structure from a crystallographic view, XRD is performed and the data are shown in Fig. 2. Copolymer exhibits two partially superimposed peaks located at 18.8 and A and 4.60 A, in 19.3 , reflecting interchain lattice spacing (d) of 4.71 accord with (110) and (200) reflection respectively [30]. The terpolymer exhibits only one peak at 18.4 with the interchain lattice space of 4.86 A (calculated from Bragg’s equation 2dsinq ¼ nl). Blends with low copolymer content (<15 wt%) also

exhibit only one narrow peak at the same position as terpolymer does. The nearly identical interchain crystal spaces of these samples indicate that the copolymer chains do not interpenetrate with the terpolymer chains and form co-crystals to change the basic crystallographic structure, and no large coherent polar crystals of P(VDFeTrFE) exists in the blends. The crystallite sizes or coherence lengths (L) perpendicular to (110,200) crystallographic plane of terpolymer, corresponding to the sizes of polar or nonpolar domains, are estimated using the Scherrer equation: L ¼ 0.9l/B cos q, where l is X-ray wavelength, B is full width at half-maximum (FWHM, in 2q), and q is angular position of the diffraction peak, respectively. The coherence lengths are listed in Table 1. Decreased coherence lengths in terpolymer crystalline phase are observed with the increasing amount of copolymer, indicating the addition of copolymer can reduce the crystallite size. Then the coherence length reaches a limit when copolymer content is larger than 20%, and further addition will not increase the FWHM, suggesting there’s certain saturation in the interaction between the

Table 1 Lattice constant and coherence length for the (110,200) reflection of the terpolymer, copolymer, and their blends. Blends Ter/Co

100/0 95/5 90/10 85/15 80/20 70/30 60/40 50/50 Copolymer

(110,200)ter

(110)co

d,  A

L, nm

d,  A

L, nm

d,  A

(200)co L, nm

4.86 4.87 4.86 4.87 4.87 4.87 4.87 4.87

40.5 34.6 31.8 24.5 31.2 24.5 25.0 26.4

4.78 4.72 4.72 4.71

11.2 11.2 11.9 12.0

4.62 4.61 4.61 4.60

10.2 13.9 13.7 14.0

2376

X.-Z. Chen et al. / Polymer 54 (2013) 2373e2381

two components. It seems that at low copolymer content, the copolymer interact with the terpolymer crystalline phases most probably through interfacial couplings because no large normal ferroelectric copolymer crystals are observed and the copolymer chains are forced to twist to adopt a mixed structure with all three conformations just as the terpolymer relaxor. Subsequently, only one characteristic reflection peak in XRD can be observed. As the copolymer content increases, a broad peak gradually grows at around 19 , evolutes into two peaks which also move slightly toward high angular position, and finally become two peaks at 18.7 and 19.2 , corresponding to the reflection of (110) and (200) crystal planes of ferroelectric phases in P(VDFeTrFE) copolymers respectively. The peak fitting results show gradually decreased d spaces and increased coherent lengths, indicating more and more perfect ferroelectric crystallites grow up with the increase of copolymer content. The co-existence of reflection peaks of the two components suggests the copolymer/terpolymer do not co-crystallize to a large extent, and both of them tend to maintain their own crystal structures at high load of copolymer. Fig. 3 shows the DSC traces of blends with different compositions acquired during the first heating run. The relaxor terpolymer has no FEePE transition and exhibits a melting peak at 128.3  C. The ferroelectric copolymer exhibits two peaks, one melting peak at 156.6  C and the other at 64  C representing FEePE transition. Two melting peaks in all blends samples suggest that the two components do not co-crystallize. Nevertheless, melting point depressions and broadening of the melting peaks in the terpolymer components indicate that the copolymer indeed influences the crystalline phases of the terpolymer. Thinner lamellae or smaller crystallites with low melting points are formed in the blends, which corroborates the decreased coherence lengths derived from XRD results. The melting peak of copolymer shifts gradually from 150.5  C to 154.4  C as the amount of the copolymer in the blends increases to

Fig. 3. DSC traces of the terpolymer, copolymer, and their blends with different composition.

