**Fabrication of Flexible, Lightweight, Magnetic Mushroom Gills and Coral-Like MXene–Carbon Nanotube Nanocomposites for EMI Shielding Application**

**Kanthasamy Raagulan 1, Ramanaskanda Braveenth 1, Lee Ro Lee 1, Joonsik Lee 2, Bo Mi Kim 3, Jai Jung Moon 4, Sang Bok Lee 2,\* and Kyu Yun Chai 1,\***


Received: 20 February 2019; Accepted: 26 March 2019; Published: 2 April 2019

**Abstract:** MXenes, carbon nanotubes, and nanoparticles are attractive candidates for electromagnetic interference (EMI) shielding. The composites were prepared through a filtration technique and spray coating process. The functionalization of non-woven carbon fabric is an attractive strategy. The prepared composite was characterized using X-ray photoelectron spectroscopy (XPS), X-ray diffraction (XRD), scanning electron microscope (SEM), energy-dispersive X-ray spectroscopy (EDX), and Raman spectroscopy. The MXene-oxidized carbon nanotube-sodium dodecyl sulfate composite (MXCS) exhibited 50.5 dB (99.999%), and the whole nanoparticle-based composite blocked 99.99% of the electromagnetic radiation. The functionalization increased the shielding by 15.4%. The composite possessed good thermal stability, and the maximum electric conductivity achieved was 12.5 S·cm<sup>−</sup>1. Thus, the composite shows excellent potential applications towards the areas such as aeronautics, mobile phones, radars, and military.

**Keywords:** MXene; oxidized carbon nanotube (CNTO); nanoparticle decoration; functionalization; electromagnetic interference (EMI) shielding

#### **1. Introduction**

Electromagnetic interference (EMI) leads to inevitable interactions in electronic devices. Smaller and high-speed electronic systems are susceptible to issues related to EMI which can affect either adjacent electronic items or humans, thus potentially affecting the security of the nation [1,2]. In war, electromagnetic pulse weapons are being used to affect systems utilizing electromagnetic radiation (EMR) such as radar systems, high tech complex electronic devices, remote control armor, aircraft, and missiles. In addition, EMI affects the functions of sensors in modern electronic vehicles as they transmit signals using weak radiation and microcomputers. Hence, protecting electronic devices from malfunction and achieving electromagnetic compatibility is an essential requirement around the globe. Further, electromagnetic compatibility should be attained by diminishing incoming and outgoing

electromagnetic radiation, ideally without affecting the function of the devices. This is because EMI not only affects electronic systems but also causes health issues in human beings [3–5].

Various substances have been used for EMI shielding, such as MXenes, graphene (GN), graphene oxide, carbon nanotubes (CNTs), nanoparticles, polymers, fabrics, textiles, composites, and metals in various frequency ranges [2,4–6]. The EMI shielding materials can be categorized in two types: Reflection and absorption domain materials. The reflection domain materials possess mobile charges while absorption domain materials contain magnetic and dielectric materials. The layered and implanted type structures influence the EMI shielding. EMI shielding can be accomplished by suppressing the incident wave, which has the three key mechanisms of absorption, reflection, and multiple reflection. The electric conductivity associated with primary shielding factor is reflection, where the mobile electron interacts with incident wave. The thickness, electric and magnetic dipole loss, magnetic permeability, defects, and structural features induce absorption. The ohmic loss can be achieved by conduction, electron hopping, and tunneling. The polarization loss occurs due to the rearrangement of the polarization while electromagnetic radiation (EMR) is passing through the shielding materials. Polarization can be induced by embedding functionalities, hybrid fillers, nanofillers, and defects in the matrix of the composite. The inhomogeneous scattering centers, layered structure, hollow structure, and interfaces generate multiple reflection, which finally leads to absorption. Further, the skin depth limits the EMI shielding effectiveness, which should be lower than that of the thickness of EMI shielding materials [2,6–11]. Certain properties, such as being lightweight, conductive, corrosion resistant, flexible, cost-effective, and high strength, are preferable for a modern EMI shielding material. Metals are conventionally exploited as shielding materials due to the fact that they possess excellent conductivity, but they are unable to fulfill the current needs of compact electronic systems. Hence, carbon-based substances are attracting attention, as their properties can be tuned by incorporating other materials like nanoparticles and polymers [1–5,8]. In addition, incorporating carbon nanotubes (CNTs), MXenes, polymers, and graphene with non-woven carbon fabric significantly increases EMI shielding. Thus, establishing a conductive network is an essential factor for good EMI shielding [5,8,12]. Different combinations of the constituents and amounts of filler loading have been used to fabricate multi-functional EMI shielding materials [13–17].

MXenes (M(n+1)XnTx) are two dimensional (2D) material derived from a corresponding three dimensional (3D) MAX phase (M(n+1)AXn), where M is early transition elements (Ti, V, Cr, Nb, Ta, Zr, and Mo), A includes group 13/14 elements, X represents carbon or nitrogen, and Tx is surface functional groups (–OH, –F, and =O) [2]. A selective etching strategy is used in the production of MXenes. The minimally intensive layer delamination (MILD) method (LiF/HCl) has recently been endorsed, as it abridges the synthetic process, and HF, NH4HF2, and FeF3/HCl have also been practiced. During the etching, the weaker M–A bonds are eradicated while the strong M–C bond remains with newly formed functionalities [2,18]. MXenes have a metal-like nature, and similar to graphene, have been used for various purposes such as in sensors, capacitors, storage, and EMI shielding materials. In general, MXenes are hydrophilic in nature, as –OH is one of the surface functional groups. Thus, MXenes can be incorporated with various materials like polymers in order to tune their properties [2]. There are various types of MXenes that have been studied such as Nb2CTx, Ti3CNTx, and Ti2CTx. In addition, MXenes are used to make hybrid composites such as TiO2–Ti3C2TX/graphene, Ti3C2TX–sodium alginate, Ti3C2Tx/PVA, cellulose nanofibers–Ti3C2Tx and Ti3C2TX/paraffin. MXenes exhibit maximum EMI shielding of 92 dB with 45 μm thickness. Thus, the loading amount of MXenes in different polymers matrices, the morphology of composite, and the thickness influence the EMI shielding of the MXene [19–23].

In this study, we developed a layered pliable composite with different surface morphology and magnetic composite. Each layer of the composite consists of a layered and magnetic domain for which we employed spray coating and filtration technique under gravity. The functionalization of carbon fabric intercalation of the MXene, CNTs, and magnetic nanoparticle dramatically changed the EMI shielding. The spray coating samples were denoted as MXCNTCx, where x (x = 10, 25, or 30) was the number of coating cycles, and Ni coated fabric was expressed as MXCNTNi25. The composite fabricated using the filtration method was denoted as MXCBCM, where M is the type of metal nanoparticle such as Ni, Co, Fe, Cu, and Fe3O4. Its corresponding composites were MXCBCNi, MXCBCCo, MXCBCFe, MXCBCCu, and MXCBCFeO, respectively. Further, MXCB and MXCS were labelled based on the surfactant used, like cetyltrimethylammonium bromide (CTAB) and sodium dodecyl sulfate (SDS), respectively. MC and FC indicated uncoated carbon fabric and functionalized carbon fabric, respectively. The parameters of EMI shielding, elementals analysis, morphology, structural analysis, electric conductivity, surface property, magnetic property, and thermal stability were investigated in detail.

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

#### *2.1. Materials*

Multiwall carbon nanotube (MWCNTs) (CM-90, 90 wt %, diameter of 20 nm, and length of 100 μm) were purchased from Applied Carbon Technology Co. Ltd. (Pohang, Korea). Carbon fiber (fiber diameter 7-micron, 6 mm length) and polyethylene terephthalate (PET) binder (fiber diameter 2.2 dtex, 5 mm) were collected from TORAY product, (Tokyo, Japan). Sodium dodecyl sulfate (SDS) (98%), lithium fluoride (LiF) (98%, 300 mesh), sodium borohydride (NaBH4), polyacrylamide (PAM), anhydrous FeCl2, FeCl3, NiCl2, CuCl2, CoCl2, and cetyltrimethylammonium bromide (CTAB) were obtained from Sigma Aldrich (Seoul, Korea). Wet laid nickel coated non-woven carbon fabric (basic density of 19.2 g/m2, thickness of 150 μm) was acquired from Clean & Science Co. Ltd., (Seoul, Korea). Nitric acid (HNO3-70%) and hydrochloric acid (HCl-35%) were obtained from Samchun Chemical Co., Ltd. (Seoul, Korea). Chitooligosaccharide (Mw 5000) was issued by biomedical polymer lab, Sunchon National University (Suncheon, Korea).

#### *2.2. Synthesis of Ti3AlC2*

We reported that TiC, Ti, and Al powders were taken with molar ration of 2:1:1 ball milled by using Pulverisette 6 Planetary Mono Mill (Fritsch, Germany) in ethanol medium at 200 rpm for 1 h in a nitrogen environment. The resultant mixture was dried at 80 ◦C for 12 h. Then, 3 g of 12 mm diameter disc was prepared by applying 27.6 MPa pressure for 5 min in a laboratory press. The resultant disc was treated at 1350 ◦C with a heating rate of 20 ◦C/minute in argon gas for 2 h, then cooled down to room temperature. The treated disc was again ball milled in ethanol medium at 300 rpm for 3 h in a nitrogen environment. The powder yield was then dried at 80 ◦C for 12 h, and the obtained product was directly used for MXene synthesis [24].

#### *2.3. Preparation of Oxidized Carbon Nanotube (CNTO)*

1 g of MWCNT (CNT) and 90 mL of HNO3 were sonicated by using a mini ultrasonic cleaner (Uil ultrasonic Co., Ltd., Gyeonggi-do, Republic of Korea) for 5 h at room temperature. The volume of the reaction mixture was doubled by adding deionized (DI) water and filtered. The acid solution and reacted CNT were separated by filtration. The resultant black color product known as CNTO was washed until reaching a neutral pH, then dried at 80 ◦C for 24 h. This CNTO was used for the coating process and synthesis decorated CNTO.

#### *2.4. Preparation of Fe3O4 Decorated CNTO*

100 mL of 0.1 M of Fe3+, 100 ml of 0.1 M of Fe2+, and 0.5 g·L−<sup>1</sup> of CNTO were stirred together at 30 ◦C for 30 min. A total of 10% of NH4OH was added dropwise until the pH of the solution reached 11–12, along with 2 mL of 0.2 M SDS. Then, the temperature of the reaction mixture was raised up to 80 ◦C and stirred for 2 h. The volume of the solution was reduced by half via evaporation and then cooled down to room temperature. The resultant product was washed until reaching a neutral pH in a vacuum filter, then dried at 105 ◦C for 12 h.

#### *2.5. Preparation of Nanoparticle Decorated CNTO*

100 mL of 0.1 M2+ of metal ion solution (Fe2+, Ni2+, Cu2+, and Co2+) and 0.5 g·L−<sup>1</sup> of CNTO were stirred for 30 min. 100 mL of 0.2 M cold NaBH4 was added drop wisely in to the mixture, along with 0.2 g·L−<sup>1</sup> of SDS. Fe, Ni, Co, and Cu decorated CNTO were successfully synthesized in a nitrogen environment. The product was filtered and washed by an ample amount of DI water. Then, the product was dried at 80 ◦C and 0.8 atm in a vacuum oven for 12 h. The obtained product was denoted as CM, where C is CNTO and M is nanoparticles. The corresponding decorated CNTOs were denoted as CFe, CNi, CCo, and CCu.

#### *2.6. Preparation of Dispersed Solutions*

0.1 g of CNTO and 0.1 g of SDS were mixed in 100 mL of deionized water and sonicated for 3 h. Then, the dispersed mixture was refluxed at 120 ◦C for 12 h. The obtained well dispersed solution was then used for the spray coating process. In addition, 0.4 g of CNTO and 0.4 g of CTAB were added together in 100 mL of deionized water and sonicated for 5 h (CTAB–CNTO). This procedure was repeated using SDS and CNTO (SDS–CNTO). The decorated nanoparticle dispersed solution was prepared accordingly; equal amounts of CTAB and CM (0.1 g) were sonicated in 25 mL of deionized water for 30 min. The obtained dispersed solutions were directly used for the filtration process.

#### *2.7. Preparation of MXene and MXene Colloidal Solution*

Equal amounts (1 g) of Ti3AlC2 and LiF were mixed together in 20 mL of 9 M HCl solution. This mixture was stirred at 35 ◦C for 24 h. The etched product was washed with DI water up to approximately pH 6 by centrifuging at 3500 rpm for 5 min, then Ti3C2Tx was dried in a vacuum oven for 12 h. Then, 0.1 g of MXene was sonicated in 10 mL of DI water in a nitrogen environment at about 15–17 ◦C for 2 h. The sonicated MXene solution was centrifuged at 3500 rpm for 30 min, and the supernatant was collected in a Teflon container stored at 5 ◦C for the coating process. The concentration of the MXene colloidal solution was 0.175 g·L−1. 100 mL of colloidal solution was filtered by using 0.45 μm of 47 mm of Nylon supported filter paper and dried at 80 ◦C (0.8 atm) until obtained constant weight. The concentration was calculated based on the weight differences.

#### *2.8. Synthesis of Carbon Fabric by Wet Laid Method (MC)*

We reported that 0.6 kg of carbon fiber, 0.15 kg of PET binder and 0.3 weight percent of dispersant (PAM) were dispersed in a sufficient amount of deionized (DI) water at 500 rpm for 10 min. The general wet laid method was used to produce the web. During the process, the drum dryer was used with a <sup>140</sup> ◦C surface temperature and 7 m·min<sup>−</sup> speed. The areal density of obtained fabric was 30 g·m−<sup>2</sup> [24].

#### *2.9. Fabrication of Fabric Composite*

#### 2.9.1. Preparation of Functionalized Carbon Fabric (FC)

0.2 g of chitooligosaccharide was dissolved in 100 mL of deionized water and stirred for 15 min. Then, carbon fabric with a dimension of 29 × 21 cm2 and a basic density of 30 g·m−<sup>2</sup> was dipped in the chitooligosaccharide solution for 2 min and dried at 100 ◦C for 12 h. This fabric was directly used for the preparation of the filtration-based composite.

#### 2.9.2. Fabrication of MXCNTCx Composite by Spray Coating

A series of MXene–CNTO composites were prepared by the spray coating process, for which 30 g·m−<sup>2</sup> of MC with a dimension of 29 × 21 cm<sup>2</sup> was used, where one CNTO layer was sandwiched between two MXene layers. The thickness of the fabric was adjusted by a spraying and drying process. The drying was done using an air-drying gun while spraying was done by using an air compressor (Keyang compressors (KAC-25), Sichuan, China). This process was repeated for Ni coated fabric in order to compare the EMI shielding of carbon fabric. The coating on carbon fabric was denoted as MXCNTCx, while the Ni coated carbon fabric was denoted as MXCNTNiCx, where x was the number of coating cycles. MXCNTC30, MXCNTC25, MXCNTC10 and MXCNTNiC25 were successfully manufactured.

#### 2.9.3. Fabrication of MXene–CNTO Composite by Filtration

100 mL of MXene colloidal solution, 100 mL of CM dispersed solution and 70 mL of dispersed CNTO were alternatively filtered through FC under gravity and dried using an air gun. The resultant composite was denoted as MXCBCM. Further, MXCB was prepared by using 100 mL of MXene colloidal solution and 70 mL of CTAB–CNTO dispersed while 100 mL of MXene colloidal solution and 70 mL of SDS–CNTO dispersed mixture were used to prepare MXCS. Finally, MXCBCFeO, MXCBCFe, MXCBCNi, MXCBCCo, and MXCBCCu were successfully prepared.

#### *2.10. Characterization*

The structural features of the composites were investigated using a high-resolution Raman spectrophotometer (Jobin Yvon, LabRam HR Evolution (Horiba, Tokyo, Japan). A Laser Flash Apparatus LFA457 (NETZSCH, Wittelsbacherstrabe, Germany) was used to measure the density of the composites. A field emission scanning electron microscope (SEM, S-4800 (Hitachi, Tokyo, Japan) was used to examine the surface morphology of the composites. XPS with a 30–400 μm spot size at 100 W of Emax (Al anode) (K-Alpha, Thermo Fisher, East Grinstead, UK) was used to analyze the chemical environment and elemental percentage of the composites. A High-power X-ray Diffractometer D/max-2500V/PC, (Ragaku, Tokyo, Japan) with Cu(Kα) was used to record the X-ray diffraction patterns of the composites. The EMI shielding effectiveness (SE) of the composite in S-band (1–3 GHz) was recorded using an EMI shielding tent ASTM-D4935-10, ASTM International (West Kentucky, PA, USA) at room temperature while X-band (8.2–12.4 GHz) EMI shielding was measured using a vector network analyzer (VNA, Agilent N5230A, Agilent Technologies, Santa Clara, CA, USA) with a sample size of 22.16 mm × 10.16 mm. The four-probe method FPP-RS8, DASOL ENG (Seoul, Korea) was used to measure the electric conductivity of the composites. The thermal stability of the composites was tested using a Thermal Analyzer DSC TMA Q400 (TA Instruments Ltd., New Castle, DE, USA). A Mitutoyo thickness 2046S dial gage (Mitutoyo, Kanagawa, Japan) was used to measure the thickness of the composites. The surface property was measured using a Phoenix-300A contact angle meter (S.E.O.Co., Ltd., Suwon, Korea). The magnetic property was measured using SQUID—VSM (Quantum Design, Inc., San Diego, CA, USA). The graphs were plotted using a Savitzky–Golay function (Origin 2017 graphing and analysis, Origin Lab (Boston, MA, USA).

