Abstract

The ability of filtration and separation media containing fibres to remove impurities from oil, water, and blood can be enhanced using magnetic fields. The ability to regulate the dielectric and magnetic behaviour of fibrous webs in terms of superparamagnetic or ferromagnetic properties by adjusting material composition is fundamental to meeting end-use requirements. Electrospun fibres were produced from PVdF (polyvinylidene fluoride) and nanomagnetite (Fe3O4 nanoparticles) from solutions of PVdF in dimethylacetamide containing Fe3O4 nanoparticle contents ranging from 3 to 10 wt%. Fibre dimensions, morphology, and nanoparticle agglomeration were characterised by environmental scanning electron microscopy (ESEM) and field emission gun transmission electron microscopy (FEGTEM). Dielectric behaviour of the fibre webs was influenced by web porosity and the Fe3O4 nanoparticle content. Impedance analysis of the webs indicated an increase in dielectric constant of 80% by the addition of 10 wt% Fe3O4 nanoparticles compared to 100 wt% PVdF. The dielectric constants of the webs were compared with those obtained from the theoretical mixing models of Maxwell and Lichtenecker. Vibrating sample magnetometer (VSM) magnetisation measurements indicated a blocking temperature above 300 K suggesting ferrimagnetic rather than superparamagnetic behaviour as a result of Fe3O4 nanoparticle agglomeration within fibres.

1. Introduction

Magnetic materials in the form of submicron diameter fibres can be used in high performance filtration and separation media to remove impurities from liquids such as oil, water, and blood. To support industrial development, it is important to understand how to control the dielectric and magnetic behaviour of the submicron fibres so that the filtration or separation device is able to function satisfactorily.

Currently, submicron fibres can be produced by various methods including fibrillation [1, 2], splitting of bicomponent fibres [3], melt blowing [4], electrospinning [57], centrifugal and force spinning, or combined processes [810]. Of these methods, electrospinning has been widely used to convert different thermoplastic engineering polymers into submicron and nanofibres [6, 11].

Processes such as sol-gel assisted electrospinning or polymer-nanocomposite electrospinning can be used to manufacture magnetic fibres. The sol-gel route has been used to prepare composite fibres of nickel ferrite [12], Ni0.5Zn0.5Fe2O4 [13], and substituted strontium ferrite SrRe0.6Fe11.4O19 (Re = La, Ce) [14]. Magnetic copper ferrite fibres prepared by electrospinning have also been reported [15] and Wu et al. [16] reported electrospun ferromagnetic fibres based upon Fe, Co, and Ni with diameters of 25 nm.

Using the polymer-nanocomposite route, a polymer matrix is combined with appropriate particulate fillers such as Fe3O4 nanoparticles to prepare films and fibres. Previously investigated polymer matrices for embedding Fe3O4 nanoparticles include polyvinyl alcohol (PVA) [17, 18], polymethyl methacrylate (PMMA) [19], polyacrylonitrile (PAN) [20], polyvinylpyrrolidone (PVP) [21], polyethylene oxide (PEO) [22], and PVdF [2325]. Specifically in relation to the manufacture of fibres, Barrera et al. [22] reported the fabrication of PEO fibres containing discrete magnetic domains using a modified electrospinning apparatus. Magnetite and cobalt ferrite nanoparticles were synthesized and coated with thiolated PEO and then suspended in a 1-2 wt% PEO-in-water solution. Polymer-nanocomposite fibres with diameters as small as 200 nm were then produced by electrospinning. PVdF has attracted considerable research interest in the preparation of submicron diameter fibre assemblies because of its ionic conductivity, electrochemical stability, and piezo-electric properties [26, 27], enabling fibrous polymer electrolytes and battery separators to be fabricated by electrospinning [28, 29]. As a relatively inert material, PVdF is extensively utilised by industry in the manufacture of membrane filters including hollow fibre membranes for liquid filtration. Although previous research has been conducted on thin films, only a few studies have been reported on the properties of electrospun PVdF and PVdF copolymer fibres containing Fe3O4 nanoparticles in respect to superparamagnetic or ferrimagnetic behaviour [23, 25, 30]. Yang et al. [23, 30] prepared electrospun webs from spinning solutions of 20 wt% of polyvinylidene fluoride-co-tetrafluoroethylene (PVdF-TeFE) dissolved into a mixture of dimethylformamide and 2-butanone with 5 wt% and 10 wt% of dispersed 20–30 nm Fe3O4 nanoparticles. Results were reported for the resulting fibre webs that showed that the Fe3O4 nanoparticle content influenced dielectric properties, but the data pertained to a copolymer of PVdF-TeFE containing 19% of the TeFE monomer rather than pure PVdF.

