Electromechanical properties of PVDF-based polymers reinforced with nanocarbonaceous fillers for pressure sensing applications

materials Article Electromechanical Properties of PVDF-Based Polymers Reinforced with Nanocarbonaceous Fillers for Pressure Sensing Applications Javier Vicente 1, P. Costa 2,3,* , S. Lanceros-Mendez 4,5, Jose Manuel Abete 6and Aitzol Iturrospe 1 1Electronics and Computing Department, Mondragon Unibertsitatea, 20500 Mondragon, Spain 2Center of Physics, University of Minho, 4710-057 Braga, Portugal 3Institute for Polymers and Composites (IPC), University of Minho, 4800-058 Guimarães, Portugal 4BCMaterials, Basque Center for Materials, Applications and Nanostructures, UPV/EHU Science Park, 48940 Leioa, Spain 5IKERBASQUE, Basque Foundation for Science, 48013 Bilbao, Spain 6Applied Mechanics Department, Mondragon Unibertsitatea, 20500 Mondragon, Spain *Correspondence: [email protected] Received: 1 October 2019; Accepted: 25 October 2019; Published: 29 October 2019   Abstract: Polymer-based composites reinforced with nanocarbonaceous materials can be tailored for functional applications. Poly(vinylidene fluoride) (PVDF) reinforced with carbon nanotubes (CNT) or graphene with different filler contents have been developed as potential piezoresistive materials. The mechanical properties of the nanocomposites depend on the PVDF matrix, filler type, and filler content. PVDF 6010 is a relatively more ductile material, whereas PVDF-HFP (hexafluropropylene) shows larger maximum strain near 300% strain for composites with CNT, 10 times higher than the pristine polymer. This behavior is similar for all composites reinforced with CNT. On the other hand, reduced graphene oxide (rGO)/PVDF composites decrease the maximum strain compared to neat PVDF. It is shown that the use of different PVDF copolymers does not influence the electrical properties of the composites. On the other hand, CNT as filler leads to composites with percolation threshold around 0.5 wt.%, whereas rGO nanocomposites show percolation threshold at ≈ 2 wt.%. Both nanocomposites present excellent linearity between applied pressure and resistance variation, with pressure sensibility (PS) decreasing with applied pressure, from PS ≈ 1.1 to 0.2 MPa −1 . A proof of concept demonstration is presented, showing the suitability of the materials for industrial pressure sensing applications. Keywords: piezoresistivity; PVDF; nanocarbonaceous; electromechanical; pressure sensibility 1. Introduction Polymer-based nanocomposites are attracting large attention in recent years both in the scientific and industrial areas. Through the inclusion of fillers, a wide range of polymer properties can be enhanced, such as mechanical [ 1 ], electrical [ 2 ], and thermal properties [ 3 ], among others. Thus, polymer composites can be tailored for specific applications. Moreover, it is possible to add new features to the polymers through the inclusion of nanoparticles, functionalizing them and enabling sensing of different parameters such as mechanical [4], temperature or humidity [5], among others physical properties. Hence, their application as sensors materials are being widely studied due to their simple manufacturing and integration into devices [ 6 , 7 ]. Some of the most used fillers for strain sensing functionalization of polymers are carbon nanoallotropes, such as carbon black (CB) [ 8 ], graphene (G) and its oxidized forms (graphene oxide (GO) and reduced GO (rGO) [ 8 , 9 ] and carbon nanotubes (CNT) [ 7 , 10 ]. With the inclusion of conductive nanoparticles, polymer nanocomposites increase their piezoresistive Materials 2019,12, 3545; doi:10.3390/ma12213545 www.mdpi.com/journal/materials Materials 2019,12, 3545 2 of 14 response, i.e., under a mechanical solicitation their resistance changes linearly with applied strength [ 11 ]. Different fillers influence the electrical and functional properties, but also the mechanical properties of the composite [ 11 , 12 ]. The percolation threshold of the polymer composites depends on the filler, matrix, and processing method, among other parameters [ 13 ]. Lower percolation thresholds around 0.1 to 0.4 vol.% can be found in polymer composites with graphene or carbon nanotubes as fillers [ 13 ]. The aspect ratio of the nanofillers as well as their intrinsic properties strongly influence the electrical and mechanical properties of the composite and, therefore, its functional response [ 12 ]. In this way, graphene and CNT are compared as reinforcement fillers due to their different intrinsic properties in order to find suitable nanofillers for specific functional devices. These materials are typically tailored to optimize their mechanical properties and piezoresistive sensibility, achieving gauge factor values up to 175 [ 14 ], two orders of magnitude higher than traditional strain gauges, and strains larger than 50 % [ 15 ]. The percentage of fillers used for optimizing functional response also vary widely [ 16 ], depending on the matrix, filler, processing method or even application. However, the larger piezoresistive sensibility in polymer composites is observed near the percolation threshold [17]. The percolation threshold is the range in which the materials undergo a transition from nearly insulating to conductive, changing several orders of magnitude their electrical conductivity for small filler content variations [ 16 ]. The percentage at which this phenomenon occurs varies widely depending