Introduction

Poly(vinylidene fluoride), PVDF, is a semi-crystalline polymer that can crystallize in four crystalline phases known as α, β, δ, and γ [1, 2]. The α and β phases are the most interesting ones for technological applications and, therefore, the most studied. The α-phase is non-polar, it has a trans and gauche-bond (TGTG′) chain conformation and it is typically obtained by cooling from the melt. The β-phase is piezo-, pyro-, and ferroelectric. It has an all-trans conformation (TTTT) resulting in a net dipole moment within the unit cell [2, 3].

Due to its electroactive properties β-PVDF finds increasing applications for sensors and actuators, batteries, and in the biomedical field [4, 5].

As the piezoelectric β-phase is the most interesting from the technological point of view and its presence in the polymer depends on the processing conditions [1, 6], different approaches have been used to obtain this phase.

Most often, the polar β-phase is obtained by uniaxial or biaxial stretching of α-phase films at temperatures between 70 and 100 °C and stretch ratios of three to five [2, 7]. It has been also obtained by crystallization from dimethyl formamide (DMF) or dimethyl acetamide (DMAc) solutions at temperatures below 70 °C [7, 8], but this method leads to porous and fragile films [7].

Other methods used to obtain piezoelectric β-phase samples with good mechanical properties are the incorporation of clays [9], zirconia [10], carbon nanotubes [11], ferrite nanoparticles [12], and silver nanoparticles [13] within the polymer matrix, among others.

Particularly interesting is the incorporation of ferrite nanoparticles as, together with the nucleation of the piezoelectric β-phase, it leads to the development of magnetoelectric (ME) composites, with interesting applications in the areas of sensor and actuator due to the interplay of the magnetostriction of the ferrites and the piezoelectricity of the polymer matrix [14, 15].

With the addition of such nanofillers the polymer thermal properties can be changed, which can in turn modify the application range of the materials [16–18]. In this way, the understanding of how thermal properties are changed due the physical and chemical interaction between nanoparticles and polymer is essential. Further, by addressing how nanoparticles induce variations in the thermal properties and thermal degradation of polymers, valuable information on the nature of the particle/polymer interactions can be obtained [19, 20]. Thermogravimetric analysis (TG) is typically used to evaluate the thermal degradation of polymers and polymer composites [21–23]. In particular, it has been proven that in PVDF the thermal degradation occurs in two steps, independent of the phase of the sample [24]. Further, the thermal stability of the material has been improved by suitable nanofillers, such as clays [25]. Similar findings were obtained in CaCO3/PVDF composites [26].

Since the thermal degradation and the nucleation of the ferroelectric phase is strongly influenced both by geometrical factors due to the nanosize of the fillers and, in particular, by the interactions in the interface between nanofillers and PVDF [27], TG appears as one of the most suitable techniques to study the nanoparticle/polymer interface effects [28].

In this work, CoFe2O4, Ni0.5Zn0.5Fe2O4, and NiFe2O4 nanoparticles were incorporated into the PVDF matrix with different filler contents as they have proven to be able to nucleate the piezoelectric phase of the polymer. The influence of filler type and content in the nucleation of the β-phase of the polymer has been correlated to the variations in the thermal degradation of the composite.

Materials and methods

Materials

Poly(vinylidene fluoride), PVDF, (Solef 1010) powder was supplied by Solvay. Ferrite nanoparticles and N,N dimethylformamide (DMF) were purchased from NanoAmor and Sigma-Aldrich, respectively, and used as received. CoFe2O4, NiFe2O4, and Ni0.5Zn0.5Fe2O4 ferrite nanoparticles have dimensions between 35–55 nm, 20–30 nm, and 10–30 nm, respectively [29].

Methods

In order to obtain a good filler dispersion, a solution of ferrite nanoparticles in DMF was first placed in an ultrasound bath for 6 h. After this period of time, PVDF powder was added to the solution. Complete dissolution of the polymer was achieved with the help of a Teflon mechanical stirrer incorporated in the ultrasound bath during 1 h.

Flexible composite films with an average thickness of ~50 μm and ferrite weight percentages (wt%) from 0.01 to 50 (from 3 × 10−5 to 0.25 in volume fraction) were obtained by spreading the solution at room temperature on a clean glass substrate. Solvent evaporation was obtained inside an oven at a controlled temperature of 210 °C [7].