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

#### *3.1. Structural Analysis*

#### 3.1.1. Scanning Electron Microscopic Analysis of Morphology

SEM was used to characterize the morphology of nanoparticles and composites, the arrangement of CNTO, nanoparticles, and MXene flakes, the structural feature of fiber, and the topography of the composites. Figure 1 illustrates the differently decorated CNTOs by different types of nanoparticles. The oxidation of the carbon nanotube mainly occurred in the tip of the carbon nanotube, as confirmed by the SEM images (Figure 1 and Figure S1), and the decoration of carbon nanotube generated a cauliflower-like structure (Figure 1a–c,e and Figure S1a,b,d). The oxidized carbon nanotube consists of a carboxylic acid functional group which can act as anchoring side of the nanoparticles. According to the Figure 1 and Figure S1, the deposition of the nanoparticle grafting occurred in the terminal of the CNTO, indicating that the oxidation predominantly happed in the tip. The oxidation of multiwall carbon nanotube produced the terminal carboxylic group which helps terminal grafting, inhibits the

aggregation and increases the solubility in water. This phenomenon leads the various morphologies in decorated CNTOs [25,26]. Fe3O4, Fe, and Cu nanoparticles behaved in a similar manner, whereas the Ni nanoparticle encircled all of the carbon nanotube and was densely packed like a cauliflower. The self-assembling of the Co nanoparticles was completely different from that of another nanoparticle used. This is because it consumed CNTO as a template and formed a structure-like bacterial chain (Figure 1d and Figure S1c) [27]. The precursor of MXene, which is Ti3AlC2, exhibited a layered structure (Figure S3h) [6]. The MILD etching created cleaves, due to the eradication of the Al layer and the evolution of the hydrogen gas (Figure 1f and Figure S3i,j). This gap is more prominent in the clay etching method (50% of HF) [18]. The etched MXene showed a layered structure with fewer gaps, and the folding of the single flake confirmed that the etching occurred. As it consisted of small gaps, it appeared like a MAX phase (Figure 1f and Figure S3i,j). Further, the presence of a single MXene flake in the composites affirmed the occurrence of effective exfoliation during the process (Figures 1f and 2b–i) [27].

**Figure 1.** SEM image of oxidized carbon nanotubes (CNTOs) decorated by (**a**) Fe3O4 (×60,000) (**b**) Fe (×30,000) (**c**) Ni (×3000) (**d**) Co (×60,000) (**e**) Cu (×50,000) and (**f**) Ti3C2Tx (×100,000).

**Figure 2.** SEM image of carbon fabric composite of (**a**) Functionalized carbon fabric (FC) (×1000) (**b**) MXCB (×500) (**c**) surface of MXCS (×10,000) (**d**) MXCBFeO (×700) (**e**) MXCBFe (×500) (**f**) MXCBNi (×500) (**g**) MXCBCCo (×10,000) (**h**) MXCBCCo (×25,000) (**i**) MXCBCCu (×200) (**j**) MXCNTC25 (×300) (**k**) MXCNTC25 (×10,000) and (**l**) MXCNTNi25 (×300).

Figure 2 illustrates the morphology of the composites. The functionalized nonwoven carbon fabric exhibited a similar morphology of nonwoven carbon fabric, where the fibers were arranged capriciously (Figure 2b,j and Figure S3d). The fibers of the carbon fabric possessed annular gaps and cracks which were occupied by CNTO, MXene, and nanoparticle [5]. This functionalization and the nanomaterials altered the property of MC and dramatically changed the structural feature of the carbon fabric (Figure 2a–l). Filtration was an effective strategy over spray coating, because filtration closed most of the gaps between fibers and interconnected the fibers with fewer defects, while MXCNTNiC25 had prominent defects (Figure 2l and Figure S3b–j) [6]. MXCB formed like a film with well interconnected fibers, whereas few pore structures remained (Figure 2b and Figure S2b). MXene, CNTO, and Fe3O4 decorated CNTOs formed a structure like roots of a tree fixed on the soil surface, and some points of the MXene flake formed an unexfoliated MXene structure. Further, a root-like nature was given by the MXene flakes (Figure 2d). MXCBFe and MXCNTC25 formed similar structures and MXCBFe generated a highly interconnected network. In addition, the MXene flakes self-assembled in a random manner and showed a similar pattern of graphene–Polyvinylidene fluoride (PVDF) coated fabric (Figure 2c,f) [6]. Furthermore, MXCBCCu exhibited a mushroom gills-like structure which was generated by MXene flakes arranged in parallel among nanoparticles and CNTO [28]. The MXCBCCo and CNi exhibited a coral like morphology (Figure 2g,h) [29]. The coral structure was formed by cubic Co nanoparticles. In order to achieve the coral structure, CCo was used as mediator (Figure S2h,i). The surface of the MXCNTC25 displayed a network of CNTO encircling the MXene flake. Furthermore, the fibers in MXCNTNiC25 were interconnected with many defects and cracks which dramatically affected the EMI shielding and conductivity (Figure 2k,l). The etching caused the introduction of new elements such as F, Cl, and O while it eradicated most of the Al from the MAX phase, leaving Ti and C (Table S1). In addition, Cl, F, and O were derived from etchant. According to the EDX analysis, the O and C were major elements present in all composites, while other metals like Co, Ni, Fe, and Cu were also present based on the precursor used to manufacture composites. As the MXene was a structural unit of the composites, the Ti and F prevailed in most of the composites, and Al and Br were also found in some composites, which are derived from MXene colloidal solution and CTAB surfactant, respectively. In addition, exfoliated MXene had more than a single layer, which consisted of the little amount of remaining Al. Further, spray coated fabric consisted of S, which was derived from the SDS surfactant used (Table S1 and Figure S2c).

#### 3.1.2. Raman Spectroscopic Analysis for Structure of Carbon-Based Material

The structural and crystalline nature of materials like MXene, graphene, and CNT can be investigated using Raman spectroscopy [2,5]. Figure 3a,b illustrate the Raman spectrum of the decorated carbon nanotube and composites as plotted between 250 and 3500 cm−<sup>1</sup> Raman shifts. The peaks at 624, 394, and 263 cm−<sup>1</sup> were attributed to the in-plane vibrational mode of surface functionalities, C, and Ti, respectively. In addition, Ti3C2Tx engendered feeble wide D and G bands at 1353 and 1568 cm−1, respectively, and the peaks at 624, 510, and 398 cm−<sup>1</sup> exhibited the presence of TiO2 anatase (Figure S4k) [30–33]. In addition, the missing peak at 263 cm−<sup>1</sup> revealed the absence of the Al layer and the fixing of new surface functionalities in the eradicated Ti–C–Al bond. The G band and G' band of the CNT and CNTO were located at 1570.9 and 2675.9 cm−1, respectively. The position of the D band slightly differed from that of the CNT peak located at 1336.3 cm−<sup>1</sup> while CNTO generated a peak at 1341.5 cm<sup>−</sup>1. In addition, CNTO gave rise to new weak peaks at 2435.5, 2916.2, and 3320.4 cm−<sup>1</sup> while CNT formed weak peaks at 2420, 2371, and 3226.2 cm−1. These differences were created due to the oxidation that was considered as oxidational effect of the carbon nanotube. Furthermore, the presence of defects and the amorphous nature of CNT generated the D band while the graphite structure induced the G band. The characteristic G' band at 2672 cm−<sup>1</sup> was formed by an overtone of the D band. Further, the level of defect present in the carbon nanotube can be explained using the ratio between ID/IG and ID/IG' [34–45]. ID/IG and ID/IG' of CNT were 0.79 and 1.39, respectively, while those of CNTO were 0.96 and 1.76, respectively. The CNTO possessing higher values of ID/IG and ID/IG' revealed that CNTO had more defect density than CNT. Hence, chemical oxidation created disorder in the carbon nanotube. The D band of the CNTO and nanoparticle decorated CNTO was located between 1340–1355 cm−<sup>1</sup> while the G and G' bands were placed between 1570–1585 cm−<sup>1</sup> and 2675–2700 cm−1, respectively. All of the carbon nanotubes and decorated CNTOs exhibited weak peaks between 2910–2943 cm−<sup>1</sup> and 3220–3240 cm−1, respectively. Further, there were extra peaks at lower Raman shift, which were due to the carbon–metal and oxygen–metal vibration modes

(Figure 3a and Table S2) [46,47]. The G band intensity of CNTO was lower than that of CNT when compared with its corresponding D band. A similar pattern was shown by CFe whereas the other decorated CNTOs exhibited a higher D band intensity, implying that the introduction of nanoparticle generates the defect. The ID/IG values of CCu, CCo, CNi, CFe, CFeO, and FC were 1.01, 1.15, 1.02, 0.92, 1.4, and 0.94, respectively, whereas the corresponding ID/IG' values were 1.84, 1.96, 1.68, 1.74, 2.71, and 4.18, respectively [47]. All of the fabric showed a similar Raman spectra pattern and a peak originated between 1339–1350 cm−<sup>1</sup> which was responsible for D band of the composites, whereas the corresponding G and G' bands laid between 1567–1586 cm−<sup>1</sup> and 2678–2688 cm−1, respectively. The non-woven carbon fabric gave rise to D and G bands at 1363 and 1592.1 cm<sup>−</sup>1, respectively, which is due to the graphite (HOPG), indicating the presence of the graphite-like structure and the generation of a feeble G' band at 2908.2 cm−1. The ID/IG values of the MC, MXCS, MXCBCCu, MXCBCCo, MXCBCNi, MXCBCFe, MXCBCFeO, and MXCB were 0.91, 1.03, 0.81, 0.87, 1.01, 0.86, 0.81, and 1.01, respectively, and the corresponding ID/IG' values were 4.13, 2.33, 1.48, 2.12, 1.88, 1.96, 2.09, and 2.04, respectively (Figure 3b and Table S2). All of the composite ID/IG values were relatively similar to MC, while the decreasing ID/IG' value of composite confirmed that the defects of the fabric were diminished significantly. The disappearing of the lower Raman shift of MXene and decorated carbon nanotube and the formation of the new peaks confirmed that the proper link occurred between the fabric, MXene, carbon nanotube, and nanoparticles.

**Figure 3.** Raman spectra of (**a**) decorated carbon nanotube and (**b**) composites.

#### 3.1.3. X-ray Diffraction (XRD) Analysis

The crystalline or amorphous nature of the material can be predicted based on XRD profile. The XRD profiles of the Ti3AlC2, Ti3C2Tx, CNT, CNTO, decorated CNTO, and composites are shown in Figure 4a,b, and drew a 2θ range between 5 to 90◦. The crystalline MAX phase generated sharp peaks at 9.52◦ (002), 19.53◦ (004), 34◦ (101), 35.1◦ (102), 36.8◦ (103), 38.99◦ (008), 41.76◦ (104), 42.54◦ (105), 48.48◦ (107), 52.36◦ (108), 56.5◦ (109), 60.16◦ (110), 52.36◦ (1011), 64.98◦ (1011), 70.34◦ (1012), 74.02◦ (118), and other miscellaneous small peaks [30–36]. Following the etching process, the corresponding MAX phase peaks vanished or shifted, and the sequence of new diffraction peaks was formed. The formed MXene held a crystalline nature and the peak at 7.14◦ (002) was a characteristic peak of MXene interplanar crystal space. In addition, the peaks originating at 14.36◦, 19.12◦, 28.98◦, 38.86◦, and 40.9◦ confirmed the crystalline nature of the MXene and attested to the occurrence of etching [32–35]. The peak shift of Ti3AlC2 from 9.59◦ to 6.96◦ and the formation of the MXene new peak at 21.57◦ indicated that the effective eradication of Al layers occurred (Figure 4a). The peak at 38.86◦ implied the remains of the layered MAX phase structure without an Al layer which confirmed the formation of MXene. Further, the separation of the layers after the etching was low; thus, the crystalline nature of the MXene remained the same as the structure of Ti3AlC2 (Figures 1f and 4a). The carbon nanotubes exhibited a crystalline nature, as confirmed by the 25.88◦ (002), 42.84◦ (100), 43.69◦ (101) 48.94◦ (102), and 54.07◦ (004) reflection peaks and implied the presence of the concentric cylindrical MWCNT [48,49]. In addition, the shifting of the position of 2θ of the corresponding MWCNT attested to the oxidation of MWCNT and increased the percentage of the sp2 hybridized carbons (Figure 4b) [36–40]. The CFeO generated peaks at 18.36◦, 30.21◦, 35.66◦, 43.33◦, 53.76◦, 57.27◦, 62.88◦, 71.37◦, and 74.42◦, and its corresponding reflection plans were (111), (220), (311), (222), (422), (511), (440), and (533), respectively [50,51]. The CFe generated peaks of zero valent iron nanoparticle at 44.73◦ (110), 64.53◦ (200), and 82.39◦ (211), and the other peaks were corresponding CNTO signals [52]. The (200) reflection peaks of CNi, CCo, and CCu were located at 51.68◦, 51.68◦, and 50.31◦, respectively. The (111) peak of CCo and CCu originated at 44.87◦ and 43.2◦, respectively [53–55]. The new peaks were raised due to the CNTOs and the aggregation of nanoparticles in the CNTOs (Figure 4b).

**Figure 4.** XRD of (**a**) MXene, MAX phase, and fabric (**b**) decorated carbon nanotubes and (**c**) composites.

The broad peaks of MC exhibited the amorphous nature of the fabric along with the 2θ peak at 25.52◦, which is similar to the peak of carbon nanotubes and confirmed the presence of a graphite structure [5,40]. Further, MC consists of various small peaks which were not above 30◦, and all small peaks disappeared with functionalization (Figure 4a). The MXene–fabric composite showed various small peaks, which were absent in FC. Hence, the introduction of the MXene, CNTOs, and nanoparticles generated many small peaks. All of the carbon fabric composites showed distinctive peaks between 20.3–21.24◦, and all of the nanocomposites showed characteristic peaks at 6.69◦ and 16.82◦, except for MXCBCFe, which showed a typical signal at 14◦. MXCBCCu, MXCBCFe, and MXCBCFeO exhibited a peak at 35.25◦ among the MXCBCCu generated shoulder peak at 38.46◦. Thus, these composites can easily be distinguished from these characteristic peaks and formed from the constitutional elements of the composites.

#### 3.1.4. X-ray Photoelectron Spectroscopy Analysis

The functionalities, surface elemental composition, structure, and bonding nature of the composites can be explained using XPS. The fitting curve of MXene was plotted using overlapping curves of the Gaussian–Lorentzian function and the overlapping curve of the composites was plotted using origin pro. The fitting curves of Ti2p, O1s, and C1s, the F1s of TI3C2TX, and the C1s of CNTO displayed different peak positions and corresponding functional groups and bonds (Figure 5 and Table S2) [56–58]. The Ti2p fitting curve disclosed five different chemical environments and the corresponding binding energies were 454.5, 456.4, 458.5, 461.3, and 464.5 eV. These binding energies indicated the presence of functional groups such as Ti–C (2p3/2), Ti2+(2p3/2), TiO2 (2p3/2), Ti2+ (2p1/2), and TiO2(2p1/2), respectively (Figure 5a). The binding energies positions of the O1s fitting curves such as 529.6, 531.1, 532.3, and 533.8 eV showed corresponding functional groups like TiO2, C–Ti–Ox, Al2O3, and H2O, respectively, (Figure 5b–d). The presence of C–Ti–Fx generated a single peak at 685.5 eV in the F1s fitting curve (Figure 5c). The C1s fitting curves at 281.1, 283.2, 284.5, and 286.1 eV confirmed the functionalities such as C–Ti–Tx, C–C, and CHx–CO. (Figure 5d) [33,35,43,59–62]. Thus, MXene formed with the formula of Ti3C2(OH, F). In the fitting curve of MWCNT, the intense peak at 284.13 eV was raised due to the C–C bond of graphite, while the presence of the oxygen generated weaker peaks between 287–291 eV, and its corresponding oxygenic species such as C=O, C–O, and carbonate were triggered. The fitting peak at 285.86 eV confirmed the presence of the defects, which was further backed by the ID/IG ratio and amount of oxygen [63,64]. The defects were comparatively low in CNT and increased by oxidation in CNTO (Figure 5e, Figure S4 and Table S3) [5]. The overlapping curve of XPS exhibited the constitutional elements and its corresponding peak positions. The constitutional elements of MXene were Cl, C, F, Ti, and O, and the corresponding binding energies were 198.7, 285.22, 685.78, 459.62, and 532.65 eV, respectively, while those given by N, Co, Ni, Fe, and Cu were 400.99, 780.1, 854.81, 710.7, and 933.65 eV, respectively (Figure 5f). In addition, the overlapping curve of the composite lying between 284.3–284.5 eV confirmed that the proper bonding occurred among the constitutional components (Figure S5) [56].

#### *3.2. Electrical Conductivity (EC) and Surface Properties*

CNT and MXene possess good EC, which is one of the factors influencing the EMI shielding effectiveness [2,57]. The electric conductivity and sheet resistance (Rs) of the fabric significantly changed due to the introduction of the MXene, CNTOs, and nanoparticles in the non-woven fabric matrix (Figure 6). The fabricated composite exhibited EC ranging from 12.5 to 2.65 S·cm−1, and Rs lay between 13.98 and 2.08 <sup>Ω</sup>·sq−1. The MXCNTNiC25 hit a maximum Rs of 13.98 <sup>Ω</sup>·sq−<sup>1</sup> and a minimum conductivity of 2.65 S·cm−1, which was due to the high surface defect (Figure 2l). The functionalization process increased EC by 10.1% and changed Rs by 20.9%. Thus, the functionalization process minimized the defect and increased the electron mobility. The filtration-based composite showed EC above 10 S·cm−<sup>1</sup> while MXCBCFeO and MXCBCFe exhibited a maximum and minimum of 12.5 and 8.81 S·cm−1, respectively. The EC of MXCB, MXCBCNi, MXCBCCo, MXCS, and MXCBCCu were 12.1, 11.3, 11.65, 11.8, and 11.22 S·cm−1, respectively, whereas the corresponding Rs were 2.08, 2.38, 2.11, 2.2, and 2.56 <sup>Ω</sup>·sq−1, respectively. Spray coated MXene–CNTO composites displayed EC below 10 S·cm−1, among which MXCNTC30 exhibited 9.55 S·cm<sup>−</sup>1. Hence, 14.62% of electric conductivity was increased by introducing MXene, CNTOs, and decorated carbon nanotubes into the MC network (Figure 6, Figure S6 and Table S6). The MXene has a high instinct electric conductivity due to its metal like nature [5]. In addition, the doping of CNT by nitrogen expands the EC while the degree of oxidation reduces the conductivity. However, the MXene–carbon fabric composite showed a maximum of 8.84 S·cm−<sup>1</sup> EC while the CNTO-carbon fabric composite displayed 16.32 S·cm−<sup>1</sup> of EC. Thus, MXene–CNTO-nanocomposite showed conductivity between the MXene and CNTO composites [2,5,24,65]. These defects limit the electron mobility and electron hopping along the fiber [66,67]. We attempted to explain this using the hydrophobic nature of the composites. The wetting ability of the surface can be explained based on the contact angle, which above 90◦ is called a water-repellent surface, while below 90◦ is considered a water loving surface. Further,

the contact angle is dependent on the surface roughness and energy. Roughness increases the defect, thus increasing the surface energy and surface roughness raise the hydrophobic nature [5,16,56]. The contact angle, wetting energy, spreading coefficient, and work of adhesion of MXCNTNiC25 were 131.3◦, −48.09 mN·m<sup>−</sup>1, −120.89 mN·m<sup>−</sup>1, and, 24.71 mN·m<sup>−</sup>1, respectively, while FC exhibited 134.18 mN·m<sup>−</sup>1, −50.7 mN·m−1, −123.54 mN·m−1, and 22.06 mN·m−1, respectively. In addition, the other composites showed no coated angle that was due to the absorption of the water or well spread on the surface of the composites. The conductivities of most of the composites were high considering the fewer surface defects while the conductivity of the MXCNTNiC25 showed the lowest conductivity, confirming that MXCNTNiC25 possessed high defect and surface roughness (Figures 2l and 6, and Figure S7).

