Electromagnetic Shielding Effectiveness of a Hybrid Carbon Nanotube/Glass Fiber Reinforced Polymer Composite

Radio-frequency radiation of electromagnetic (EM) waves is regarded as an invisible electronic pollution. Aircraft electronic systems, communication devices, electronic medical appliances, and scientific electronic instruments are just few examples of applications prone to the EMI pollution.

where A is the absorption loss, related to the energy attenuation due to energy transmission within the conductive sheet, R is the reflection loss, corresponding to the energy reflected from the sheet's recipient face, and M is the multiple-reflection loss, related to repetitive reflections between the two internal surfaces of the conductive sheet, all expressed in decibels.

In general, a good EMI shield should possess a high electrical conductivity and a low magnetic permeability [3]. Traditional structural materials, i.e., metals, are well-known to be good EMI shielding materials for most of the applications. However, limited physical flexibility, heavy weight, possible lack of resistant against corrosion and wear, as well as difficulty of tuning their EMI SE are some drawbacks of using metals for shielding purposes.

Intrinsically conductive polymers (ICPs) such as polyaniline (PANI) and polypyrrole exhibit high electrical conductivity and dielectric constants while retaining lightweight, good flexibility, corrosion resistance, and facile processability [1,4]. Tuning of the SE of ICPs is feasible to some extent via various chemical treatments [1]. Due to their aforementioned properties, they have evinced much interest in potential application as EMI shielding screens, coatings for flexible conductors, and as broadband microwave absorbers [5].

Despite their attractive properties, it is difficult to achieve an extremely high level of shielding efficiency by using conducting polymers only. For example, the crystallinity needs to be enhanced significantly to attain proper dielectric constant [6–9]. Furthermore, significant volume fraction of PANI is needed to obtain sufficient electrical conductivity to achieve acceptable EMI shielding effectiveness. Luo and Chung [10] investigated the EMI shielding properties of PANI polymeric blends between 50 and 100 MHz frequencies. Changing the weight fraction of PANI from 1% to 40% altered the SE from 10 dB to 70 dB. Although ICPs possess interesting shielding capabilities, conductive polymer–metal complexes have been shown to further enhance the SE of the polymers [6,7,11]. However, the additive metal weight penalty is a paramount challenge. Alternatively, carbon-based materials have been used as conductive fillers for the polymers in EMI shielding applications. Carbon nanofibers, CNTs, and graphite are some examples of carbon-based materials used as fillers to enhance the EMI SE of polymers [8–10,12,13]. The EMI SE of polymers filled with carbon-based fillers increases as the fill factor and/or the filler aspect ratio increases [8,9]. Although elevated volume fractions of filler could depreciate the mechanical performance of the host matrix via deterioration of its inherent morphology [14], higher filler content is necessary to achieve higher SE.

As fiber-reinforced polymer (FRP composites lack sufficient EMI SE, incorporating CNTs as an additive to the FRP structure can boost their electrical conductivity and consequently, their EMI SE. There are several methods for incorporating the CNTs into the FRPs. The most straightforward is to shear-mix the CNTs with the polymer matrix prior to fabricating the FRP and curing the polymer [15]. Besides improved conductivity, improvements in the matrix-dominated mechanical properties of FRPs have been observed utilizing this method [16–19]. However, there are two major hurdles in utilizing this method. First, the inherent tendency of CNTs to agglomerate due to the Van der Waals interactions [18] limits their dispersibility within the matrix. Another obstacle arises from the excessive viscosity of the CNT–polymer mixture, even with very small CNTs weight fractions (e.g., 2 wt.%) [18]. The excessive viscosity of the polymer/CNT slurry restricts sufficient impregnation of the fibers during the composite manufacturing process, leading to potential delamination.

Alternatively, to overcome these challenges, CNTs can be controlled-grown directly on the fibers surface prior to fabricating the FRPs to form hybrid reinforcements. Catalytic chemical vapor deposition (CCVD) has been utilized to grow CNTs, on the fibers [20–27]. However, since the CCVD method usually requires elevated temperatures (700–1200 °C), the substrate carbon fibers could be damaged and, therefore, the fiber-dominated properties of the resulting FRPs will be reduced due to the fibers degradation [28–30]. Despite degradations in fiber-dominated mechanical performance of the CCVD processed FRPs, fiber/matrix bond dominated properties, e.g., interlaminar shear strength, have shown some improvements [31–33].

Graphitic structure by design is a recently developed method for growing multiwalled carbon nanotubes (MWCNTs) on fibrous materials [28,34–37]. Since this method utilizes a relatively low temperature (450–550 °C), the destructive effects of the elevated temperature would be minimized in lieu of the CCVD approach [22].