on fillers, matrices, and processing methods [ 18 , 19 ]. The percolation threshold is reported to occur when the first conductive paths spanning all the nanocomposite are formed due to the proximity of the conductive fillers embedded in the isolating matrix. This phenomenon has been explained by different theoretical models [19–21]. The selection of the polymer matrix used for the development of a functional sensor depends on the stimulus needed for sensor response. Flexibility or stretchability, force, and environmental conditions influence the host polymer to use in view the overall properties to select. Soft polymer matrices such as natural rubbers or thermoplastic elastomers are reported to provide the nanocomposite high strain capability from low to large strains [ 15 ], and a wide range of sensitivity in functional response [ 22 ]. For applications in which mechanical solicitations can compromise the structural integrity of the material or for large force applications, stiffer matrices are employed. Among the most used thermoplastic polymers for force and deformation sensor development, literature reports on polypropylene (PP) [ 23 ], poly(vinylidene fluorine) (PVDF) [ 23 , 24 ] and poly-ether-ether-ketone (PEEK) [ 25 ], among others [ 25 , 26 ], though thermosetting such as polyepoxides [ 27 ] and elastomers such as thermoplastic polyurethane (TPU) [ 8 , 9 ], triblock styrene-butadiene-styrene (SBS) [ 15 ] or styrene-ethylene/butylene-styrene (SEBS) [23] have been also used. PVDF and its copolymers are excellent materials for functional applications, such as sensors, actuators, energy harvesting and biomaterials in the biomedical field [ 28 ]. PVDF shows excellent electroactive properties, being used as host polymer for large number of applications [ 28 ]. PVDF is semi-crystalline material with five distinct crystalline phases, the most investigated and used for applications being the non-polar α -phase and the polar β -phase [ 11 , 27 ]. Furthermore, PVDF presents excellent mechanical and chemical properties, weather resistance, and outstanding properties associated with their polar crystalline forms [ 29 ]. In this way, functional composite applications is an interesting material with large potential for force sensor, due to its mechanical and chemical resistance properties [12,27]. Polymer nanocomposites can be manufactured using laboratory and industrial techniques. At the laboratory level, they can be processed by solvent casting [ 30 ], for example, whereas at industrial level typical processing methods include hot pressing [ 31 ], extrusion [ 32 ], or injection [ 33 ]. Through the different fabrication processes, the overall properties of the PVDF composites can be tuned, including mechanical and electrical properties, with large influences on the functional performance of the composite. In particular, thin-film materials attract increasing attention based on their simple integration in to devices [ 33 – 36 ]. Hence, the integration of these films in components or processes enables an inexpensive sensor conformation. Materials 2019,12, 3545 3 of 14 In this work, a comprehensive study of PVDF-based materials for force compression sensing is presented with excellent performance and linearity [ 37 , 38 ]. Different PVDF based polymers (PVDF-HFP (poly(vinylidene fluoride-co-hexafluropropylene)), PVDF 6010 and PVDF 5130) and nanocarbonaceous fillers (carbon nanotubes and reduced graphene oxide) were used to tailor composites in view their mechanical, electrical and electromechanical properties. Further, proof of concept application is presented, submitting the sensor to different pressures. In order to develop a functional material with specific responses to the applications, the focus on the different properties of the polymer matrices and fillers (type and content) will be evaluated to tailor the overall properties of the composite to work as piezoresistive sensible material under mechanical compression. Host matrices were selected from 400 MPa to 2.5 GPa of tensile modulus, as detailed in experimental part and CNT and rGO have been selected as functional fillers based on their different dimensions and intrinsic properties. 2. Experimental 2.1. Materials The selected polymers were poly(vinylidene fluorine) with reference 5130, 6010, and PVDF-HFP, all supplied by Solvay. PVDF 5130 is characterized by ultra-high viscosity with excellent adhesion, a density of 1.75 g/cm3and a tensile modulus between 1 to 1.5 GPa. PVDF 6010 is a homopolymer with medium viscosity, density between 1.75 to 1.8 g/cm 3 and a tensile modulus between 1.7 to 2.5 GPa. Poly(vinylidene fluoride-co-hexafluropropylene) (PVDF-HFP), with reference Solef 21,216 and VDF/HFP mole ratio of 88/12, shows a density of 1.78 g/cm3and a tensile modulus between 400 to 600 MPa. The solvent used to disperse the nanofillers and dissolve the PVDF was N,N 0 -dimethylpropyleneurea (DMPU) and was purchased from LaborSpirit. Multi-walled carbon nanotubes were supplied by Nanocyl with reference NC7000, showing an average length of 1.5 µ m, an outer mean diameter of 9.5 nm and 90% purity. Reduced graphene oxide was obtained from The Graphene Box (Spain) with >99% of purity, 1–5 µm of length and 1–2 layers. 