**Figure 5.** XPS fitting curves of (**a**) Ti2p, (**b**) O1s, (**c**) C1s, (**d**) F1s, (**e**) C1s of MWCNT, and (**f**) survey of the composites.

**Figure 6.** Electric conductivities and sheet resistances of the composites.

#### *3.3. Magnetic Properties of the Composites*

The magnetic properties of the composites at 300 K were studied using hysteresis loop measured at 300 K (Figure 7 and Figure S8). The saturation magnetization (Ms), remanence (Mr), coercivity (Hc), and coefficient of squareness of hysteresis loops (Kp) can be determined using hysteresis loop [68–71]. In addition, the Kp can be calculated using the Mr/Ms ratio. Alborzi et al. and co-workers showed that decreasing the Kp value enhanced the super magnetic property, whereas diminishing the particle size decreased the Hc and increased the Ms. In addition, super magnetic material can be created by minimizing the Hc [69]. The formation of a cluster structure improves the Ms and dropping the cluster size leads to lower magnetic energy and super magnetic behavior [71]. Thus, the magnetic properties of the materials were influenced by various factors such as the geometry, size, functional groups, morphology, and crystallinity [72]. Further, precursor salt also affects the magnetic properties of the material and the synthesis of Fe3O4 by using ferrous and ferric sulfate generates 46.7 emu·g−<sup>1</sup> of Ms while ferrous and ferric chloride produce 55.4 emu·g−<sup>1</sup> of Ms and the magnetization of bulk Fe3O4 is 93 emu·g−<sup>1</sup> [69]. All the composites possessed nonlinear behavior against applied field and showed the hysteresis loop (Figure 7 and Figure S8). The Ms of the composites ranged from 0.45–0.009 emu·g−<sup>1</sup> and the saturated magnetic strength was placed between 9.95–3284.4 Oe. The Kp ranging from 0.022 to 1.128 and Mr fluctuated between 2.2 × <sup>10</sup><sup>−</sup>4–0.187 emu·g<sup>−</sup>1, whereas a 1.27–232.3 Oe range of Hc was given by the composites (Figure 7, Figure S8 and Table 1). According to Figure 7b, the magnetization of the FC approached almost zero, compared with other composites, and the hysteresis loop was not smooth, like other composites. It was considered to be due to the irregular arrangement of the fibers (Figure S8c and Figure 2a), and a study by Lu et al. showed that carbon fibers are non-magnetic materials. Thus, functionalization induced the magnetic behavior of the MC [7]. Furthermore, the Ms, Mr, and Kp values were lower than those of the others. This confirmed that interconnecting fibers using nanomaterials alters the magnetic property of the nonwoven fabric (Table 1 and Figure 2b–l). MXCBCNi and MXCBCCo behaved differently when applied magnetic field increased the magnetization and also increased while others exhibited constant magnetization (Figure 7b). A part of the loop of MXCBCFe pass through the origin, did not have negative Mr and positive Hc value, and had more than one loop, whereas MXCBCFeO did not have a positive Hc value, and instead the loop was located near the Mr point. Further, MXCBCFeO and MXCBCFe possessed high Kp values of 1.128 and 0.095, where 0.095 was lowest among all composites. Additionally, MXCBCFeO and MXCBCFe contained the lowest Ms, Hc, and saturated magnetic strength, which

affect the EMI shielding of the nanoparticle-based composites. According to this study, increasing Kp increased the EMI shielding.

**Figure 7.** Magnetization against applied field at 300 K: (**a**) Fe- and Fe3O4-based composites and (**b**) Ni, Co, Cu, and non-nanoparticle-based composites.

**Table 1.** Comparison of saturation magnetization, retentivity, and coercivity of the composites.


#### *3.4. Electromagnetic Shielding Effectiveness (EMI-SE) of Composites*

The study employed filtration and spray coating techniques, for which 30 g·m−<sup>2</sup> of areal density fabrics were used, while Ni coated fabric was 20 g·m−<sup>2</sup> of areal density. For the spray coating, MXene colloidal solution and dispersed CNTO and SDS solution were used, while each coating used 40 mL of solution. A maximum of 100 mL of colloidal solution was used in the filtration process. In spray coating composites, one MXene layer was sandwiched between two layers of CNTO, while one MXene layer placed between CNTO and nanoparticles decorated CNTO in filtrated composites. For the spray coating process, SDS was used as surfactant and CTAB was used to disperse nanoparticles decorated carbon nanotube. The spraying and filtration process changed the thickness and pore size of the fabric. The EMI shielding was performed in the region of the X band and S band. X band (8–12.4 GHz) EMI-SE was performed for all composite, while S band (1–3 GHz) EMI-SE was only measured for spray coated composites (Figure 7a). The intercalation of nanomaterials with FC and MC significantly increased the EMI shielding. The spray coated composite showed a higher EMI-SE between 1.5 to 2.6 GHz and non-woven carbon fabric preferable over Ni coated carbon fabric as it showed a lower EMI-SE than the other composites (Figure 8a,b). The non-fabrics were flexible and contained physically interconnected fibers with a three-dimensional reticular structure. When the incident wave hit the surface of the shielding materials, the reflection, multiple reflection, absorption, and transmission occurred. The strength of this mechanism varies based on the materials used for EMI-SE. According to the Simon formalism, EMI-SE depends on the electric conductivity and thickness of the material, and the length of the carbon fiber has no influence on EMI shielding. Further, hiking areal and volume density increase the electric conductivity and EMI shielding. A study by Lu et al. showed that the EMI-SE of 50 g·m−<sup>2</sup> of carbon fabric is 30.2 dB, while 30 g·m−<sup>2</sup> produces 23.1 dB, and the EMI shielding of carbon fabric is independent of the frequency range [7,73,74]. Hence, increasing the areal density of the carbon fabric increases the EMI-SE. The 30 g·m−<sup>2</sup> of MC gave rise to a maximum of 28.5 dB of EMI-SE in the S-band whereas 31.7 dB of EMI-SE was generated in the X-band region, and this was further enhanced by functionalization up to 43.9 dB of maximum (Table S5 and Figure 8a,b). This result was almost consistent with that of the Lu et al. study that EMI-SE was independent of frequency. This is despite the fact that the areal density and electric conductivity is low while functionalization had a greater effect on EMI-SE. After the functionalization, the MC possessing the magnetic property was advanced criteria for the high EMI-SE.

All the FC-based nanocomposites' minimum, maximum, and averaged shielding were above 99.99% of incident wave in the X-band region while spray coated composites' maximum shielding was just below 99.99% and the minimum and average laid between 99–99.9%. For the FC-based composite of SE, reflection (SER) and absorption (SEA) prevented 90% and 99.9%, respectively. MXCNTC30 and MXCNTC25 displayed 99.99% of maximum, minimum, and average shielding whereas the others showed 99.9% of shielding in the X-band region. In addition, spray coated composites SEA were in the range of 99–99.9% and 90% of reflection. The spray coated composite showed maximum shielding of 99.9% (Figure 8a–d and Table S5). The maximum EMI-SE shown by MXCNTC30, MXCNTC25, MXCNTNiC25, MXCNTC10, and MC in the S band were 39.6, 39.9, 34.1, 33.2, and 28.5 dB, respectively, whereas 47.3, 47.1, 34.9, 39.9, and 39.6 dB were given in the X-band region. Thus, the shielding of composites comparatively increased when the measurement frequency reached from the S band to the X band. Further, it was obvious that above 25 coating cycles, the shielding of composite was significantly reduced in the S-band region. It was considered that increasing the amount of MXene and CNTO reduced the conductivity and shielding ability of composite in the S band. When MXCNTC30 went from the S band to the X band, the EMI-SE increased by 19.4%, while it increased by 18% for MXCNTC25. (Figure 7a and Table S5). Thus, EMI-SE for nanocomposites were studied in the X-band region. By contrast, MXCNTNiC25 showed the lowest shielding in the S and X bands, due to surface defects and low electric conductivity. The defect was due to the reduced adhesive ability of MXene–CNTO composites with Ni coated carbon fabric, and the carbon fabric showed low defects (Figures 2l and 8 and Figure S3f). The MXCNTNiC25 exhibited higher EMI shielding than MC as

the porosity was diminished. Nonetheless, all the coated carbon fabric showed higher EMI shielding than MXCNTNiC25 [1–3]. The maximum EMI-SE of MXCB, MXCBCFeO, MXCBCFe, MXCBCNi, MXCBCCo, MXCBCCu, and MXCS were 47.6, 45.9, 46.7, 45, 46, 43.6, and 50.5 dB, respectively. The nanoparticle-free composites displayed higher blocking ability than others and the mushroom gills like structure significantly reduced the EMI-SE as compared with FC, whereas the coral like structure showed a higher EMI-SE than FC (Figures 2g–i and 8b and Table S5). Despite this, all showed above 45 dB, except for MXCBCCu (Figure 7 and Table S5). The absorption was dominant over reflection, and reflection stayed nearly constant for all composites. Thus, the changing of the shielding effectiveness was due to the absorption of the composites. The composite showed electric conductivity in the range of 2–13 S·cm−1, which was not sufficient to produce good SER. The effective percolation of the conductive network facilitates the electron mobility and reflection and enhances the ohmic loss [1–3,7]. The intercalation of the MXene, CNTO, and nanoparticle decorated CNTO increase the electric conductivity, though the lack of an effective conductive network minimizes the SER [7]. In this case, part of the incident wave was reflected, while most of the remaining part underwent absorption and multiple reflection. The maximum SEA of MXCB, MXCBCFeO, MXCBCFe, MXCBCCo, MXCS, MXCNTC30, and MXCNTC25, were all above 33 dB, while the SER of all the composites were above 10 dB (Figure 8c,d and Table S5). In addition, the Kp value above 0.09 led to shielding above 99.99 % (Figure 7a,b and Table 1). Thus, absorption can be achieved by different internal geometry, functionalization, ohmic loss, defects, and magnetic property of materials. In addition, the CNTO contributed inner tube scattering leads to a higher SEA because the blend of different components establishes diverse phases which enhance multiple reflection and absorption [1]. The functionalization considerably increased the specific shielding effectiveness (SSE) and absolute effectiveness (SSE/t), and the SSE and SSE/t of the FC were 401.93 dB cm−3·g−<sup>1</sup> and 12639.95 dB cm−2·g−1, respectively. The intercalations of zero, one-, and two-dimensional materials in the three-dimensional fiber network significantly reduced the SSE and SSE/t. Further, iron-based composited displayed lower SSE and comparatively lower SSE/t (Figure 8e and Table S6).

Figure 8f and Table S4 compared the EMI-SE with previous corresponding work. The dip coated CNTO non-woven fabric (basic weight 20 g·m<sup>−</sup>2) showed a maximum of 33 dB, which is substantially lower than that of the MXene–CNTO–nanoparticle composite [5]. The polystyrene–MWCNT composite showed a maximum EMI-SE of about 22 dB. The Polyvinylidene fluoride (PVDF)–MWCNT composite exhibited 28.5 dB of EMI-SE with a thickness of 0.2 cm while the segregated carbon nanotube–polypropylene composite showed an EMI-SE of 48.3 dB with 0.22 cm of thickness and interconnected MWCNT polymeric matrix exhibited 27 dB at 18 GHz [1,12,13,74,75]. In addition, the spongy carbon nanotube composite disclosed 54.8 with thickness of 0.18 cm in the X–band region [16]. Most of the study showed the EMI shielding range of the CNT composite was 20–30 dB with higher thickness. The different types of fillers and geometry altered the shielding ability of the MWCNT. Thus, we analyzed various combinations of materials with different geometries in order to increase the shielding effectiveness of the non-woven fabric and MWCNT. MXene film crammed between Polyethylene terephthalate polymer film showed an excellent EMI shielding of 92 dB with 4600 S·cm−<sup>1</sup> electrical conductivity [2]. Cao et al. and coworkers highlighted that nacre-like MXene–cellulose nanofiber composite showed an EMI shielding range of 5.3–25.8 dB based on the percentage of MXene added to one dimensional cellulose fiber [76]. The MXene–carbon fabric composite exhibited an EMI-SE of 43.2 dB [24]. Hence, a combination of the MXene–CNTO-based composite dramatically increased the EMI-SE. Further, the carbon fabric-based composite showed lower shielding; therefore, the manufactured composite was considered to have excellent EMI-SE ability (Figure 8f).

**Figure 8.** Electromagnetic interference (EMI) shielding of the composites: (**a**) Total EMI shielding (S band), (**b**) total EMI shielding (X band), (**c**) absorption (X band), (**d**) reflection (X band), (**e**) specific shielding effectiveness (SSE) and absolute effectiveness (SSE/t) of the composites, and (**f**) EMI shielding comparison of different composites.

#### *3.5. Thermal Stability and Thermo Gravimetric Analysis of Composites.*

Thermogravimetric analysis (TGA) and differential thermal analysis (DTA) were used to investigate the thermal stabilities of the composites. TGA graphs were plotted in the temperature range of 30–1000 ◦C while DTA graphs were plotted in the range of 30–900 ◦C (Figure 9a–d). The TGA and DTA analysis were carried out using Al2O3 crucible in a nitrogen environment with a heating rate of 10 ◦C·min−1. The composites exhibited an outstanding thermal stability over 100 ◦C. Most of the composites had a starting point of degradation above 140 ◦C, and the degradation beginning lay between 120–255 ◦C, where MXCNTC30 showed 120 ◦C. In addition, MXCB, MXCBCFeO, MXCBCFe, MXCBCNi, MXCBCCo, and MXCNTNiC25 prohibited degradation above 150 ◦C among mushroom gill like MXCBCCu decomposed at lower temperature of 127.5 ◦C, whereas the degradation temperatures of the other composites were below 149 ◦C and above 127.5 ◦C (except MXCNTC30 at 120 ◦C). The rapid mass changing percentage of composites occurred between 120–897.5 ◦C where spray coated composite hit the higher value (above 700 ◦C), while filtered gave rise below 653 ◦C (Figures 2i and 9a,c and Table S7). The temperature between the 30–1000 ◦C range of weight losing percentage of the composites were 6–36%, in which MXCBCFeO gave a maximum of 35.71% while 8.85% of weight loss occurred in MXCNTC25 (Figure 9a,c and Table S7).The MC and FC degradations of the whole temperature range were 6.84% and 6.4%, respectively, and the corresponding degradation beginning temperatures were 190 ◦C and 255.5 ◦C, respectively (Figure 9a,c and Table S7). The functionalization of MC by using chitiooligosaccharide increased the thermal stability. This was attributed to the fact that the chitiooligosaccharide contains lots of the hydroxyl groups, thus creating the hydrogen bonding which led to higher thermal stability [68]. According to Raagulan et al., rapid mass changes occurred due to the loss of surface functionalities, water, and decomposition of fabric [5,6,63,64]. In addition, the MXene–carbon fabric composite prohibited degradation up to 235 ◦C and a Gamage et al. study showed CNTO–carbon fabric composites prohibited degradation until 284 ◦C. The combination of MXene, CNTO, and nanoparticles considerably diminished the degradation temperature (Table S7). The DTA analysis revealed various peaks raised due the degradation of the composites (Figure 9b,d). Strong peaks were placed in the region of the rapid changing of TGA curve [77]. Further, we observed that the slope of the TGA curve depends on the intensity of the corresponding peaks in DTA analysis, and that the high intensity of the DTA peaks lead to rapid degradation. The more intense peaks were located between 209–934.2 ◦C temperature range (Figure 9b,d, Figure S9a–m and Table S8). It was obvious that the introduction of the nanomaterials in the fabric network shifted the peaks to lower temperatures and diminished the decomposition temperature.

**Figure 9.** TGA and differential thermal analysis (DTA) analysis of the composites (**a**) TGA of filtration-based composite, (**b**) TGA of filtration-based composite, (**c**) TGA of spray coated composite, and (**d**) DTA of spray coated composite.