Accordingly, the goal of this paper is to generate a more detailed understanding of the dielectric and specific magnetic saturation properties of pure PVdF and Fe3O4 electrospun fibre webs that contain different nanoparticle loadings and to determine the link between measured and calculated dielectric values, specifically in relation to porous media, rather than films. This data is required to enable the assembly of improved filter and separation media in the future.

2. Materials and Methods

2.1. Materials

A medium viscosity (Mw = 352 × 103 g mol−1) polyvinylidine fluoride (PVdF) homopolymer was industrially sourced (1010 grade, Solef, Tavaux, France). Nonmagnetised nanoparticles of ≥98% Fe3O4 determined by trace metal analysis (Fe3O4, Sigma-Aldrich, Poole, UK) had a mean particle size of 28.4 nm with minimum and maximum particle sizes of 10 nm and 50 nm, respectively, as determined by TEM. These particles were free from ligands or stabilising ions over the surface. Spinning solutions were prepared from dimethylacetamide (DMAc) (Sigma-Aldrich, Poole, UK), which is a known solvent for PVdF.

2.2. Preparation of Linear Nanocomposite Fibre Webs

A 20 wt% PVdF solution in DMAc was prepared into which Fe3O4 nanoparticles of 3 wt%, 5 wt%, and 10 wt% were dispersed. Spinning solutions were sonicated for 60 min at a frequency of 40 KHz to reduce nanoparticle agglomeration. Electrospinning was conducted using a polymer flow rate of 0.6 mL/hr at an applied voltage of 20 kV with a capillary tip to collector distance of 0.13 m. The fibres were spun onto a target of 0.1 m × 0.1 m aluminium foil to form a self-supporting web. The ambient temperature and humidity within the fume cupboard were 20°C and 60%, respectively. Environmental scanning electron microscopy (Phillips XL-30 ESEM, Eindhoven, Netherlands) was utilised to verify the uniformity of the as-spun fibres and freedom from structural defects such as beads and fibre bundles.

2.3. Dimensional Characterisation of Nanoparticles

The nanoparticles and as-spun fibres were probed using a thermally assisted field emission gun (FEG) TEM (TECNAI TF20, FEI, Eindhoven, Netherlands) operating at 197 kV (point resolution of 0.24 nm) equipped with a UTW Oxford Instruments EDX detector. Mean nanoparticle size distributions were determined directly from the TEM images. After dimensional calibration, measurement of nanoparticle sizes was conducted by image analysis with the inbuilt scale option (Image Pro V 6.2) and the results were directly exported to Microsoft Excel.

2.4. Web Porosity

Web porosity was approximated from ESEM (Phillips XL-30 ESEM) micrographs of each web sample. Binary images (.bmp format) were prepared by image thresholding such that porosity (P) could then be determined from the mean intensity of the image, distinguishing between the solid and air fractions of the web represented by either black or white pixels (1): where n is the number of white pixels (pores) and N is the total number of pixels in the binary image. Micrographs were obtained at an accelerating voltage of 20 kV with a spot size of 4 nm and a working distance of 5 mm.

2.5. Dielectric Behaviour

The dielectric constant for each sample was calculated using where is the capacitance, d is the plate separation, A is area of the plates, is the dielectric constant of free space (8.854 ×  F/m), and is the dielectric constant of the bulk structure. The capacitance of each electrospun web sample was measured by an impedance analyser (Solartron 1260, Solartron Analytical, Farnborough, UK) at room temperature. The plate separation was determined by the mean thickness of the web as measured directly by ESEM. Electrospun samples of 100 wt% PVdF and PVdF polymer nanocomposites containing 3 wt%, 5 wt%, and 10 wt% of Fe3O4 nanoparticles were analysed.