2.2. Sample Preparation Carbon nanofillers were dispersed in DMPU within an ultrasonic bath (ATU, Model ATM40-3LCD) for an average time of 4 h, assuring a correct de-agglomeration and homogeneous dispersion of the fillers in the solvent. Then, PVDF (5130, 6010, or HFP) was added to the filler/solvent solution and completely dissolved through magnetic stirring for approximately 3 h at 30 ◦ C. It is to notice that this processing method has demonstrated to ensure good filler dispersion for both graphene [ 12 ] and CNT [ 15 ]. Then, the solution was spread on a clean glass substrate by doctor blade method with a 100 µ m blade thickness. Finally, films were melted in an oven at 210 ◦ C for 25 min, promoting the crystallization of the PVDF in the α -phase and achieving complete solvent evaporation [ 39 ]. The thicknesses of the films after complete evaporation of the DMPU solvent ranges from 20 to 60 µm. Films with the different polymer matrices and carbonaceous filler percentages were prepared, as indicated in Table 1. Table 1. Poly (vinylidene fluoride) (PVDF)-based polymers, nanofillers and solvent/polymer ratio used in the processing of the nanocomposites. DMPU/PVDF Vol/Vol Carbon Nanotubes (wt.%) Nomenclature Reduced Graphene Oxide (wt.%) Nomenclature PVDF 5130 95/50 0.25 0.5 1 PVDF5130 0 0.5 1 2 PVDF5130 0.25CNT/5130 0.5rGO/5130 0.5CNT/5130 0.1rGO/5130 1CNT/5130 2rGO/5130 PVDF 6010 90/10 0 0.25 0.5 PVDF6010 0.25CNT/6010 0.5CNT/6010 PVDF-Hexafluropropylene 90/10 0 0.25 0.5 PVDF-HFP 0.25CNT/HFP 0.5CNT/HFP Materials 2019,12, 3545 4 of 14 2.3. Sample Characterization Fourier-transform infrared (FTIR) spectroscopy analysis was carried out in transmission mode at room temperature from 4000 cm −1 to 600 cm −1 with a resolution of 4 cm −1 employing a Jasco FT/IR-4100 spectrometer with a TGS detector. Differential scanning calorimetry (DSC) tests were performed with a Netzsch DSC 200F3 Maia set up. Samples were placed into Al pan crucibles, stabilized at 30 ◦ C and then subjected to a 20 ◦ C.min −1 heating rate up to 200 ◦C under nitrogen atmosphere. Electrical conductivity was obtained after I–V measurements. Samples were submitted to a voltage sweep from − 10 to +10 V and the current was measured with a Keithley 6430 SourceMeter. Gold electrodes were previously deposited in both sides of the films by magnetron sputtering with a Polaron SC502 sputter coater. The electrical conductivity was obtained after Equation (1) considering sample thickness and electrode area: σ=ρ−1=RA L−1 (1) where Ris the electrical resistance, Athe electrode area and Lthe sample thickness. Electrical conductivity tests were performed in two sample points and repeated three times each. The mechanical characterization of the nanocomposites was performed by tensile tests up to failure with a Shimadzu AG-IS universal testing machine with a 50 N load cell, repeated for 5 measures for each sample. For the tensile tests, samples of 30 × 10 mm of the area and 20 to 60 µ m of thickness were cut from the fabricated films, placed in the universal testing machine through a clamping system leaving an approximate interspace between clamps of 10 mm to assure a proper grip of the sample. The measurements were performed at speed of 1 mm/min and force and test time were recorded at a 100 ms rate. Engineering stress and strain (Equations (2) and (3)) were obtained according to: σ= F A(2) ε= ∆L L0 (3) where Fis the tensile force, Athe transversal area of the sample considering the thickness and width of the sample, ∆ Lthe recorded displacement and L 0 the initial interspace between clamps. Tensile tests were repeated for three times, ensuring that measurements were consistent within materials. The samples were considered as macroscopically homogeneous materials, without considering non-local effects [ 40 ] due to the low filler content. Electromechanical tests were performed under cyclic compression on samples with initial minimum pre-load of 10 N corresponding to 0.09 MPa to forces of 50, 100, 200, and 400 N, corresponding to 0.43, 0.86, 1.72, and 3.45 MPa, respectively, considering the contact area of the compression tests. The piezoresistive tests were performed in a Shimadzu AG-IS universal testing machine with a load cell of 500 N at speed of 0.5 and 1 mm/min, for 10 and 50 cycles and replicated twice. Force and displacement were recorded with a 500 ms time span. The acquisition of the electrical resistance between the compression pieces was performed with an Agilent 344401A multimeter. Piezoresistive tests were repeated twice. The quantitative evaluation of the pressure sensitivity (PS in Equation (4)) was performed according to: PS = ∆R R0 P(4) where ∆ Rrepresents the resistance, variation measured during the test and R 0 the initial resistance of the sample under the minimum pre-load and Pis the pressure on the sample. Materials 2019,12, 3545 5 of 14 3. Results and Discussion 3.1. Chemical and Thermal Characterization Figure 1presents the FTIR spectra of the different neat semi-crystalline PVDF polymer (Figure 1A) and the corresponding composites reinforced with CNT and rGO (Figure 1B). The presence of α -phase in the samples is confirmed by the corresponding bands at 614, 763, 795 and 975 cm −1 [ 28 ], whereas vibration bands corresponding to the β , γ , and δ [ 28 ] phases are detected neither in the neat polymers or the corresponding composites. In fact, the α-PVDF phase is the one obtained when the polymer is crystallized from the melt [ 28 ]. No significative variations in the FTIR spectra were detected among the different composites. Materials 2019, 11, x FOR PEER REVIEW 5 of 16 corresponding to 0.43, 0.86, 1.72, and 3.45 MPa, respectively, considering the contact area of the compression tests. The piezoresistive tests were performed in a Shimadzu AG-IS universal testing machine with a load cell of 500 N at speed of 0.5 and 1 mm/min, for 10 and 50 cycles and replicated twice. Force and displacement were recorded with a 500 ms time span. The acquisition of the electrical resistance between the compression pieces was performed with an Agilent 344401A multimeter. Piezoresistive tests were repeated twice. The quantitative evaluation of the pressure sensitivity (PS in Equation (4)) was performed according to: =∆    (4) where ΔR represents the resistance, variation measured during the test and R0 the initial resistance of the sample under the minimum pre-load and P is the pressure on the sample. 