**Comparative Analysis of Properties of PVA Composites with Various Nanofillers: Pristine Clay, Organoclay, and Functionalized Graphene**

#### **Jin-Hae Chang**

Department of Polymer Science and Engineering, Kumoh National Institute of Technology, Gumi 39177, Korea; changjinhae@hanmail.net

Received: 7 January 2019; Accepted: 18 February 2019; Published: 1 March 2019

**Abstract:** Poly(vinyl alcohol) (PVA) nanocomposites containing three different nanofillers are prepared and compared in terms of their thermal properties, morphologies, and oxygen permeabilities. Specifically, pristine saponite (SPT) clay, hydrophilic organically modified bentonite (OMB), and hexadecylamine-functionalized graphene sheets (HDA-GSs) are utilized as nanofillers to fabricate PVA nanocomposite films. The hybrid films are fabricated from blended solutions of PVA and the three different nanofillers. The content of each filler with respect to PVA is varied from 0 to 10 wt%, and the changes in the properties of the PVA matrices as a function of the filler content are discussed. With respect to the hybrid containing 5 wt% of SPT, OMB, and HDA-GS, each layer in the polymer matrix consists of well-dispersed individual nanofiller layers. However, the fillers are mainly aggregated in the polymer matrix in a manner similar to the case for the hybrid material containing 10 wt% of fillers. In the thermal properties, SPT and OMB are most effective when the filler corresponds to 5 wt% and 7 wt% for HDA-GS, respectively, and the gas barrier is most effective with respect to 5 wt% content in all fillers. Among the three types of nanofillers that are investigated, OMB exhibits optimal results in terms of thermal stability and the gas barrier effect.

**Keywords:** poly(vinyl alcohol); nanocomposite; nanofiller; film

#### **1. Introduction**

Poly(vinyl alcohol) (PVA) is a water-soluble synthetic polymer with high hydrophilicity, biocompatibility, and non-toxicity [1,2]. The high capacity of PVA to simultaneously form both intra- and inter-chain hydrogen bonds make it a unique polymer that can interact with nanofillers such as clay, graphene, and functionalized-graphene [3,4]. For example, incorporation of graphene oxide (GO) into PVA facilitates good dispersion and interfacial interaction due to the presence of OH-bonds at the end due to the interaction of OH bonding. The content of clay can readily affect the thermo-mechanical properties and gas permeability [5,6].

Clay and graphene exhibit high aspect and high load transfer results in the agglomeration of layers in polymer matrix, and this makes dispersion difficult. In order to avoid this practical challenge, it can be exfoliated or organically modified to achieve the desired properties by dispersing nanofillers in compatible solvent via sonification [7].

In the wet state or especially after mild drying, clay layers are distributed and embedded in the PVA gel to yield a true nanoscale hybrid material. However, drying in vacuo can cause the re-aggregation of the clay layers. The steric constraints created by the PVA matrix impede the re-aggregation of the clay layers, and, thus, a few clay layers remain in the dispersed state. Ideally, useful nanocomposites are fabricated to create amorphous domains with uniformly distributed mineral layers. However, the preparation of PVA/clay nanocomposite materials from a solution is challenging due to the re-aggregation of the layers [8].

Clays with sandwich-type structures that typically consist of an octahedral Al sheet and multiple tetrahedral Si sheets are referred to as phyllosilicates [9,10]. There are several types of phyllosilicates including kaolinite, montmorillonite, hectorite, saponite, bentonite, and synthetic mica. In the present study, we selected saponite (SPT) [11,12] and hydrophilic organically modified bentonite (OMB) [13,14] as clays for the synthesis of clay/PVA polymer nanocomposites. Specifically, SPT and bentonite consist of stacked silicate 1-nm-thick sheets with lengths of approximately 165 nm and 68 nm, respectively. Currently, SPT and bentonite are widely employed as reinforcing fillers in polymeric matrices due to their excellent mechanical, electrical, and thermal properties and their low cost [15–17].

Generally, with the exception of SPT, several types of pristine clays are not compatible with most polymers, and, thus, require organic treatments to render them as organophilic. A common method for this type of organic treatment is based on the ion exchange of the cations within the clay with organic ammonium cations [18–20]. Thus, we selected hydrophilic organo-clay OMB as nano-filler for the synthesis of hybrid polymer films.

Graphene tends to aggregate or restack due to its strong stacking tendency and high cohesive energy. Additionally, it is insoluble in a variety of organic solvents due to its significantly hydrophobic nature and high specific surface area of graphene. Therefore, a key challenge in the preparation and processing of graphene-based composites corresponds to the prevention of aggregation. The functionalization of the graphene surface can introduce reactive moieties that disrupt the bundle structure and can potentially obtain individual sheets [21–23]. This type of functionalization involves the attachment of functional moieties to the open ends and walls of graphene to improve the solubility and dispersibility of graphene sheets (GSs) [24]. Hence, an optimal method to achieve a homogeneous graphene dispersion throughout a polymer matrix is the use of functionalized graphene sheets (FGSs) [25] that exhibit improved dispersion in solvents and polymers. Furthermore, covalent functionalization can provide the means to engineer the GS/polymer interface and, thereby, optimize the properties of the composite material.

Typically, traditional composite structures typically contain a significant content (≈ 40 wt%) of a filler bound within a polymer matrix. However, significant changes in the properties of the materials are also possible at low loadings (< 10 wt%) of nano-fillers, such as exfoliated pristine clays, organoclays, and FGSs, in the hybrid materials [26,27]. The improvements in material performance are achieved as a result of the inherent properties of the nanofillers and also by optimizing the dispersion, interface chemistry, and nanoscale morphology. This is completed to utilize the advantages of the tremendous surface area per unit volume exhibited by nano-fillers (the theoretical limits correspond to 760 m2/g for clay [28] and 2,630 m2/g for graphene [29]).

In the present study, we prepared hybrid films containing PVA and an appropriate amount of filler (≤ 10 wt%) and examined their properties as a function of the filler content and type. We examine and compare the thermal properties, morphologies, and oxygen permeation capabilities of PVA nanocomposites containing three different nanofillers, such as pristine clay SPT, hydrophilic organoclay OMB, and hexadecylamine-functionalized GSs (HDA-GSs). The thermal and oxygen barrier properties of the hybrids are also examined as a function of the nanofiller type and content in the PVA polymer matrix. Lastly, we investigate the effects of filler loadings on the morphologies of the PVA hybrid films.

#### **2. Materials and Methods**

#### *2.1. Materials*

The source clays, SPT, and OMB were obtained from Kunimine Ind. Co. (Tokyo, Japan) and Nanomer Co. (Seoul, Korea), respectively. The clays were passed through a 325-mesh sieve to remove impurities to yield an SPT clay with a cationic exchange capacity of 100 meq/100 g and bentonite with a cationic exchange capacity of 145 meq/100 g. Additionally, PVA with >99% saponification (*M*<sup>w</sup> = approximately 89,000–98,000), graphite, and HDA were purchased from Aldrich Chem. Co. These materials were used in an as-received condition. Commercially available solvents were purified via distillation, and common reagents were used without further purification.

#### *2.2. Synthesis of HDA-GS*

Graphene oxide (GO) was synthesized from natural graphite using a multi-step route known as the Hummers method [30]. Furthermore, HDA-GS was synthesized from hexadecylamine (HDA) and GO based on the following procedure: GO (1 g) was dissolved in 1.5 L of distilled water. HDA (2.00 g; 8.28 × <sup>10</sup>−<sup>3</sup> mol) was added to 25 mL of ethanol, and the mixture was stirred at 25 ◦C under a steady stream of N2, and subsequently added to the GO/water mixture. The resulting mixture was heated for 12 h at 25 ◦C under a steady stream of N2, cooled to 25 ◦C, washed twice with a mixture of distilled water and ethanol (1:1, v/v), and dried under vacuum at 70 ◦C for 24 h to obtain HDA-GS. The synthetic route for HDA-GS is shown in Scheme 1.

**Scheme 1.** Chemical structures of GO and HDA-GS.

#### *2.3. Preparation of PVA Hybrid Films*

The synthetic procedures used to produce the polymer hybrids were identical for all filler contents used in this experiment. Therefore, the preparation of 5-wt% SPT/PVA is detailed in this paper as a representative example. Specifically, SPT (0.1 g) was added to distilled water (20 mL) in a 100-mL beaker, and the mixture was stirred at 80 ◦C for 1 h. The resulting mixture was subjected to ultrasonication three times for 5 min to obtain a homogeneously dispersed clay solution. In a separate beaker, PVA (1.9 g) and distilled water (140 mL) were mixed at 80 ◦C for 3 h and subsequently added dropwise to the SPT/water system with vigorous stirring for 3 h to obtain a homogeneously dispersed system. The solution was cast on poly(ethylene terephthalate) (PET) films and evaporated in a vacuum oven at 35 ◦C for two days. After the removal of the solvent, the hybrid film was dried for a second time in a vacuum oven at 80 ◦C for a day.

