1 Introduction

As the most abundant organic compound on earth, cellulose is the common structural component in herbal cells and tissues featuring a natural long chain polymer which plays an important role in human food cycle indirectly. The major advantage of cellulose is its possible accessibility and availability as a raw material, a renewable and biodegradable resource. Recently regenerated cellulose (RC) has been receiving much attentions in pharmaceutical, wood and paper industries, fiber-based products as well as electronics [1,2,3,4,5,6]. As its own structural uniqueness, cellulose can be modified using three hydroxyl groups per monomeric to react with numerous chemicals. Additionally, RC chemical structure enables the sol–gel process which is a potential highly scalable route to fabricate films for diverse applications. By chemical modification of cellulose, transparent cellulose-POSS-silica/titania hybrid material synthesized by an in situ sol–gel process for organic–inorganic composite, showed antimicrobial behaviors against the pathogenic bacterial effect on B. cereus and E. coli. [7]. Additionally, cellulose and silica/gold nanomaterial with covalent bonding was fabricated via sol–gel process and showed potential application for heavy metal absorption and electronic applications [8]. Concomitantly, cellulose properties was shown to be enhanced by interactions with nano-carbon particles such as carbon nanotubes and graphene nanoplatelets (GnPs) [2, 9,10,11,12,13], which leads to eco-friendly functional nanocomposites [14].

Pristine graphene is a one-atom thick sp2-bonded carbon arranged in a honeycomb, whereas GnP consists of a few layers of single layer graphene. The π-conjugation in graphene can result to outstanding composite properties [15]. Therefore, graphene-based regenerated cellulose (RC) composites are receiving an increase in research interest due to their eco-friendly nature, added to the tunable property and functionality enhancements from low GnP loadings [16,17,18].

Dielectric and polarization behavior have been explored for graphene-based polymeric nanocomposites in the sake to use them for energy storage applications [19, 20]. Recently designed flexible energy and memory storage materials using cellulose modified graphene oxide nanocomposites were fabricated by simple casting/solvent evaporation method [21]. It was revealed that the electric field inside the graphene oxide fillers can be a controlling factor of the nanocomposites polarizability. Moreover, the polarization of the nanocomposites increased as the applied electric field increases. But the relatively low loading of graphene resulted in composites that could not be used for energy storage applications. On the other hand, 3D porous structures of reduced graphene oxide (70)/cellulose(100) (rGO:cellulose = 70:100 in weight) showed high supercapacitive potential due to its high specific surface area and fast charge propagation [22], but with very high composite thickness. Low thickness of RC-graphene oxide made from the sol–gel process was reported as viable for energy storage applications [13]. In that work, authors used the porous composite cross-section for measurements rather than the lateral surface of the composite, which is suitable for practical applications. Additionally, the amount of reduced graphene oxide used was lower than 4 wt% of the overall nanocomposites mass. To fabricate eco-friendly supercapacitors made from the highly scalable sol–gel process, the graphene loading has to be increased while their dielectric and polarization behaviors need to be clarified.

Herein, we report the dielectric and polarization behavior of RC loaded commercial exfoliated graphene nanoplatelet (GnP) nanocomposites with higher loading of graphene. 90, 70 and 50% in weight loading of Gnp nanocomposites were fabricated. The nanocomposites were prepared using the sol–gel process. The coagulation effect of cellulose was used to maintain the 3D structures of the nanocomposites. To understand the mechanism of significant improvement of the dielectric constant and the polarization, we characterized the fabricated RC-GnP composites using different material characterization techniques.

2 Experimental details

Commercial GNP from XG Sciences (xGnP C-750) was used as-obtained for the composites preparation. Cotton pulp with 98% purity and degree of polymerization 4500 was purchased from Buckeye Technology, USA. Dimethyl acetamide (DMAc) and lithium chloride (LiCl), solution was used to dissolve cellulose and fabricating cellulose nanocomposites. Further, 99.5% isopropyl alcohol was purchased from Daejeon, South Korea. Cellulose solution was prepared using the common DMAc–LiCl method [23]. The cotton pulp and LiCl were dried in an air oven at 1000 C to remove water molecules prior to its use. The mixture of 1.5/8.5/90 cotton pulp, LiCl and anhydrous DMAc were heated at 155 °C and stirred by mechanical stirring for 4 h to obtain a viscous solution and centrifuged at 11,000 rpm to eliminate the undissolved cellulose fibers. All reagents were of analytical graded and used without further treatment. Distilled water was used in relevant situations.