50 wt%, indicating the growth of copolymer crystallites when they become richer in the blends. The total melting heat and the normalized melting heat of each component (the actual heat of melting divided by the weight ratio) are summarized in Table 2. The total melting heat increases after addition of copolymer compared with that of pure terpolymer, and so does the normalized heat of melting (DHm(ter)). One possible explanation is that the ability to crystallize for copolymer is higher than that for terpolymer. The copolymer will first crystallize and then serve as nuclear center so as to favor the crystallizing process and increase the crystallinity of terpolymer, and thus the normalized melting heat of terpolymer is increased. Another possible reason is that some of thin copolymer lamellae are confined between terpolymer chains and the interfacial couplings between terpolymer and copolymer crystallites make them only can grow into very thin lamellae, and therefore, they melt at low temperature as the tiny terpolymer crystallites do. From the data obtained by peak fitting, we believe it should be mostly the latter case in two reasons: first, the normalized DHm(ter) increases while the normalized DHm(co) decreases, and second, there is no large copolymer crystals with characteristic reflection peak of ferroelectric phase found in the blends with low copolymer content from the XRD profiles. Another evidence of this supposed case is that the ferroelectric-paraelectric phase transition of P(VDFeTrFE) 55/45, which is supposed at around 65  C, is not detected at low content copolymer blends. Only when the copolymer weights over 20%, the FEePE transition does gradually increase, because the interface effect saturates and crystallites of copolymer grow up. The interfacial couplings can also be corroborated by the DMA data. Fig. 4 shows the mechanical loss tangent of terpolymer, copolymer and their blends. For terpolymer, two relaxation peaks are observed. The shoulder peak at low temperature (w20  C) is supposed to be related to the segmental motions in the amorphous phase, i.e., the glass transition, while the large peak at higher temperature (w10  C) is believed to relate with the molecular relaxation at amorphousecrystalline interface boundaries [42]. With increase of copolymer, the mechanical loss representing interfacial molecular motions becomes smaller and smaller, and meanwhile, the peak maximum shifts toward lower temperature, which means the addition of copolymer can produce more disordered interfacial regions with lowered activation energy. No further shift of peak maximum is observed when the copolymer weighs over 20%, indicating the interfacial coupling effect saturates, which is consistent with the XRD and DSC data. Besides, the glass transition peak (w20  C) shifts toward high temperature, and a new peak above 40  C representing the FEePE transition gradually comes up, which means increased crystallinity and large coherent ferroelectric crystals form in the blends with high copolymer content. From the structural data, e.g. the absence of reflection peak of ferroelectric phase in XRD and phase transition peak in DSC, it is reasonable to infer that in the blends with low copolymer content, no normal ferroelectric copolymer phase exists. Furthermore, the copolymer is converted to relaxor with a mixed structure composed of TGTG0 , T3GT3G0 and Tm>3 conformations through interfacial couplings. Besides, the incorporation of copolymer increases the blends’ crystallinity, resulting in an enhanced polarization and thus the electrocaloric response, as will be discussed in later sections. In the blends with high copolymer content, the normal ferroelectric copolymer phase arises and becomes more and more distinguished with addition of copolymer, as can be seen in XRD. However, the defects in terpolymer still affect the copolymer, mainly in the perfectness of ferroelectric domains in crystalline regions, which can be employed to tailor the ferroelectric properties of the system.

X.-Z. Chen et al. / Polymer 54 (2013) 2373e2381

2377

Table 2 Ferroelectriceparaelectric phase transition temperature and heat, melting temperature and heat of the terpolymer, copolymer, and their blends. Blends Ter/Co 100/0 95/5 90/10 85/15 80/20 70/30 60/40 50/50 0/100