3. Results and Discussion 3.1. Chemical and Thermal Characterization Figure 1 presents the FTIR spectra of the different neat semi-crystalline PVDF polymer (Figure 1A) and the corresponding composites reinforced with CNT and rGO (Figure 1B). The presence of αphase in the samples is confirmed by the corresponding bands at 614, 763, 795 and 975 cm−1 [28], whereas vibration bands corresponding to the β, γ, and δ [28] phases are detected neither in the neat polymers or the corresponding composites. In fact, the α-PVDF phase is the one obtained when the polymer is crystallized from the melt [28]. No significative variations in the FTIR spectra were detected among the different composites. 600 800 1000 1200 1400 Absorbance (a.u.) PVDF-HFP PVDF6010 PVDF5130 Wavenumber ( cm-1 ) A 600 800 1000 1200 1400 025CNT/HFP 025CNT/6010 1rGO/5130 Absorbance (a. u.) Wavenumber (cm -1 ) PVDF5310 1CNT/5130 B Figure 1. Fourier-transform infrared (FTIR) spectra of (A) neat polymers and (B) composites with carbon nanotubes (CNT) or reduced graphene oxide (rGO) nanofillers. Literature reports that some nanoparticles such as carbon nanotubes can influence the crystallization phase of PVDF [41]. The neat conductive fillers CNT and rGO, without surface modification, do not influence the crystallization phase of the polymers as observed in the FTIR spectra (Figure 1B) for the PVDF composite materials, as reported previously [12]. Neat nanocarbonaceous materials do not act as nucleating agents [41] in PVDF, leading just to α-phase crystallization independently of filler type and content. Thermal properties were evaluated by DSC thermograms for neat PVDF 5130, 6010 and HFP (Figure 2A) and their corresponding nanocomposites with CNT or rGO with different filler content (Figure 2B). Higher variability on the melting temperature (Tm) can be observed between the different PVDF matrixes, with PVDF-HFP presenting lower crystallization temperature, Tc ≈ 132.3 °C, than PVDF5130 and 6010, with Tm ≈ 158.2 and Tm ≈ 168.9 °C, respectively. The thermograms of the neat Figure 1. Fourier-transform infrared (FTIR) spectra of ( A ) neat polymers and ( B ) composites with carbon nanotubes (CNT) or reduced graphene oxide (rGO) nanofillers. Literature reports that some nanoparticles such as carbon nanotubes can influence the crystallization phase of PVDF [ 41 ]. The neat conductive fillers CNT and rGO, without surface modification, do not influence the crystallization phase of the polymers as observed in the FTIR spectra (Figure 1B) for the PVDF composite materials, as reported previously [ 12 ]. Neat nanocarbonaceous materials do not act as nucleating agents [ 41 ] in PVDF, leading just to α -phase crystallization independently of filler type and content. Thermal properties were evaluated by DSC thermograms for neat PVDF 5130, 6010 and HFP (Figure 2A) and their corresponding nanocomposites with CNT or rGO with different filler content (Figure 2B). Higher variability on the melting temperature (T m ) can be observed between the different PVDF matrixes, with PVDF-HFP presenting lower crystallization temperature, T c≈ 132.3 ◦ C, than PVDF5130 and 6010, with T m≈ 158.2 and T m≈ 168.9 ◦ C, respectively. The thermograms of the neat polymers and the nanocomposites show a single peak corresponding to the melting of the α -phase of PVDF [42]. Materials 2019, 11, x FOR PEER REVIEW 6 of 16 polymers and the nanocomposites show a single peak corresponding to the melting of the α-phase of PVDF [42]. 60 80 100 120 140 160 180 PVDF-HFP PVDF6010 Heat Flow (Endo up) Temperature (ºC) PVDF5130 A 60 80 100 120 140 160 180 025CNT/HFP 025CNT/6010 1rGO/5130 Heat Flow (Endo Up) Temperature (ºC) 05CNT/5130 B Figure 2. Differential scanning calorimetry (DSC) scans of (A) neat PVDF 5310, 6010 and HFP (hexafluropropylene) and (B) corresponding nanocomposites with CNT and rGO for different filler contents. PVDF composites with CNT or rGO lead to crystallization temperatures slightly higher than the neat polymers, increasing less than 2 °C in all cases, demonstrating that the effect of the fillers in the melting behavior of the samples is quite small and independent of filler type and content, confirming the low interaction between fillers and polymers observed in the FTIR spectra (Figure 1), where the introduction of the fillers lead to no new chemical bonds or variations in the polymer bonds. 3.2. Electrical Characterization The volume of electrical conductivity for the different composites is shown in Figure 3. Electrical conductivity increases with the inclusion of both nanofillers for all PVDF copolymers, showing a percolation threshold lower than 0.5 wt.% for CNT/PVDF composites and near 2 wt.