After the dissolved cellulose process, the regenerated cellulose-GnP nanocomposite was prepared as shown in Fig. 1. A fixed amount of GnP was first dispersed in the DMAc solution by horn-sonication during 10 min to obtain a homogenous graphene suspension. Thereafter, 10, 30, and 50% w/w ratio of pre-prepared cellulose solution with respect to the amount of GnP materials was added to the suspension and mixed well by mechanical mixing method. The obtained suspensions were again horn-sonicated for an additional 5 min before homogenization process for an additional 1 h by a rotor/stator homogenizer. Finally, the composite solutions were casted on a glass plate with a doctor blade and cured using deionized water/isoprapanol. This curing process eliminated the remnants, namely, LiCl and DMAc. Cured wet nanocomposite was rinsed in two different deionized water baths and dried in a vacuum oven at 60 °C. Finally, we obtained thin RC-GnP nanocomposite film.

Fig. 1
figure 1

Fabrication process of RC-GnP nanocomposites

Full size image

For detailed understanding of RC-GnP composite, we characterized the fabricated films by scanning electron microscope (SEM, JEOL, JSM-840-A) and x-ray diffraction analysis (XRD, Riguku co D/max-3C). For the measurement of thermal property, thermogravimetric analysis (TGA, TA instruments 2050 universal V4.1D) was carried out. For TGA analysis, a ceramic hybrid samples weighing 9.69 mg were heated up to 1000 °C with ramping rate of 10 °C/min.

Because the dielectric constant of material is related to its own polarization behavior, it can provide further detailed information on the electrical dielectric behavior of nanocomposites from the capacitance measurement. To investigate the polarization and electrical behaviors of GNPs nanocomposites, we deposited thin metal electrodes on both surfaces using a thermal evaporator. The electrode was 15 mm × 15 mm. The capacitance of the nanocomposite sample was measured using an LCR meter (HP 4284A) in the frequency range from 20 Hz to 1 MHz with a parallel mode. The dielectric constant was calculated from the capacitance measurement using the relation \(\varepsilon = t \cdot C/\varepsilon_{0} A\), where C is the measured capacitance, t is the composite film thickness, A is the area of the metal electrode and ε0 is the permittivity of free space. To obtain more detailed polarization behavior of RC-GnP composite films, a ferroelectric tester (Radiant Technology Premier II) was used under applied voltage range from − 2 to 2 V.

3 Results and discussion

The morphology and structural analysis of the RC-GnP composites were characterized by observing their surfaces and cross-sections. The case of 10 and 30% in weight of RC are shown in Fig. 2a, b, respectively. It is confirmed by SEM observation that Gnps and RC chains are randomly dispersed. This is better observed in the case of 30% RC where an obvious dual component structure of GnP sheets intercalated by RC chains was observed. Illustrations of these RC chains are shown in Fig. 2b by black arrows. The as-received graphene powder micrograph is shown in Fig. 2d with the nanoplatelet structure of individual flakes that are aggregated. The surface micrographs in the case of 10 and 30% are shown in Fig. 2e, d. An obvious multi-planes structure made of graphene sheets intercalated and covered by RC is observed since the graphene content is higher. This intercalation effect is much obvious in the case of 30% RC.

Fig. 2
figure 2

SEM micrographs images of a 10% RC-GnP composite, b 30%RC-GnP composite, c of xGnP-750 and GnP/cellulose nanocomposites. a, b Cross-section of nanocomposites in the case of 10 and 30%, respectively. c Magnified cross-section in the case of (a). d Pristine GnP graphene. d, e Surface view of 10% RC-GnP and 30% RC-GnP, f surface of nanocomposites in the case of 10 and 30%, respectively

Full size image

X-ray diffractogram of as-received GnP (Fig. 3a) showed a sharp peak at 26.30 which indicates that its interlayer distance is around 0.33 nm. This is attributed to the graphitic-like structure of GnPs. In the case of RC (Fig. 3a), a peak at 2θ ~ 22.1° was observed, which corresponds to the 002 crystallographic plane of the cellulose [14]. After GnP was loaded into RC and vice versa, the RC peak disappeared from 50 to 30% in Fig. 1b. This may be due to the low amount of RC introduced. Besides, GnPs may have hindered the polymerization process which caused reduction in polymer crystallinity. Moreover, the intensity of the major peak of GnP decreased as the amount of RC increased with a saturation at around 30 wt% of RC since the peak intensities in the case of 30 and 50% are almost at the same level. Accordingly, the RC chains may disturb the GnP aggregation and crystallinity as their interactions increased which is evidenced by SEM micrographs (Fig. 2b). So Higher concentrations of GnP result in sharper and more intense diffraction peaks. In Fig. 3b, the XRD peak revealed gradual shifts toward low angle when the wt% of RC was increased from 10 to 50%. This is attributed to the improved GnP dispersion which reduced the GnP aggregates sizes. In fact, the intercalation of GnP by RC chains increased with the increase of RC wt%, which was interpreted as the decrease of the mean graphene plate thickness in the XRD curves.