Corresponding dielectric constants were also calculated based upon Maxwell’s rule of mixtures [31] (see, (3)) and Lichtenecker’s logarithmic mixing model [32] (see (4)). Of the various mixture models available those of Maxwell and Lichtenecker have been reported to be particularly applicable to porous materials such as electrospun webs [33]. Other models including those of Bruggeman and Maxwell-Garnett [34] are applicable for composite materials containing spherical inclusions. The Maxwell model has been previously applied to materials with a uniformly connected pore structure, while the Lichtenecker model is claimed to be more applicable to those that are randomly connected [32]. Since the electrospun webs produced in the present work were expected to be porous and randomly oriented both models were evaluated. The Maxwell mixture equation can be written as where and are the volume fractions of the solid material and voids (air), respectively, and and are the corresponding dielectric constants for these components. The Lichtenecker logarithmic equation may be expressed as where is the dielectric constant of the entire material, , , are the volume fractions of the polymer, iron oxide, and voids, respectively, and , , are the corresponding dielectric constants. The void fraction (porosity) was determined from the weight and thickness of the web. The thickness was measured directly from sections of the web observed by ESEM.

2.6. Magnetic Behaviour

Electrospun webs were characterised by means of a vibrating sample magnetometer (VSM) (Oxford Instruments, Abingdon, UK) operating at 55 Hz with vibration amplitude of 1.5 mm. The magnetic field is generated by a pair of superconducting coils immersed in a liquid helium reservoir, and the sample temperature was controlled via a PID loop with a heating element and a valve connecting the sample space with the reservoir. Typical noise levels were below 10 μemu or 1% of the signal, with temperature stability better than 50 mK and a maximum sample moment of the order of several hundred memu.

3. Results and Discussion

3.1. Incorporation of Fe3O4 Nanoparticles

All three (3 wt%, 5 wt%, and 10 wt%) Fe3O4 nanoparticle contents were found to disperse in the 20 wt% PVdF polymer solution such that electrospun webs could be produced containing continuous fibres with few discernible bead defects (Figure 1). Previously it has been reported that fibre diameter and fibre morphology can be sensitive to increasing Fe3O4 nanoparticle content in electrospun fibres. Yang et al. [23, 30] reported a decrease in mean diameter of PVdF-TeFE/Fe3O4 electrospun fibres as the nanoparticle content increased. The same trend was observed in the PVdF/Fe3O4 electrospun samples produced in the production of the present set of samples [35]. Furthermore Wang et al. [36] have shown that the morphology and diameter of PAN/Fe3O4 (particle size = 8 nm) electrospun fibres were strongly dependent upon the PAN concentration and salt additives in the spinning solution. In Figure 1, it is evident that the continuous PVdF/Fe3O4 fibres obtained did not appreciably change in terms of morphology as the fraction of Fe3O4 nanoparticles increased from 0 to 10 wt%.

Prior to electrospinning, the mean size of the Fe3O4 nanoparticles was 28.4 nm. Electrospun samples were analysed to determine the degree to which mean particle size within the fibre corresponded to that of the original size prior to spinning. TEM analysis revealed distributed agglomerations of Fe3O4 nanoparticles with a mean diameter of 52.7 nm and 38.2 nm for unsonicated and sonicated spinning solutions, respectively (Table 1). Individual agglomeration sizes up to 93 nm were also observed some of which extended beyond the external boundaries of the fibres, creating irregular fibre morphology (Figure 2). Uneven fibre morphologies and difficulties in dispersing ca. 8 nm diameter Fe3O4 nanoparticles in PVA nanofibres were also reported by Chung et al. [17]