#### **4. Conclusions**

The filtration-based nanocomposite and spray coated nanocomposite were successfully prepared. The effect of the functionalization of non-woven carbon fabric was analyzed as well. The composites were lightweight and thinner. The density of the composites ranged from 0.108 to 0.288 g·cm−<sup>3</sup> while the thickness of the composites ranged between 0.266–0.408 mm. The ranges of Ms and Mr were 0.0099–0.86 emu·g−<sup>1</sup> and 0.00022–0.44 emu·g−1, respectively. The MXCBCNi displayed a higher Ms of 0.86 emu·g−<sup>1</sup> and a 0.44 emu·g−<sup>1</sup> of Mr was shown by MXCBCFeO. The MXCCFeO exhibited a maximum Kp of 1.128 while FC was 0.022. The functionalized carbon fabric displayed a hydrophobic nature with a 134.18◦ contact angle, whereas MXCNTC25 displayed a contact angle of 131.3◦. The electric conductivities of the composites varied between 2.65–12 S/cm, whereas the surface resistances were in the range of 2.08–13.98 Ω/sq. The composites showed good thermal stability and resisted complete thermal degradation above 120◦. In addition, functionalization increased the thermal stability and prevented degradation up to 255.5◦. The maximum EMI-SE shown by MXCS was 50.5 dB, and the EMI shielding range of the composite was 50.5–28.5 dB. The maximum SER and SEA were shown by MXCB and MXCS, respectively. The ranges of SSE and SSE/t were 149.37–401.93 dB cm3·g−<sup>1</sup> and 4330.82–12,639.35 dB cm2·g<sup>−</sup>1, respectively, where FC showed maximums of both. Hence, the manufactured fabric composite presented high EMI shielding, magnetic behavior, low density, smaller thickness, and flexibility.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2079-4991/9/4/519/s1, Figure S1: SEM image of CNTO decorated by (a) Fe3O4 (b) Fe (c) Ni (d) Cu; Figure S2: SEM image of carbon fabric composite of (a) MC (500) (b) MXCB (500) (c) surface of MXCS (10000) (d) MXCBCFeO (500) (e)MXCBCFe (500) (f)MXCBCNi (450) (g) MXCBCCo (150) (h) MXCBCCo (100000) (i) MXXBCCu (45000) (j) MXCNTC30 (100) (k) MXCNTNi25 (1000) (l) MXCNT10 (400); Figure S3: SEM image of (a) MC (×500), (b) MC (×2000), (c) cracks on fiber (×300000), (d) MXene-CNTO coated carbon fabric (×500), (e) MXene-CNTO coated carbon fabric (×300), (f) MXene-CNTO coated fabric (Ni coated fabric) (×300), (g) MXene and CNTO on the surface of the fabric (×120000), (h) Ti3AlC2 (×22000), (i) Ti3C2Tx (×50000) and (j) Ti3C2Tx (×80000); Figure S4: Normalized curve of Raman spectrum of (a) decorated CNT, (b) composites (c) CNT, (d) CNTO, (e) CCu, (f) CCo, (g) CNi, (h) CFe, (i) CFeO, (j) FC (k) comparison of MXene, MAX phase and MC, (l) MC, (m) MXCS, (n) MXCBCCu, (o) MXCBCCo, (p) MXCBCNi, (q) MXCBCFe, (r) MXCBCFeO and (s) MXCB; Figure S5: The c1s fitting curve of the MXene, functionalized fabric and composites; Figure S6: Resistivity profile of the composites; Figure S7: Contact angle of (a) MXCNTNiC25 (b) FC and (c) represent other composites; Figure S8: The magnetization of composites against the applied field at 300 K (a) MXCBCCu, (b) MXCS, (c) FC, (d) MXCBCCo, (e) MXCBCNi, (f) MXCBCFe, (g) MXCBCFe and (h) MXCBCFeO; Figure S9: Comparison of the TG and DTA of all the composites (a) MXCB, (b) MXCBCFeO, (c) MXCBCFe, (d) MXCBCNi, (e) MXCBCCo, (f) MXCS, (g) MXCBCCu, (h) FC, (i) MXCNTC25, (j) MXCNTNiC25, (k) MXCNTC10, (l) MXCNTC30 and (m) MC; Table S1: Elemental percentage of composites from EDX analysis; Table S2: Raman spectra band positions; Table S3: Atomic percentage MXene, MC, CNT, CNTO and composites from XPS analysis; Table S4: Comparison of EMI SE with thickness; Table S5: Comparison of maximum (MAX), minimum (MINI), average (AVE) shielding, SSE and SSE/t of the composite in each case; Table S6: Density of the composites; Table S7: Comparison of mass changes with different temperature range from TGA analysis; Table S8: Comparison of different peak positions range from DTA analysis.

**Author Contributions:** K.Y.C. and R.B. designed the project; K.R., B.M.K. and J.J.M. performed the experiments; L.R.L. and J.L. analyzed the data; S.B.L. supervised the analysis; K.R. wrote the manuscript.

**Funding:** This research was supported by the Leading Human Resource Training Program of Regional Neo industry through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and future Planning (grant number) (NRF-2017H1D5A1043865).

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

#### **References**


© 2019 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* **Improved Catalytic Durability of Pt-Particle/ABS for H2O2 Decomposition in Contact Lens Cleaning**

**Yuji Ohkubo 1,\*, Tomonori Aoki 1, Satoshi Seino 1, Osamu Mori 2, Issaku Ito 2, Katsuyoshi Endo <sup>1</sup> and Kazuya Yamamura <sup>1</sup>**


Received: 29 January 2019; Accepted: 18 February 2019; Published: 3 March 2019

**Abstract:** In a previous study, Pt nanoparticles were supported on a substrate of acrylonitrile– butadiene–styrene copolymer (ABS) to give the ABS surface catalytic activity for H2O2 decomposition during contact lens cleaning. Although the Pt-particle/ABS catalysts exhibited considerably high specific catalytic activity for H2O2 decomposition, the catalytic activity decreased with increasing numbers of repeated usage, which meant the durability of the catalytic activity was low. Therefore, to improve the catalytic durability in this study, we proposed two types of pretreatments, as well as a combination of these treatments before supporting Pt nanoparticles on the ABS substrate. In the first method, the ABS substrate was etched, and in the second method, the surface charge of the ABS substrate was controlled. A combination of etching and surface charge control was also applied as a third method. The effects of these pretreatments on the surface morphology, surface chemical composition, deposition behavior of Pt particles, and Pt loading weight were investigated by scanning electron microscopy (SEM), X-ray photoelectron spectroscopy (XPS), cross-sectional SEM, and inductively coupled plasma atomic emission spectroscopy (ICP-AES), respectively. Both etching and controlling the surface charge effectively improved the catalytic durability for H2O2 decomposition. In addition, the combination treatment was the most effective.

**Keywords:** catalytic durability; nanoparticle; supported catalyst; radical reactions; platinum (Pt); H2O2 decomposition; contact lens cleaning

#### **1. Introduction**

The number of contact lens wearers is estimated to be approximately 140 million all over the world [1]. Contact lens materials have been improved based on demands from the wearers [2]. Contact lenses are divided into two types: disposable and extended wear. Extended-wear contact lenses can be used repeatedly, which offers long-term cost advantages. However, to prevent eye troubles, a repeatable-use-type contact lens requires daily cleaning and sterilization. Three methods are used to clean and sterilize contact lenses: boiling, and cleaning in either H2O2 or multipurpose solution (MPS). When cleaning with a MPS, a single solution plays the roles of cleaning, sterilizing, and preserving lenses. Thus, MPS cleaning is a simple method, and about 70% of contact lens wearers currently use an MPS to clean their lenses [3]. However, if contact lens wearers are not careful while using the MPS, eye troubles are likely to occur due to inadequate sterilization. Therefore, the number of wearers using H2O2 cleaning, which has higher sterilization performance, has gradually increased in recent years [3]. In H2O2 cleaning, a 35,000 ppm H2O2 solution is used to clean and sterilize contact lenses. Although the H2O2 solution exhibits high sterilizing performance, it involves the risk of eyes becoming

bloodshot or painful, and can even lead to blindness if the H2O2 solution enters the eyes without decomposing. Thus, a Pt catalyst is used to promote H2O2 decomposition to lower the residual H2O2 concentration below 100 ppm [4]. A Pt film plated on an acrylonitrile–butadiene–styrene copolymer (ABS) container using electroless plating is usually used to catalyze H2O2 decomposition. Thus, each ABS container needs a Pt loading weight of 1.5 mg. However, Pt is an expensive material. In addition, the Pt-film/ABS container must be thrown away after repeated use for a month. Therefore, a technique that decreases the amount of Pt used to clean contact lenses is needed.

In a previous report [5], we proposed replacing the Pt film with Pt nanoparticles. Several methods can synthesize and immobilize metal nanoparticles, for example: impregnation [6–8], polyol [9–11], and sonolytic methods [12–15]. These methods have the disadvantages of high processing temperature, long processing time, nonuniform deposition of the metal ions, and low producibility. Therefore, we selected a radiolytic synthesis method that uses a high-energy electron beam (EB) to synthesize and immobilize Pt nanoparticles on the ABS container. This method is called the electron-beam irradiation reduction method (EBIRM), and it offers the advantages of a low processing temperature, short processing time, highly uniform deposition, and high producibility [16–19]. We successfully decreased the Pt loading weight from 1.5 mg/substrate for Pt-film/ABS to 5.9 μg/substrate for Pt-particle/ABS and synthesized a Pt-particle/ABS catalyst having considerably higher specific catalytic activity for H2O2 decomposition than the Pt-film/ABS catalyst. However, the catalytic activity of the Pt-particle/ABS catalyst decreased with increasing number of repeated uses, although the catalytic activity of the Pt-film/ABS catalyst did not change. The decrease in catalytic activity of the Pt-particle/ABS catalyst was caused by decreasing the Pt loading weight with increasing numbers of repeated uses, not by poisoning Pt particles [5]. This problem of catalytic durability for H2O2 decomposition remains unsolved. Thus, in this study, we attempted to improve the catalytic durability using two pretreatments before EB irradiation—etching and controlling the surface charge—and a combination of both. The effects of pretreatments on ABS surface, catalytic activity, and catalytic durability were investigated.

#### **2. Results and Discussion**

#### *2.1. Effect of Pretreatment on ABS Substrate*

To examine the effect of etching on the surface morphology of the ABS substrate, etched ABS surfaces not containing Pt particles were observed using a scanning electron microscope (SEM). Figure 1 shows the SEM images of the surface of the ABS substrate before and after etching (the samples are hereafter labeled with their pretreatment and whether they contain Pt). Although no holes were present before etching, as shown in Figure 1a, many holes with a diameter of 100–500 nm appeared after etching, as shown in Figure 1b. The etching process was confirmed to dissolve butadiene rubber, thereby increasing the surface area of the ABS substrate.

To examine the effects of etching and surface charge control on the chemical composition of an ABS substrate, the chemical compositions of the pretreated ABS surfaces not containing Pt particles were investigated using X-ray photoelectron spectroscopy (XPS). Figure 2a,b shows the C1s-XPS spectra of the surface of the ABS substrate before and after etching. When the ABS substrate was etched, the intensity of the peak indexed to C–H and C–C (285 eV) decreased whereas the intensity of the peaks indexed to C=O–O (289 eV), C=O (287.5 eV), C–N and C–O (286.5 eV) increased. These results indicate that etching not only dissolves butadiene rubber, but also introduces oxygen-containing functional groups. Figure 2a,c shows the C1s-XPS spectra of the surface of the ABS substrate before and after the surface charge control treatment. When the surface charge of the ABS substrate was controlled, the intensity of the peak indexed to C–H and C–C (285 eV) decreased, whereas that of the peaks indexed to C–N and C–O (286.5 eV) increased, indicating that the ABS surface was covered with surface charge controllers. When both etching and surface charge control were performed, the C1s-XPS spectrum for ABS-Etch&Charge was shaped similarly to that for ABS-Charge, as shown in Figure 2c,d, respectively. This similarity indicates that the ABS-Etch&Charge substrate was also covered with surface charge controllers.

**Figure 1.** Scanning electron microscope (SEM) images of the surface of the acrylonitrile–butadiene–styrene copolymer (ABS) substrate not containing Pt particles before and after etching: (**a**) ABS-untreated and (**b**) ABS-Etch.

**Figure 2.** C1s-X-ray photoelectron spectroscopy (XPS) spectra of the ABS surface before (dotted lines) and after surface (solid colored lines): (**a**) ABS-untreated, (**b**) ABS-Etch, (**c**) ABS-Charge, and (**d**) ABS-Etch&Charge.

The effects of etching and surface charge control on the deposition behavior of Pt particles on four types of the ABS samples were examined. Figure 3 shows the field-emission (FE) SEM images of the surface morphology for the four types of Pt/ABS samples: Pt/ABS-untreated, Pt/ABS-Etch, Pt/ABS-Charge, and Pt/ABS-Etch&Charge. Main Pt particles with diameter of <20 nm and partial Pt particles with a diameter of 20–60 nm were observed on a surface of all four types of Pt/ABS samples. The size of the Pt particles was almost the same whether the surface was etched or its surface charge was controlled.

**Figure 3.** Field-emission (FE) SEM images of the surface morphology of four types of Pt/ABS samples: (**a**) Pt/ABS-untreated, (**b**) Pt/ABS-Etch, (**c**) Pt/ABS-Charge, and (**d**) Pt/ABS-Etch&Charge.

Cross-sectional FE-SEM images confirmed where Pt particles were deposited on etched ABS samples. Figure 4 shows the cross-sectional backscattered electron images of the Pt/ABS-Etch and Pt/ABS-Etch&Charge samples. The Pt particles were observed both on the ABS surface and in the holes opened by etching. These results indicated that etching increased not only the specific surface area of ABS, but also the number of sites for Pt deposition. In addition, the deposition behavior of Pt particles in the holes was confirmed to be almost the same as that of Pt particles on the ABS surface.

The effects of etching and surface charge control on the Pt loading weight of Pt/ABS samples were also examined. Figure 5 shows the Pt loading weights of the four types of Pt/ABS samples. The Pt loading weights for the Pt/ABS-Etch, Pt/ABS-Charge, and Pt/ABS-Etch&Charge samples were higher than that for the Pt/ABS-untreated sample. This result indicated that both types of pretreatments and their combination increased the Pt loading weight. In addition, the Pt loading weights for the Pt/ABS-Etch and Pt/ABS-Etch&Charge samples were higher than that for the Pt/ABS-Charge sample. When an ABS substrate is etched, the butadiene rubber component dissolves, which results in a larger surface area. This larger surface area would increase the Pt loading weight because of the increase in sites for the immobilization of Pt nanoparticles. The Pt loading weight per unit area of the ABS substrate covered with an electroless-plated Pt film (Pt-film/ABS), which was calculated in the previous report [5], and the maximum Pt loading weight per unit area of Pt/ABS samples were 2240 and 18.2 ng/mm2, respectively. Thus, the amount of Pt consumed for the Pt/ABS samples prepared in this study was at least 120 times less than that for the Pt-film/ABS.

**Figure 4.** Cross-sectional backscattered electron images for etched samples: (**a**) Pt/ABS-Etch and (**b**) Pt/ABS-Etch&Charge.

**Figure 5.** Pt loading weights of four types of Pt/ABS samples prepared using an electron-beam irradiation reduction method (EBIRM): Pt/ABS-untreated, Pt/ABS-Etch, Pt/ABS-Charge, and Pt/ABS-Etch&Charge.

#### *2.2. Catalytic Activity for H2O2 Decomposition*

To evaluate the catalytic activity for H2O2 decomposition, the residual H2O2 concentration was measured after employing the four types of Pt/ABS samples. Thus, the untreated ABS, Pt/ABS-untreated, Pt/ABS-Etch, Pt/ABS-Charge, and Pt/ABS-Etch&Charge samples were immersed in a 35,000 ppm H2O2 solution for 360 min. The catalytic activities of the four types of Pt/ABS samples are compared in Figure 6. The untreated ABS sample did not decompose H2O2 at all within 360 min, whereas all the Pt-supported ABS samples significantly decreased the residual H2O2 concentration from 35,000 to less than 400 ppm. Moreover, the residual H2O2 concentrations for the Pt/ABS-Etch, Pt/ABS-Charge, and Pt/ABS-Etch&Charge samples became lower than that of the Pt/ABS-untreated sample. Therefore, the two types of pretreatments and their combination improved the catalytic

activity for H2O2 decomposition. The difference in catalytic activity could be explained by the increase in Pt loading weight. The Pt/ABS-Etch&Charge sample exhibited the highest catalytic activity for H2O2 decomposition, successfully reaching the target value of 100 ppm.

**Figure 6.** Catalytic activity of the untreated ABS and Pt/ABS samples with various pretreatment: residual H2O2 concentration after immersion for 360 min.

#### *2.3. Catalytic Durability in H2O2 Decomposition*

To examine the effects of etching and surface charge control on the catalytic durability, the relation between the number of repeated uses and residual H2O2 concentration was examined. Figure 7 shows the catalytic durability of the four types of Pt/ABS samples. After the Pt/ABS-untreated catalyst was used 10 times, the residual H2O2 concentration was 3056 ppm. This result suggests that much of the Pt remained on Pt/ABS-untreated, and more than 90% of H2O2 was decomposed after using it 10 times. However, this residual H2O2 concentration increased from 305 to 3056 ppm after repeated usage, thus demonstrating insufficient durability. For the pretreated Pt/ABS samples, the residual H2O2 concentrations after using Pt/ABS-Etch, Pt/ABS-Charge, and Pt/ABS-Etch&Charge samples 10 times were 851, 713, and 479 ppm, respectively. Thus, the residual H2O2 concentration decreased in the order: Pt/ABS-untreated > Pt/ABS-Etch > Pt/ABS-Charge > Pt/ABS-Etch&Charge. This result indicates that both etching and surface charge control effectively improved the catalytic activity for H2O2 decomposition. Moreover, the combination of etching and surface charge control was the most effective. However, the residual H2O2 concentration for Pt/ABS-Etch&Charge gradually increased with the number of repeated uses, suggesting that the catalytic durability was insufficient for use in practical applications, although the catalytic durability steadily improved upon etching and surface charge control. Therefore, the desorption of Pt nanoparticles must be further prevented.

**Figure 7.** Catalytic durability of the Pt/ABS samples: relation between the number of repeated uses and the residual H2O2 concentration.

#### **3. Materials and Methods**

#### *3.1. Pretreatment and Synthesis of Pt-Particle/ABS*

Pt nanoparticles were immobilized on an ABS substrate using EBIRM according to previous studies [16,20]. The methods for washing the ABS substrate and the radiolytic synthesis of Pt-particle/ABS samples were the same as those reported in the previous article [5]. The main difference between this and previous reports was the use of pretreatments to improve the catalytic durability for H2O2 decomposition.

A commercially available 1-mm-thick ABS sheet (2-9229-01, AS-ONE, Nishi-ku, Osaka, Japan) with dimensions of 20 mm × 15 mm × 1 mm was used as the ABS substrate. First, ABS substrates were sequentially washed with ethanol (99.5%, Kishida Chemical, Chuo-ku, Osaka, Japan) and pure water for 10 min each using an ultrasonic cleaner (USK-1R, AS-ONE). Then, they were dried using an N2 gun (99.99%, Iwatani Fine Gas, Amagasaki, Hyogo, Japan). Prior to immobilizing the Pt nanoparticles, the washed substrates were pretreated via either etching, surface charge control, or both. Table 1 shows the sample conditions and IDs.