Tc,  C

62.9 62.4 62.7 64

DHc, J/g

1.2 2.2 3.8 5.8

Tm1,  C 128.3 126.8 126.7 126.6 125.4 125.6 126.1 125.3

Tm2,  C

150.5 150.7 153.3 153.8 153.8 154.1 154.4 156.5

3.2. Dielectric properties in weak field Fig. 5(a) shows weak field dielectric constant of the blends with various compositions. Dielectric constants of all samples decrease with increasing frequency due to dielectric relaxation [43]. The dielectric constant for the terpolymer at 1 kHz is about 48. It is believed at lower frequency, high dielectric constant of the terpolymer is caused by the response of dipoles in the randomly distributed polar nano-region [43,44]. For the blends, the dielectric constant first increases by w10% when the copolymer content is 5 wt%, and then gradually decreases but still is higher than that of pure terpolymer as the copolymer content increases to 15 wt%. The enhancement of dielectric constant can be observed through all frequencies where the data are measured. It is interesting that an increase in dielectric constant is obtained by adding a low dielectric constant component (w17 at 1 kHz at room temperature for the copolymer). As has been discussed earlier, the copolymer is converted to relaxor through interfacial couplings and thus enhances the overall dielectric response of the blends. When the copolymer weighs over 20%, the dielectric constant of the blends becomes lower than that of pure terpolymer. However, it is noted that for the blend with 20wt% copolymer, its dielectric constant (w45) is 10% higher than the weighted average of the two components (w41), indicating the interface effect still play its role in the blends. With further increase of the copolymer content, the dielectric constants decrease monotonously in the whole frequency range. Fig. 5(b) shows the coleecole plot of dielectric properties (ε00 vs ε0 ) of the blends. The data is analyzed using the ColeeCole equation:

Fig. 4. Mechanical loss tangent for terpolymer, copolymer, and terpolymer/copolymer blends.

DHm, J/g

DHm(ter), J/g

22.8 24.4 25.2 23.7 25.2 25.2 25.4 25.8

22.8 22.8 22.6 20.1 20.4 18.2 15.8 12.6


ε* u ¼ εi þ

DHm(co), J/g

1.5 2.5 3.5 4.8 7.0 9.6 13.2 30.6

Dε 1 þ ðiusÞa

Normalized

DHm(ter), J/g 22.8 24.0 25.1 23.6 25.6 26.0 26.3 25.2

Normalized DHm(co), J/g 30.0 25.0 23.3 23.8 23.3 24.0 26.4 30.6

(1)

where Dε ¼ εsεi is the dielectric relaxation strength, with εs the static dielectric constant and εi the dielectric constant at “infinite” frequency; u is the frequency, s is characteristic relaxation time, and a is the parameter describing the distribution of the relaxation time. a ¼ 1 corresponds to the Debye model of monodispersive relaxation. While in most cases, a varies between 0 and 1, indicating a broad distribution of the relaxation times in the system. The solid curves are the fitting results and the parameters are listed in Table 3. Both the static dielectric constant and the dielectric relaxation strength Dε first go up and then decline with the copolymer content in the blends. The increased Dε in the blends

Fig. 5. (a) Dielectric constant of the terpolymer, copolymer, and terpolymer/copolymer blends at room temperature. (b) ε00 vs ε0 at room temperature. Solid lines are curve fitting obtained by the ColeeCole expression.