% for rGO/PVDF5130 composites. The conductive network formed by the CNT is more effective than for rGO for similar filler contents into the PVDF matrix, the intrinsic conductivity of the CNT being higher than the one of the rGO nanofillers. Further, the aspect ratio of the CNT is larger than the one of the 2D materials, leading to lower percolation threshold composites. The conductivity of the neat polymers is in agreement with the manufacturer specifications and is similar among PVDF copolymers [11,42,43]. Among CNT nanocomposites, PVDF6010 shows the higher electrical conductivity for filler content between 0.25 and 0.5 wt.% when compared to PVDF-HFP and 5310 composites. The electrical conductivity for CNT/6010 is higher for 0.25 wt.% CNT (near 2 orders of magnitude) but for composite with 0.5 wt.% filler content the electrical conductivity is in the same order of magnitude for all polymer matrices. Their percolation threshold is thus around 0.5 wt.% CNT and the maximum conductivity is σ ≈ 5 × 10−1 (Ω.m)−1, as reported in literature for CNT/polymer materials [43,44]. It is to notice that percolation thresholds below 0.1 wt.% of CNT have been reported for PVDF matrix composites [44], which is not verified in the present work. With respect to rGO/5130 nanocomposites, they show lower conductivity when compared to CNT at the same filler content, which is attributed to the lower aspect ratio and intrinsic conductivity of rGO when compared to CNT, leading to an increase of percolation threshold of the composite. The rGO/5130 up to 1 wt.% filler content shows similar electrical conductivity than neat PVDF5130 and the percolation threshold is ≈ 2 wt.% rGO, with an electrical conductivity of σ ≈ 1 × 10−5 (Ω.m)−1. The intrinsic properties of CNT lead to low percolation thresholds in polymer-based composites. To tailor polymer-based nanomaterials with functional properties, low nanofiller content is typically Figure 2. Differential scanning calorimetry (DSC) scans of ( A ) neat PVDF 5310, 6010 and HFP (hexafluropropylene) and ( B ) corresponding nanocomposites with CNT and rGO for different filler contents. Materials 2019,12, 3545 6 of 14 PVDF composites with CNT or rGO lead to crystallization temperatures slightly higher than the neat polymers, increasing less than 2 ◦ C in all cases, demonstrating that the effect of the fillers in the melting behavior of the samples is quite small and independent of filler type and content, confirming the low interaction between fillers and polymers observed in the FTIR spectra (Figure 1), where the introduction of the fillers lead to no new chemical bonds or variations in the polymer bonds. 3.2. Electrical Characterization The volume of electrical conductivity for the different composites is shown in Figure 3. Electrical conductivity increases with the inclusion of both nanofillers for all PVDF copolymers, showing a percolation threshold lower than 0.5 wt.% for CNT/PVDF composites and near 2wt.% for rGO/PVDF5130 composites. The conductive network formed by the CNT is more effective than for rGO for similar filler contents into the PVDF matrix, the intrinsic conductivity of the CNT being higher than the one of the rGO nanofillers. Further, the aspect ratio of the CNT is larger than the one of the 2D materials, leading to lower percolation threshold composites. The conductivity of the neat polymers is in agreement with the manufacturer specifications and is similar among PVDF copolymers [ 11 , 42 , 43 ]. Among CNT nanocomposites, PVDF6010 shows the higher electrical conductivity for filler content between 0.25 and 0.5 wt.% when compared to PVDF-HFP and 5310 composites. The electrical conductivity for CNT/6010 is higher for 0.25 wt.% CNT (near 2 orders of magnitude) but for composite with 0.5 wt.% filler content the electrical conductivity is in the same order of magnitude for all polymer matrices. Their percolation threshold is thus around 0.5 wt.% CNT and the maximum conductivity is σ≈ 5 × 10 −1 ( Ω .m) −1 , as reported in literature for CNT/polymer materials [ 43 , 44 ]. It is to notice that percolation thresholds below 0.1 wt.% of CNT have been reported for PVDF matrix composites [ 44 ], which is not verified in the present work. Materials 2019, 11, x FOR PEER REVIEW 7 of 16 required in order not to affect other properties of the polymer, such as thermal or mechanical. In terms of higher conductivity and lower threshold, CNT appears as more interesting nanocarbonaceous filler than rGO for conductive polymer nanocomposites. 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 10 -10 10 -8 10 -6 10 -4 10 -2 CNT/6010 CNT/HFP rGO/5130 Conductivity ( Ω .m)-1 Nanofiller content (wt.%) CNT/5130 Figure 3. Electrical conductivity of the fabricated samples as a function of filler type and content. The lines are for guiding the eyes. 3.3. Mechanical Measurements Mechanical measurements were performed to evaluate the stress-strain response of the several PVDF polymers and composites, as a function of the filler content and type. As it is shown in Figure 4, CNT tends to reinforce the PVDF matrix, leading to a higher stiffness for the composites when compared to neat matrix [45–48]. However, at the higher CNT contents, maximum stress increases, while the maximum strain of the composites decreases. This effect is ascribed to the heterogeneity of the composites caused by filler agglomeration [49] that limits the mechanical strain for all PVDF matrices, PVDF-HFP, 6010, and 5130. The maximum stress of the CNT/PVDF