The problem of nanoparticle agglomeration is a well-known phenomenon leading to problems such as reduced fibre tensile strength. Minimisation of nanoparticle agglomeration is not straightforward [3741]. In relation to nanoparticles that can be magnetised, approaches for improving nanoparticle dispersion include sonication of the polymer solution [42, 43], silica addition [44], steric functionalisation of magnetite in an organic solvent [19], and the placement of magnets in close proximity to the collector during electrospinning [45]. Silica aids magnetic nanoparticle dispersion by reducing the interparticle attraction (residual magnetic attraction) between the nanoparticles but after spinning it remains as an unwanted contaminant within the fibres. The propensity to agglomerate is influenced by the viscosity of polymer solution and by surface functionalisation. Since the viscosity of the PVdF polymer solution was relatively high (1200 cP), particles could be retained in suspension for 2 to 3 hr and the nanoparticles were not premagnetised or surface-functionalised. In the present study, no stabilising agents were used and extended sonication (t = 60 min) was adopted to promote nanoparticle dispersion in the solution. Before electrospinning, the Fe3O4 nanoparticles sizes varied from 10 nm to 50 nm with a mean size of 28.4 nm. Sonication of the spinning solution reduced the mean Fe3O4 particle size in the as-spun fibres from 52.7 nm to 39.2 nm (Table 1). Sonication resulted in improved dispersion of particles, effectively redistributing the particles throughout the fibre volume (Figure 3).

In terms of fibre morphology, nanoparticle agglomeration is normally considered disadvantageous, but in the present work it is conceivable that agglomeration could also influence the ferrimagnetic properties of samples by holding greater magnetic moment than a single particle. Thus, the degree of agglomeration may be utilised as a means of controlling dielectric and magnetic behaviour in the fibre web. Agglomeration of nanoparticles can be affected by the use of high filler concentrations even when the fillers are coated to prevent mutual interaction [46]. Interparticle attraction and agglomeration can also be influenced by magnetic behaviour. Superparamagnetism in ferrites below a critical diameter has previously been reported to prevent self-agglomeration [47].

3.2. Dielectric Constant

The dielectric constant of Solef PVdF polymer ranges from four to twelve in the frequency range 100 Hz to 100 kHz. Figure 4 shows that markedly lower mean dielectric constants were obtained for the as-spun electrospun PVdF webs of 1.25 (100 wt% PVdF)–2.25 (PVdF/10 wt% Fe3O4) compared to the solid PVdF polymer. This is due to the high porosity of the webs ( to 0.83, Figure 1). It is evident that the dielectric constant of electrospun PVdF/Fe3O4 webs will be sensitive to large changes in web porosity. This is because the webs can be viewed as multiphase materials consisting of three phases: polymer, nanoparticles, and a large void or air fraction. As is the case with the majority of nonwoven assemblies the void fraction has a large influence on the dielectric behaviour and industrially, control of void fraction is a commonly utilised means for modifying the dielectric behaviour of electrically conductive materials. The influence of porosity on the dielectric constant of electrospun PVdF/Fe3O4 nanoparticle fibre webs is illustrated in Figure 5 based upon the theoretical mixing models of (3) and (4). Linear and exponential relationships between dielectric constant and porosity were obtained for the Maxwell and Lichtenecker models, respectively (Figure 5), which are in accordance with the results obtained for nanocomposite films reported by Nunomura et al. [48].

The experimental dielectric constant for each electrospun web was compared with corresponding calculated data obtained from the mixing models of Maxwell (see (3)) and Lichtenecker (see (4)). The dielectric constant of the bulk PVdF material was found to be 9.0 at 10 kHz, and therefore all calculated values were compared with the corresponding measured data at 10 kHz. The relationship between the experimental and theoretical data was of the form (Figure 6):

The absolute dielectric constants obtained from the Lichtenecker model were slightly lower than that of Maxwell and were found to be in closest agreement to the experimental data. Previously, the Maxwell model has been reported to be well suited to materials with uniformly distributed spherical pores [32] as compared to the more randomly connected pores present in the as-spun webs. The experimentally observed trend of increasing dielectric constant with increasing Fe3O4 content (Figure 3) was also found to be in reasonable agreement with the model of Lichtenecker as compared to that of Maxwell (Figure 7).