**Table 1.** Sample condition and IDs.

A potassium permanganate solution (KMnO4, 0.2 M, Fujifilm Wako Pure Chemical, Chuo-ku, Osaka, Japan) and concentrated sulfuric acid (H2SO4; 97%, Kishida Chemical) were used for etching. An etching solution of molar ratio KMnO4/H2SO4 = 0.16/3.6 prepared according to patent [21] was used to dissolve butadiene rubber on the ABS surface. The ABS substrates were immersed in this etching solution for 20 min at room temperature. Then, the etched ABS substrates were washed with pure water for 10 min using an ultrasonic cleaner, followed by drying with an N2 gun.

Hexadecyltrimethylammonium chloride (Condiriser FR Conc, Okuno Chemical Industries, Chuo-ku, Osaka, Japan) was used to control surface charge, as shown in Figure 8. The hydrocarbon part adsorbs onto the ABS surface, whereas the ammonium end group modifies the surface charge of the ABS surface to be positive. Hexachloroplatinic acid hexahydrate (H2PtCl6·6H2O; 98.5%, Wako Pure Chemical Industries) was used as the Pt precursor. H2PtCl6·6H2O becomes PtCl6 <sup>2</sup><sup>−</sup> in aqueous solution—that is, it becomes negatively charged—which is why Condiriser FR was selected. A 5% *v*/*v* solution of Condiriser FR was prepared, and ABS substrates were immersed in this surface charge controller solution for 5 min at 40 ◦C while stirring at 300 rpm using a magnetic hot stirrer (RCT Basic, IKA, Staufen, Baden-Württemberg, Germany) and a PTFE stirring bar. Excess surface charge controllers were washed away from the ABS substrates with pure water for 20 s. Then, the substrates were dried naturally at room temperature.

To immobilize the Pt nanoparticles, 4 mM solutions of H2PtCl6 were separately prepared in cylindrical polystyrene (PS) containers (diameter = 33 mm and height = 16 mm), and 2-propanol (IPA; 99.7%, Kishida Chemical) was added to the solutions to be controlled at 1% *v*/*v*. Then, the pretreated ABS substrates were immersed in the Pt precursor solutions. These PS containers with the Pt precursor solutions and the pretreated ABS substrates were then irradiated for 7 s with a high-energy EB of 4.8 MeV using the Dynamitron® accelerator at SHI-ATEX Co. Ltd., in Osaka, Japan. After EB

irradiation, the substrates were removed from the solution and were washed with pure water using an ultrasonic cleaner for 10 min to remove the unsupported Pt nanoparticles. Finally, they were dried using the N2 gun. Figure 9 schematically shows the entire process.

**Figure 8.** (**a**) Chemical formula of the surface charge controller (hexadecyltrimethylammonium chloride) and (**b**) schematic of electrostatic interactions between the surface charge controllers and Pt precursors (PtCl6 <sup>2</sup>−).

**Figure 9.** Schematic of the processes for pretreating and preparing a Pt-particle/ABS-Etch&Charge sample using an EBIRM through pretreatment: (**a**) KMnO4/H2SO4 etching to dissolve butadiene rubber on the ABS surface; (**b**) surface charge control; (**c**) immersion of the pretreated ABS substrate in the Pt precursor solution; (**d**) irradiation with an electron beam; and (**e**) removing it from the solution, washing using an ultrasonic cleaner, and drying by blowing with N2 gas.

#### *3.2. Characterization*

To confirm that butadiene rubber dissolved from the ABS surface, the ABS surface was observed before and after etching using a SEM (JCM-6000, JEOL, Akishima, Tokyo, Japan) at an accelerating voltage of 10 kV. Prior to observation, a thin layer of Au was sputtered on the ABS surfaces using a Smart Coat DII-29010SCTR (JEOL) to prevent from electrostatic charge buildup during SEM observation.

To investigate the effects of etching and surface charge control on the chemical composition of the ABS surface, XPS measurement was performed using a Quantum 2000 (Ulvac-Phi, Chigasaki, Kanagawa, Japan) attached to an Al-*K*α source at 15 kV. The area of X-ray irradiation was ∅100 μm, the pass energy was 23.50 eV, and the step size was 0.05 eV. The XPS spectra were recorded at take-off angles of 45◦. A low-speed EB and an Ar ion beam were irradiated on the measured samples during the XPS measurement to neutralize their charges.

To investigate the deposition behavior of the Pt particles on the ABS surface, the ABS surfaces of the four types of Pt/ABS samples were observed using a FE-SEM (JSM-7800F, JEOL) at an accelerating voltage of 8 kV. Prior to observation, Os was coated on the Pt/ABS surfaces via plasma chemical vapor deposition using an Osmium Coater HPC-20 (VACUUM DEVICE, Mito, Ibaragi, Japan) to prevent electrostatic charge buildup during FE-SEM observation. Cross-sectional samples of Pt/ABS-Etch and Pt/ABS-Etch&Charge were prepared using a cross section polisher (IB-09020CP, JEOL) with a broad Ar ion beam source. Cross-sectional backscattered electron images were also obtained using the same FE-SEM (JSM-7800F, JEOL).

To measure the Pt loading weight on the Pt/ABS samples, inductively coupled plasma atomic emission spectrometry (ICP-AES; ICPE-9000, Shimadzu, Kyoto, Kyoto, Japan) was utilized. Pt nanoparticles on the Pt/ABS substrates were dissolved using aqua regia (volume ratio of HCl/HNO3 = 3/1). Then, the diluted aqua regia solutions were sprayed into a plasma torch in the ICPE-9000. The amount of Pt in the Pt/ABS samples was calculated from a calibration curve of a Pt standard solution (1000 ppm, Wako Pure Chemical Industries), as shown in the Supporting Information of the previous report [5].

H2O2 decomposition is accelerated by the platinum catalysts, then water and oxygen gas are generated, as shown in Equation (1):

$$\text{H}\_2\text{H}\_2\text{O}\_2 \rightarrow 2\text{H}\_2\text{O} + \text{O}\_2\tag{1}$$

The generation of O2 bubbles were observed during the H2O2 decomposition test in the present study as well as the previous study [5]. It was clear that H2O2 decomposition was accelerated by the Pt/ABS samples. To evaluate the catalytic activity for H2O2 decomposition, the residual H2O2 concentration was measured after immersing the Pt/ABS samples in a 35,000 ppm diluted solution of H2O2 (30% *w*/*w*, Kishida Chemical) at 25 ◦C for 360 min in an incubator (i-CUBE FCI-280, AS-ONE). In the previous study, a H2O2 decomposition curve was obtained by collecting the data of the residual H2O2 concentration with different H2O2 decomposition times of 2, 5, 10, 20, 30, 60, 120, 240, and 360 min, and it was confirmed that the residual H2O2 concentration steadily decreased with increasing H2O2 decomposition times [5]. Therefore, in the present study, the residual H2O2 concentration was measured after immersion of the Pt/ABS samples for only 360 min. The method for measuring the H2O2 concentration was the same that reported in the previous article [5]. A 5% *w*/*w* diluted solution of titanium sulfate (Ti(SO4)2; 30% *w*/*w*, Wako Pure Chemical Industries) was added to the H2O2 solution to color the H2O2 solution. Then, the optical absorbance of the colored H2O2 solution was measured using a deuterium-halogen and tungsten lamp (DH-2000, Ocean Optics, Largo, FL, USA), fiber multichannel spectrometer (HR-4000, Ocean Optics), and optical fiber (P600-1-UV/VIS, Ocean Optics). The absorbance at 407 nm was used to calculate the residual H2O2 concentration from the calibration curve, as shown in the Supporting Information of the previous report [5].

To evaluate the catalytic durability for H2O2 decomposition, the residual H2O2 concentration was measured after Pt/ABS samples were repetitively used 1, 3, 5, and 10 times.

#### **4. Conclusions**

We prepared four types of Pt/ABS catalysts with EBIRM and investigated the effects of two types of pretreatments—etching, surface charge control, and the combination of both—on the ABS surface, catalytic activity, Pt loading weight, and durability of these catalysts. Etching increased the Pt loading weight because of the increase in surface area of the ABS substrate, which in turn increased the catalytic activity. Etching also increased the catalytic durability, which could be attributed to the holes created by etching, which partially prevented Pt particles from detaching from the ABS surface. Surface charge control increased the Pt loading weight, which increased both the catalytic activity and durability. These improvements could be explained by electrostatic interactions between the Pt nanoparticles and surface charge controllers on the ABS substrate. The effects of etching on the Pt loading weight and catalytic activity were larger than those of the surface charge control. In contrast, the effect of surface charge control on the catalytic durability was higher than that of etching. Finally,

the combination of etching and surface charge control most effectively improved both the catalytic activity and durability. Thus, we successfully improved the catalytic durability through either etching, surface charge control, or both before EB irradiation. Although the catalytic durability was insufficient for cleaning contact lenses in practical applications, these pretreatments would be useful for improving the adhesion between metal nanoparticles and resin substrates or microparticles except in severe conditions such as in H2O2 solution.

**Author Contributions:** Y.O., K.E. and K.Y. supervised the work. T.A. and S.S. prepared the Pt-particle/ABS samples. Y.O. and T.A. performed SEM observation, AFM observation, and XPS analysis. Y.O., T.A. and S.S. measured and calculated the Pt loading weights of the samples using ICP-AES. T.A. evaluated the catalytic activity and catalytic durability. O.M. and I.I. helped the evaluations. All authors contributed to the scientific discussion and manuscript preparation. Y.O. wrote the manuscript.

**Funding:** The research was funded by the Japan Science and Technology Agency, grant number JST No. MP27215667957 and JST No. VP29117941540.

**Acknowledgments:** We thank the staff of the SHI-ATEX Co. Ltd. for their assistance with the electron-beam irradiation experiments. We thank the staff of the Okuno Chemical Industries because hexadecyltrimethylammonium chloride (Condiriser FR Conc) was provided by them. We also thank the staff of the JEOL for their assistance with the SEM observations.

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

#### **References**


© 2019 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* **Effect of SiO2 Nanoparticles on the Performance of PVdF-HFP/Ionic Liquid Separator for Lithium-Ion Batteries**

#### **Stefano Caimi, Antoine Klaue, Hua Wu \* and Massimo Morbidelli \***

Department of Chemistry and Applied Biosciences, Institute for Chemical and Bioengineering, ETH Zurich, 8093 Zurich, Switzerland; stefano.caimi@chem.ethz.ch (S.C.); antoine.klaue@chem.ethz.ch (A.K.)

**\*** Correspondence: hua.wu@chem.ethz.ch (H.W.); massimo.morbidelli@chem.ethz.ch (M.M.);

Tel.: +41-446-320-635 (H.W.); +41-446-323-034 (M.M.)

Received: 10 October 2018; Accepted: 5 November 2018; Published: 8 November 2018

**Abstract:** Safety concerns related to the use of potentially explosive, liquid organic electrolytes in commercial high-power lithium-ion batteries are constantly rising. One promising alternative is to use thermally stable ionic liquids (ILs) as conductive media, which are however, limited by low ionic conductivity at room temperature. This can be improved by adding fillers, such as silica or alumina nanoparticles (NPs), in the polymer matrix that hosts the IL. To maximize the effect of such NPs, they have to be uniformly dispersed in the matrix while keeping their size as small as possible. In this work, starting from a water dispersion of silica NPs, we present a novel method to incorporate silica NPs at the nanoscale level (<200 nm) into PVdF-HFP polymer clusters, which are then blended with the IL solution and hot-pressed to form separators suitable for battery applications. The effect of different amounts of silica in the polymer matrix on the ionic conductivity and cyclability of the separator is investigated. A membrane containing 10 wt.% of silica (with respect to the polymer) was shown to maximize the performance of the separator, with a room temperature ionic conductivity of of 1.22 mS cm−1. The assembled half-coin cell with LiFePO4 and Li as the cathode and the anode exhibited a capacity retention of more than 80% at a current density of 2C and 60 ◦C.

**Keywords:** lithium-ion battery; ionic-liquid-based separator; hot-pressing; inorganic nanoparticle; nanocomposite; fractal cluster

#### **1. Introduction**

Recent large-scale power applications of lithium-ion batteries (LIB) including hybrid vehicles and smart grids require long cycle life, low impact on the environment and high reliability and safety [1–6]. Current technologies are based on the use of organic liquid electrolytes, which guarantee high ionic conductivity at low temperatures and long cycle stability. However, they are also highly volatile, toxic and thermally unstable and may leak out of the battery under abnormal operations [7–14]. One of the most promising alternatives to replace liquid electrolytes is the employment of ionic liquids (ILs), which possess high ion density and are characterized by high thermal stability [6,15–24]. Among the several existing ILs, those based on pyrrolidinium, and in particular Pyr13TFSI, are often considered since they exhibit low viscosities and are chemically inert towards the cell components [10,19,25,26]. The main drawback of the use of ILs is the limited ionic conductivity at low temperature, especially when mixed with the host polymer to form the separator. One possibility to increase the ionic conductivity at low temperatures is the addition of inorganic nanoparticles (NPs) such as SiO2, Al2O3, TiO2 and CeO2, to the polymer matrix [27–31]. These fillers can improve the conductivity by reducing the polymer crystallinity and by interacting with the ionic species in the electrolyte through Lewis acid–base interactions [29,32–35]. Moreover, it is well

established that the addition of inorganic NPs into the polymer matrix improves its mechanical stability, thus preventing thermal shrinkage and mechanical breakdown of the separator [35–37]. In order to maximize the effect of the added NPs, it is essential to disperse them in the polymer matrix uniformly and at the nanoscale level [38,39]. To achieve this, in this work, we start from a dispersion of silica NPs in water and mix it with an aqueous dispersion of PVdF-HFP NPs. The binary dispersion is then subjected to shear-driven gelation by passing through a microchannel where, if present alone, the polymer NPs undergo gelation whereas the silica NPs are shear-inactive (i.e., they are stable and do not aggregate). As the process occurs in few milliseconds the silica NPs cannot escape from the polymer gel network and remain entrapped and dispersed uniformly in the polymer matrix [40–42]. Poly(vinylidenefluoride-*co*-hexafluoropropylene) (PVdF-HFP) is chosen as it possesses high dielectric constant, it is chemically compatible with the electrode materials, and it is characterized by low crystallinity [27,28,39,43]. The method to form a freestanding, uniform and transparent membrane through hot-pressing, starting from the polymer/filler clusters and the IL has been developed earlier in our group and described elsewhere [44]. The effect of the presence of silica NPs in the IL-based membrane is investigated in terms of ionic conductivity and electrochemical cyclability.

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

#### *2.1. Materials*

The following chemicals have been employed without further treatments: Sodium dodecyl sulphate (SDS, purity 99%) and *N*-Propyl-*N*-Methylpyrrolidinium bis(trifluoromethane-sulfonyl)-imide (Pyr1308b, purity 99.5%) were purchased from Apollo Scientific (Bredbury, UK) and Solvionic (Toulouse, France), respectively. The water dispersion of PVdF-HFP NPs, the amorphous silica powder Tixosil 365 and bis(trifluoro methane)sulfonamide lithium salt (LiTFSI) were provided by Solvay (Bollate, Italy). The ion-exchange resin (Dowex MR-3) was purchased from Sigma-Aldrich (Steinheim, Germany). The cathode material lithium iron phosphate (LFP) is commercial grade Life Power P2 from Clariant (Muttenz, Switzerland). Before assembling, LFP and the IL-based separator were dried overnight under vacuum at 130 ◦C and 60 ◦C, respectively.

#### *2.2. Methods*

To form a suitable dispersion, Millipore water is added to the silica powder to reach a solid fraction of 30% and the mixture is mechanically stirred and repeatedly sonicated using a digital sonifier from Branson. Eventually, the dispersion is centrifuged at 2000 rpm for 10 min to remove the remaining large clusters. The binary dispersion of PVdF-HFP and SiO2 NPs is prepared by adding the silica NP dispersion dropwise to the PVdF-HFP NP dispersion under agitation. As the addition of SiO2 NPs may destabilize the PVdF-HFP dispersion due to a repartition of the adsorbed surfactant between the two species, some additional SDS is added (0.2% with respect to the solid content of the dispersion). The content of the added silica is evaluated as mass percentage with respect to the mass of polymer.

A high-shear device, HC-5000 (Microfluidics, Westwood, MA, USA), connected to a L30Z microchannel with a width of 300 <sup>μ</sup>m, a rectangular cross section of 5.26·10−<sup>8</sup> <sup>m</sup><sup>2</sup> and a length of 5.8 mm was used to perform shear-driven gelation of the PVdF-HFP/SiO2 NPs. The operating conditions for the gelation and subsequent drying to obtain the polymer-silica clusters (PSiCs) are reported elsewhere [44].

The IL solution consisted of a 0.5 M solution of LiTFSI salt dissolved into Pyr13TFSI.

The obtained clusters were mixed with the IL solution at a mass fraction of PSiC/IL equal to 30/70 wt.%, following the procedure previously reported [44]. The slurry was then transferred between two aluminum sheets in a preheated hydraulic hand-press (Rondol, Strasbourg, France) and hot-pressed at 120 ◦C and 10 kN to form the separator. The cooling phase was performed while holding the pressure. The formed separator is here referred to as the PSiCIL membrane.

Small-angle light scattering (SALS) measurements were taken using a Mastersizer 2000 (Malvern, UK) equipped with a laser having a wavelength *λ* = 633 nm to characterize the obtained clusters in terms of their average radius of gyration, *Rg* , and their fractal dimension, *df* , following the procedure reported elsewhere [45,46]. Measures of dynamic light scattering and zeta potential were conducted using Zetasizer Nano ZS 3600 from Malvern Instruments (Malvern, UK).

Differential scanning calorimetry (DSC) measurements were conducted using Q1000 instrument (TA Instruments, New Castle, DE, USA) using 40 μL crucibles in aluminum and a heating and cooling rate of 5 ◦C min−<sup>1</sup> in a nitrogen atmosphere in the temperature range from 80 to 200 ◦C. The solid content is measured using a HG53 Halogen Moisture Analyzer (Mettler-Toledo, Columbus, OH, USA). Powder XRD measurements were carried out with a X'Pert PRO-MPD diffractometer (Malvern PANalytical, Malvern, UK). Data were recorded in the 5–70◦ 2*θ* range with an angular step size of 0.05◦ and a counting time of 0.26 s per step. The peaks at 2*θ* = 18.2, 20.0, 26.6 and 38.8 correspond to the (100), (020), (110) and (021) crystalline peaks of PVdF-HFP, respectively [47].