#### *2.4. Characterization*

Fourier-transform infrared (FT-IR) spectra were obtained via an FT-IR 460 (JASCO, Tokyo, Japan) instrument in the range of 4000 to 600 cm−<sup>1</sup> with KBr pellets. Wide-angle X-ray diffraction (XRD) measurements were performed at room temperature via a Rigaku (D/Max-IIIB, Tokyo, Japan) X-ray diffractometer with Ni-filtered CoKα radiation. The scan rate corresponded to 2◦/min over the 2*θ* range of 2◦ to 10◦. Differential scanning calorimetry (DSC 200F3, Berlin, Germany) was performed on a NETZSCH instrument, and a thermogravimetric analyzer (AutoTGA 1000, New Castle, USA)) was employed as the thermogravimetric analysis instrument with a heating rate of 20 ◦C/min under the flow of N2.

Atomic force microscopy (AFM, Multimode, NanoScope III, Digital instruments Inc. NY, USA) images were obtained on an AutoProbe CP/MT scanning probe microscope. The GO samples were dispersed in water and HDA-GS samples in toluene. The suspensions were ultrasonicated for 3 h and subsequently spin-coated at 5000 rpm on silicon wafers.

The morphologies of the fractured surfaces of the extrusion samples were investigated via a Hitachi S-2400 scanning electron microscope (SEM). In order to enhance the conductivity, the fractured surfaces were sputter-coated with gold via an SPI sputter coater. Transmission electron microscopy (TEM) samples were prepared by placing the PVA hybrid films on epoxy capsules and curing them at 70 ◦C for 24 h under vacuum. The cured epoxy capsules containing the PVA hybrids were microtomed into 90-nm thick slices and positioned on a 200-mesh copper net, and a layer of carbon (approximately 3-nm thick) was deposited on each slice. The TEM images of the ultrathin sections of the polymer hybrid samples were obtained via an EM 912 OMEGA TEM instrument with an acceleration voltage corresponding to 120 kV.

The O2 transmission rates (O2TRs) of the films were measured based on ASTM E96 with a Mocon DL 100 instrument. The O2TRs were obtained at 23 ◦C, 0% relative humidity, and 1 atm pressure.

#### **3. Results and Discussion**

#### *3.1. FT-IR Spectroscopy*

Figure 1 shows the FT-IR spectra of PVA, graphite, GO, HDA-GS, and OMB. The spectrum of pure graphite does not exhibit any peaks while that of PVA and GO exhibit significant and broad absorption peaks characteristic for the OH and COOH functional groups [31]. Details of each substance are specified below.

**Figure 1.** FT-IR spectra of PVA, graphite, and nanofillers.

The characteristic absorption peaks of the PVA are observed at 3330 cm−<sup>1</sup> (O–H stretching), 2940 cm−<sup>1</sup> and 2910 cm−<sup>1</sup> (asymmetric stretching CH2), 1730 cm−<sup>1</sup> (due to water absorption), and 1256 and 1090 cm−<sup>1</sup> (C-H bending and C-O stretching). In the case of the GO, the characteristic absorption peaks of the O-H are observed at 3220 cm−<sup>1</sup> (stretching) and asymmetric epoxy appears in 3063 cm<sup>−</sup>1, but does not appear to overlap with OH, 1732 and 1616 cm−<sup>1</sup> (C=O stretching), and 1040 cm−<sup>1</sup> (C–O stretching). The epoxide ring and C=C bond are much weaker compared to the others, whereas the O–H stretching is more intense. Figure 1 also shows the spectrum of HDA-GS: 3160 cm−<sup>1</sup> (O–H stretching), 2921 and 2852 cm−<sup>1</sup> (aliphatic C–H stretching), 1375 cm−<sup>1</sup> (aromatic C–N–C symmetric stretching), and 790 cm−<sup>1</sup> (N–H out-of-plane stretching). The OMB also exhibits a peak at 1100 cm−<sup>1</sup> (C-O stretching).

#### *3.2. Morphology of HDA-GS*

The SEM images of unmodified natural graphite, GO, and HDA-GS are shown in Figure 2. The natural graphite exhibits a lamellar structure similar to that of graphene sheets with stacks denser than those observed for the other materials (Figure 2a). Figure 2b shows the translucent GO sheets that are wrinkled and folded in a manner resembling thin paper. Additionally, HDA-GS is prepared from chemically modified GO and exhibit an entangled morphology and a random distribution (Figure 2c).

**Figure 2.** FE-SEM photographs of (**a**) graphite, (**b**) GO, and (**c**) HDA-GS at different magnifications (×10,000 and ×50,000).

Generally, a significant volume expansion and high porosity are observed in the FGSs, which results in low FGS bulk densities that can cause feed problems during melt compounding. Normally, master batches are required to solve the problem. However, in the study, solution blending of FGSs with polymers is successfully employed.

The AFM imaging provides more reliable information on the sheet dimensions and can also be used to probe the surface topology, defects, and bending properties. Additionally, stepped-height scans can also allow us to determine the lateral sizes and thicknesses of the particles lying on the substrates. Thus, AFM is used to indicate that the carbon sheets obtained in the present study are comprised of only a single atomic layer. Figure 3 shows the AFM image of HDA-treated GO sheets (HDA-GSs) on mica and a profile plot that reveals the average sheet thickness corresponding to 1.76 nm (Figure 3) when the thickness of the bare graphene sheet is approximately 1 nm [32], and the thickness of the layer of substituted HDA organic groups is approximately 0.76 nm. However, the layer of the substituted HDA organic groups exhibits a thickness of approximately 0.38 nm when the organic HDA groups on both sides of the graphene sheet in the synthesized FGS are significantly tilted. The change in thickness is associated with different orientations that are adopted by the long alkyl chains in the chemically modified HDA-GSs.

**Figure 3.** (**a**) Noncontact-mode AFM image of HDA-GS sheets with (**b**) three height profiles acquired in different locations.

#### *3.3. Dispersion*

The XRD traces of the pure nanofillers and their PVA hybrid films are shown in Figure 4. The *d*<sup>001</sup> reflection of pristine SPT is present at 2*θ* = 6.62◦, and this corresponds to an interlayer spacing *d* of 13.54 Å, as shown in Figure 4a. Specifically, Figure 4a also shows the XRD curves of the SPT/PVA hybrid films with clay contents in the range of 0–10 wt%. With respect to the PVA hybrids with a clay content ≤10 wt%, clay peaks did not appear in the XRD traces, which indicates that the clay particles are homogeneously dispersed in the hybrid polymer matrices. Figure 4b also shows the XRD curves of OMB in the region 2*θ* = 2◦–10◦. The *d*<sup>001</sup> reflection of OMB is observed at 2*θ* = 6.64◦, and this corresponds to an interlayer distance (*d*) of 13.30 Å. In the case of PVA hybrids containing 3 wt% of OMB, only a very slight peak appeared at 2*θ* = 4.78◦, and this corresponds to an interlayer spacing of 18.46 Å (Figure 4b). The result indicates that a small amount of the clay is not aggregated in the PVA matrix. However, a significant increase in aggregation is observed for samples with OMB loadings reaching 10 wt%, as shown by the intensities of the XRD peaks. This suggests that perfect exfoliation of the layered structure of the clay did not occur. Further evidence of clay dispersion into PVA on a nanometer scale is obtained via TEM.

The XRD diffractograms of pure HDA-GS and HDA-GS/PVA hybrid films are shown in Figure 4c. The *d*<sup>001</sup> reflection for HDA-GS is observed at 2*θ* = 2.73◦ and corresponds to an interlayer spacing (*d*) of 32.32 Å. With respect to PVA hybrid films containing up to 7 wt% of HDA-GS, the peak observed at 2*θ* = 2.73◦ for GS is observed to disappear from the diffraction patterns. The result indicates that the graphene layers are exfoliated and homogeneously dispersed throughout the PVA matrix and provide supporting evidence for the nanocomposite character of the HDA-GS/PVA hybrids. However, the intensities of the XRD peaks at 2*θ* = 2.64 (*d* = 33.42 Å) and 2*θ* = 5.35 (*d* = 16.50 Å) increase suddenly when the HDA-GS loading increases from 7 wt% to 10 wt%, which suggests that the dispersion is more effective at lower loadings as opposed to higher loadings of graphene.