samples, near the yielding of the PVDF and composites, increases with increasing CNT content. This means that the CNT effectively mechanically reinforces the composites, as presented in Table 2, and that the CNT agglomerates can act as mechanical interlocking between polymer chains and the filler [49]. In fact, it has been shown in different graphene/PVDF [12] and CNT/PVDF [15] composites with low filler content into the PVDF matrix, that their presence do not influence the spherulitic size and the kinetics of crystallization, the cross-section SEM images demonstrating a homogeneous dispersion of the nanocarbonaceous fillers, independently of the filler type and content [12]. Morphological analysis of the nanocarbonaceous/polymer composites have been intensively studied in literature [7,50,51], the present results being in agreement with the reported literature (data not shown). Among the different polymers, PVDF6010 shows the highest yield strength (larger than 40 MPa) and PVDF-HFP the lowest yield stress (lower than 20 MPa), being inversely for strain at rupture, where PVDF-HFP shows the larger strain, near 70%, and the 6010 rupture is near 12%. At rupture, composites with CNT embedded in PVDF-HFP and 5130 present larger maximum strain, reaching 200% of strain for 025CNT/5130 and 300% for 025CNT/HFP samples. In all the composite samples, except the ones including rGO, the inclusion of CNT fillers yields to higher stiffness and elongation at break. From the behavior observed in the rGO/PVDF samples, it can be concluded that the matrixfiller wetting is considerably weaker than the PVDF-CNT one, leading to brittle fracture of the Figure 3. Electrical conductivity of the fabricated samples as a function of filler type and content. The lines are for guiding the eyes. With respect to rGO/5130 nanocomposites, they show lower conductivity when compared to CNT at the same filler content, which is attributed to the lower aspect ratio and intrinsic conductivity of rGO when compared to CNT, leading to an increase of percolation threshold of the composite. The rGO/5130 up to 1 wt.% filler content shows similar electrical conductivity than neat PVDF5130 and the percolation threshold is ≈2 wt.% rGO, with an electrical conductivity of σ≈1×10−5(Ω.m)−1. The intrinsic properties of CNT lead to low percolation thresholds in polymer-based composites. To tailor polymer-based nanomaterials with functional properties, low nanofiller content is typically required in order not to affect other properties of the polymer, such as thermal or mechanical. In terms of higher conductivity and lower threshold, CNT appears as more interesting nanocarbonaceous filler than rGO for conductive polymer nanocomposites. Materials 2019,12, 3545 7 of 14 3.3. Mechanical Measurements Mechanical measurements were performed to evaluate the stress-strain response of the several PVDF polymers and composites, as a function of the filler content and type. As it is shown in Figure 4, CNT tends to reinforce the PVDF matrix, leading to a higher stiffness for the composites when compared to neat matrix [ 45 – 48 ]. However, at the higher CNT contents, maximum stress increases, while the maximum strain of the composites decreases. This effect is ascribed to the heterogeneity of the composites caused by filler agglomeration [49] that limits the mechanical strain for all PVDF matrices, PVDF-HFP, 6010, and 5130. The maximum stress of the CNT/PVDF samples, near the yielding of the PVDF and composites, increases with increasing CNT content. This means that the CNT effectively mechanically reinforces the composites, as presented in Table 2, and that the CNT agglomerates can act as mechanical interlocking between polymer chains and the filler [ 49 ]. In fact, it has been shown in different graphene/PVDF [ 12 ] and CNT/PVDF [ 15 ] composites with low filler content into the PVDF matrix, that their presence do not influence the spherulitic size and the kinetics of crystallization, the cross-section SEM images demonstrating a homogeneous dispersion of the nanocarbonaceous fillers, independently of the filler type and content [ 12 ]. Morphological analysis of the nanocarbonaceous/polymer composites have been intensively studied in literature [ 7 , 50 , 51 ], the present results being in agreement with the reported literature (data not shown). Table 2. Mechanical parameters obtained from the stress-strain measurements for the different PVDF matrices and the corresponding composites with CNT and rGO. Sample Initial Modulus (MPa) Strain at Rupture (%) Stress at Rupture (MPa) Yield Strain (%) Yield Stress (MPa) PVDF-HFP 356 ±15 70.8 ±4 15.5 ±3 12.7 ±3 18.1 ±4 025CNT/HFP 372 ±16 328.2 ±16 17.4 ±4 19.3 ±5 19.1 ±4 05CNT/HFP 439 ±18 309.5 ±15 22.4 ±4 18.3 ±4 22.3 ±5 PVDF6010 1065 ±45 11 ±2 42.8 ±8 9.6 ±2 44.7 ±10 025CNT/6010 1293 ±49 11.1 ±2 49.5 ±10 8.9 ±2 51.9 ±11 05CNT/6010 1388 ±51 13.6 ±3 51.9 ±10 8.3 ±2 55.9 ±12 PVDF5130 870 ±40 23.4 ±5 27.3 ±6 9.8 ±2 34.4 ±7 025CNT/5130 863 ±40 212.9 ±13 30.0 ±6 13.3 ±3 42.0 ±8 05CNT/5130 1244 ±54 99.2 ±5 32.0 ±7 13.2 ±3 43.6 ±9 1CNT/5130 1220 ±53 85.9 ±4 38.6 ±8 14.4 ±4 46.6 ±9 PVDF5130 870 ±41 23.4 ±4 27.3 ±6 9.8 ±2 34.4 ±7 05rGO/5130 1151 ±52 12.3 ±3 39.6 ±8 8.7 ±2 43.2 ±9 1rGO/5130 1327 ±55 11.9 ±3 53.4 ±11 8.9 ±2 56.1 ±11 2rGO/5130 1265 ±54 4.5 ±1 44.1 ±9 4.5 ±1 44.1 ±10 Materials 2019, 11, x FOR PEER REVIEW 8 of 16 samples. In fact, literature has reported on the decreasing of the maximum strain with low graphene content as reinforcement for PVDF 1010 [12]. 