AC (alternating current) impedance spectroscopy was used to measure the ionic conductivity of the PSiCIL membranes using a conductivity cell consisting of two stainless steel blocking electrodes. The measurement was carried out under PEIS conditions (impedance under potentiostatic mode), Δ*V* = 5 mV and frequency range from 300 kHz to 1 Hz. The resistance of the polymer electrolyte was measured and the ionic conductivity (*σ*) was obtained as follows:

$$
\sigma = \frac{d}{R\_b S} \tag{1}
$$

where *d* is the thickness of the separator, *Rb* the bulk resistance and *S* the area of the stainless steel electrode.

SEM images were taken with a Zeiss Leo 1530 (Zeiss, Oberkochen, Germany) microscopy with a field emission gun of 5 kV and platinum coating. TEM images were performed using a Morgagni 268 from FEI equipped with a tungsten emitter operated at 100 kV.

The battery consisted of CR2032 coin cells assembled using lithium iron phosphate (LFP) and lithium metal as the cathode and the counter electrode, respectively, and the PSiCIL membrane as the separator. The electrochemical tests were performed with a Land CT2001A battery tester and with a mass loading of the active material per cell equal to 4 mg. The battery assembly was performed under argon atmosphere. The cells were cycled in the voltage range 2.5–4 V (vs. Li/Li+) at 60 ◦C, at current densities from 0.1 to 2C. For charge/discharge performance characterization, 1C is defined as 170 mAh g<sup>−</sup>1.

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

#### *3.1. Preparation of the PSiCIL Separators*

The procedure to obtain the silica dispersion is reported in the experimental section. Typical properties of the obtained dispersion of SiO2 NPs in water are summarized in Table 1.

**Table 1.** Properties of the dispersion of SiO2 NPs in water.


For the application at hand, the silica NPs have to be well-dispersed in water and should be negatively charged to maintain their stability while mixed with the negatively charged PVdF-HFP NPs. As reported in Table 1, these requirements are met by the silica NPs, showing an average diameter smaller than 200 nm and a negative potential larger than 40 mV (absolute value). In order to investigate the effect of pH, Figure 1a reports the average diameter and the zeta potential of the silica NPs in the

pH range 6–12. To better appreciate the morphology of the silica NPs, Figure 1b reports a TEM image of the dried silica dispersion. From Figure 1a, it is seen that the dispersed SiO2 NPs have a constant average diameter smaller than 200 nm and are negatively charged with the zeta potential ranging from −48 to −42 mV. Moreover, from Figure 1b, it is possible to recognize that the each silica NP is a nanocluster made of 10 nm silica primary particles.

**Figure 1.** (**a**) Average diameter and zeta potential of the silica NPs as a function of pH in the silica dispersion. (**b**) TEM image of the dried silica NPs (scale 200 nm).

The chemical and physical properties of the used PVdF-HFP NP dispersion are reported in the Supplementary Materials (Table S1). The anionic surfactant was extracted from the polymer dispersion by repeated washing with ion-exchange resin Dowex MR-3 to reduce the colloidal stability of the polymer NPs and facilitate the gelation under shear.

The polymer and filler dispersions are mixed as described in the experimental section and the binary system is subjected to intense shear by forcing it to pass through a microchannel so as to have rapid gelation of the polymer NPs with the typical fractal characteristics. Since the formation of the polymer gel network occurs in few milliseconds and the silica NPs are shear-inactive (i.e., they do not aggregate under the given shear rate), the fillers, silica NPs, have no time to escape and remains entrapped in the formed matrix at a nanoscale level. In order to have complete capture of the silica nanoparticles, it is of utmost importance to obtain a compact gel after a single passage through the microchannel, as discussed elsewhere [40,48]. This depends on the solid content of the polymer/silica dispersion: the higher the solid content, the greater the compactness of the formed gel. In order to analyze the distribution of the silica NPs inside the polymer matrix, SEM pictures of the gel obtained from the microchannel are shown in Figure 2.

From Figure 2 it is evident that the shear-induced gelation is capable of dispersing uniformly and randomly the silica NPs as fillers within the polymer matrix, where the silica NPs with respect to the PVdF-HFP NPs are clearly and easily distinguishable (whiter and smaller ones, encircled in violet in Figure 2). It is worth noticing that the fillers are uniformly dispersed at the nanoscale level (i.e., the silica NPs are smaller than 200 nm). Moreover, it is possible to observe that also the PVdF-HFP NPs maintain their original identity. In order to investigate the physical properties of the pure polymer and of the composite material, DSC is performed. Figure 3 reports the results of the heating and cooling ramps of the dried gel at different contents of SiO2. The melting and crystallization temperatures as well as the crystallinity of the composite derived from the DSC heating and cooling curves are reported in Table 2.

**Figure 2.** SEM picture of the silica NPs entrapped in the polymer matrix after a single passage through the microchannel (scale 300 nm). Some silica NPs are encircled in violet to be better visualized.

**Table 2.** Properties of the composite material at increasing amount of SiO2, derived from the DSC curves in Figure 3. *Tm*: melting temperature. *Tc*: crystallization temperature. Δ*Hm*: enthalpy of melting. *Xc*: crystallinity.


The results reported in Table 2 show that the amount of SiO2 NPs affects only slightly the melting and crystallization temperatures. The melting temperature of the pure polymer PVdF-HFP results in approximately 134 ◦C as derived from the maximum of the heating curve and progressively decreases with increasing amount of SiO2 to reach approximately 131 ◦C at 30 wt.% content of silica. The crystallization temperature, on the other hand, is measured from the minimum of the cooling curve, which is approximately 98 ◦C for the pure polymer and 100 ◦C for the composite material, independently of the amount of silica. Furthermore, the enthalpy of melting and the crystallinity significantly decrease with increasing content of SiO2 NPs. These results are expected and are related to the hindered reorganization of the polymer chains due to the cross-linking centers formed by the interaction of Lewis acid groups with the polar groups (i.e., the -F atoms of the polymer chains). This interaction can stabilize the amorphous structure and facilitates the transport of Li<sup>+</sup> ions (i.e., the ionic conductivity), as observed by several authors [29,33,34,49,50].

In order to measure the crystallinity of the polymer and to investigate the effect of the introduction of the SiO2 NPs on it, XRD measurements are performed and the results are shown in Figure 4.

It is seen that the spectrum of the pure polymer confirms the partial crystallization of PVdF units in the copolymer and gives a semi-crystalline structure of PVdF-HFP [27]. Moreover, the crystallinity of the polymer has been considerably reduced upon the addition of silica NPs. The intensity of the crystalline peaks, indeed, decreases and broadens when increasing the amount of SiO2. This reduction in crystallinity is attributed to the changes of the chain conformation due to the presence of the silica NPs, which again facilitate higher ionic conduction [51,52].

As described in the experimental section, the dried PSiCs containing different percentages of SiO2 are mixed with the IL solution at a mass fraction PSiC/IL equal to 30/70 wt.% and left at rest overnight to allow full impregnation of the pores of the PSiCs by the IL solution, before being hot-pressed. After the hot-pressing, it is possible to obtain a freestanding, homogeneous and transparent 50-μm

thick IL-based membrane (referred as PSiCIL membrane) containing the silica NPs at a nanoscale level. A SEM picture showing the preservation of the internal morphology after hot-pressing has been reported in our previous work [44]. To prevent water absorption, the PSiCIL membranes are stored in a nitrogen atmosphere.

**Figure 3.** DSC of the composite gel obtained after one passage through the microchannel at different percentages of the fillers, silica NPs. Blue curve: PVdF-HFP; Red curve: PVdF-HFP + 10% SiO2; Black curve: PVdF-HFP + 20% SiO2; Green curve: PVdF-HFP + 30% SiO2.

**Figure 4.** XRD of the composite gel obtained after one passage through the microchannel at different percentages of the fillers, silica NPs. Blue curve: PVdF-HFP; Red curve: PVdF-HFP + 10% SiO2; Black curve: PVdF-HFP + 20% SiO2; Green curve: PVdF-HFP + 30% SiO2.

#### *3.2. Electrochemical Properties of the PSiCIL Separators*

The ionic conductivity of the PSiCIL membranes was measured by AC impedance spectroscopy in the temperature range from 25 to 80 ◦C, at increasing percentages of silica (with respect to the polymer) in the PSiC, and the results are reported in Figure 5 and Table S2 in the Supplementary Materials. As can be seen in Figure 5, the ionic conductivity improves as the temperature increases. This can be understood by taking into account the effect of the temperature on the viscosity of the IL,

which decreases as the temperature rises. The introduction of SiO2 NPs within the polymer matrix substantially improves the ionic conductivity of the PSiCIL membrane. In particular, the room temperature conductivity increases from 0.51 mS cm−<sup>1</sup> for the pure polymer to 1.04, 1.22 and 0.71 mS cm−<sup>1</sup> with 5%, 10%, and 15% of SiO2, respectively. The same trend is observed in the entire temperature range, where the membrane containing 10% of SiO2 NPs shows the highest values of ionic conductivity reaching 1.77, 2.51, 2.95 and 3.23 mS cm−<sup>1</sup> at 40, 55, 70 and 80 ◦C, respectively (the conductivity data for the other samples are summarized in Table S2 in the Supplementary Materials). Interestingly, it is observed that the increase in conductivity is not a linear function of the SiO2 content. The conductivity values corresponding to the SiO2 content of 15% are lower than those at an SiO2 content of 5% and 10%. This behavior has been previously observed in the literature and attributed to the fact that at low filler concentrations the interaction between polymer matrix and SiO2 NPs facilitates the transport of Li<sup>+</sup> ions. However, when the SiO2 concentration reaches a certain level, the dilution effect predominates and the ionic conductivity decreases [27,53]. As reported by Stephan et al. [27], the highest conductivity is reached when the filler content ranges from 8 to 10 wt.%. The values of the ionic conductivity obtained in this work are in line with those previously reported in the literature for polymer/IL separators [44,54].

**Figure 5.** Ionic conductivity at 25, 40, 55, 70 and 80 ◦C of the PSiCIL membranes containing 70 wt.% of IL and 0% (blue squares) [44], 5% (green diamonds), 10% (red triangles), and 15% (black circles) of SiO2, respectively.

The membrane with an SiO2 content of 10%, showing the highest ionic conductivity, was used as separator to assemble CR2032 coin cells, with LFP and lithium metal as electrodes. The corresponding discharge capacity was measured as a function of the applied current density and compared in Figure 6 with the values obtained using the membrane made of the PVdF-HFP/IL composite without silica [44]. It is seen that the discharge capacities of the two membranes are almost identical up to a current density of 1C, with a capacity retention higher than 90% with respect to the initial cycles at 0.1C. At a current density of 2C, a clear difference between the two membranes is observed, where the normalized discharge capacity is 79.5% and 83.3% for the separators with 0 and 10% of SiO2, respectively. Such a superior high capacity retention can be attributed to the fractal structure of the polymer clusters and to the bicontinuous morphology of the separator. Indeed, the high and well-controlled porosity formed via shear-gelation is preserved during the hot-pressing phase, where the IL solution fully impregnates the pores forming a multitude of channels through which the ions can flow, thus showing limited loss of capacity at high current density. On the other hand, the positive contribution of the silica NPs can be attributed to two factors. Firstly, the silica NPs might hinder the reorganization, even if

already limited, of the polymer particles during the membrane formation process, thus leaving more channels open to the ions transfer. Secondly, as discussed in the previous paragraph, the performance at high C-rates are improved because of the same silica/polymer interactions which favor the ionic conductivity. At low current densities this is not observed as the existing channels, given the fractal geometry of the polymer clusters, are well-developed and the ion transfer is not limited by them. It is worth mentioning that after the cycles at 2C, the battery is tested again at 0.2C showing a recovery for the membranes containing 0 and 10% of SiO2 of 97.5% and 99%, respectively, thus showing limited performance loss during the cycles at high current densities. It is also worth pointing out that the cycle efficiency remained close to unity during all cycles. Moreover, the very low dependence of the discharge capacity on the applied current density is not commonly observed when considering earlier literature results [36,52,55].

**Figure 6.** Discharge capacity, normalized with respect to the initial cycle at 0.1C, at 60 ◦C of the PSiCIL membranes containing 70 wt.% of IL with 0% (blue squares) and 10% (red triangles) of SiO2, respectively.

#### **4. Conclusions**

In this work, we have analyzed the effect of dispersing silica NPs into PVdF-HFP/IL membranes on the ionic conductivity and discharge capacity of lithium-ion batteries. In particular, starting from the corresponding powder, we have formed a stable water dispersion of silica NPs, which could be mixed with a PVdF-HFP NP dispersion, to form a binary dispersion which was then subjected to intense shear-driven gelation. As the gelation occurs extremely fast, the silica NPs cannot escape during the gel network formation and remain entrapped and dispersed into the polymer matrix at the nanoscale level. The introduction of silica NPs into the polymer matrix was shown via DSC and XRD to reduce the crystallinity of the polymer, thus stabilizing the amorphous structure and facilitating the transport of Li<sup>+</sup> ions.

The so-produced PVdF-HFP-SiO2 composite clusters (PSiCs) were mixed with an IL solution and hot-pressed to form a membrane, so as to analyze the effect of the silica NPs on its electrochemical performance. It was observed that the ionic conductivity increases as the SiO2 content increases. The ionic conductivity reaches a maximum at an SiO2 content of 10%, being 1.22 mS cm−<sup>1</sup> at room temperature, and then decreases as the SiO2 content further increases. The membrane formed with 10% SiO2 was used to assemble coin cells and tested for cyclability at different C-rates at 60 ◦C. At low current densities, no significant differences between the membranes with 0% and 10% silica were observed and the measured discharge capacities at 1C were higher than 90% of the ones measured at 0.1C, showing excellent capacity retention even at high current densities. At 2C, the membrane

containing 10% silica performed better, showing a discharge capacity of 83.3%, compared to 79.5% of the membrane containing no silica. This can be attributed to the positive effect of the dispersed SiO2 NPs, which, on one side hinder the reorganization of the polymer NPs, thus reducing the crystallinity and increasing the amorphous phase, and on the other side, favor the transfer of ions because of their interaction with the polymer matrix.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2079-4991/8/11/926/s1, Table S1: Properties of the aqueous dispersion of PVdF-HFP NPs, Table S2: Ionic conductivity results.

**Author Contributions:** Conceptualization, S.C., H.W. and M.M.; Investigation, S.C., A.K.; Writing—Original Drat Preparation, S.C.; Writing—Review & Editing, S.C., A.K., H.W. and M.M.; Supervision, S.C., H.W. and M.M.

**Funding:** This research was funded by the Swiss National Science Foundation grant number 200020\_165917.

**Acknowledgments:** The authors would like to express their special thanks to Lu Jin for the TEM image, Christine Hamon and Riccardo Pieri for their hints, and to Maurizio Biso for the electrochemical tests. The PVdF-HFP NP dispersion was supplied by Solvay (Italy).

**Conflicts of Interest:** The authors declare no conflict of interest. The founding sponsors had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **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* **Multifunctional Platform Based on Electroactive Polymers and Silica Nanoparticles for Tissue Engineering Applications**

**Sylvie Ribeiro 1,2, Tânia Ribeiro 3,4, Clarisse Ribeiro 1,5,\*, Daniela M. Correia 6,7, José P. Sequeira Farinha 3,4, Andreia Castro Gomes 2, Carlos Baleizão 3,4 and Senentxu Lanceros-Méndez 7,8**


Received: 25 October 2018; Accepted: 6 November 2018; Published: 9 November 2018

**Abstract:** Poly(vinylidene fluoride) nanocomposites processed with different morphologies, such as porous and non-porous films and fibres, have been prepared with silica nanoparticles (SiNPs) of varying diameter (17, 100, 160 and 300 nm), which in turn have encapsulated perylenediimide (PDI), a fluorescent molecule. The structural, morphological, optical, thermal, and mechanical properties of the nanocomposites, with SiNP filler concentration up to 16 wt %, were evaluated. Furthermore, cytotoxicity and cell proliferation studies were performed. All SiNPs are negatively charged independently of the pH and more stable from pH 5 upwards. The introduction of SiNPs within the polymer matrix increases the contact angle independently of the nanoparticle diameter. Moreover, the smallest ones (17 nm) also improve the PVDF Young's modulus. The filler diameter, physico-chemical, thermal and mechanical properties of the polymer matrix were not significantly affected. Finally, the SiNPs' inclusion does not induce cytotoxicity in murine myoblasts (C2C12) after 72 h of contact and proliferation studies reveal that the prepared composites represent a suitable platform for tissue engineering applications, as they allow us to combine the biocompatibility and piezoelectricity of the polymer with the possible functionalization and drug encapsulation and release of the SiNP.

**Keywords:** nanostructures; polymer matrix composites (PMCs); mechanical properties; thermal properties

#### **1. Introduction**

The development of advanced multifunctional materials is essential for the development of society [1]. Nanocomposites are among the most important materials for an increasing number of applications due to the possibility of designing materials with tailored properties meeting specific application demands in areas ranging from the automotive industry [2,3] to food packaging [4,5] and tissue engineering [6,7], among others. The introduction of inorganic nanomaterials into polymers allows for the combination of the rigidity and high thermal stability of the inorganic material with the ductibility, flexibility and processability of the organic polymers [8], as well as the introduction/tuning of further functionalities such as magnetic [9], mechanical [10] or electrical properties [11,12]. Typical nanomaterials include nanoparticles, nanotubes, nanofibres, fullerenes and nanowires [13]. Among nanomaterials, silica is widely present in the environment and has several key features [14].