This study investigates the effect of the GSD-grown MWCNTs on the electromagnetic SE of a hybrid glass fiber reinforced epoxy polymer composite. In contrast to the effects of each parameter in the GSD method, different composite configurations based on different fiber surface treatments were fabricated and tested. In order to protect the fibers against the GSD thermal environment, a thermal barrier coating (SiO₂) was deposited on the fiber surfaces prior to the GSD process. The quality of the grown MWCNTs was examined via Raman spectroscopy as well as scanning electron microscopy (SEM). In-plane and through thickness electrical conductivities of the different GFRP configurations were also measured to contrast the effects of the GSD-grown MWCNTs on the EMI SE of the hybrid GFRPs. The mechanical performance of the fabricated GFRPs was also probed utilizing quasi-static tension tests.

Glass fiber 4 HS woven fabrics with an average fiber diameter of 7 μm, manufactured by BFG industries, Inc., Greensboro, NC, were employed as the reinforcement. To protect the fibers against the thermal degradation during the CNTs growth, a 75 nm thick thermal barrier coating (SiO₂) was sputtered on the glass fiber fabrics using a magnetron, ATC Orion, high-vacuum sputtering system by AJA International, Inc. (Scituate, MA). The sputtering process was performed under 3 mTorr Argon pressure and 300 W power on the SiO₂ target. To add the catalyst for the CNTs growth, the sputtering system was utilized to deposit a 4 nm thick nickel film.

Following the GSD protocol [28,34], the GSD growth procedure was performed in a quartz tube furnace reactor, capable of adjusting a temperature–time profile, and equipped with three gas mass flow controllers. The input gasses were ultrahigh purity (UHP) N₂, C₂H₄, and UHP H₂. The first step comprises the reduction of the metal catalyst and it entails constant flow of a N₂/H₂ mixture on the fabrics under the atmospheric pressure while maintaining the temperature at 550 °C for 2 hrs. This step allows for removing the oxide from the nickel catalyst and breaks the catalyst film into particles; a step necessary for growth of nanofibers. In the second step, the tube furnace is flushed with N₂ in order to get rid of the byproducts of the reduction step. Right after the flushing step, the growth step takes place by introducing the deposition mixture, i.e., N₂/H₂/C₂H₄, into the reactor at 550 °C for another 2 hrs. In this step, MWCNTs start to grow on top of the fibers initiating from the nickel particles.

To establish a reference sample, the same procedures were carried out on another fabric except for not including the hydrocarbon (C₂H₄) gas into the deposition mixture, i.e., no growth. In total, four fabric configurations were utilized for making distinct composites: Raw (fabric as is), Sp (raw fabric sputtered with SiO₂ and nickel), Sp + HT (Sp sample underwent the GSD process without flowing the hydrocarbon gas), and GSD (Sp sample underwent the full GSD procedure).

Three different composites layups were manufactured for the EMI testing, the mechanical testing, and the electrical resistivity measurements with 2-, 2-, and 20-layer laminates, respectively. Composites were fabricated utilizing a vacuum- and hydraulic pressing-assisted hand layup system. The polymer matrix was a structural epoxy, Aeropoxy^TM PR2032 by PTM&W Industries, Inc. (Santa Fe Springs, CA). This epoxy system has been successfully utilized by the authors to manufacture FRPs [19,22,38,39] and nanocomposites based on single wall carbon nanotubes (SWCNTs) [40] and MWCNTs [41]. The 2- and 20-layered composite laminates possessed 65.0 ± 1.0% and 58.0 ± 2.0% fiber volume fractions, respectively. Sample sizes of 125.0 × 12.5 × 0.44 mm³ and 50.0 × 50.0 × 0.44 mm³ were used for tensile and EMI shielding tests, respectively. For the through thickness resistivity measurements, 4.0 × 3.0 × 3.0 mm³ samples from different locations, and for the in-plane resistivity tests, narrow strips of 30.0 × 4.0 × 3.0 mm³ were diced from the fabricated 20-ply composite laminates.

To examine the quality of the GSD-grown CNTs on the glass fiber fabrics, a 5200 Hitachi scanning electron microscope (SEM) operated at 5 keV was utilized. Raman spectra of all the sample configurations were produced using a ProSeek Raman Spectroscopy system from the Raman System, Inc. The scanning was performed via a 5 mW and 785 nm wavelength exciting laser beam for a total integration time of 60 s.