0 50 100 150 200 250 300 350 0 5 10 15 20 25 05CNT/HFP 025CNT/HFP Stress (MPa) Strain (%) PVDF-HFP A 0 5 10 15 0 10 20 30 40 50 60 Stress (MPa) 05CNT/6010 025CNT/6010 PVDF6010 Strain (%) B 0 50 100 150 200 0 10 20 30 40 50 025CNT/5130 Stress (MPa) 05CNT/5130 1CNT/5130 PVDF5130 Strain (%) C 0 5 10 15 20 25 0 10 20 30 40 50 60 Stress (MPa) 1rGO/5130 2rGO/5130 05rGO/5130 PVDF5130 Strain (%) D Figure 4. Stress-strain response for PVDF and the corresponding composites for (A) PVDF-HPF, (B) PVDF6010 and (C) PVDF5010 for different CNT contents. (D) PVDF5130 reinforced with rGO. The initial modulus for neat PVDF is larger for 6010 near E ≈ 1 GPa being E ≈ 870 and E ≈ 350 MPa for 5310 and HFP, respectively. Reinforced with CNT, all PVDF matrices increase the initial modulus with increasing CNT content, excepting for the 1CNT/5310 sample. Similar behavior is found in rGO/5310 composites, where the initial modulus increases with filler content up to 1 wt.%, decreasing for samples with 2 wt.%. The 05CNT/6010 sample shows an initial modulus near E ≈ 1.4 GPa. The larger initial modulus and yield stress of the PVDF 6010 composites are correlated with the percolation threshold, which depends on the filler dispersion, but also rigidity of the matrix [52], decreasing the threshold below 0.5 wt.%. Figure 4. Cont. Materials 2019,12, 3545 8 of 14 Materials 2019, 11, x FOR PEER REVIEW 8 of 16 samples. In fact, literature has reported on the decreasing of the maximum strain with low graphene content as reinforcement for PVDF 1010 [12]. 0 50 100 150 200 250 300 350 0 5 10 15 20 25 05CNT/HFP 025CNT/HFP Stress (MPa) Strain (%) PVDF-HFP A 0 5 10 15 0 10 20 30 40 50 60 Stress (MPa) 05CNT/6010 025CNT/6010 PVDF6010 Strain (%) B 0 50 100 150 200 0 10 20 30 40 50 025CNT/5130 Stress (MPa) 05CNT/5130 1CNT/5130 PVDF5130 Strain (%) C 0 5 10 15 20 25 0 10 20 30 40 50 60 Stress (MPa) 1rGO/5130 2rGO/5130 05rGO/5130 PVDF5130 Strain (%) D Figure 4. Stress-strain response for PVDF and the corresponding composites for (A) PVDF-HPF, (B) PVDF6010 and (C) PVDF5010 for different CNT contents. (D) PVDF5130 reinforced with rGO. The initial modulus for neat PVDF is larger for 6010 near E ≈ 1 GPa being E ≈ 870 and E ≈ 350 MPa for 5310 and HFP, respectively. Reinforced with CNT, all PVDF matrices increase the initial modulus with increasing CNT content, excepting for the 1CNT/5310 sample. Similar behavior is found in rGO/5310 composites, where the initial modulus increases with filler content up to 1 wt.%, decreasing for samples with 2 wt.%. The 05CNT/6010 sample shows an initial modulus near E ≈ 1.4 GPa. The larger initial modulus and yield stress of the PVDF 6010 composites are correlated with the percolation threshold, which depends on the filler dispersion, but also rigidity of the matrix [52], decreasing the threshold below 0.5 wt.%. Figure 4. Stress-strain response for PVDF and the corresponding composites for ( A ) PVDF-HPF, ( B ) PVDF6010 and (C) PVDF5010 for different CNT contents. (D) PVDF5130 reinforced with rGO. Among the different polymers, PVDF6010 shows the highest yield strength (larger than 40 MPa) and PVDF-HFP the lowest yield stress (lower than 20 MPa), being inversely for strain at rupture, where PVDF-HFP shows the larger strain, near 70%, and the 6010 rupture is near 12%. At rupture, composites with CNT embedded in PVDF-HFP and 5130 present larger maximum strain, reaching 200% of strain for 025CNT/5130 and 300% for 025CNT/HFP samples. In all the composite samples, except the ones including rGO, the inclusion of CNT fillers yields to higher stiffness and elongation at break. From the behavior observed in the rGO/PVDF samples, it can be concluded that the matrix-filler wetting is considerably weaker than the PVDF-CNT one, leading to brittle fracture of the samples. In fact, literature has reported on the decreasing of the maximum strain with low graphene content as reinforcement for PVDF 1010 [12]. The initial modulus for neat PVDF is larger for 6010 near E ≈ 1 GPa being E ≈ 870 and E≈350 MPa for 5310 and HFP, respectively. Reinforced with CNT, all PVDF matrices increase the initial modulus with increasing CNT content, excepting for the 1CNT/5310 sample. Similar behavior is found in rGO/5310 composites, where the initial modulus increases with filler content up to 1 wt.%, decreasing for samples with 2 wt.%. The 05CNT/6010 sample shows an initial modulus near E ≈ 1.4 GPa. The larger initial modulus and yield stress of the PVDF 6010 composites are correlated with the percolation threshold, which depends on the filler dispersion, but also rigidity of the matrix [52], decreasing the threshold below 0.5 wt.%. The strain at rupture for composites with PVDF 6010 is similar for the different CNT contents, increasing in the remaining PVDF composites, from 70 to 300% in HFP and 23 to 210% in 5130. PVDF5130 with rGO filler suffers a decrease of the strain at rupture from 23 to 4.5%, from neat polymer to the composite with 2 wt.