Given the periodic arrangement of the graphite layers in the virgin GS and in intercalated hybrids, the XRD offers a convenient method to determine the interlayer spacing. However, although the XRD enables precise routine measurements of the GS layer spacings, it neither allows for the determination of the spatial distributions of GSs or the detection of any inhomogeneous sections of the hybrids. Initially, a few layered GSs do not exhibit well-defined basal reflections, and, thus, it is difficult to systematically follow peak broadening and reductions in intensity. Therefore, all conclusions on the mechanisms of hybrid formation and microstructure, based solely on XRD results, are only tentative. Therefore, further evidence of the GS dispersion in the PVA films on a nanometer scale is obtained via SEM and TEM as described in the next section.

**Figure 4.** XRD patterns of pristine clay and PVA hybrid films with various nanofiller contents: (**a**) SPT, (**b**) OMB, and (**c**) HDA-GS in PVA.

#### *3.4. Morphologies of PVA Hybrids*

In addition to using XRD to measure the *d* spacings of the nanocomposites, SEM and TEM are used to evaluate the degree of intercalation and amount of aggregation in the nanofiller clusters. The morphologies of the aggregated fillers are characterized via SEM. Large filler aggregates can be easily imaged by SEM due to the difference between the scattering densities of the filler and the PVA matrix [33].

The morphologies of the hybrid films containing up to 10 wt% of SPT in the PVA matrix are examined by observing their fracture surfaces by SEM (Figure 5a). The PVA hybrid films containing 3 wt% and 5 wt% of SPT display uniform and dispersed phases. Conversely, the films containing 7 wt% and 10 wt% of SPT exhibit large particles and some deformed regions that can result from the coarseness of the fractured surface. A similar type of behavior was observed for PVA/OMB hybrid films. For example, the fractured surface of the 5-wt% OMB hybrid film (Figure 5b) exhibit uniform and dispersed phases. However, increased agglomeration is observed in the PVA matrix with clay content exceeding 7 wt% in the OMB system.

**Figure 5.** FE-SEM micrographs of PVA hybrid films with various clay contents: (**a**) SPT and (**b**) OMB in PVA (×10,000).

The fractured surfaces display increased levels of deformation for samples with higher clay contents. The trend is most likely linked to increases in the agglomeration of clay particles and indicate the lack of interfacial interactions between the clay and matrix polymers. Thus, several defects and significant agglomeration occur in interphase areas in high clay content nanocomposite PVA films. Figure 6 shows a comparative analysis of the SEM micrographs obtained for the PVA hybrids with different contents of HDA-GSs that exhibit platelet-orientation distribution morphology.

Graphene dispersions are readily observed in the SEM images due to the differences in the scattering densities of graphene and the matrix polymer. The SEM images of the fractured surfaces of PVA hybrid films containing 0–10 wt% of HDA-GS are compared in Figure 6. The hybrid film with 5 wt% of HDA-GS display a morphology consisting of graphene domains that are well dispersed throughout the continuous PVA phase. However, deformed surfaces and voids are observed in the case of the 7-wt% FGS hybrid film. Conversely, the micrographs of the 10-wt% HDA-GS/PVA hybrid films exhibit increased levels of voids and deformed regions when compared to the 3–7 wt% HDA-GS/PVA hybrid films due to the coarseness of the fractured surface. Overall, the comparison reveals that increases in the graphene content in the hybrid films increase the level of deformation of their fractured surfaces. The finding potentially results from the agglomeration of graphene particles. A comparison of the micrographs indicate that the fractured surfaces of the hybrid films with higher graphene content are more deformed than those of hybrid films with lower graphene content, and this is possibly due to

the agglomeration of graphene particles [34,35]. It should be noted that Figure 6 also shows that most of the graphene remains in the form of straight and rigid platelets in the composite, which indicates that the graphene sheets are extremely stiff.

**Figure 6.** SEM micrographs of PVA hybrid films with various HDA-GS contents.

We extend the morphological analysis via TEM to evaluate the degree of intercalation and degree of aggregation of the nanofiller clusters. Additional direct evidence for the formation of a true nanocomposite is provided by the TEM analysis of ultra-microtomed sections. Figure 7 shows the micrographs of the PVA hybrid films with identical contents of the three different nanofillers. The dark lines in the photographs denote the intersections of the clays and GSs (1-nm thick) while the length between the dark lines denotes the interlayer distance. Figure 7a shows the morphologies of the PVA hybrids with 5 wt% and 10 wt% of SPT. With respect to the hybrid containing 5 wt% of SPT, each layer in the polymer matrix consists of well-dispersed individual clay layers (dark lines), and a few of the clays aggregate to a thickness of approximately 10 nm. In a manner similar to the case for the hybrid material containing 10 wt% of SPT, these clays are mainly aggregated in the polymer matrix. However, the average particle size is observed to be below 20 nm, as calculated from the TEM images. The presence of agglomerated particles in the SEM micrographs of hybrid materials with higher SPT contents (see Figure 5a) is attributed to the formation of aggregated layers.

**Figure 7.** TEM micrographs of PVA hybrid films with different nanofiller contents: (**a**) SPT, (**b**) OMB, and (**c**) HDA-GS in PVA.

Typical TEM images of PVA hybrid films containing 5 and 10 wt% OMB are shown in Figure 7b. Evidently, the clays are well dispersed within the polymer matrix irrespective of the clay content. In contrast to the clay layers in the hybrids containing SPT, the clay layers in OMB hybrids are exfoliated within the matrix polymer.

The TEM micrographs of the 5 and 10 wt% HDA-GS hybrid films are shown in Figure 7c. The TEM micrographs indicate that the GS in the 5 wt% HDA-GS hybrid is dispersed in the polymer matrix, which indicates the formation of a nanocomposite. The findings suggest that GS breaks down into nanoscale building blocks during the intercalative polymerization process and is homogeneously dispersed in the polymer matrix to yield a polymer/GS nanocomposite. As in the case of the 10 wt% HDA-GS (see Figure 7c), the GSs are mostly agglomerated in the polymer matrix. In contrast to the hybrids containing 5 wt% of HDA-GS, the graphene layers of the 10 wt% hybrid exhibited agglomeration of the dispersed graphene phase and are not intercalated into the matrix polymer. The agglomeration of the dispersed graphene phase visibly increases with increases in graphene content, and the outcome is consistent with the XRD and SEM data shown in Figures 4c and 6.

The XRD, SEM, and TEM results indicate that the fillers are well dispersed throughout the PVA matrix at low filler contents while aggregated structures are present at higher filler contents. Additionally, the dispersions of SPT and OMB exceed that of HDA-GS in the PVA matrix (see Figures 4–7). The unusual thermal and gas barrier properties of these hybrid films are discussed in the following sections with respect to the dispersion of the nanofillers.

#### *3.5. Thermal Properties*

A comparison of the DSC results for pure PVA and PVA hybrids with approximately 3–10 wt% of clays (SPT and OMB) and graphene (HDA-GS) are listed in Table 1. The glass transition temperature (*Tg*) of pure PVA corresponds to 69 ◦C. The *Tg* values of PVA hybrids containing various clay contents are virtually unchanged in the DSC results when compared to that of pure PVA irrespective of the filler loading, i.e., approximately 68–71 ◦C for the three different fillers. Generally, *Tg* increases when the filler content increases up to a critical concentration. The increase in *Tg* is potentially due to the confinement of the intercalated polymer chains within the filler galleries that prevents the segmental motion of the polymer chains [36,37]. However, the *Tg* values of the three types of PVA hybrid films remains constant irrespective of the filler content, which suggests that the variation in the filler content does not affect the confinement of the PVA chains.


**Table 1.** Thermal properties of PVA hybrid films with various nano-filler contents.

<sup>a</sup> At a 2 % initial weight-loss temperature. <sup>b</sup> Weight percent of residue at 600 ◦C.