The vibrational modes of the polymer used to determine the polymer phase and phase content were recorded by Fourier transformed infrared spectroscopy (FTIR) using a Perkin-Elmer Spectrum 100 in ATR from 650 to 4000 cm−1 with a resolution of 4 cm−1. 32 scans were performed for each sample.

The thermal behavior of the samples was determined by TG. Samples were transferred to open ceramic crucibles with capacity of 60 μL and analyzed using a Pyris 1 TG Perkin-Elmer thermobalance operating between 50 and 850 °C. A heating rate of 10 ± 0.2 °C min−1 and a nitrogen flow rate of 50 mL min−1 were used.

The existence of a nanoparticle/polymer interphase located at the interface was investigated by transmission electron microscopy (TEM) imaging using a JEOL JEM-1210 electron microscope operating at 200 keV. The samples were embedded in an epoxy resin and cut into thin films of about 100 nm using a Leica Ultracut UCT Ultramicrotome.

Theoretical background

PVDF has characteristic infrared bands such as the ones at 766 and 840 cm−1 that have been identified to correspond to the α- and β-phase, respectively [3, 30]. The variation of those bands as a function of ferrite nanoparticle wt% was analyzed and the phase content was quantified applying Eq. 1 [6, 31]:

$$ F\left( \upbeta \right) = \frac{{X_{\upbeta } }}{{X_{\alpha } + X_{\upbeta } }} = \frac{{A_{\upbeta } }}{{\left( {{{K_{\upbeta } } \mathord{\left/ {\vphantom {{K_{\upbeta } } {K_{\alpha } }}} \right. \kern-0pt} {K_{\alpha } }}} \right)A_{\alpha } + A_{\upbeta } }}, $$
(1)

where F(β) represents the β-phase content; A α and A β are the absorbencies at 766 and 840 cm−1, corresponding to the α and β-phase material; K α and K β are the absorption coefficient at the respective wavenumber; and X α and X β are the degree of crystallinity of each phase. The value of K α is 6.1 × 104 and the value of K β is 7.7 × 104 cm2 mol−1 [32, 33].

The influence the ferrite nanoparticles in the polymer thermal degradation was determined by two different methods [34, 35]. First, the Broido method was used, assuming n = 1 and considering the specific heating rate \( \upbeta = {{\partial T} \mathord{\left/ {\vphantom {{\partial T} {\partial t}}} \right. \kern-0pt} {\partial t}} \) [36]:

$$ \ln \left[ { - \ln \left( {1 - \alpha } \right)} \right] = - \frac{{E_{\rm a} }}{RT} + {\text{const}}, $$
(2)

where \( \alpha \) represents the degree of conversion of the sample under degradation, defined by \( \alpha = {{w_{0} - w\left( t \right)} \mathord{\left/ {\vphantom {{w_{0} - w\left( t \right)} {w_{0} - w_{\infty } }}} \right. \kern-0pt} {w_{0} - w_{\infty } }} \). Here, \( w_{0} \), \( w_{t} \) and \( w_{\infty } \) are the weights of the sample before degradation, at time t and after complete degradation, respectively. \( E_{\text{a}} \) is the activation energy of the process, T is the temperature, and R is the gas constant (8.314 J mol−1 K−1).

Further, the Coats–Redfern method [37] was also used, considering a constant heating rate:

$$ \ln \left[ {\frac{{ - \ln \left( {1 - \alpha } \right)}}{{T^{2} }}} \right] = - \frac{{E_{\rm a} }}{RT} + \ln \frac{AR}{{\upbeta E_{\rm a} }}, $$
(3)

where \( A \) is constant, \( \upbeta \) is the heating rate and \( \alpha \) represents the degree of conversion.

The dependence of the maximum weight loss degradation temperature of the polymer on the ferrite nanoparticle concentration is described by Eq. 4 [28]:

$$ T_{\text I} = T_{1} + T_{2} \exp (Cx), $$
(4)

where \( T_{1} \), \( T_{2} \) and \( C \) are fitting constants and \( x \) is the weight fraction of ferrite nanoparticles. Here, \( T_{1} + T_{2} = T_{\rm I}^{0} \), where \( T_{\rm I}^{0} \) is the temperature at which the weight loss of the pristine polymer is maximum.