A system equipped with an R&S^®ZVA Vector Network Analyzer, two waveguides, and two coaxial to waveguide adapters was utilized to measure the EMI SE of the samples. The utilized waveguide types for low and high frequency ranges were WR430 and WR159, respectively. Figure 1 shows a schematic of the test setup. The samples were individually clamped down between the waveguides using four screws. Prior to the actual measurements, the waveguides were calibrated for both the amplitude and phase. EMI SE of the samples was calculated following Eq. (1).

The in-plane and through thickness electrical resistivities of the fabricated hybrid GFRP composites were examined using a Keithley 4200 system. In order to exclude the resistance of the contact joints, measurements were conducted at different lengths of the samples and the resistivity was calculated using a curve-fitting method over the obtained resistance–length curves.

An Instron^® 4400R frame was utilized to perform the tensile tests under a 0.20 mm/min constant speed of the crosshead following the ASTM D3039/D3039M-08 standard. A total of four tensile coupons per composite configuration were tested in order to obtain their ultimate tensile strengths, elastic moduli, and strain-to-failures. The strain was measured using an MTS extensometer.

The signature for crystalline carbon materials is identified by two distinct bands in Raman spectra. The principal band is called G-band exhibiting the tangential mode of vibrations in the crystalline structure of a carbonaceous material excited by the Raman laser beam. The G-band peak appears at Raman shifts of 1581.2 cm⁻¹ for MWCNTs and between 1595 and 1605 cm⁻¹ for SWCNTs [42]. The second important band related to disorder-induced vibrations is called D-band, which appears in the Raman shifts of 1354.7 cm⁻¹ in the MWCNTs spectrum and between 1330 and 1390 cm⁻¹ for the SWCNTs [42]. Figure 2 shows the Raman spectra for the samples with different configurations. The GSD sample spectra exhibit the two aforementioned distinct peaks, which confirms the existence of a graphitic structure. The SEM micrographs of the glass fiber undergone the full GSD process, illustrated in Fig. 3, confirmed that the grown graphitic structures are MWCNTs.

Figure 4 shows the results of the EMI SE tests. Both the glass fiber and epoxy matrix are highly electrical insulating materials and do not contribute much to the EMI shielding. This is evident by the results of the raw sample in Fig. 4. From the figure, it is also inferred that the thermal barrier coating (SiO₂) and the catalyst layer (nickel) did not improve the EMI SE of the composite laminate. The almost identical EMI SEs of the Sp + HT sample and the raw sample confirm this observation. This can be attributed to the fact that SiO₂ is an electrically insulting material and the very thin (4 nm) layer of the nickel is not contributing significantly to the electrical conductivity of the laminate. In contrast, the GSD composite configuration, comprising the conductive MWCNTs grown on the glass fibers, displays a higher SE than the raw sample. For the frequency range of 1.70–2.80 MHz, the surface-grown MWCNTs have improved the EMI SE of the GFRP by almost 0.8 dB. This increment in the EMI SE of the GFRP is more pronounced at higher frequency range (5.3–7.10 MHz), where the SE was improved by almost 1 dB. Since the EMI SE is a logarithmic quantity, this result indicates that one order of magnitude increase in the EMI SE was achieved by the GSD sample. Such improvement in EMI SE originates from the dense and long (2–3 μm) MWCNT forest grown on the glass fibers. The dense MWCNTs forests contribute extra electrical conductivity to the hybrid FRP.

To quantify the improvements of the electrical conductivity, the electrical resistivity was measured along both the in-plane and through the thickness directions. It is worth mentioning that the other samples, other than the GSD, exhibited very high resistivity fallen beyond the measuring device range. The only sample with a measurable conductivity (or resistivity) was the GSD sample with the surface-grown MWCNTs. The in-plan electrical resistance versus length of the GSD sample is plotted in Fig. 5. The resistance of the sample along the in-plane direction was calculated to be 0.3667 MΩ/cm, while the resistance of the contact joints was 0.8 MΩ. Multiplication of the cross-sectional area of the specimens by the in-plane resistance gives the in-plane electrical resistivity of the GSD specimen as 44.0 kΩ cm. Figure 6 shows the electrical resistivity through the thickness measurements. Due to the small thickness of the sample, only two measurements were conducted through the thickness. Assuming the resistance of the contact joints to be the same as the in-plane case (0.8 MΩ), the linear trend line was assumed to cross the vertical axis at 0.8 MΩ. The slope of the trend line, exemplifying the resistivity per length, reads 64.40 MΩ/cm. Using the same approach as in the in-plane resistivity, the electrical resistivity of the GSD sample was calculated as 7.7 MΩ cm. The through thickness electrical resistivity of the hybrid CNT–FRP was more than two orders of magnitude higher than the in-plane resistivity. These results were expected since the long carbon nanotubes grown on the surface entangle with one another forming continuous conductive pathways, which translates to better conductivity, while the through the thickness MWCNT forests are separated by the insulating layers of thermal barrier coating and the highly resistive glass fiber fabric. However, MWCNTs radially grown on the glass fibers can penetrate through the woven fiberglass fabric pores and to some extent improve the electrical conductivity of the hybrid composite laminate. From EMI shielding perspective, the hybrid CNT–FRP (GSD sample) capitalizes on the enhanced conductivity to dissipate more EMI energy compared to the FRPs without surface-grown MWCNTs.