The endothermic peak of pure PVA appears at 165 ◦C and corresponds to its melting transition temperature (*Tm*) (see Table 1). The *Tm* values of the hybrid films are observed to increase from 165 to 176 ◦C when the SPT loading is increased from 0 to 5 wt% and, subsequently, decreases to 166 ◦C at an SPT content of 10 wt%. The increase in the *Tm* of the hybrid film potentially occurs as a result of the insulation effect of the clays and the interactions between the clay and PVA chains [38]. However, the decrease in the *Tm* value of the 10-wt% SPT hybrid suggests that its domains are more poorly dispersed in the PVA matrix than those in the 5-wt% SPT hybrid. Hence, increases in clay content led to the aggregation of clay particles, and this reduces the heat-insulation effect of the clay layers in the polymer matrix.

A similar trend is observed for both the OMB and HDA-GS hybrids. Specifically, the *Tm* of the PVA hybrids increases from 165 to 184 ◦C and from 165 to 173 ◦C when the filler loadings increase to 5 wt% for OMB and 7 wt% for HDA-GS. As observed for the SPT hybrid material, the maximum transition peaks of the PVA hybrids increase with the addition of the nanofiller only up to a certain content level and, subsequently, decreases when the content is increased above this point. For example, when the filler content of PVA reaches 10 wt%, the *Tm* decreases to 178 ◦C and 156 ◦C for materials employing OMB and HDA-GS nanofillers, respectively. The DSC thermograms of the PVA hybrids with various HDA-GS contents are shown in Figure 8. When the nanofiller is included, the peak is stronger than that of the pure PVA, and the more the amount of the filler increased up to 7 wt%, the greater the intensity of the peak becomes. Hence, the HDA-GS appears to act as a nucleating agent [39]. However, when the amount of HDA-GS reaches 10 wt%, the degree of dispersion decreases in conjunction with decreases in the peak intensity.

**Figure 8.** DSC thermograms of PVA hybrid films with various HDA-GS contents.

In a manner similar to the results for *Tm*, the initial thermal degradation temperatures (*TD<sup>i</sup>* ) of the PVA hybrid films also increases linearly from 227 ◦C to 249 ◦C while increasing the SPT loading from 0 to 5 wt%, as shown in Table 1. With respect to the hybrids containing OMB, the *TD<sup>i</sup>* value varies from 227 to 252 ◦C when the content of the OMB organoclay increases from 0 to 5 wt% in the PVA hybrids. The highest increase of 25 ◦C in the *TD<sup>i</sup>* relative to that of pure PVA is observed for the 5 wt% OMB hybrid (252 ◦C). The clay enhances the *TD<sup>i</sup>* by acting as an insulator and a mass-transport barrier to the volatile products generated during decomposition [40,41]. The increase in thermal stability is attributed to the high thermal stability of the clay and interactions between the clay particles and polymer matrix. In contrast to the behavior observed for clay contents ranging from 0 to 5 wt%, the *TD<sup>i</sup>* values of the hybrids decreases when the clay content increases from 5 to 10 wt%. For example, the *TD<sup>i</sup>* values of PVA hybrid film containing 10 wt% of clay loadings were 11 ◦C (238 ◦C) and 6 ◦C (246 ◦C) lower than those of PVA hybrids containing 5 wt% of SPT and OMB, respectively. The decrease in *TD<sup>i</sup>* appeared to correspond to the result of clay aggregation that occurs when the clay content in the polymer matrix exceeds a critical value. Similar results are observed in samples containing HDA-GS, as shown in Table 1. The *TD<sup>i</sup>* values of the PVA hybrid films also increases linearly from 227 to 245 ◦C when the HDA-GS loading is increased from 0 to 7 wt%. At 10 wt% content of HDA-GS in the PVA, the *TD<sup>i</sup>* decreases again to 238 ◦C.

An analysis of the weight residue at 600 ◦C (wtR600) indicates that the weight increases with growth in the filler loading from 0 to 10 wt% and corresponds to a range from 3% to 15% for SPT, from 3% to 14% for OMB, and from 3% to 15% for HDA-GS (Table 1). The enhancement in char formation with increases in filler content is ascribed to the high heat resistance of the clays and graphene.

#### *3.6. Gas Permeation*

The mobility of the polymer chain segments in the polymer nanocomposite clearly differs from that of the pure polymer due to the confined environment, and this also affects the gas permeability. The following two main factors are responsible for permeability reduction [42–44]: (i) polymer chain-segment immobility and (ii) detour ratio, which is defined as the ratio of the film thickness in the nominal diffusion flow direction to the average length of the tortuous diffusion distance between nanolayers.

High aspect ratio nanolayers also lead to properties that are not possible for larger-scaled composites. The impermeable nano-sized layers mandate a tortuous pathway for a permeant to transverse the nanocomposite. Enhanced barrier characteristics, chemical resistance, reduced solvent uptake, and flame retardance of polymer hybrids benefit from the hindered diffusion pathways through the nanocomposite [43,44].

In order to further characterize the barrier properties of the PVA hybrids fabricated by the intercalation of polymer chains in the galleries of clays and graphene, the permeability of the resulting PVA hybrid films to O2 is evaluated for various filler loadings in the range of 0–10 wt%. The results are summarized in Table 2. The thicknesses of all the films subjected to the gas permeation measurements are in the range of 20 to 24 μm. We discuss our results in terms of relative permeability, *Pc/Pp*, where *Pp* denotes the permeability of the pure polymer and *Pc* denotes the permeability of the composite. The results confirm that the mass-transfer process for O2 as the penetrant is highly dependent on the level of filler loading. For example, the addition of only 7 wt% of SPT results in a 95% reduction in the permeability rate of O2 (0.24 cc/m2/day) relative to that of the pure PVA film (5.13 cc/m2/day). With respect to the PVA hybrid films containing 3 to 7 wt% of OMB, the relative O2 permeability rate is close to zero (see Table 2). The outcome is attributed to the increase in the length of the tortuous paths followed by the gas molecules and interactions between the O2 and clay molecules. Furthermore, films containing higher amounts of clay appear as significantly more rigid, and this decreases their gas permeability. In a manner similar to the results observed for the clay hybrids, the addition of 5 wt% of HDA-GS results in an 81% reduction in the permeability rate of O2 (0.98 cc/m2/day) relative to that of the pure PVA film (5.13 cc/m2/day). However, when the clay content increases from 7 to 10 wt%, the O2 permeability rate slightly increases from 0.24 to 0.88 cc/m2/day for SPT and more significantly from <10−<sup>2</sup> to 3.18 cc/m2/day for OMB. The increase in the HDA-GS loading from 5 to 10 wt% led to a similar increase in permeability from 0.98 to 4.27 cc/m2/day. The increases in permeability values are primarily due to the aggregation of the filler particles in materials employing loadings that exceed the critical filler content levels. The present results are further corroborated via the electron micrographs shown in Figures 5–7.


**Table 2.** Oxygen permeations of PVA hybrid films with various nanofiller contents.

<sup>a</sup> Oxygen transmission rate. <sup>b</sup> Composite permeability/polymer permeability (i.e., relative permeability rate).

Collectively, the results of the gas permeation analysis reveal that the strongest gas barrier effect of OMB is observed with respect to the O2TR among the three types of nanofillers that are studied. The enhanced gas barrier capacity of OMB stems from its hydrophilic character, and this allows the formation of hydrogen bonds between OMB molecules and the PVA polymer matrix as well as the reinforcement of chain packing, which significantly reduces the gas permeability of the material [45].

#### **4. Conclusions**

In the present study, we investigated the dispersibilities of three nanofillers including pristine clay SPT, hydrophilic organoclay OMB, and functionalized graphene sheets HDA-GSs, in PVA to improve the properties of the PVA hybrid films. Specifically, PVA hybrid films with varying filler contents ranging from 0 to 10 wt% were synthesized via the solution intercalation method. Their thermal properties, morphologies, and gas permeabilities were compared. The present results confirmed that the properties were dependent on the type and quantity of the nano-filler incorporated in the PVA polymer matrix.

The morphologies of the hybrid materials were examined via TEM, which confirmed that OMB exhibited better dispersion properties than SPT and HDA-GS with respect to the PVA matrix. This observation agreed with the thermal stabilities and gas barrier capabilities of these hybrid materials with the same filler loading levels. Furthermore, the results indicated that the addition of a small amount of nanofiller can sufficiently improve the properties of the PVA. Overall, the addition of OMB was more effective than the addition of SPT and HDA-GS to improve the thermal stability and O2TR of the PVA hybrid composite due to the interactions that formed between the OMB and hydrophilic PVA.

In summary, we demonstrated a simple and effective method to fabricate PVA nanocomposites using the solution intercalation method. Improvements in the thermal property and gas barrier of the obtained composites were observed. It is expected that the use of the proposed methods will allow the widespread use of PVA hybrids in various applications such as permeation membranes, polymer electrolyte fuel cell, packaging films, and drug delivery. Application of the technique to the nano-sized fillers in other polymer composite materials can enhance the various advantages of polymeric materials. The technique can be utilized to further improve thermo-mechanical properties by using clay and other types of fillers based on carbon.