2378

X.-Z. Chen et al. / Polymer 54 (2013) 2373e2381

Table 3 Summary of ColeeCole parameters. Blends Ter/Co

εs

εi

Dε

a

100/0 95/5 90/10 85/15 80/20 70/30 60/40 50/50 0/100

48.0 55.5 53.4 49.1 46.2 42.0 33.5 26.8 17.3

4.7 4.9 4.8 4.2 3.8 3.8 3.6 3.4 3.2

43.3 50.6 48.7 44.9 42.5 38.2 29.9 23.4 14.1

0.575 0.575 0.568 0.558 0.580 0.558 0.542 0.533 0.456

with low copolymer content (<15wt %) indicates that the relaxor behavior is indeed enhanced. Besides, all the samples have nearly same εi z 4, suggesting the polar processes that govern the dielectric response tends to diminish at high frequency, as εi only accounts for high frequency nonpolar processes. The enhancement observed in the blends with low copolymer content is believed to be related with the interfacial couplings between the terpolymer and copolymer [25]. Just as shown in XRD and DSC data, the defects in the terpolymer can hinder the formation of large ferroelectric domains in P(VDFeTrFE) copolymer, and convert the copolymer into relaxor, leading to higher dielectric constant. On the other hand, the copolymer will become normal ferroelectric at high copolymer content, which has lower dielectric constant. Consequently, the total dielectric response becomes smaller. The observed decrease of a with increased copolymer content also indicates an increased normal ferroelectric response in the blends [45]. Fig. 6(a) presents the temperature dependent dielectric properties of all the samples measured at 1 kHz. For the terpolymer there is one broad peak observed near room temperature, and for the copolymer, a dielectric anomaly is observed at the ferroelectriceparaelectric (FEePE) phase transition temperature (w65  C). For the blends with the copolymer content <15 wt%, no FE-PE transition is observed as shown in Fig. 6(b) and (c). The broad dielectric constant peak of the neat terpolymer shifts progressively towards higher temperature with frequency, typical relaxor characteristics. It should also be noted that there is an additional peak, near 20  C, observed in the low frequency dielectric loss data, whose peak position moves progressively towards higher temperature with frequency, which is related to glass-transition in the amorphous region of terpolymer. For the blends with 10 wt% copolymer, besides the enhanced dielectric constant at room temperature, there is a shoulder appearing at a temperature higher than the peak of neat terpolymer, as indicated by the arrows. This new peak is very clear for the dielectric constant at low frequencies, and gradually merges with the terpolymer relaxation peak at high frequencies. One possible explanation for this new higher temperature peak is that a new phase may form in the blends with low copolymer content (<15 wt%) and that the new phase is neither as disorder as the relaxor phase nor as order as the normal ferroelectric phase. Thus, the most probable assignment of the new peak is the dielectric response of disordered interface regions. The interfacial couplings with the terpolymer perturb or disrupt the correlation of the ferroelectric domains, and convert the nanocrystallites of the copolymer from a normal ferroelectric to a relaxor-like ferroelectric. This interpretation is consistent with the results of XRD and DSC data of the blends with the copolymer content <15 wt%, as has been discussed earlier. Fig. 6(a) also reveals that the blends with the copolymer content > 30 wt% exhibit characteristics of both components: one relaxor peak locates around 20  C whose peak position is independent of terpolymer/copolymer ratio, and an FEePE transition

Fig. 6. (a) Temperature dependent dielectric properties of blends measured at 1 kHz. Dielectric properties as a function of temperature and frequency in neat terpolymer (b) and blends with 10 wt% copolymer (c). Curves from left to right: 102e106 Hz at intervals of 101 Hz.

peak around 60  C. The latter moves slightly towards higher temperature with the P(VDFeTrFE) content, indicating that the influence of defects introduced by the terpolymer to the normal ferroelectric copolymer becomes weaker, as can be corroborated by growing ferroelectric phase peak in XRD and phase transition peak in DSC. In addition, there is another peak observed around 40  C between the relaxor peak and the FEePE transition peak, which can be interpreted as a less ordered ferroelectric phase. Such a peak may be caused by relatively more ordered phases in terpolymer which can be induced under certain circumstances (for here, it may be induced through interactions with P(VDFeTrFE)), or less ordered ferroelectric states of the copolymer. It is reported that there is a “CL” phase exists in P(VDFeTrFE) 55/45 copolymer, which embraces more “tilt” defects and thus less ordered ferroelectric phases [30]. The interaction with the terpolymer may enhance this phase so that it can be observed in the dielectric data. 3.3. Polarization response and electrocaloric effect Fig. 7(a) summarizes the maximum unipolar polarization as functions of the applied field amplitude of each blend composition measured at 10 Hz. The blend films with the copolymer content