Figure 7 shows representative stress–strain curves for the two-ply composite laminates. Figure 8 summarizes the average mechanical properties of the FRPs. Inferred from Fig. 7, all the FRPs exhibit a linear elastic behavior at the earlier stages of strain up to almost 0.2% strain due to the high volume fraction of the stiff carbon fibers. At higher strain levels, the stiffness of the FRPs decreases due to the initiation of microcracks in the matrix [17]. The elastic moduli of the FRPs are known to be dominated by the fibers stiffness [17], and therefore, different surface treatments on the fibers are not expected to affect the elastic modulus profoundly. However, for the sputtered sample a phase change in SiO₂ layer from amorphous to crystalline (upon heating) can stiffen the corresponding FRP to some extent. Figure 8 suggests that the sheathing of the glass fibers with MWCNTs can stiffen the hybrid FRP further. Despite the precoating with a thermal barrier layer, the high sensitivity of the glass fibers to elevated temperatures has induced reductions in the tensile strengths of all the heat-treated samples (i.e., Sp + HT and GSD samples). Comparing the results of the raw and Sp samples, it is evident that the thermal barrier coating, even with no heat treatment, has reduced the FRP strength. This could be attributed to the possibility that the SiO₂ coating has negatively affected the adhesion at the fiber/matrix interface. Poor adhesion in the Sp samples, consequently, leads to a reduced strain-to-failure.

To elaborate in detail, the stress transfer mechanism is responsible for ductility of the composite laminates and allows the FRP to carry load even after individual fiber breakage occurs. Better adhesion between the fibers and matrix enables an FRP to capitalize on this mechanism more and to exhibit higher ductility. Growing the MWCNTs on the fibers potentially can improve the adhesion in the fiber/matrix interface. The activated CNTs walls can adhere to polymer chains and improve the contact between the hybrid fibers and polymer matrix. However, very long and dense CNTs do not allow for sufficient polymer infusing in between the CNTs and, therefore, weaken the fiber/matrix interface. This was the case for the tested GSD samples, and therefore, reductions in the ultimate strength and strain-to-failure were observed for this configuration as well. Nevertheless, while the long and dense CNT forest on the glass fibers did not enhance the mechanical properties of the FRP, they still facilitate the formation of a connected conductive network, which is crucial for adding the EMI shielding functionality to the glass fiber composites.

This study utilized GSD technique to grow CNTs on the surface of glass fiber fabrics toward enhancing their electrical conductivities. A thermal barrier coating was employed to reduce thermal degradation of the glass fibers under GSD protocols. Raman spectroscopy and SEM microscopy revealed that the GSD yielded surface-grown MWCNTs.

Long and dense forest of MWCNTs grown on the surface of the glass fiber fabrics generated well-connected pathways for electrons transport. The GSD-grown MWCNTs improved the electrical conductivity of the FRP along both the in-plane and the through the thickness directions. The conductivity of the hybrid CNT–FRP along the in-plane direction was two orders of magnitude higher than that along the through thickness direction. The network percolation through the thickness was not as strong as the in-plane direction leading to more resistivity along the out-of-plane direction.

Per the enhanced conductivities in both directions, the EMI energy attenuation of the MWCNT/FRP configuration, over the measured frequency range, improved by one order of magnitude compared to the raw glass fibers based FRP.

Despite the enhancement in the electrical transport properties, the excessive sensitivity of the glass fibers to high temperatures has caused the sample with GSD surface-grown MWCNTs to attain lower tensile strength and strain-to-failure compared to the composite sample based on the raw glass fibers. Nevertheless, the slightly higher elastic modulus of the GSD sample indicates that the glass fiber core was immune to the deteriorating effects of the elevated temperature. While this work shows the feasibility for growing CNTs toward enhancing the EMI shielding of fiber glass based composites, further optimization of the GSD process is needed to ensure improvements on both the electrical conductivity, and consequently, EMI shielding, together with unaltered/enhanced mechanical properties.

This work was supported by the Office of Naval Research (ONR) Grant No. 10960991 and the National Science Foundation (NSF) Award Nos. CMMI-0846589 and CMMI-1200506.