**Funding:** The study was supported by the Research Fund of Kumoh National Institute of Technology (2018-104-105).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Abbreviations**

The following abbreviations are used in the manuscript:


#### **References**


© 2019 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Magnesium Oxide Nanoparticles: Dielectric Properties, Surface Functionalization and Improvement of Epoxy-Based Composites Insulating Properties**

**Jaroslav Hornak 1,\*, Pavel Trnka 1, Petr Kadlec 1, Ondˇrej Michal 1, Václav Mentlík 1, Pavol Šutta 2, Gergely Márk Csányi <sup>3</sup> and Zoltán Ádám Tamus <sup>3</sup>**


Received: 9 May 2018; Accepted: 23 May 2018; Published: 30 May 2018

**Abstract:** Composite insulation materials are an inseparable part of numerous electrical devices because of synergy effect between their individual parts. One of the main aims of the presented study is an introduction of the dielectric properties of nanoscale magnesium oxide powder via Broadband Dielectric Spectroscopy (BDS). These unique results present the behavior of relative permittivity and loss factor in frequency and temperature range. Following the current trends in the application of inorganic nanofillers, this article is complemented by the study of dielectric properties (dielectric strength, volume resistivity, dissipation factor and relative permittivity) of epoxy-based composites depending on the filler amount (0, 0.5, 0.75, 1 and 1.25 weight percent). These parameters are the most important for the design and development of the insulation systems. The X-ray diffraction patterns are presented for pure resin and resin with optimal filler amount (1 wt %), which was estimated according to measurement results. Magnesium oxide nanoparticles were also treated by addition of silane coupling agent (*γ*-Glycidoxypropyltrimethoxysilane), in the case of optimal filler loading (1 wt %) as well. Besides previously mentioned parameters, the effects of surface functionalization have been observed by two unique measurement and evaluation techniques which have never been used for this evaluation, i.e., reduced resorption curves (RRCs) and voltage response method (VR). These methods (developed in our departments), extend the possibilities of measurement of composite dielectric responses related to DC voltage application, allow the facile comparability of different materials and could be used for dispersion level evaluation. This fact has been confirmed by X-ray diffraction analyses.

**Keywords:** broadband dielectric spectroscopy; dielectric strength; loss factor; magnesium oxide; nanocomposite; relative permittivity; surface functionalization; voltage response

#### **1. Introduction**

Magnesium oxide, often called periclase [1] (from Greek word periklao, peri—"around", klao—"to cut"), is white hygroscopic solid mineral. Its empirical formula is MgO and its lattice consist of Mg2<sup>+</sup> ions and O2<sup>−</sup> ions, together bonded by ionic bond (Figure 1). Magnesium oxide is

generally produced by the calcination of magnesium hydroxide Mg(OH)2 or magnesium carbonate MgCO3. Thermal treatment, used when calcination process occurs, affects the surface area and pore size and also the final reactivity of formed magnesium oxide. Used temperature can be divided into three groups, 700 ◦C to 1000 ◦C, where caustic calcined magnesium oxide is formed, 1000 ◦C to 1500 ◦C, where lower chemical activity magnesium oxide is formed and calcination over 1500 ◦C, where reduced chemical activity type of refractory magnesium oxide is formed, that is mostly used for electrical and refractory applications [2].

**Figure 1.** Structure of magnesium oxide crystal (Redrawn and adapted from [2]).

Physical properties (see [3]) make magnesium oxide a good candidate for various applications. It is colorless to brown or black (based on the presence of iron or other foreign element). Considering the surface structure, it is visible that MgO has simplest oxide structure, called Rock-Salt structure. Its density is around 3.579 g/cm3 and hardness around 5 on Mohs scale. Thermal conductivity value of sintered magnesium oxide is defined at *T* = 100 ◦C as 36 W/(mK). Due to refractory properties, the melting and also boiling points of magnesium oxide are very high (melting point: 2800 ◦C, boiling point: 3600 ◦C). Value of electrical resistance is depended on the purity of magnesium oxide. For high purity magnesia, the values of electrical resistivity can reach 1016 <sup>Ω</sup>·m. Specific resistance is mostly depended on chemical purity, but for higher values of temperature, i.e., 2000 ◦C and more, the purity of magnesia does not have any influence on values of electrical resistivity. The dielectric constant of magnesium oxide is in the range from 3.2 to 9.8 at 25 ◦C and under frequency 1 MHz, also values of dielectric loss for same conditions are around 10<sup>−</sup>4.

Chemical properties and surface composition of magnesium oxide are also influenced by the calcination procedure [4,5] (used temperature and used medium, i.e., air or vacuum) and also by the source of the precursor. Based on the various result, physical adsorption of water only occurs if MgO contains surface defects, such as high quantity of pores [6].

Applications of magnesium oxide includes various industry sectors. For their refractory properties, it is a valuable fireproofing ingredient in construction materials. Also in applications where corrosion [7] is not acceptable such as nuclear, chemical or superalloy industries. It has a usage in medical applications [8], where MgO is used for relief of heartburn and sour stomach, as an antacid, magnesium supplement, and as a short-term laxative. Other applications include insulators [9], fertilizers [10], water treatment [11], protective coating [12], etc.

Currently, there are trends to use nanoscale fillers [13]. In general, the nanotechnology is the production of functional structures in the range of 0.1–100 nm by various physical or chemical methods [14]. This fact also applies to magnesium oxide. The sol-gel technique [15] or hydrothermal technique [16] could be used for the production of nanoscale magnesium oxide. For an electrical application, e.g., in high voltage insulation, the MgO represents a prospective filler. Especially, due to wide band gap (7.8 eV) and high volume resistivity (1017 <sup>Ω</sup>·m). It is the highest value of volume resistivity from commonly used nanoscale oxides [17]. For this research, the MgO (supplied by NanoAmor [18]) with average diameter 20 nm, specific surface area more than 60 m2/g and density 0.3 g/cm3, was used.

A lot of studies have shown the effect of different types of nanofillers dispersed in epoxy-based composites on the mechanical [19,20], thermo-mechanical [21,22] or electrical [23] properties. The use of nanofillers has been also demonstrated to impove mechanical characteristics of many biomedical materials, mainly used in orthopedics [24] and dentistry [25]. However, this paper presents the unique results of nanoscale MgO dielectric properties itself, in the temperature and frequency range. These results are complemented by changes of dielectric properties of epoxy-based composites depending on the filler amount and by the effect of surface modification. These effects of surface functionalization were studied by two special measurement and evaluation techniques, i.e., reduced resorption curves (RRCs) [26] and voltage response method (VR) [27], which may be used as indirect method for evaluation of filler dispersion.

#### **2. Dielectric Properties of MgO Nanoparticles via Broadband Dielectric Spectroscopy**

Broadband Dielectric Spectroscopy (BDS) is a modern diagnostic method which allows interconnecting several measurement techniques to obtain a comprehensive view of the material behavior under an electric field with a frequency in very wide range. For this investigation, the main diagnostic unit of the Alpha-A measuring device (Novocontrol Technologies) has been used. It contains a frequency response analyzer with a sinusoidal signal generator and allows analyzing in the frequency range from 3 × <sup>10</sup>−<sup>6</sup> to 4 × <sup>10</sup><sup>7</sup> Hz [28]. Used electrode system (ZGS type) is an active (incorporate a block for diagnostic signal processing) electrode system whose sample cell consists of two parallel cylindrical gold-plated electrodes and the tested flat sample is placed between these electrodes. The diameter of electrodes which are in direct contact with the sample (MgO pellet—see Section 2.1) is 10 mm and therefore the plate capacitor with a diameter of 10 mm is considered in calculations.

The aim of the BDS analysis is primarily to find out general trends of development (and some selected values) of dielectric constant and loss factor as results of measurement at variable frequency of the electric field and variable ambient temperature which is regulated by nitrogen vapor in the cryostat with inserted sample cell. Presented analysis was performed in the temperature range from 25 ◦C to 150 ◦C and in the frequency range from 0.5 Hz to 1 MHz. These ranges of set-up measurement parameters were chosen as sufficient with reference to the intended application of MgO as a filler for electrical insulating composites with a polymer matrix [29]. Entire measurement consists of two phases. The first phase called heating represents the period when the temperature in the cryostat was gradually increased from 150 ◦C to 25 ◦C with a step of 5 ◦C. Then, the temperature was decreased from 150 ◦C to 25 ◦C with same temperature step in the next phase called cooling. The frequency of the measuring voltage with amplitude of 1 V was gradually decreased in the chosen range for each selected temperature in both phases [30]. Several tens of pair of and values for different frequencies were obtained for each temperature (for heating and cooling) after the processing of measured data as a final result of dielectric analysis via BDS.