X.-Z. Chen et al. / Polymer 54 (2013) 2373e2381

2379

polarization becomes smaller, the polarization at fields <150 MV/m is still higher than that of neat terpolymer. Changing the measuring field frequency does not change the trend, indicating the enhancement of polarization is intrinsic and is not caused by conduction loss (data is not shown here). With further addition of the copolymer, the maximum polarization drops monotonously at all electric field. Fig. 7(b) shows the variation of maximum polarization measured at 200 MV/m regarding to the composition of the blends, and the insets show the typical PeE loops of these samples. It can be seen that a small amount of the copolymer can enhance the relaxor polarization behavior, while further increase the copolymer content will reduce it. As has been discussed earlier, small amount of the copolymer added into the blends can be transformed into relaxor through interfacial couplings with the terpolymer. The crystallinity can also be increased as shown by DSC. These factors contribute to the enhancement of relaxor polarization response. With increased copolymer content (15 wt%w20 wt%), the interface effect becomes less effective, as evidenced by an increase in the broad dielectric constant peak towards higher temperature. Further addition of the copolymer will form the normal ferroelectric crystalline phase, as observed by the DSC and XRD data presented earlier. An outcome of the tunable polarizations is the change in electrocaloric effect (ECE) in these polymer blends. The ECE is a result of direct coupling between the thermal properties such as entropy and electric properties such as electric field induced polarization in a dielectric material. Fig. 7(c) presents the adiabatic temperature change (DT) of the blends as a function of electric field measured at room temperature. Analogous to the polarization response in Fig. 7(a), DT exhibits a composition-dependent behavior. For example, a DT of 5.1 K is observed for neat terpolymer under 100 MV/m. The blends with 5wt%, 10wt%, and 15wt% copolymer have DT of 6.5 K, 7 K, and 5.7 K, corresponding to w25%, w35%, w10% enhancement respectively. While for blends with 20e50 wt% copolymer, the DT is 4.4 K, 3.9 K, 3 K, and 2 K, respectively. To make quantitative comparison, we use the relation [22,46]

DS ¼ bD2 =2;

(2)

where D is the electric displacement and b is a coefficient. Applying cEDT ¼ TDS yields the adiabatic temperature change

DT ¼ bD2 =ð2TcE Þ;

Fig. 7. (a) Maximum polarizations of the terpolymer and its blends measured at 10 Hz from 50 MV/m to 300 MV/m. (b) Variation of maximum polarizations with respect to the composition of the blends measured at 10 Hz and 200 MV/m. The inset shows typical unipolar PeE loops of blends measured at 200 MV/m. (c) Shows adiabatic temperature change as a function of electric field at room temperature.

<40 wt% exhibit a fast increase of polarization with field amplitude below 100 MV/m and then there is a gradual saturation in polarization at higher electric field. The blends with 5 wt% and 10 wt% copolymer exhibit higher polarization level than that of neat terpolymer at all electric field measured. For example, the polarization of the blends with 10 wt% of the copolymer is 0.095 C/m2 at 300 MV/m, higher than 0.088 C/m2 of neat terpolymer. Although for the blends with 15 wt% of the copolymer, the enhancement in

(3)

where cE is the specific heat capacity. Using the data in Fig. 7(c) for DT at room temperature (25  C) and Fig. 7(a) for D, b in Eq. (4) can be deduced for the blends. Within the experimental error, b does not show marked change for blends with the copolymer content <15 wt% (inset of Fig. 8), indicating that the enhancement in ECE in blends is a direct consequence of the enhanced polarization response. However, a reduction in b is found when the copolymer content increases further, indicating that the structure changes and formation of ferroelectric phase may be the reason of lowered electrocaloric effect. It is noteworthy that another merit of blends is their enhanced mechanical properties. At room temperature, the modulus of pure terpolymer is only 103 MPa; adding just 10 wt% copolymer will increase the elastic modulus of the blends to 203 MPa, twice of that of neat terpolymer. The increased mechanical strength will greatly facilitate post-processing of the polymer films, for example, fabricating multi-layered ECE device by lamination. The ferroelectric responses of the blends with the copolymer content >20 wt% are further investigated. Fig. 9(a) shows the bipolar polarization loops of the blends measured under an electric field of 200 MV/m at room temperature. The data show that both

2380

X.-Z. Chen et al. / Polymer 54 (2013) 2373e2381

the composition in between, both the coercive field and remnant polarization can be tuned between these two extremes, which can be attractive for device applications. 4. Conclusions

Fig. 8. DS vs P2 for the terpolymer and its blends with the copolymer. The fitted straight lines are drawn to guide eyes. The inset shows changes in b with respect to the copolymer content.

the polarization and coercive field can be tuned by adjusting the blend composition the terpolymer/copolymer blends. Since the maximum polarization of the terpolymer and copolymer differs little from each other, there is only minor difference between maximum polarizations in the polarization hysteresis loops of Fig. 9(a) among different blends, which is different from the polarization data acquired under the unipolar electric fields (Fig. 7). On the other hand, significant variations are found in the coercive field and remnant polarization, as summarized in Fig. 9(b), for the polarization hysteresis loops in Fig. 9(a). The copolymer film exhibits a near square polarization hysteresis loop with a large remnant polarization (Pr w 6 mC/cm2), while a very slim hysteresis loop with very low remnant polarization and coercive field, which is typical for ferroelectric relaxor, is observed for the terpolymer. In