Fig. 3
figure 3

XRD spectra of cellulose, GnP and GnP/cellulose nanocomposites. a The case of neat cellulose and GnP. b The case of nanocomposites as compared to neat GnP

Full size image

The thermal stability of the RC-GnP nanocomposites was measured by TGA. As shown in Fig. 4, the GnP sample shows very high thermal stability resulting in less than 10% of total weight loss at temperature over 900 °C. This minor weight loss can be attributed to the absence of most oxygen functional groups in GnP. However, RC decomposed in two steps (Fig. 4). The first step is assigned to the evaporation of water around 120 °C while the second and larger one is due to the thermal pyrolysis of the cellulose skeleton [14]. Specially, 10, 30, and 50 wt% of RC-GnP composites samples did not show the first step decomposition, which may stem from the interfacial interaction of RC and GnPs, resulting a better thermal stability at low temperature. At the temperature over 500 °C, it is observed that the decomposition pathway of composites with only 10% loading of RC is similar to that of pure GnP, whereas 30 and 50% RC contents in nanocomposite tends to behave as pristine RC. The residue of GnP was 87 wt% while residues of 10, 30 and 50% % RC loaded samples were 60, 73 and 72 wt%. This indicates a lower thermal stability as the amount of RC increased. In fact, the first step decomposition of 30 wt% was clearly distinct from 10 and 50 wt% cases. Thus, the fact that 10 and 30 wt% showed similar residues is mostly due to the less cellulose thermal pyrolysis of the 30 wt% sample compared to 50 wt%. It means that part of the cellulose skeleton was protected by the GNP at higher temperature, and the protection decreased with the increase of the RC amount to 50 wt%. The curves in Fig. 4 do not show any significant change of the onset of decomposition temperature of cellulose. This result is in agreement with the previous report [22].

Fig. 4
figure 4

The TGA of RC, and RC/GnP nanocomposites

Full size image

The dielectric constant can help to evaluate the storage capability of materials. The measurements of the dielectric constant for all the three RC loading levels were frequency-dependent. Figure 5a shows that the dielectric constant increased as the loading fraction of RC decreased. In other words, the interfacial polarization inversely varied with the loading fraction. Moreover, Fig. 5a also shows that the dielectric constant increases as the amount of RC decreases throughout the frequency sweep. These two observations can be explained only by understanding the GnP/RC nanocomposites-wave interaction. Also, as our nanocomposites here are condensed phase, the influence of the molecular surroundings has to be taken into account. The interaction of electromagnetic waves with any material causes volumetric electronic polarization and forces electrons to vibrate with the wave frequency. It is thus expected that the contribution of the electronic polarization to the dielectric constant should be sensibly the same for all the three composites systems fabricated here because the loading material (RC) is essentially an insulator. By increasing the loading fractions of RC from to 10, 30, and 50 wt%, we basically increased the interfacial area as the number of free electrons brought around by the π–π interactions on the graphene basal plane remained intact. This leads to a buildup of charges at the interface between the RC chains and GnP as an alternating field is applied resulting in interfacial polarization in the nanocomposites. This type of polarization is usually observed at lower frequencies where the half cycle is of sufficient time. Furthermore, as an alternating field is applied, the charge movements of dipoles and dipoles molecules in the RC-GnP composites are limited in extend to less than a molecular diameter resulting in dipolar polarization. So finally, we attributed the high dielectric constant at lower frequencies here to mainly interfacial polarization. Additionally, XRD (Fig. 3) and SEM (Fig. 2) results confirmed that GnP was intercalated by RC chains. It is expected that the amount of interface increases as the RC loading level increases which in turn increases the resistive flow in the composite. So 30 wt% RC has the highest internal resistance. The higher the internal resistance, the lower the electronic polarization which adversely affects the dielectric constant. This may explain why the highest dielectric constant is the case of 10% loading of RC in Fig. 5a.