Properties of silica nanoparticles (SiNPs), such as high mechanical strength, permeability, thermal and chemical stability, relatively low refractive index, high surface area, and the fact of being used for coatings of other particles, such as magnetic and quantum dots [15–17], make these nanoparticles highly interesting for various applications [18]. Furthermore, their biocompatibility and the different possibilities of functionalizing them are the basis of their potential for biomedicine and tissue engineering applications [19]. Silica nanostructures have been extensively used as supports or carriers in drug delivery [20,21], nanomedicine [22,23] and bioanalysis [24]. Their characteristics can be tuned during synthesis to obtain a wide range of particle diameters ranging from 20 to 500 nm, different pore sizes and the incorporation of molecules such as drugs or fluorophores [24], as well as magnetic nanostructures [25]. Mesoporous silica nanoparticles (MSNPs) [20,26] have attracted particular attention for their functionalization versatility. Silica-based mesoporous nanoparticles, due to the strong Si-O bond compared to niosomes, liposomes and dendrimers, are more resistant to degradation and mechanical stress, obviating the need for any external stabilization of the MSNPs [27,28].

With respect to tissue engineering, different tissues require different microenvironments for suitable regeneration [29]. Thus, muscle tissue has electromechanical responses and needs electrical stimulation to support ionic exchange, mainly sodium by calcium ion [30]. In this context, electroactive polymers such as magnetoelectric [31,32], piezoelectric and conductive polymers [33], among others [34], show strong potential for tissue engineering applications. Among the different electroactive polymers, piezoelectric polymers have already shown their suitability for tissue engineering [6,29] due to their capacity to vary surface charge when a mechanical load is applied or vice versa. These materials can play a significant role because electric stimulation can be found in many living tissues of the human body, namely nerves [35] and bones [36,37], and it can provide the electromechanical solicitations for muscle tissue [38]. Poly(vinylidene fluoride) (PVDF) is the biocompatible piezoelectric polymer with the highest piezoelectric response. It can crystallize in four differentiated crystalline phases, α, β, γ, δ, with the β-phase being the one with the highest piezoelectric coefficient. Furthermore, it can be processed in different morphologies, including fibres, spheres, membranes and 3D scaffolds [29,39], providing a suitable platform for tissue engineering.

In order to further exploit the applicability of PVDF in regenerative medicine, polymer nanocomposites based on PVDF using silica nanoparticles with different diameters were prepared, improving the electroactive characteristics of PVDF with the aforementioned characteristics of MSNPs for biomedical applications. Together with the physico-chemical characteristics of the developed composites, their biocompatibility was evaluated in murine myoblast cells.

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

#### *2.1. Materials*

PVDF (Solef 1010) was purchased from Solvay, *N*,*N*-dimethylformamide (DMF) from Merck. Absolute ethanol (EtOH, Panreac, Barcelona, Spain, 99.5%), ammonium hydroxide solution (NH4OH, 28% in water, Fluka, Carnaxide, Portugal) and tetraethyl orthosilicate (TEOS, Sigma-Aldrich, Sintra, Portugal, 99%) were used as received. Deionized water from a Millipore (Oeiras, Portugal) system

Milli-Q ≥ 18 MΩ cm was used in the synthesis of the silica nanoparticles. Perylenediimide derivative (PDI) was synthesized according to the literature [40].

#### *2.2. Silica Nanoparticles*

#### 2.2.1. Preparation of the Silica Nanoparticles

Fluorescent silica nanoparticles, doped with PDI were prepared by a modified Stöber method [41,42]. Water, absolute ethanol, and PDI (previously dispersed in ethanol, 1 × <sup>10</sup>−<sup>6</sup> M) were mixed and after 30 min the ammonia solution was added to the mixture, followed by TEOS. The reaction was kept under stirring at constant temperature for 24 h. After that time, the nanoparticles were recovered and washed with ethanol (three cycles of centrifugation). The nanoparticles were redispersed in ethanol and dried at 50 ◦C in a ventilated oven. The experimental details are provided in Table 1.

**Table 1.** Experimental details used for the preparation of the SiNPs.


#### 2.2.2. Characterization of the SiNPs

Transmission electron microscopy: Transmission electron microscopy (TEM) images were obtained on a Hitachi (Krefeld, Germany) transmission electron microscope (model H-8100 with a LaB6 filament) with an acceleration voltage of 200 kV. One drop of the dispersion of particles in ethanol was placed on a carbon grid and dried in air before observation. The images were processed with the Fiji software (Madison, WI, USA).

Zeta potential: The surface charge of the nanoparticles was estimated with the use of zeta potential (Zetasizer NANO ZS-ZEN3600, Malvern). The zeta potential of the fluorescent SiNPs with different diameters were evaluated at different pH values (3, 5, 7, 11, 13). To adjust the pH, it was used a solutions of HCl (1M) and NaOH (1M). The average value and standard deviation for each sample were obtained from six measurements.

#### *2.3. Nanocomposite Samples*

#### 2.3.1. Preparation of the SiNPs/PVDF Nanocomposites

SiNPs/PVDF nanocomposites with 16 wt % of SiNPs were prepared by dispersing the respective mass of SiNPs in the DMF solvent within an ultrasound bath for 4 h at room temperature. The filler concentration was selected based in [43], as it shows a suitable filler content without compromising the mechanical characteristics of the polymer matrix and allowing a suitable dispersion of the filler. After we obtained a good dispersion of the nanoparticles, PVDF was added ay a concentration of 15% (*w*/*w*) and the solution was magnetically stirred at room temperature until the complete dissolution of the polymer. The materials were then prepared by different production methods [39].

First, SiNPs/PVDF samples (porous and non-porous films) were prepared by solution casting on a clean glass substrate and, in some cases, melted at different temperatures for different times (Table 2). The different preparation conditions allowed us to tailor the porosity and to study the possibility of the nucleation of the electroactive β-phase of the polymer by the fillers [44]. The thickness of the films ranges from 30 to 50 μm.


**Table 2.** Denomination, relevant preparation conditions and morphology of the PVDF and nanocomposite samples prepared in this work.

For SiNPs/PVDF electrospun fibre mats, the solution was placed in a plastic syringe (10 mL) fitted with a steel needle with inner diameter of 0.5 mm. After an optimization procedure, electrospinning was conducted with a high-voltage power supply from Glasman (model PS/FC30P04, Radeberg, Germany) at 14 kV with a feed rate of 0.5 mL·h−<sup>1</sup> (with a syringe pump from Syringepump, Porto, Portugal). The electrospun fibres were collected in an aluminium plate (placed 20 cm from the needle) and in a rotating drum (1500 rpm) to obtain random and oriented nanofibres, respectively.

Table 2 summarizes the main characteristics of the samples and the corresponding denomination that refers the type of sample and processing temperature, the nanoparticle diameter and the composite morphology. For example, F90-17NP is a film (F) obtained at 90 ◦C (90) with nanoparticles with a diameter of 17 nm (17), which is non-porous (NP).

#### 2.3.2. Characterization of the Nanocomposite Samples

Scanning electron microscopy: A desktop scanning electron microscope (SEM) (Phenom ProX, Eindhoven, The Netherlands) was used to observe the morphology and microstructure of the PVDF and SiNPs/PVDF nanocomposites. This technique was also used to observe the cell morphology seeded on the different fibrous samples. All the samples were added to the aluminium pin stubs with electrically conductive carbon adhesive tape (PELCO TabsTM, Redding, CA, USA). The aluminium pin stub was then placed on a phenom Charge Reduction sample Holder. All results were acquired using the ProSuite software (Waarschoot, Belgium). The images were obtained with an acceleration voltage of 10 kV.

Laser scanning confocal fluorescence microscopy: Laser scanning confocal fluorescence microscopy (LSCFM) images were obtained with a Leica TCS SP5 laser scanning microscope (Leica Microsystems CMS GmbH, Mannheim, Germany) using an inverted microscope (DMI6000), a HCX PL APO CS 10× dry immersion objective (10× magnification and 0.4 numerical aperture) and a HC PL FLUOTAR 50× dry immersion objective (50× magnification and 0.8 numerical aperture). Imaging used the 488 nm line of an argon ion laser.

Contact angle measurements: Water contact angle (CA) measurements (sessile drop in dynamic mode) were performed at room temperature in a Data Physics OCA20 (Data Physics, Filderstadt, Germany) setup using ultrapure water as the test liquid. The samples wettability was determined by using water drops (3 μL) placed onto the surface of the samples. Each sample was measured at six different locations and the mean contact angle and standard deviation were calculated.

Fourier transform infrared spectroscopy: Fourier transform infrared spectroscopy (FTIR) measurements in attenuated total reflectance (ATR) were performed at room temperature, using a Nicolet Nexus 670 FTIR-spectrophotometer (ThermoFisher Scientific, Porto Salvo, Portugal) with

Smart Orbit Accessory equipment (ThermoFisher Scientific, Porto Salvo, Portugal). The analysis was performed from 4000 to 600 cm−1, after 64 scans with a resolution of 4 cm−1. The spectra of each sample was used to determine the relative content of the electroactive β-phase in the composite samples, by using the method presented in [44]. In short, the ®-phase content (F®) was calculated by Equation (1):

$$F\_{\beta} = \frac{A\_{\beta}}{\left(\frac{K\_{\beta}}{K\_{a}}\right) \times A\_{a} + A\_{\beta}} \, ^{\prime} \tag{1}$$

where *<sup>A</sup><sup>β</sup>* are the absorbance at 840 cm−<sup>1</sup> and *<sup>K</sup><sup>β</sup>* = 7.7 × <sup>10</sup><sup>4</sup> cm2·mol−<sup>1</sup> is the absorption coefficients and correspond to the <sup>β</sup> phase. A<sup>α</sup> is the absorbance at 760 cm−<sup>1</sup> and *<sup>K</sup><sup>α</sup>* = 6.1 × 104 cm2·mol−<sup>1</sup> is the absorption coefficient, and correspond to the α phase.

Thermal properties: Differential scanning calorimetry (DSC) was carried out with a DSC 6000 Perkin Elmer (Mettler Toledo, Columbus, OH, USA) instrument. The samples were heated from 30 to 200 ◦C at a rate of 10 ◦C·min−<sup>1</sup> under a flowing nitrogen atmosphere. Samples were cut from the middle region of the samples and placed in aluminium pans.

From the melting in the DSC thermograms, the degree crystallinity (*Xc*) of the samples was calculated by the following equation [44]:

$$X\_{\varepsilon} = \frac{\Delta H\_f}{x \Delta H\_a + y \Delta H\_\beta} \tag{2}$$

where Δ*Hf* is the melting enthalpy of the sample, *x* and *y* represent the *α* and *β* phase contents present in the sample, respectively, and Δ*H<sup>α</sup>* and Δ*H<sup>β</sup>* are the melting enthalpies for a 100% α-PVDF (93.04 J·g<sup>−</sup>1) and <sup>β</sup>-PVDF (104.4 J·g<sup>−</sup>1) crystalline samples respectively.

Mechanical characterization: Mechanical measurements were performed with a universal testing machine (Shimadzu model AG-IS, Kyoto, Japan) at room temperature, in tensile mode at a test velocity of 1 mm·min−1, with a load cell of 50 N. The tests were performed on rectangular samples (30 × 10 mm) with a thickness between 30 and 50 μm (Fischer Dualscope 603-478, digital micrometer, Windsor, CT, USA). The mechanical parameters were calculated from the average of triplicate measurements. Hook's law was used to obtain the effective Young's modulus (E) of PVDF and SiNPs/PVDF nanocomposite samples in the linear zone of elasticity between 0 and 1% strain.

#### *2.4. Cell Culture Experiments*

#### 2.4.1. Sample Sterilization

The samples were sterilized by multiple immersions into 70% ethanol for 30 min each and to remove any residual solvent, they were washed five times in a phosphate buffered saline (PBS) 1× solution for 5 min each. Each side of the samples was then exposed to ultraviolet (UV) light for 1 h.

#### 2.4.2. Cell Culture

Murine myoblasts (C2C12 cell line) were cultivated in Dulbecco's Modified Eagle's Medium (DMEM, Gibco, Porto Salvo, Portugal) with 4.5 g·L−<sup>1</sup> containing 10% of Foetal Bovine Serum (FBS, Biochrom, Cambridge, UK) and 1% of Penicillin/Streptomycin (P/S, Biochrom). The cells were grown in a 75 cm2 cell-culture flask at 37 ◦C in a humidified air containing 5% CO2 atmosphere. Every two days, the culture medium was changed. The cells were trypsinized with 0.05% trypsin-EDTA when they reached 60–70% confluence. For the cytotoxicity assays, SiNPs/PVDF nanocomposites with different morphologies were cut according to the ISO\_10993-12. The extraction ratio (surface area or mass/volume) was 6 cm2.mL−1. To analyse cell morphology and viability, the materials were cut into 6 mm diameter. PVDF films without nanoparticles were used as the control.

#### 2.4.3. Cytotoxicity Assay by the Indirect Contact

C2C12 cells were seeded at the density of 2 × 104 cells·mL−<sup>1</sup> in 96-well tissue culture polystyrene plates. Cells were allowed to attach for 24 h, after which the culture medium was removed and the conditioned medium (the medium that was in contact with the samples) was added to the wells (100 μL). Afterwards, the cells were incubated for 24 or 72 h, and the number of viable cells was quantified by (3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) (MTT) assay. The cells received MTT solution (5 mg·mL−<sup>1</sup> in PBS dissolved in DMEM in proportion of 10%) and were incubated in the dark at 37 ◦C for 2 h. The medium was then removed and 100 μL of DMSO/well were added to dissolve the precipitated formazan. The quantification was determined by measuring the absorbance at 570 nm using a microplate reader. All quantitative results were obtained from four replicate samples and controls and were analysed as the average of viability ± standard deviation (SD).

#### 2.4.4. Direct Contact and Proliferation

Since MTT interferes with the materials, we chose the MTS as having the same theoretical basis but a soluble reaction product. C2C12 cells (4000) were seeded on each sample. After 24 h and 72 h, the viable cell number was determined using the (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl) -2-(4-sulfophenyl)-2H-tetrazolium) (MTS) assay. At the desired time points, the MTS reagent was added into each well in a 1:5 proportion of DMEM medium, and incubated at 37 ◦C for 2 h. The absorbance was detected at 490 nm with a microplate reader. Experimental data were obtained from four replicates.

#### 2.4.5. Immunofluorescence Staining

Using the same time points as in the proliferation assays, the nanocomposite samples were subjected to immunofluorescence staining to analyse the cytoskeleton morphology of the cells, also verifying the cell viability and adhesion. At each time point, the medium of each well was removed, the samples were washed with PBS and the cells fixed with 4% formaldehyde for 10 min at 37 ◦C in a 5% CO2 incubator. After fixation, the samples were washed with PBS 1× (three times) and incubated for 45 min at room temperature in 0.1 μg mL−<sup>1</sup> of green phalloidin (Sigma-Aldrich, Sintra, Portugal). Then, the samples were incubated for 5 min with 1 μg mL−<sup>1</sup> of 4,6-diamidino-2-phenylindole (DAPI, Sigma). Afterwards, the samples were washed again with PBS 1× (three times) and one time with distillate water. Finally, the samples were visualized with fluorescence microscopy (Olympus BX51 Microscope, Lisboa, Portugal).

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

#### *3.1. Silica Nanoparticles*

#### 3.1.1. Morphology and Size of the Nanoparticles

The morphology and the size of the SiNPs were analysed from TEM images (Figure 1). The spherical nanoparticles prepared by the Stober method [45] were prepared in four different diameters: 17 ± 2, 100 ± 18, 160 ± 17 and 300 ± 37 nm. The corresponding histograms are presented as insets in Figure 1.

**Figure 1.** TEM images of SiNPs-PDI with different particle size: (**a**) 17 ± 2 nm, (**b**) 100 ± 1 m, (**c**) 160 ± 1 m and (**d**) 300 ± 3 m.

#### 3.1.2. Surface Charge of the Nanoparticles

Figure 2 shows the zeta potential of aqueous dispersions of the different SiNPs at different pH to analyse the periphery charge of the particles.

**Figure 2.** Zeta potential of the different SiNPs nanoparticles at different pH.

The particles are considered more stable with a zeta potential above +30 mV or below −30 mV. This fact is due to the electrostatic repulsions between the nanoparticles that prevent their aggregation. Figure 2 shows that all nanoparticles are more stable at pH ≥ 5, independently of their average diameter. On the other hand, nanoparticles with higher average diameters are more stable. The isoelectric point of SiNPs is close to pH 2 so, from this pH upwards, the silica nanoparticles are negatively charged in acidic, neutral and basic environments, which can be taken advantage of as it has been demonstrated that the interactions between negatively charged nanoparticle surfaces and the positive charge density

of the CH2 groups of the PVDF polymer can promote the nucleation of the electroactive β-phase of the polymer [46].

#### *3.2. SiNPs/PVDF Nanocomposite Samples*

#### 3.2.1. Morphology of the Nanocomposites

The morphology of the nanocomposites was assessed by SEM. Figure 3 shows the different morphologies obtained after the different processing methods as well as the variations due to the introduction of fillers with different diameters. Figure 3 shows the cross section (Figure 3a–c) of the nanocomposites and electrospun fibres samples (Figure 3d) with 16 wt % of SiNPs. Figure 3a,b present the differences between the samples obtained at 90 ◦C with SiNPs of different diameters, showing that the higher diameter particles are well-dispersed in the PVDF polymer matrix, in contrast to the SiNPs with lower diameter that present particle agglomerates. Furthermore, a small porosity is observed (Figure 3a), which is in agreement with the literature [47].

It is important to note that the nanoparticles act as nucleation agents for crystallization in PVDF composites [48], which can be verified with the results obtained, indicating a good interfacial interaction between the PVDF chains and silica nanoparticles.

Figure 3a,c shows the differences in composite morphology due to the crystallization process. The samples obtained at 90 ◦C (Figure 3a,b) present a slightly more porous morphology than the ones obtained at 210 ◦C (Figure 3c).

**Figure 3.** Cross section SEM micrographs of SiNPs/PVDF nanocomposite samples with nanoparticles of different diameters and different processing conditions: (**a**) F90-17NP, (**b**) F90-300NP, (**c**) F210-17NP, (**d**) R-17P.

Once the SiNPs of 17 nm do not show a suitable dispersion in the films, electrospinning was used in order to produce fibres with well-dispersed particles. Relative to the fibres (Figure 3d), smooth randomly oriented fibres with encapsulated particles are observed, with no particles at the surface.