#### *2.1. Preparation of MgO Pellet*

Used electrode system as a part of the Broadband dielectric spectroscope do not enable measurement of a powder filler in the delivered state. Measurement is possible only with a sample with defined shape and dimensions. In particular, it is necessary to prepare a pellet with a structure, which is as homogeneous as possible, and with a defined thickness. Optimally prepared pellet can be placed between measuring electrodes without pellet fragmentation. The preparation of the pellet represents a homogenization of the MgO powder (more than 200 mg) in a ShakIR sample grinder (PIKE Technologies, Fitchburg, WI, USA) in the first step. The second step is a pressing of the powder whereas

the amount of 200 ±1 mg of powder was loaded into the evacuable pellet press (PIKE Technologies) with a pressing chambers diameter of 13 mm. Air is evacuated from the pressing chamber during a compression. The evacuable pellet press is inserted between parallel pressing plates of the hydraulic press H-62 (Trystom, Olomouc, Czech Republic). It was necessary to optimize the maximum applied pressure in order to avoid a excessive deformation and a fixation of filler particles in pellet volume. This pressure was set to a value of approximately 340 MPa (force of 45 kN applied to anvils) to minimize the adhesion of the pellet surface to the anvils surface that may cause a damage of the pellet during removal from the chamber [31]. This pressure was determined on the basis of self-optimization of pellet preparation from MgO.

#### *2.2. Comprehensive Analysis of Relative Permittivity*

The final result of dielectric analysis performed via BDS are interpreted primarily as complex 3D view (Origin®, OriginLab, Northampton, MA, USA) of frequency and temperature dependencies of dielectric constant and loss factor. These dependencies are shown in Figure 2, in which phases of the measurement under increasing and subsequently under decreasing temperature are distinguished in color.

**Figure 2.** 3D interpretation of frequency-temperature dependencies of (**a**) dielectric constant and (**b**) loss factor for heating (red scale) and cooling (blue scale) of MgO pellet.

The results of the analysis show significant increase in and values with the decreasing of measuring voltage frequency and also with the increasing of temperature when the material is heated in the delivered state. This trend of and development is visible for cooling too, but with a different character. The results for the heating also denote the formation of noticeably visible peak in the temperature dependencies of dielectric constant and loss factor. The 3D interpretation of results of cooling shows a significantly smoother surface without visible peaks. In general, lower values of and are recorded always for cooling, whereas primarily the loss factors decline by several orders of magnitude compared with heating is detected for lower temperatures.

The increase of and caused by the frequency decreasing is primarily the effect of the electrical conductivity of the tested material, which is commonly visible in the low frequencies area of 3D interpretation of similar analysis results. The general cause of the most pronounced increase of dielectric constant and loss factor in the highest temperatures area is a disordered thermal movement of particles in the MgO pellet which is become more apparent under increasing temperature not only for this material. The peak occurrence in characteristics is a result of changes in material structure which are caused by the temperature rise during the measurement and which significantly influence dielectric properties of MgO. Specifically, changes in structure are related to a process of MgO dehydration. This material evidently contains a significant amount of water molecules in the

delivered state. However, these water molecules are with high probability only absorbed or very weakly bonded in the volume of MgO powder if they are released at temperatures below 100 ◦C. The effect of dehydration of MgO (differences of and between heating and cooling) is significant in the case of lower frequencies and mean temperatures which are the most important for the intended application in electrical engineering. Differences in values of dielectric constant and loss factor for selected temperatures and industrial frequency of 50 Hz are shown in Table 1. Results in this table prove the fact that the usage of MgO without dehydration in composites with thermoset matrix (with values of the loss factor lower by orders of magnitude than for MgO at evaluated temperatures) cured at room temperature can have a significantly negative effect on dielectric properties of the composite. On the other hand, after dehydration, MgO exhibits very similar or even lower values of the loss factor than the pure thermoset at evaluated temperatures.

**Table 1.** Selected values of dielectric constant and loss factor for the industrial frequency of 50 Hz.


Numbers before brackets are values for heating and in brackets are values for cooling.

#### **3. Improvement of Epoxy Based Composites Insulating Properties**

The industrial epoxy resin (composition according to supplier safety sheet: Bisphenol-A and epichlorohydrin 50–70% (Figure 3); 1,4-Bis(2,3-epoxypropoxy)butane: 10–20%; Alkyl (C12–C14) glycidyl ether: 5–10%) with low processing viscosity and high bond strength was used for this experiment. This epoxy resin is curable at elevated temperatures (140 ◦C, 4–6 h or 160 ◦C 3–6 h) without additional hardener and it is commonly used in industry due to low viscosity and this is the main reason of our choice. Because lower viscosity of basic material ensures better dispersion of nanoparticles [32]. This resin is evaluated for thermal class H (IEC Standard 60085:2007 [33]), it is free of solvents and it is recommended for vacuum pressure impregnation (VPI) of rotating machines [34]. The density of selected epoxy resin is 1.12 g/cm3.

$$\mathbf{C}^{\mathsf{F}} \overset{\mathsf{O}}{\underset{\mathbf{C}}{\operatorname{\bf}}} \overset{\mathsf{O}}{\operatorname{\bf}} - \mathbf{C}^{\mathsf{H}^{2}} \mathbf{C} - \mathbf{C}^{\mathsf{H}^{2}} \left[ \mathbf{O} - \sqrt{\underset{\mathbf{C}}{\operatorname{\bf}}} \right] - \mathbf{C}^{\mathsf{H}^{2}} \sqrt{\underset{\mathbf{C}}{\operatorname{\bf}}} \overset{\mathsf{O}}{\operatorname{\bf}} - \mathbf{C}^{\mathsf{H}^{2}} \mathbf{C} - \mathbf{C}^{\mathsf{H}^{2}} \left[ \underset{\mathbf{C}}{\operatorname{\bf}} \right] - \mathbf{C}^{\mathsf{H}^{2}} \overbrace{\overset{\mathsf{O}}{\operatorname{\bf}}} - \mathbf{C}^{\mathsf{H}^{2}} \overbrace{\overset{\mathsf{O}}{\operatorname{\bf}}} - \mathbf{C}^{\mathsf{H}^{2}} \overbrace{\overset{\mathsf{O}}{\operatorname{\bf}}}$$

**Figure 3.** Structure of resin based on on Bisphenol-A diglycidyl ether.

#### *3.1. Production of Epoxy-Based Nanocomposites Samples*

In this part, the sample production of pure epoxy resin and resin with dispersed nanoparticles will be described. Amount of 60 g of epoxy resin was used for the creation of a collection of the samples (5 pcs). In the case of pure epoxy, only magnetic stirring of the epoxy resin was carried out together with the vacuum venting (8 mbar) for 3 h. After this time, the epoxy resin was placed in the preheated Teflon molds with a silicone frame determining the height of the sample. The resin was then cured in a hot-air oven (140 ◦C, 6 h). In case of matrices with dispersed particles, nanoparticles (0.5, 0.75, 1 and 1.25 wt %) were added to the already heated resin (75 ◦C, 600 rpm, 3 h). An ultrasonic thorn was then used to break agglomerates (30 min) with simultaneous magnetic stirring and heating (70 ◦C, 300 rpm). Further, the vacuuming (8 mbar) process was combined with magnetic stirring and simultaneous heating (90 ◦C, 300 rpm, 3 h). The epoxy resin mixture was further placed in dried Teflon molds and cured under the same conditions as in the previous case.

#### *3.2. Dielectric Properties of Epoxy-Based MgO Nano-Composites*

Four experimental measurements were performed for observation of MgO effect and finding the optimal filler ration to achieve the best dielectric properties in comparison with unfilled matrix (Figure 4). Namely dissipation factor and relative permittivity (IEC 60250:1969 [35]), dielectric strength (IEC 60243-1:2013 [36]) and volume resistivity (IEC 62631-3-2:2015 [37]) were measured according to mentioned standards. All measurements were performed according to Standard conditions given by IEC 60212:2010 [38]. Because of the basics of these parameters, no further information about step-by-step measurement procedure is provided in this article.