Here we show that the ferroelectric responses of the blends can be tailored and tuned effectively through the interfacial coupling between the relaxor terpolymer and normal ferroelectric copolymer. At the copolymer content <15 wt%, the random defects in the terpolymer influence the ferroelectric response in the copolymer through interfaces between the two polymers and convert the nano-phase copolymer into relaxor. Consequently, the blends exhibit an enhanced relaxor polarization response and a marked increase (>30%) in ECE compared with the neat terpolymer. At high copolymer content, the two crystalline phases become more distinguished, and each component tends to keep its own structures, leading to intermediate ferroelectric properties. This blend provides a model system to study how random defects influence the polarization response in the copolymer, and also provide a means to understand the relationship between the polarization response and ECE in the blends. The results demonstrate the promise of nano-composite approaches in tailoring and enhancing ECE and ferroelectric properties in PVDF-based ferroelectric polymers. Acknowledgments We thank Zoubeida Ounaies and Payam Khodaparast for their assistance in the experiment. This research was supported by Army Research Office under Grant No. W911NF-11-1-0534 (X. Chen, S.G. Lu, M. Lin, and Q. Zhang) and by the U.S. DoE, Office of Basic Energy Sciences, Division of Materials Science and Engineering under Award No. DE-FG02-07ER46410 (X. Li and X. Qian). X. Chen was also partially supported by Nanjing University, China and H. Gu was in part supported by EE Department of PSU. References

Fig. 9. (a) Bipolar polarization hysteresis loops and (b) The variation of maximum polarization (Pmax), remnant polarization (Pr), coercive field (Ec) of the terpolymer, copolymer, and their blends measured at a maximum electric field of 200 MV/m.

[1] Naber RCG, Tanase C, Blom PWM, Gelinck GH, Marsman AW, Touwslager FJ, et al. Nat Mater 2005;4(3):243e8. [2] Asadi K, De Leeuw DM, De Boer B, Blom PWM. Nat Mater 2008;7(7):547e50. [3] Hu ZJ, Tian MW, Nysten B, Jonas AM. Nat Mater 2009;8(1):62e7. [4] Zhang QM, Li HF, Poh M, Xia F, Cheng ZY, Xu HS, et al. Nature 2002;419(6904): 284e7. [5] Wang JW, Shen QD, Bao HM, Yang CZ, Zhang QM. Macromolecules 2005;38(6):2247e52. [6] Wang CC, Song JF, Bao HM, Shen QD, Yang CZ. Adv Funct Mater 2008;18(8): 1299e306. [7] Huang C, Zhang QM, Su J. Appl Phys Lett 2003;82(20):3502e4. [8] Wang JW, Shen QD, Yang CZ, Zhang QM. Macromolecules 2004;37(6):2294e8. [9] Bobnar V, Levstik A, Huang C, Zhang QM. Phys Rev Lett 2004;92(4):047604. [10] Zhang QM, Bharti V, Zhao X. Science 1998;280(5372):2101e4. [11] Huang C, Zhang QM. Adv Funct Mater 2004;14(5):501e6. [12] Xia F, Cheng ZY, Xu HS, Li HF, Zhang QM, Kavarnos GJ, et al. Adv Mater 2002;14(21):1574e7. [13] Huang C, Klein R, Xia F, Li HF, Zhang QM, Bauer F, et al. IEEE Trans Dielectr Electr Insul 2004;11(2):299e311. [14] Chu BJ, Zhou X, Ren KL, Neese B, Lin MR, Wang Q, et al. Science 2006;313(5785):334e6. [15] Chu BJ, Zhou X, Neese B, Zhang QM, Bauer F. IEEE Trans Dielectr Electr Insul 2006;13(5):1162e9. [16] Li JJ, Seok SI, Chu BJ, Dogan F, Zhang QM, Wang Q. Adv Mater 2009;21(2):217e21. [17] Chu BJ, Lin MR, Neese B, Zhou X, Chen Q, Zhang QM. Appl Phys Lett 2007;91(12):122909. [18] Chu BJ, Neese B, Lin MR, Lu SG, Zhang QM. Appl Phys Lett 2008;93(15): 152903. [19] Guan FX, Pan JL, Wang J, Wang Q, Zhu L. Macromolecules 2010;43(1):384e92. [20] Chen XZ, Li ZW, Cheng ZX, Zhang JZ, Shen QD, Ge HX, et al. Macromol Rapid Commun 2011;32(1):94e9. [21] Wu S, Lin MR, Lu SG, Zhu L, Zhang QM. Appl Phys Lett 2011;99(13):132901. [22] Neese B, Chu BJ, Lu SG, Wang Y, Furman E, Zhang QM. Science 2008;321(5890):821e3.