In polymer nanocomposites, the interface between the polymer and nanoparticles has an important role in determining the final properties of the material. Through TG it is possible to evaluate the mass fraction of the polymer located at the interface, \( m_{\rm I} \), using the following equation [28]:

$$ m_{\rm I} = \frac{{m\left( x \right)_{{\rm I}0} - m_{{\rm I}0} }}{{m_{{\rm I}0} }} \times 100, $$
(5)

where \( m_{{\rm I}0} \) is the mass of the pristine polymer at the temperature at which the mass loss rate is maximum and \( m\left( x \right)_{{\rm I}0} \) is the mass of the composite containing a given wt% of nanoparticles that has not degraded at the temperature at which the mass loss rate of the pristine polymer is maximum.

Results and discussion

FTIR spectra (not shown) have been previously used to quantify the crystalline phase type and content in the polymer as described in detail in [12]. Figure 1 presents the variation of β-phase content with increasing ferrite concentration.

Fig. 1
figure 1

Evolution of the β-phase content with increasing filler concentration for, a CoFe2O4/PVDF, b Ni0.5Zn0.5Fe2O4/PVDF, and c NiFe2O4/PVDF nanocomposites

Full size image

Figure 1 reveals that the β-phase content depends not only on the ferrite content but also on the type of ferrite nanoparticle. CoFe2O4 and Ni0.5Zn0.5Fe2O4 nanoparticles induce an almost complete β-phase formation for ferrite weight contents lower than 5 %. However, in the case of NiFe2O4 nanoparticles, the β-phase nucleation just starts for ferrite contents higher than 5 wt%.

The β-phase nucleation depends on the nanoparticle-polymer electrostatic interactions and it is also related to the formation of an interphase region as explained in [29].

TG thermograms and the corresponding differential TG plots (DTG) for different CoFe2O4 nanoparticle concentrations, ranging from 0 to 50 wt%, are shown in Fig. 2.

Fig. 2
figure 2

TG thermograms of CoFe2O4/PVDF nanocomposites for a small ferrite contents and b large ferrite contents. An arrow identifies an additional degradation step with respect to the pristine polymer in the DTG representation

Full size image

In all cases, the typical two step thermal degradation characteristic of PVDF [24] is verified. The first step of degradation of the polymer occurs between 400 and 500 °C, the polymer maximum degradation temperature being influenced by the nanoparticle content present in the nanocomposite. In this first degradation step the decomposition mechanism is chain-stripping where carbon-hydrogen and carbon fluorine scission occurs and the presence of both hydrogen and fluorine radicals leads to the formation of hydrogen fluoride [24, 38]. The formation of this compound is responsible for the weight loss observed in the first degradation step.

The second degradation step occurs between 500 and 850 °C, and the differences observed in the plots comparative to the pure α-phase PVDF are to be ascribed to the presence of ferrite nanoparticles, as the different phases of PVDF show similar thermal degradation characteristics [39]. This second step is a complex degradation process resulting in poly(aromatization). The polyenic sequence formed previously on the first degradation step is unstable and, as a consequence, the macromolecules formed undergo further reactions leading to scission followed by the formation of aromatic molecules [24, 40].

The residual weight that remains at 850 °C corresponds mainly to the ferrite nanoparticles together with the residual char from the previous degradation steps.

An additional degradation step (identified as 3rd step in Table 1) with respect to the pure polymer is also identified in the multiferroic nanocomposite samples. This step is particularly well identified in the derivative of the TG curves (DTG) (Fig. 2, inset). The emergence of a new stage of degradation in comparison to the pure polymer is related to the increase of an interphase in the interface volume between nanoparticles and polymer, related to the nucleation of the polymer β-phase [29].

Table 1 Parameters determined by TG analysis for the different nanocomposites
Full size table

The onset temperature for neat PVDF (458 °C) is lower than those of CoFe2O4/PVDF nanocomposites with ferrite contents higher than 1 wt% (≈475 °C), indicating that the thermal stability of the matrix has been improved with the addition of CoFe2O4 nanoparticles [41].

TG plots presented in the Fig. 3a, b show the thermal stability of the polymer with increasing NiFe2O4 and Ni0.5Zn0.5Fe2O4 ferrite nanoparticle content, respectively.