In the study of the dielectric behaviour of PVdF-TeFE/Fe3O4 electrospun webs, Yang [23] also found reasonable agreement with the model of Lichtenecker. Note that the dielectric constant and electrical conductivity of Fe3O4 nanoparticles at high frequency differ from the bulk values obtained for the web because of the insulating environment of the surrounding voids. However, the dielectric constant can be expected to remain roughly constant with small electrical losses up to several GHz in the microwave region [49].

3.3. Magnetic Behaviour

Zero field cooled-field cooled measurements at 25 mT revealed the typical magnetisation response of magnetic nanoparticles with a blocking temperature above 300 K (Figure 8). In agreement with the hysteresis loops measured at 285 K, this indicates that the material remained ferrimagnetic, rather than superparamagnetic, at room temperature.

This is in agreement with the measured coercive field of some 8 mT (see inset in Figure 9). The corresponding particle size would be anticipated to be of the order of 40 nm [50]. This further suggests that there was agglomeration and dipolar interaction between the nanoparticles in the constituent fibres, and the estimated 40 nm particle size based on the VSM data was in good agreement with the mean particle size value of 39.2 nm obtained from the TEM measurements. The VSM results in Figure 9 indicate the values of specific magnetic saturation of PVdF and the electrospun materials containing 3–10 wt% Fe3O4. As expected, the increase in saturation magnetisation was roughly linear with the Fe3O4 content. The specific magnetic saturation was relatively low for these nanoparticles when compared with the bulk because of the dipolar interaction and the formation of larger magnetic domains in bulk.

Previously, Fe3O4 nanoparticles have been found to be single domain below 50 nm and not superparamagnetic until <23 nm [51]. Superparamagnetism has been observed in electrospun polymer nanocomposite fibres of PVdF-TeFE with embedded Fe3O4 [22] particles of 20–30 nm diameter. Similarly, work by Wang et al. [36] on electrospun PAN nanofibre webs containing 8 nm Fe3O4 nanoparticles demonstrated superparamagnetic behaviour at room temperature. PVP-coated Fe3O4 nanoparticles electrospun directly to make nanofibre webs with constituent fibre diameters of 200–400 nm displayed superparamagnetic behavior and the Fe3O4 nanoparticles showed a saturation magnetisation of 36.6 emu/g [52].

The increase in embedded mean particle size within the fibres due to agglomeration is likely to mitigate against a transition from ferrimagnetic to superparamagnetic behavior. The increased mean size of the nanoparticles as a result of agglomeration therefore had the effect of maintaining residual ferrimagnetism in the electrospun PVdF/Fe3O4 web assembly. The potential for relatively high magnetic field strength in ferrimagnetic PVdF/Fe3O4 webs could provide significant potential for their integration within cost-effective filtration and separation media to increase the efficiency of suspended particle removal from fluid streams.

4. Conclusions

The purpose of this study was to develop a detailed understanding of the dielectric and specific magnetic saturation properties of pristine PVdF and Fe3O4 electrospun fibre webs to inform the assembly of improved liquid filtration and separation media in the future. The dielectric behaviour of the fibre web was found to be influenced by the Fe3O4 nanoparticle content and the overall web porosity. PVdF/Fe3O4 nanocomposite fibre webs were prepared by electrospinning using a dimethylacetamide solvent system. The Fe3O4 nanoparticles were successfully embedded into the fibres, but particle agglomerations of up to 90 nm diameter were observed. Agglomeration of the Fe3O4 nanoparticles in the fibres promoted ferrimagnetic rather than superparamagnetic behaviour in the fibre web and the magnetic field saturation increased linearly by increasing the Fe3O4 nanoparticle content. The dielectric constant of the electrospun webs increased up to 80% by the addition of a 10 wt% of Fe3O4 nanoparticles and was also influenced by web porosity. Of the Lichtenecker and Maxwell mixing models, the former was found to give values that were in closest agreement with experimentally determined dielectric constants of the electrospun PVdF/Fe3O4 webs.

Conflict of Interests

The authors declare that there is no conflict of interests regarding the publication of this paper.