% rGO, respectively. The yield stress and strain are comparable for both fillers (CNT and rGO) in PVDF 5130 materials. The load transfer efficiency and interfacial shear stress of the composites as a function of the dimensions of the fillers have been theoretically calculated [ 53 ]. The CNT and rGO fillers employed in the present work show average lengths of 1.5 µ m and 1 to 5 µ m for CNT and rGO, respectively. The CNT diameter is 9.5 nm and the rGO thickness is 1–2 layers that correspond to 2–3 nm [ 54 ]. Thickness and length of the employed nanocarbonaceous fillers are similar, being different the width of both materials, from 9.5 nm to 1–5 µ m, respectively, for CNT and rGO. In this way, theoretical and experimental results are in agreement, the larger nanofillers leading to better reinforcement of the composites [ 55 ], as observed in the rGO/5130 composites, which show slightly higher initial modulus when compared to CNT/5130 composites. Those results reflect not just the different mechanical characteristics of the polymers but also the different wettability between polymer and fillers, depending on macromolecule characteristics. 3.4. Electromechanical Measurements PVDF composites with CNT or rGO as filler are excellent candidates for electromechanical sensors, leading to high sensibility composites [ 11 , 24 ]. Electromechanical compression tests (in Figure 5) were Materials 2019,12, 3545 9 of 14 performed in the composites around the percolation threshold to evaluate the sensitivity and linearity of the different materials. The mechanical stress-strain cycles applied to the samples up to 3.45 MPa are shown in Figure 6. It can be observed a slight nonlinear response and that the mechanical hysteresis is very low for the 05CNT/5130 composites, being similar for the different nanocomposites. Materials 2019, 11, x FOR PEER REVIEW 10 of 16 0.00 0.03 0.06 0.09 0.12 0.15 0.0 0.6 1.2 1.8 2.4 3.0 3.6 0.00 0.03 0.06 0.09 0.0 0.6 1.2 1.8 Stress (MPa) Strain (mm) 0.43 MPa 0.86 MPa 1.72 MPa 3.45 MPa Stress (MPa) Strain (mm) 0.43 MPa 0.86 MPa 1.72 MPa 3.45 MPa Figure 5. Stress-strain mechanical response for the 05CNT/PVDF composite, as representative for the rest of the nanocomposites. Figure 6 shows 10 loading-unloading tests performed at a maximum load of 3.5 MPa (400 N of force) for the different composites with 0.5 wt.% CNT content and the 2rGO/5130 sample. Electromechanical tests show good linearity between electrical resistance variation and applied pressure for the different matrices and fillers, as a function of the pressure. Similar electromechanical linearity and cycling performance have been reported in [56,57] with different materials and experimental approaches. 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Pressure (MPa) A 05CNT/HFP 0 20406080100120140160 0.0 0.2 0.4 0.6 0.8 Δ R/R 0 Pressure Δ R/R 0 Time (s) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Pressure ΔR/R0 Time (s) B 0 20 40 60 80 100 120 140 160 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Δ R/ R 0 Pressure (MPa) 05CNT/6010 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0 20 40 60 80 100 120 140 160 180 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 05CNT/5130 Pressure Δ R/R0 Pressure (MPa) Δ R/R 0 Time (s) C 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Pressure (MPa) D 0 20 40 60 80 100 120 140 160 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 d) 2rGO/5130 Pressure Δ R/R 0 Time (s) Δ R/R0 Figure 5. Stress-strain mechanical response for the 05CNT/PVDF composite, as representative for the rest of the nanocomposites. Materials 2019, 11, x FOR PEER REVIEW 10 of 16 0.00 0.03 0.06 0.09 0.12 0.15 0.0 0.6 1.2 1.8 2.4 3.0 3.6 0.00 0.03 0.06 0.09 0.0 0.6 1.2 1.8 Stress (MPa) Strain (mm) 0.43 MPa 0.86 MPa 1.72 MPa 3.45 MPa Stress (MPa) Strain (mm) 0.43 MPa 0.86 MPa 1.72 MPa 3.45 MPa Figure 5. Stress-strain mechanical response for the 05CNT/PVDF composite, as representative for the rest of the nanocomposites. Figure 6 shows 10 loading-unloading tests performed at a maximum load of 3.5 MPa (400 N of force) for the different composites with 0.5 wt.% CNT content and the 2rGO/5130 sample. Electromechanical tests show good linearity between electrical resistance variation and applied pressure for the different matrices and fillers, as a function of the pressure. Similar electromechanical linearity and cycling performance have been reported in [56,57] with different materials and experimental approaches. 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Pressure (MPa) A 05CNT/HFP 0 20406080100120140160 0.0 0.2 0.4 0.6 0.8 Δ R/R 0 Pressure Δ R/R 0 Time (s) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Pressure ΔR/R0 Time (s) B 0 20 40 60 80 100 120 140 160 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Δ R/ R 0 Pressure (MPa) 05CNT/6010 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0 20 40 60 80 100 120 140 160 180 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 05CNT/5130 Pressure Δ R/R0 Pressure (MPa) Δ R/R 0 Time (s) C 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Pressure (MPa) D 0 20 40 60 80 100 120 140 160 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 d) 2rGO/5130 Pressure Δ R/R 0 Time (s) Δ R/R0 Figure 6. Electromechanical performance of the ( A ) PVDF-HFP, ( B ) 6010, and ( C ) 5130 with 0.5 wt.% of CNT for 10 cycles from unloading to 3.5 MPa of pressure. ( D ) PVDF 5130 reinforced with 2 wt.% CNT in cycles up to 3.5 MPa. Figure 6shows 10 loading-unloading tests performed at a maximum load of 3.5 MPa (400 N of force) for the different composites with 0.5 wt.% CNT content and the 2rGO/5130 sample. Electromechanical tests show good linearity between electrical resistance variation and applied pressure for the different matrices and fillers, as a function of the pressure. Similar electromechanical linearity and cycling performance have been reported in [56,57] with different materials and experimental approaches.