Fig. 3
figure 3

TG thermograms a function of the ferrite content for a NiFe2O4/PVDF and b Ni0.5Zn0.5Fe2O4/PVDF. The inset shows the corresponding DTG curves. An arrow identifies an additional degradation step

Full size image

Similar to the CoFe2O4/PVDF composites, the typical two steps thermal degradation characteristic of the polymer is observed for the samples with low ferrite nanoparticle content. With increasing ferrite content an additional degradation step emerges at temperatures around 600 °C, which is related to the interaction between the nanoparticles and the polymer matrix [29]. In this case, for temperature above 800 °C and high filler loading contents, residuals are about 4 % lower than the original ferrite nanoparticle loading, indicating some mass loss with respect to the original nanofiller content which is to be ascribed to impurities related to the nanoparticle preparation.

In the same way as observed for the CoFe2O4/PVDF composites, the onset temperature for neat PVDF (458 °C) is again lower than those for NiFe2O4/PVDF and Ni0.5Zn0.5Fe2O4/PVDF nanocomposites with ferrite contents higher than 1 wt% (≈480 °C), indicating that the thermal stability of the matrix has been improved with the addition of the nanoparticles [41].

This behavior can be explained by the presence of the char formed from the PVDF matrix, which is further stabilized through π–π electronic interactions with the ferrite nanoparticles [10, 42]. Due to the fine dispersion and good thermal properties, the ferrite nanoparticles might strongly hinder the volatility of the decomposed products and limit the continuous decomposition of the PVDF [43].

A small loss mass at around 200 °C (Figs. 2, 3) in the nanocomposites that does not exist in the pure polymer matrix is also noted. This mass loss is commonly found in polymer/nanoparticle composites hybrid films [44] and it is ascribed to the evaporation of residual solvent and/or physically absorbed water [45]. In the present case, the amount of mass loss seems to be too large to be just attributed to the aforementioned situations and, therefore, some catalytic degradation seems to be induced at ~200 °C by the presence of the nanoparticles. This degradation can be induced by the different thermal characteristics of ferrite fillers and polymer will lead to larger local temperatures in the polymer/nanoparticle interface than in the rest of the polymer and, therefore, to earlier degradation of the polymer at that region when compared to the average sample temperature. The increase in the onset temperature and a small decrease in thermal stability around 200 °C found in the ferrite/PVDF nanocomposites has also been reported by Pal et al. [10] in Zirconia-coated carbon nanotube/PVDF nanocomposites. Due to this lower temperature mass loss, it can be stated that the composites show an overall lower thermal stability with respect to mass loss than the pure polymer, even when the main thermal degradation of the polymer matrix is increased.

The thermal degradation was studied by the Broido method, in which a straight line is observed in the plot of ln[−ln(1 − α)] versus T −1 with a slope of −E a R −1, from which the activation energy is determined [46] (Fig. 4a).

Fig. 4
figure 4

Representation of the data of Fig. 2 for the sample with 1 wt% of CoFe2O4 in order to apply the a Broido method and the b Coats–Redfern method

Full size image

In an analogous way, analysis of the TG curves was performed with the Coats–Redfern method by plotting ln[−ln(1 − α)/T 2] versus T −1 and determining the activation energy of the nanocomposites [47] (Fig. 4b).

Table 1 shows a compilation of the data obtained from the TG measurements for the different composites, e.g., the degradation temperature, the interface value determined through Eqs. 4 and 5, and the kinetic parameters obtained from Broido (Eq. 2) and Coats–Redfern (Eq. 3) methods.

In low lower ferrite content CoFe2O4/PVDF composites (0.01–1 wt%), nanoparticles act as polymer matrix defects decreasing the value of the activation energy. For all nanocomposites with ferrite contents above 5 %, the activation energy increases with increasing ferrite content until reaching a maximum and then decreases. It is known that the addition of nanofillers can lead to thermal stabilization of polymers during their decomposition [48]. In the present investigation, a thermal stability enhancement has been observed at intermediate nanoparticle loadings while at higher contents thermal stability becomes progressively lower [49]. At the lower concentrations, the α to β polymer phase transformation induced by the electrostatic interactions between the C–F bands of PVDF and the ferrite nanoparticles improves the thermal stability of PVDF [50]. On the other hand, the formation of aggregates at high nanoparticle concentrations and the existence of high ferrite content will fade the beneficial effects of nucleation of the β-phase with respect to thermal stability as the presence of such large amounts of fillers leads to variations in the connectivity of the polymer phase as well as strong effects in the internal heat transfer kinetics due to the different thermal characteristics of the nanoparticles and the polymer.