This result is confirmed by the confocal images represented in Figure 4. It was observed that the introduction of the particles increases the fibre diameter (243 ± 89 nm to 339 ± 92 nm). Oriented fibres with SiNPs were also produced (data not shown), verifying the particles' encapsulation within the fibres and a fibre diameter of 683 ± 140 nm. The increase of fibre diameter with the incorporation of the SiNPs is attributed to the higher viscosity of the solution, with also hinders fibre stretching by the applied field. The higher diameter of the oriented fibres relative to the randomly oriented fibres is attributed to the merging of aligned fibrils that crystallize simultaneously [49].

**Figure 4.** Representative confocal images of SiNPs/PVDF nanocomposites with different morphologies: (**a**) F210-17NP, (**b**) F90-17NP, (**c**) Ftrt-17P, (**d**) O-17P and (**e**) R-17P.

#### 3.2.2. Confocal Fluorescence Microscopy of the Nanocomposites

The incorporation of PDI in the silica nanoparticles can increase their application range, in particular, for biomedical applications, as it allows their tracking and localization [42,50]. In Figure 4, the green identifies the fluorescence of the nanoparticles; a higher colour intensity indicates a higher number of nanoparticles present. Figure 4a–c shows that, as the processing temperature decreases, a larger aggregation of nanoparticles is observed. In Figure 4a, where the temperature is higher, more homogeneous samples were obtained.

Relative to the oriented and random fibres (Figure 4d,e, respectively), it is observed that the nanoparticles are present and included within the fibres.

#### 3.2.3. Wettability of the Nanocomposites

Material surface characteristics are essential in determining cell response in tissue engineering applications. For this reason, the static CA was measured on the different SiNPs/PVDF nanocomposites and the values are presented in Figure 5.

**Figure 5.** Contact angle of the SiNPs/PVDF nanocomposites: (**a**) PVDF with the SiNPs with different diameters processed at 90 ◦C and (**b**) SiNPs/PVDF samples with silica nanoparticles (17 nm) with different morphologies.

The introduction of the Si nanoparticles increases the CA values, independently of the diameter of the silica nanoparticles [18], to around 100◦ excepting for the samples with silica nanoparticles with the highest diameter (F90-300NP). This increase is attributed to the hydrophobic properties of the silica nanoparticles [18]. Samples with nanoparticles with the highest diameter show a higher range of CA values, which is explained by the variation in the diameter of the nanoparticles, as observed in Figure 1. Regarding Figure 5b, the CA for the composite samples with the smallest silica nanoparticles show that the CA of PVDF fibres increases significantly compared to the one of PVDF films, and the CA of the oriented PVDF fibres is slightly higher than the one for randomly oriented PVDF fibres, showing a contact angle of 146.0 ± 7.2◦. These results support the idea that the increase in the hydrophobicity of electrospun samples is mainly related to the membrane morphology [8], with the fibres being significantly more hydrophobic than films. In the case of PVDF films, the CA is also higher for films with higher porosity, as already reported for pristine films [43].

#### 3.2.4. Structural Properties and Electroactive Phase Content of the Nanocomposites

FTIR-ATR spectra allow us to identify and quantify (Equation (2)) the polymer phase present in the samples and, therefore, to evaluate possible modifications induced by the introduction of silica nanoparticles (Figure 6).

Figure 6a shows the FTIR spectra of the different samples prepared at 90 ◦C as well as the corresponding quantification of the β-phase content (Figure 6c, calculated after Equation (1)). The characteristic bands of β PVDF (840 cm−1) is present in all samples, with low traces of α-PVDF (bands at 766, 855 cm−1), with the exception of F210-17NP. This is mainly attributed to the processing temperature [47], which mainly governs the solvent evaporation kinetics and the polymer crystallization in the β phase for processing at temperatures below 90 ◦C [44]. The introduction of SiNPs in PVDF does not significantly change the β-phase content, independently of the SiNPs content and average diameter. The β-phase value of pristine PVDF is 83 ± 3.3% and for the nanocomposites F90-17NP, F90-100NP, F90-160NP and F90-300NP, is 62 ± 2.5, 91 ± 3.6, 79 ± 3.1 and 74 ± 3, respectively (Figure 6c). On the other hand, Figure 6d shows that, depending on the nanocomposites' morphology, the polymer crystallizes in different phases, mainly due to the different processing conditions. Thus, electrospinning involves room-temperature solvent evaporation and polymer stretching during jet formation, both favourable conditions for the crystallization of the polymer fibres in the β phase [49]. With respect to the films, the F210-17NP nanocomposite, which is processed by a melting and recrystallization process, crystallizes in the α-phase and shows that the addition of SiNPs does not induce the nucleation of the electroactive β-phase of the polymer, as observed in previous study with Fe3O4 spherical nanoparticles [51]. On the other hand, the porous samples, as well as the fibres, are prepared after solvent evaporation at room temperature, conditions leading to the crystallization in the β-phase. This fact is not affected by the introduction of the nanoparticles. Thus, it is concluded

that the presence of the nanoparticles does not induce strong interactions with the polymer chain, leading to the nucleation of a specific phase, as observed with other fillers such as CoFe2O4 [52] and NaY zeolite [43]. Thus, processing temperature and solvent/polymer ratio remain the main factors determining polymer phase content in those composites [39,44].

**Figure 6.** FTIR spectra of (**a**) neat PVDF and SiNPs/PVDF nanocomposites with silica nanoparticles of different diameters processed at 90 ◦C and (**b**) different morphologies of SiNPs nanocomposites prepared with the smallest nanoparticles. The β-phase content for the different sample is represented in (**c**,**d**).

#### 3.2.5. Thermal Behaviour of the Nanocomposites

The DSC scans allow us to determine the melting temperature and the degree of polymer crystallinity (Figure 7).

**Figure 7.** (**a**) DSC thermographs and (**b**) degree of crystallinity of the SiNPs/PVDF nanocomposites with different morphologies and with the fillers of lowest average diameter.

All the samples show an endothermic peak around 168 ◦C corresponding to the polymer melting of the crystalline phase [44], thus, processing conditions and the incorporation of the filler do not affect the melting temperature. The degree of crystallinity was calculated (Equation (2)) from the enthalpy of the melting peak of the DSC thermograms. It was noticed that the samples prepared by solvent evaporation at 90 ◦C and after melting and recrystallization showed a lower degree of crystallinity than the samples prepared by solvent evaporation at room temperature, which also includes electrospun samples (Figure 7b). The pristine PVDF film processed at 90 ◦C shows a degree of crystallinity of ≈40%, which slightly increases with the introduction of the SiNPs and with the size of the SiNPs, being 43% for F90-17NP and 55% for F90-160NP (data not shown). Relative to the different morphologies (Figure 7a), the endothermic peak value is lower for the sample processed at 210 ◦C, indicating a lower degree of crystallinity if the sample, attributed to the fillers acting as defects during the crystallization from the melt [53]. Inclusion of the nanoparticles in the fibres does not significantly alters the crystallinity degree of the O-17P (52%) and R 17P (49%) with respect to the pristine polymer-oriented fibres (50% [8]).

The latter is ascribed to the combined effect of solvent evaporation at room temperature and stretching during the crystallization process that overcomes the effect of the presence of NP.

#### 3.2.6. Mechanical Properties of the Nanocomposites

The mechanical properties of the materials are essential parameters to design a scaffold suitable for tissues with different mechanical characteristics. The characteristic mechanical strain-stress curves of samples with different morphology, filler type and content are presented in Figure 8.

Figure 8a shows the stress-strain curves for the nanocomposites prepared with fillers with different average diameter after a melting process and Figure 8b refers to the nanocomposites with the same SiNPs (17 nm) after different processing conditions. Independently of the filler average diameter or processing conditions all samples show the typical mechanical behaviour of PVDF [54] characterized by the elastic region, yielding and plastic region, i.e., the typical behaviour of a thermoplastic elastomer.

**Figure 8.** Stress-strain curves for (**a**) SiNPs/PVDF nanocomposites with different SiNPs average diameters within the PVDF matrix and (**b**) for nanocomposites obtained after different processing conditions.

The Young's modulus of the samples was calculated from the linear zone of elasticity between 0 and 1% strain, as presented in Figure 9.

**Figure 9.** Young's modulus of the SiNPs/PVDF nanocomposites varying (**a**) the processing method and (**b**) the average diameters of the SiNPs. The values are shown as mean ± SD.

The characteristic features of the strain-stress curves are similar for all the materials, demonstrating that the mechanical characteristics are not strongly dependent on nanoparticle diameter. Furthermore, the introduction of particles with different diameters does not significantly affect the Young's modulus of the pristine PVDF (F210-NP) −0.94 ± 0.04 GPa. However, a slight improvement in the Young's modulus is observed for samples prepared with smaller silica nanoparticles (F210-17NP): 1.05 ± 0.06 GPa; this is in line with reports showing that the modulus increases as the particle size decreases [55]. Relative to the different production methods for the polymer films, F210-NP, F90-17NP and Frt-17P, it is observed that the more porous the structure, the lower the Young's modulus, 0.83 ± 0.16 GPa for FTrt-17P. On the other hand, oriented fibres (O-17P) show a higher Young's modulus (0.082 ± 0.012 GPa) than the random fibre samples (R-17P) (0.032 ± 0.002 GPa) due to the larger number of fibres along the stretch direction [8].

Relative to the other samples, the production method has a relevant influence on their mechanical response, as the samples prepared at room temperature by solvent evaporation showed a lower Young's modulus than those obtained at 210 ◦C due to the porous nature of the former and the compact structure of the latter, as was also visible in the SEM images (Figure 3).

#### *3.3. Cell Culture Studies*

In order to explore the potential use of the developed materials in tissue engineering applications, it is necessary to evaluate the putative cytotoxicity of the samples. The study of metabolic activity of C2C12 myoblasts, evaluated with the MTS assay, was applied to all samples and the results for 24 and 72 h are presented in Figure 10. Thus, the effects associated with introducing a fluorescent SiNP with different sizes are analysed, as well as the effect of the different microstructures/morphologies.

**Figure 10.** Cytotoxicity indirect test of (**a**) samples prepared with nanoparticles of different diameters and prepared by solvent evaporation at 90 ◦C and (**b**) samples prepared with SiNPs of 17 nm diameter after different processing methods and therefore with different morphologies.

It has already been reported that PVDF is biocompatible and shows no cytotoxicity to C2C12 cells for 24 or 72 h [29,38]. The SiNPs are also biocompatible for many cells including C2C12 myoblasts [56–58]. It is important to notice that in the polymer composites silica nanoparticles are within the polymer films, avoiding any possible cytotoxic effects from the particles themselves. This is confirmed by the results of the cytotoxic assays of Figure 10. Once PVDF is a non-biodegradable polymer, there is also no risk of the particles leaching out from the films.

Thus, Figure 10 shows that none of the samples are cytotoxic, independently of the nanoparticle diameter and of the material morphology. It is important to note that, despite both materials being biocompatible, the result is not evident, as polymer-filler interface effects or solvent retained in the nanoparticles or in the interface areas, can lead to cytotoxic effects. According to the ISO standard 10993-5, samples are considered cytotoxic when cells suffer a viability reduction larger than 30%. The measured cell viability values are all higher than 70%, confirming the cytocompatibility of the SiNPs/PVDF nanocomposites.

C2C12 myoblasts were used in previous studies to analyse cell proliferation of cultures grown on porous [59] and non-porous [38] PVDF films as well as fibres [38], with the verification that C2C12 cells proliferate better on piezoelectric β-PVDF "poled" samples. The samples obtained in this work were studied to determine the suitability for tissue engineering applications, namely muscle tissue.

MTS (Figure 11), immunofluorescence (Figure 12) and SEM (Figure 13) assays were used to assess cell viability and morphology in the different samples. Relative to the proliferation results (Figure 11), the cell viability has been obtained in relation to the sample of F90-NP at 24 h.

$$\text{Cell Velocity} (\%) = \left( \frac{\text{Absorbance of samples at 72 h}}{\text{Absorbance of F90-NP at 24 h}} \times 100 \right) - \text{cell viability of F90-NP at 24 h} \tag{3}$$

Figure 10 shows that the cell viability of the samples increases after 72 h of cell culture, independently of the SiNPs' diameters (Figure 11a) and the morphology of the materials (Figure 11b), when compared with the sample without particles (F90-NP). No significant differences are observed between the samples and the negative control (F90-NP), revealing that C2C12 myoblast proliferation is not affected by the presence of SiNPs in the PVDF matrix. In fact, it has been reported that SiNPs being included in different polymers improves the cell attachment and proliferation, and enhances cellular processes [60,61], which is in agreement with the obtained results.

**Figure 11.** Cell proliferation of C2C12 cells seeded on (**a**) SiNPs/PVDF samples prepared at 90 ◦C with different sized nanoparticles and (**b**) SiNPs/PVDF samples with different morphologies.

Cell cytoskeleton morphology, viability and adhesion were analysed by fluorescence microscopy for porous and non-porous films and SEM for fibre samples.

Independently of the nanoparticles' diameters and the sample morphology, it is observed that the cell behaviour is similar. Bigger cell agglomerates (and larger nanoparticle agglomerates) are observed with the increasing nanoparticle diameter of the samples (Figure 12a–d). This fact is associated with the interaction between serum proteins and nanoparticles present on the PVDF matrix, as it has been reported that a negative surface charge enhances the adsorption of proteins with isoelectric point more than 5.5 such as immunoglobulin G (IgG) that can be important for C2C12 myoblasts [62,63]. Cell cultures on PVDF fibres prepared with the smaller silica nanoparticles were analysed by SEM and Figure 13 shows the cell morphology of C2C12 cells after 72 h of cell culture on oriented and random PVDF fibre nanocomposites.

**Figure 12.** Representative images of C2C12 myoblast culture after 72 h on (**a**) F90-17NP, (**b**) F90-100NP, (**c**) F90-160NP, (**d**) F90-300NP, (**e**) F210-17NP and (**f**) FTrt-17P samples (nucleus stained with DAPI—blue and cytoskeleton stained with FITC—green). Scale bar = 100 μm for all the samples.

These representative images demonstrate that, in the presence of a fibrillar microstructure, the muscle cells orientate their cytoskeleton along the fibres, which is in agreement with the literature [38]. In this way, in the presence of oriented fibres, the cells share a similar architecture to the natural muscle cells in living systems.

**Figure 13.** Cell morphology obtained by SEM of C2C12 myoblasts seeded on PVDF fibres: (**a**) O-17P and (**b**) R-17P, after three days of culture. The scale bar is 200 μm for all samples.

Thus, the overall results prove the potential of the use of SiNPs/PVDF piezoelectric nanocomposites for muscle tissue engineering. Physical and chemical stimuli are important factors to obtain tissues with characteristics similar to those of natural living tissues in the human body, developing therefore specific biomimetic microenvironments for different tissues according to their specific biophysico-chemical needs. The developed platform presents nanocomposites with different morphologies (membranes and fibres), piezoelectric β phase and SiNPs diameter (from 17 to 300 nm), which makes it an interesting and complete platform for tissue engineering.

Furthermore, this platform will allow further studies applying mechanical stimuli on the nanocomposites obtained in this work with specific bioreactors [36] applying mechanical and/or mechanoelectrical stimuli. It may also take advantage of the SiNPs' capacity to include specific biomolecules or to develop drug delivery systems, or, more specifically, differentiation factors to promote directed myogenic differentiation. This will not only allow a deeper knowledge of the stimuli necessary for muscle tissue regeneration, but also lead to more effective therapies.

#### **4. Conclusions**

Different parameters important for tissue engineering, such as materials morphology, porosity and the PVDF electroactive phase, are modified in the obtained membranes.

Different diameters of silica nanoparticles have been introduced within the PVDF polymer matrix to obtain multifunctional samples for tissue engineering applications.

It is observed that the introduction of the SiNPs fillers in the PVDF matrix decreases its wettability. Furthermore, it is shown that the filler diameter does not significantly affect the properties of the polymer matrix, such as physico-chemical, thermal and mechanical properties.

Cytotoxicity assays with C2C12 cells show no cytotoxicity associated with neat PVDF and composites with different SiNPs diameters and sample morphologies.

Thus, it is demonstrated that the developed platform of PVDF materials with silica nanoparticles demonstrates potential for tissue engineering applications, allowing us to develop electromechanically active microenvironments with different morphologies with SiNPs, allowing protein functionalization and/or controlled release of specific drugs and/or growth or differentiation factors according to the targeted application.

**Author Contributions:** S.L.-M. conceived and designed the project. S.R. and T.R. contributed to the processing and characterization of the particles. J.P.S.F. and C.B. contributed to the characterization of the nanoparticles. S.R. and D.M.C. contributed to the processing and characterization of the samples in the different morphologies. S.R. was in charge of the cell culture assays and their characterization and interpretation. C.R. contributed to the cell culture assays and the interpretation of the cell culture assays. All authors contributed to the evaluation and interpretation of the data, as well as to the writing of the manuscript. All authors agree with the paper submission.

**Funding:** This work was supported by the Portuguese Foundation for Science and Technology (FCT) in the framework of the Strategic Funding UID/FIS/04650/2013 and UID/BIA/04050/2013 (POCI-01-0145-FEDER-007569) and project POCI-01-0145-FEDER-028237 funded by national funds through Fundação para a Ciência e a Tecnologia (FCT) and by the ERDF through the COMPETE2020-Programa Operacional Competitividade e Internacionalização (POCI); and also under the scope of the strategic funding of UID/BIO/04469 unit and COMPETE 2020 (POCI-01-0145-FEDER-006684) and BioTecNorte operation (NORTE-01-0145-FEDER-000004) funded by the European Regional Development Fund under the scope of Norte2020-Programa Operacional Regional do Norte. The authors also thank the FCT for the SFRH/BD/111478/2015 (S.R.), SFRH/BPD/96707/2013 (T.R.), SFRH/BPD/90870/2012 (C.R.) and SFRH/BPD/121526/2016 (D.C) grants. The authors acknowledge funding from the Spanish Ministry of Economy and Competitiveness (MINECO) through the project MAT2016-76039-C4-3-R (AEI/FEDER, UE) and from the Basque Government Industry and Education Departments under the ELKARTEK, HAZITEK and PIBA (PIBA-2018-06) programs, respectively.

**Acknowledgments:** The SEM measurements have been conducted at the Centre of Biological Engineering (CEB), Braga, Portugal. The authors thank CEB for offering access to their instruments and expertise.

**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/).