X.-Z. Chen et al. / Polymer 54 (2013) 2373e2381 [23] Lu SG, Rozic B, Zhang QM, Kutnjak Z, Neese B. Appl Phys Lett 2011;98(12): 122906. [24] Li XY, Qian XS, Gu HM, Chen XZ, Lu SG, Lin MR, et al. Appl Phys Lett 2012;101(13):132903. [25] Chen XZ, Qian XS, Li XY, Lu SG, Gu HM, Lin MR, et al. Appl Phys Lett 2012;100(22):222902. [26] Fatuzzo E, Merz WJ. Ferroelectricity. Amsterdam: North-Holland Publishing Company; 1967. [27] Mitsui T, Tatsuzaki I, Nakamura E. An introduction to the physics of ferroelectrics. London: Gordon and Breach; 1976. [28] Lines ME, Glass AM. Principles and applications of ferroelectrics and related materials. Oxford: Clarendon Press; 1977. [29] Lovinger AJ. Science 1983;220(4602):1115e21. [30] Nalwa HS. Ferroelectric polymers: chemistry, physics, and applications. New York: Marcel Dekker; 1995. [31] Lu SG, Zhang QM. Adv Mater 2009;21(19):1983e7. [32] Xu HS, Shanthi G, Bharti V, Zhang QM, Ramotowski T. Macromolecules 2000;33(11):4125e31. [33] Bharti V, Zhang QM. Phys Rev B 2001;6318(18):184103. [34] Cheng ZY, Olson D, Xu HS, Xia F, Hundal JS, Zhang QM, et al. Macromolecules 2002;35(3):664e72.

2381

[35] Bobnar V, Vodopivec B, Levstik A, Kosec M, Hilczer B, Zhang QM. Macromolecules 2003;36(12):4436e42. [36] Li XY, Qian XS, Lu SG, Cheng JP, Fang Z, Zhang QM. Appl Phys Lett 2011;99(5): 052907. [37] Bao HM, Song JF, Zhang J, Shen QD, Yang CZ, Zhang QM. Macromolecules 2007;40(7):2371e9. [38] Klein RJ, Runt J, Zhang QM. Macromolecules 2003;36(19):7220e6. [39] Bauer F, Fousson E, Zhang QM, Lee LM. IEEE Trans Dielectr Electr Insul 2004;11(2):293e8. [40] Lu SG, Rozic B, Zhang QM, Kutnjak Z, Li XY, Furman E, et al. Appl Phys Lett 2010;97(16):162904. [41] Tashiro K, Kobayashi M, Tadokoro H. Macromolecules 1981;14(6):1757e64. [42] Bharti V, Xu HS, Shanthi G, Zhang QM, Liang KM. J Appl Phys 2000;87(1): 452e61. [43] Zhang SH, Klein RJ, Ren KL, Chu BJ, Zhang X, Runt J, et al. J Mater Sci 2006;41(1):271e80. [44] Hahn B, Wendorff J, Yoon DY. Macromolecules 1985;18(4):718e21. [45] Bharti V, Cheng ZY, Zhang QM. In: Streiffer SK, Gibbons BJ, Tsurumi T, editors. Proceedings of the 2001 12th IEEE international symposium on applications of ferroelectrics, vols. I and II. New York: IEEE; 2001. p. 801e6. [46] Tuttle BA, Payne DA. Ferroelectrics 1981;37(1e4):603e6.