The observed difference in the activation energy values obtained from the Broido and Coats–Redfern methods is attributed to the different mathematical approaches used to calculate the kinetic parameters [35, 51]. The correlation coefficients are 0.984 and 0.975, respectively, and the observed trend obtained by the two methods is the same.

Further, variations in the DTG temperature as a function of ferrite nanoparticle content are represented in Fig. 5. The fitting for each ferrite/PVDF nanocomposite with Eq. 4 is also shown.

Fig. 5
figure 5

Maximum degradation temperature versus ferrite nanoparticle content and the respective fitting curves with Eq. 4

Full size image

For all nanocomposites, the addition of ferrite nanoparticles leads to the enhancement of the thermal stability of the main degradation of the polymer which is demonstrated by the increase of the polymer DTG temperature with increasing ferrite nanoparticle loading.

It is well known that the polymer thermal degradation begins with the generation of free radicals, which are transferred to the adjacent chains by intermolecular and intramolecular chain reaction, followed by a termination step [52]. Ferrite nanoparticles may act as radical scavengers that largely suppress these chain transfer reactions and prevent polymer chains from decomposing, that is to say, the rate of bubble nucleation decreases and it needs more time for the volatiles to reach a critical concentration. Therefore, DTG temperature of a nanocomposite is higher in comparison to neat PVDF [53].

In Fig. 6, the variations in the volume fraction of the polymer at the interface region, obtained with Eq. 5, as a function of ferrite concentration are represented.

Fig. 6
figure 6

Evolution of the interface volume between nanoparticles and polymer as a function of ferrite nanoparticles content

Full size image

An increase of the interface value with increasing ferrite loading is observed for all composites. The size of the interphase between ferrite nanoparticles and polymer is significantly affected by the quantity and type of ferrite nanoparticle.

The interface value increases with increasing ferrite loading and it is larger for CoFe2O4 and Ni0.5Zn0.5Fe2O4 ferrite nanoparticles in comparison to NiFe2O4 ones, indicating that the quantity of the interface also depends on the ferrite type.

With increasing interface value, the interaction between the partially positive CH2 bonds of the PVDF chains and the electrostatically negative charged ferrites will be promoted, explaining the fact that the electroactive β-phase nucleation is higher for the nanocomposites that have higher interface values [29].

The existence of a nanoparticle/polymer interphase is distinctly confirmed by TEM in the CoFe2O4/PVDF (95/5 wt%) nanocomposites (Fig. 7).

Fig. 7
figure 7

TEM images of ferrite/PVDF (95/5 wt%) nanocomposites with: a CoFe2O4 nanoparticles and b NiFe2O4 nanoparticles. The scale bar corresponds to 10 nm

Full size image

Figure 7 reveals a significant dissimilarity between the two nanocomposites: while a well distinguished interphase between the ferrite nanoparticle and the polymer is observed for the CoFe2O4/PVDF nanocomposites (a), the existence of such interphase is not observed in the NiFe2O4/PVDF nanocomposites (b). This observation is in agreement with the results presented in Figs. 1 and 6, since the nucleation of the electroactive β-phase in NiFe2O4/PVDF nanocomposites starts from the 5 wt% and the interface value is higher for CoFe2O4 in comparison to NiFe2O4 ones. It is noticed here that the interface volume, as defined by Eq. 5 is not necessarily equal to the interphase region observed by TEM, as they are based on different physical principles. It is relevant to notice that a larger calculated interface region (Fig. 6), e.g., larger interaction volume, is correlated to larger observed interphase volume (Fig. 7).

Conclusions

Composite films of different ferrites nanoparticles, CoFe2O4, Ni0.5Zn0.5Fe2O4, and NiFe2O4, and PVDF fluoride have been prepared in order to obtain multiferroic composites through the nucleation of the piezoelectric β-phase of the polymer by the ferrite fillers. TG was used to study the thermal stability of the composites and to quantify the interface region leading to the β-phase nucleation.

It was found that the size of the interface between the ferrite nanoparticles and the polymer is significantly affected by the content and type of ferrite nanoparticles. The interface value and the β-phase content increases with increasing ferrite loading and it is higher for the CoFe2O4 and Ni0.5Zn0.5Fe2O4 fillers in comparison to the NiFe2O4 ones, indicating that the value of the size of the interface is intimately related to the piezoelectric β-phase formation and depends on the ferrite type. The composite shows lower thermal stability due to the existence of mass loss processes at lower temperatures than the main degradation of the polymer induced by the presence of the ferrite fillers.