From presented results is visible that the addition of MgO nanofiller causes the changes of selected parameters. Possible reasons are discussed in following text. There is a slight increase of relative permittivity and dissipation factor, respectively. These parameters characterize the degree of polarizability of the matrix, filler and their interfaces and losses caused by their interactions with electric field. Different behavior can be explained by changing the curing reaction of the whole composite and by changing the degree of crosslinking [39].

**Figure 4.** Changes of dielectric properties depending on the filler amount. (**a**) dissipation factor—500 V AC, 50 Hz; (**b**) relative permittivity—500 V AC, 50 Hz; (**c**) Volume resistivity—500 V DC; (**d**) Dielectric strength—increase 1.5 kV/s AC.

Study [40] highlighted the positive effect of magnesium oxide of nanometric dimensions on the reduction of trapped charge in the internal structure of the material at a fill volume in the range 0.5–2%. Due to the nature of the particles, when their volume resistivity is in the order of 1017 <sup>Ω</sup>·m, the resistivity of whole composite may increased at low filling rates. This fact was confirmed by performed measurements. An increase of volume resistivity of the composites can be attributed to an increase in resistance to injection of the charge carriers and their generation in the internal dielectric structure [41]. Dielectric breakdown phenomena of nanodielectrics is affected mainly due to low quantity of agglomeration at low filler loadings [42]. Some studies [43,44] also shown the changes of relation between the dielectric breakdown and free-volume in polymers. In these cases, also a percolation threshold could plays a role, but not more works have been presented for percolation threshold estimation in the case of dispersed nanoscale insulating particles, where the character of added filler causes improvement of the electro-insulation properties of whole composite. On the other hand, the behavior of dispersed conductive fillers is very-well known [45–48] and percolation threshold could be estimated based on the significant increase of conductivity. However, the dielectric parameters of basic material could be improved by addition of relatively low amount of nanofiller, as is evident from our previous studies [49–51]. In connection with these claims, the study [52] describes a theory of percolation and interfacial characterisation via breakdown voltage measurement. It may be also used in this case for the confirmation of presented results (Figure 4d). In the case of pure epoxy resin, deeper traps generally exist. It results in relatively easy charge capture. If the charge carriers are released, the breakdown occurs due to their energy. The increase of breakdown voltage can be attributed to increasing of shallow traps inside the material by addition of nanofiller up to percolation threshold. The expected decrease will occur if the double-layers [53] on the particles surface are overlapped. It leads to the easier movement of the charge carriers in double-layers. Due to this fact, the conductive path will be formed [54] and breakdown can occur more easily. These effects are better noticable on the values of volume resistivity (Figure 4c), which goes hand in hand with the breakdown voltage measurement. According to the presented results, the percolation threshold could be estimated greater than 1 wt %.

Taking into account the preliminary measurements of the basic electrical properties (Figure 4), especially the volume resistivity and breakdown voltage, together with the the above-presented data, the optimal weight ratio for further investigation was set to 1 wt %.

#### *3.3. X-ray Diffraction of Epoxy Resin and Epoxy Resin with Dispersed MgO Nanoparticles*

The X-ray difraction has been used for the characterisation of internal structure of the tested material. These measurements were performed on a Panalytical X'Pert Pro (Malvern Panalytical) automated powder X-ray diffractometer using an X-ray lamp (IKa1 = 0.154 nm, 40 kV, 30 mA) and a semiconductor ultra-rapid PIXcel detector in the geometric Bragg-Bretan arrangement. The results from the diffractometer were aligned with the Pearson VII curves. From the X-ray diffraction analysis is visible the character of the amorphous material. This symbolizes very wide diffraction (6–8 degrees in the diffraction angles 2-theta), which are shown in Figure 5.

These diffractions, resp. their positions on the x-axis, corresponding with the results of presented studies [55,56]. On this diffractogram, only the diffraction pattern of MgO (200), (220) and (222) are noticed. Other lines are weak, or we do not notice them at all. Analysis of the profile of diffraction lines (200) showed that the size of the coherent dispersion region of X-ray crystallization (crystallite) is in all cases about 23–25 nm and the micro deformation is relatively low (0.0022–0.0025).

**Figure 5.** X-ray diffraction pattern of investigated materials.

#### **4. Surface Functionalization and Effect on Dielectric Properties**

With regard to the fact that the fillers used for electro-technical applications are in more cases of inorganic origin, it is very difficult to achieve sufficient dispersion in the organic matrix under normal conditions [57]. One possible solution is the use of silane-based coupling agents. The most popular ones are *γ*-Glycidyloxypropyltrimethoxysilane (GLYMO) and *γ*-Aminopropyltriethoxysilane (APTES) which provide covalent interface links to prevent a phase separation [58].

A lot of studies have shown the effect of functionalization on different material properties. The surface of inorganic particles [59], glass fibers [60] and also the natural fibers e.g., jute fibers [61] or hemp fibers [62] have been already modified by addition of GLYMO or APTES, respectively. In general, the formula of silane coupling agent can be written as R(CH2)nSiX3 where the silane molecule is silicon (Si) and two functional substituents (R, X) that provide a bonding effect between the inorganic filler and the organic matrix [63]. The substituent X represents hydrolyzable groups (e.g., methoxy, ethoxy, alkoxy), and R represents an organofunctional group attached to the silicon atom by a hydrolytically stable bond. Most of the coupling agents comprise three hydrolyzable groups X and one organofunctional group R [64]. Coupling agents and their linear formulas are shown in Table 2.



The reaction of the *γ*-Glycidoxypropyltrimethoxysilane with the magnesium oxide filler can be explained as follow and is illustrated in Figure 6. The corresponding silanol molecules are formed after hydrolysis of the hydrolysable groups [68]. Furthermore, the process of chemisorption is going. The hydrogen bonds are formed between silanol and –OH groups on the surface of magnesium oxide. A polysiloxane layer bonded with covalent bonds to the surface of the magnesium oxide is formed while water is released due to the condensation reaction [69,70].

**Figure 6.** Simplified illustration of reaction of *γ*-Glycidoxypropyltrimethoxysilane with magnesium oxide surface (Redrawn and adepted from: [71–73]).

The determination of the correct ratio [67] of the coupling agent can be based on the relationship (1)

$$X = \frac{A}{w} \cdot f\_{\prime} \tag{1}$$

where *X* (g) is the amount of coupling agent to form the minimum cover layer, *A* (m2/g) is the specific surface area of the nanoparticle, *w* (m2/g) is the wetting specific area of the coupling agent and *f* (g) is the weight of nanoparticles. Formic acid or hydrofluoric acid may be applied first to the nanoparticle to increase the electro-kinetic potential of its surface [74]. For the ability to react with different types of matrices, *γ*-Glycidoxypropyltrimethoxysilane was chosen for this experiment. The density of the selected coupling agent is 1.07 g/cm<sup>3</sup> and the wetting specific area is 331 m2/g. It is an epoxysilane coupling agent, in particular, an organofunctional trialkoxysilane having a high reactivity between epoxide rings and amino groups [63]. The amount of GLYMO was determined as 18.12% from the total weight of nanoparticles according to Equation (1) and above-mentioned parameters.

#### *4.1. Production of Epoxy-Based Nano-Composites Samples with Treated Surface of MgO Filler*

In this case, the dried nanoparticles (1% of total weight of epoxy resin) were first added to a solution of 96% ethanol 4% H2O (10 mL) and ultrasonic mixed (30 min). The coupling agent GLYMO (18.2% of total weight of nanoparticles) was then added to the mixture with re-application of ultrasonic mixing (2 h). Treated nanoparticles were added to already heated resin. The following procedure is the same as in the previous case.

#### *4.2. Dielectric Properties of Epoxy-Based Nanocomposites with Treated Surface of MgO Filler*

For comparison of the effect of the surface treatment, the measurement of dissipation factor (500 V AC, 50 Hz), relative permittivity (500 V AC, 50 Hz), volume resistivity (500 V DC) and dielectric strength (increase 1.5 kV/s AC) were repeated. The average values are shown in Table 3.


**Table 3.** Comparison of selected parameters after surface treatment.

From the measurement results, it is clearly visible that the addition of coupling agent improved dielectric properties of the whole composite in comparison previous case. The relative permittivity was reduced approx. by 8%, due to the surface modification. The lower value of relative permittivity also guaranties a lower level of local stress inside the electrical insulation system. For example, in the case of an imperfect technological process during manufacturing of the insulating system of electrical machines and equipment. Mentioned decrease of relative permittivity could be caused by a changes of the degree of crosslinking due to the reaction of polymeric groups. They react with the coupling agent molecules on the nanoparticle surface and form a linear polymer chains in the interphase region [75]. The results further show that the dissipation factor of the composite is not negatively affected by the addition of the coupling agent, as confirmed other studies [39,75], as well. Addition of the coupling agent results in higher volume resistivity values, which can be attributed to a higher degree of the filler dispersion in the matrix and also to an increase of the energy levels of the electron traps [57], which results in a higher resistance to charge accumulation in the inner structure of the material.

#### *4.3. Dielectric Response Measurement*

Different optical methods and measurement techniques [76–79] are used for evaluation of the surface treatment effect or dispersion level, respectively. However, the idea of this paper is to evaluate the effect of surface treatment and particle dispersion by measurement of dielectric responses by special measurement techniques, i.e., reduced resorption curves (RRCs) and voltage response method (VR), which evaluate the conditions of the dielectric materials during charging and discharging process.

#### 4.3.1. Reduced Resorption Curves Analyses

Dielectric absorption is a non-stationary phenomenon in dielectric materials after dc voltage application. Dielectric material is not able to follow the step change of the applied voltage. It means that the dielectric is charged for a certain time interval which is given by the relaxation time. This also applies to the discharging phenomenon. Both effects are caused by slow polarizations [80]. Here, the attention of our investigation was focused on the resorption characteristics. Resorption current can be used for reduced resorption curves (RRC) [26] determination. This method is based on the mathematical processing of time variable resorption current, which is transformed to relative resorption characteristic. The mathematical process of this methodology is expressed by Equations (2) and (3)

$$\mathbf{x} = \ln(t) - \ln(15),\tag{2}$$

$$y = \ln[ABS(i\_l)] - \ln[ABS(i\_{15})],\tag{3}$$

where *x, y* are transformed axes (-), *t* (s) is time, *it* (A) is current in time *t, i*<sup>15</sup> (A) is current in 15th seconds. The main parameter for this investigation is the slope of the linear fit from transformed data. The higher slope of the curve generally means better resistance to charge trapping. The adequate interval length is important for appropriate linear fitting. In general, the interval between 15 and 300 s is usable for this determination. This procedure is illustrated in Figure 7.

**Figure 7.** Transformation of resorption current to RRCs.

The electrometer KEITHLEY 6517A with suitable electrode system KEITHLEY 8009 Resistivity fixture was used for this measurement. The flat samples were conditioned (25 ◦C, 35% RH) and short-circuited in shielding room for 24 h before the measurement. After that, samples were charged by DC voltage 1000 V for 3600 s. After charging, the resorption current was measured up to 600 s and was recorded by the developed script in VEE Pro software. Transformation interval was set in the range 15...300 s. Average values of resorption currents were transformed to reduced resorption curves (RRCs) according to Equations (2) and (3). The RRCs are compared in Figure 8.

**Figure 8.** Reduced resorption curves for individual sample sets.

From the results is clearly visible the slope increase of the trendline due to addition of filler resp. filler and coupling agent. In the first case, the slope increase may be caused by the reduction of bulk charge accumulation due to better resistance to charge injection and ionic carriers generation in the bulk of dielectric [41]. The increase of the slope in the case of addition of silane coupling agent is caused

by better dispersion and by a higher level of miscibility between organic matrix and an inorganic filler. The other reason is the increase of the trap depth which may contribute to charge recombination [81]. It means that the charge suppression is more effective. From this point of view, there is the higher ability of the material to discharge the charge accumulated in the inner structure after dc voltage charging, because a higher level of energy is needed for its trapping.

#### 4.3.2. Voltage Response Analyses

Originally, the Voltage Response measurement method (VR) was developed for investigation of oil-paper insulated cables and measures the initial slopes of the decay and return voltages [27]. The timing diagram of the measurement can be seen in Figure 9.

After a long duration (100. . . 1000 s) charging period (*tch*) the discharge voltage (*Vd*(*t*)) is measured on the insulation for *tidp* time (<0.5 s). After a few seconds of short-circuiting (*tdch*) return voltage (*Vr*(*t*)) is measured on charged insulation for *trvp* time (0.1. . . 2 s).

**Figure 9.** Timing diagram of Voltage Response measurement.

The initial slope of *Vd*(*t*) (marked with *Sd*) is directly proportional to the conductivity of the insulation and the initial slope of *Vr*(*t*) (marked with *Sr*) is directly proportional to the polarization conductivity, in other words to the intensity of the slow polarization processes. Therefore, the separate investigation of conductive and polarization processes is ensured by the measurement of *Sd* and *Sr* since they have the same information content as *Ic* conductive and *Ip* polarization component of the leakage current, respectively [82]. The measured values can be seen in Table 4.



According to the measurement results, the pure resin had the highest conductivity and polarization conductivity from all the samples. By adding MgO filler, the conductive and polarization processes decreased significantly. However, the best results were measured after the addition of the silane coupling agent.

#### *4.4. X-ray Diffraction of Epoxy-Based Nanocomposites with Treated Surface of MgO Filler*

In this case, also the X-ray diffraction analyse has been performed for confirmation of our uttered assumptions. The X-ray diffraction signals of investigated materials have been deconvolute and diffraction pattern of epoxy matrix has been removed (Figure 10).

**Figure 10.** X-ray diagram of treated and untreated MgO nanoparticles.

There is clearly visible that addition of *γ*-Glycidoxypropyltrimethoxysilane do not contribute to additional chemical reactions inside the material and the phase structure has not been changed. Hovewer, on the first view, there are a differences between the heights of intensity peaks. The lower intensity peaks may be described by theory of presence of polysiloxane layer on the surface of the magnesium oxide powder [83]. It means that coupling agent occupies a part of volume. This measurement indirectly supports our previous results and statements on the suitability of dielectric response diagnostic methods for evaluation of nanofiller dispersion level.

#### **5. Conclusions**

This paper presents a few unique results. From the measurement results is visible the effect of dehydration on parameters of complex permittivity. This material evidently contains a significant amount of water molecules in the delivered state. However, these water molecules are with high probability only absorbed or very weakly bonded in the volume of MgO powder if they are released at temperatures below 100 ◦C. For this reason, it is necessary to dry nanoparticles to remove the surface moisture content before their applications.

Following on the first measurement, the effect of filler loading has been tested for improvement of epoxy matrix properties. The optimal filler loading was set to 1 wt % after dielectric parameters measurement. There are significant changes in investigated parameters, especially in the case of volume resistivity. The increase of volume resistivity is a consequence of the increase in resistance to injection of the charge carriers and their generation in the internal dielectric structure.

The effect of silane coupling agent was also investigated on the optimal filler loading. For the ability to react with different types of matrixes, *γ*-Glycidoxypropyltrimethoxysilane, has been used. The verification of improvement of dielectric parameters were also carried out. The main reason, why the parameters were improved, are due to a change in the degree of crosslinking due to the reaction of polymeric groups which react with the coupling agent molecules on the nanoparticle surface and form linear polymer chains in the interphase region. Also due to higher degree of dispersion of the filler in the matrix and also to an increase in the energy levels of the electron traps.

Two measurement and evaluation techniques (RRCs, VR) have been taken together in this paper. The possibility of interconnection between these two different techniques is clearly visible from the experimental results. In both cases, the effect of surface treatment of nanofiller was observed on the dielectric response. From this point of view, this methodology can be used not only for evaluation of dielectric parameters, such as conductivity or relaxation time of polarization mechanisms. According to presented results and previous presumptions, these methods may be used for observation of the proper dispersion of nanofiller in a polymer base.

**Author Contributions:** Conceptualization, J.H., P.T. and V.M.; Formal analysis, J.H., P.K. and G.M.C.; Investigation, J.H., P.K., P.Š. and G.M.C.; Methodology, J.H.; Supervision, P.T., V.M. and Z.Á.T.; Validation, P.T., V.M. and Z.Á.T.; Visualization, J.H., O.M., P.Š. and G.M.C.; Writing—original draft, J.H., P.T., P.K., O.M. and G.M.C.; Writing—review & editing, P.T., V.M. and Z.Á.T.

**Acknowledgments:** This work is supported by the Ministry of Education, Youth and Sports of the Czech Republic under the RICE-New Technologies and Concepts for Smart Industrial Systems, project No. LO1607 and by the Student Grant Agency of the West Bohemia University in Pilsen, grant No. SGS-2018-016 Diagnostics and materials in electrotechnics.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