Fig. 5
figure 5

Dielectric properties of GNP/cellulose composites. a The case of dielectric constant. b The case of dielectric loss tangent

Full size image

To better understand the dielectric constant behavior per frequency, we plotted the dielectric loss for the three nanocomposites (Fig. 5b). The ratio of imaginary part to the real part of the permittivity is defined as a dissipation factor, DF, also known as a loss tangent as it is proportional to dielectric loss. So Fig. 5b accounts for the loss that occurred in the different nanocomposites. It was observed that the dielectric loss decreased as the loading fraction increased which is opposed to the behavior of dielectric constant. The dielectric loss did not vary linearly with the frequency. Higher dielectric loss at lower frequencies was observed for all the curves. Dielectric loss of 50% loading was higher than 30% which was in turn higher than 10%, comforting again the interfacial-related phenomenon. We believe that this behavior is due to an excessively polarized interface induced by the dissimilar electronic properties of graphene and RC in the composites. Specifically, the movement or rotation of the atoms, molecules, or charge carriers in an alternating field are compromised leading to heat generation as a result of non- elastic polarization. In general, dielectric loss results from the sum of space charge distortion, dipolar, interfacial motion, and conduction losses. Distortional losses are related to electronic polarization mechanisms [24]. These losses prevailed on the storage mechanism when the RC loading fraction increased due to the structure of nanocomposites.

To deepen our understanding on the polarization behavior, we measured the polarization P as function of voltage. The rotational movement of hydroxyl groups and the difficult-to-avoid existing water in RC at room temperature account for the cellulose contribution to the overall polarization, all of which can be referred as dipolar polarization. The electronic polarization of Gnp through the free π–π electronic cloud mostly accounts for its contribution to the overall polarization. Finally the dipolar interfacial polarization has also to be taken in consideration. From the polarization curves shown in Fig. 6a, b, it is observed that the polarization behavior of 10 and 30 wt% of RC are the same and have an ideal resistor response since the curves are all elliptic [25], which means the resistance elements of RC-GnP increased more than the capacitance parts. This is not surprising since the layer-by-layer structure of cellulose can easily trap space charges [26, 27], which may play a resistive role against the internal current flow during the relaxation of the space charge in the composite [28]. It is also observed that the polarization increased with the applied voltage. This observation also applied to the remnant polarization, but at a higher rate in the case of 30%, which showed that there were more trapped charges in the case of 30% loading resulting in a higher loss. Interestingly, after negative biasing, the negative remnant polarization (Prem) showed electric field dependent behavior. In both positive and negative biased conditions, 10 and 30 wt% did not show equal remnant polarization values (positive and negative polarization at V = 0) with the maximum polarization value located near zero bias. This means that all polarization mechanisms cited above, did not follow the direction of the external field spontaneously as the voltage was alternated due to the trapped charges in RC-GnP. Perfect hysteresis are usually encountered in typical ferroelectric materials [29]. Trapped charges in RC have shown to play an important role in governing its polarization behavior [28]. Therefore, compared to the neat RC in [28], the presence of GnP has drastically increased the overall polarization level of RC-GnP, but its behavior is still similar to that of neat RC (Fig. 6).

Fig. 6
figure 6

Polarization behavior of nanocomposites in the case of 10% (a) and 30% loading (b)

Full size image

The polarization behavior revealed that our nanocomposites have an ideal resistor response as observed in the through thickness direction. It was then essential to measure the electrical resistivity for each composite in the same direction as function of frequency. The results are shown in Fig. 7. It is seeing that all the curves showed a strong frequency-dependent resistivity response. The resistivity decreased as the frequency increased for all the three nanocomposites. This is consistent with the dielectric measurement as well. As the dielectric constant decreased with the increase of frequency, some trapped carriers escaped and served to boost an eventual current. This may explain why we obtained high resistivity at low frequency compared to high frequency range. All measurements were done in a through thickness direction even at extremely high fraction of the GnP (10% RC). Surface images of SEM (Fig. 2) reveals a crumpled surface made by GnP in all nanocomposite cases. These GnPs were covered by thin insulating layer of RC, which rendered the nanocomposites alternative-current-nonconductive in the through thickness direction [30].

Fig. 7
figure 7

Through thickness resistive properties of the prepared nanocomposites. b is the magnification of (a)

Full size image

4 Conclusions

In the paper, we prepared nanocomposite materials made by the incorporation of a dielectric material namely the RC into a stable amount of GnP. We varied the amount of RC from 50, 30, and 10%. The electrical characterizations of the composites were conducted. It was shown that by varying the amount of RC, one can tune the loss and permittivity over a wide range of values in the 50 Hz to 1 MHz frequency range. Both the dielectric and loss tangent showed higher values at low frequencies, and decreased gradually as the frequency increased. The dielectric constant increased as the loading fractions of RC decreased, whereas the loss tangent decreased as the loading fractions increased. Ferroelectric investigation revealed an ideal resistor response for the fabricated nanocomposites which was later confirmed by the resistivity measurement with frequency-dependent response. According to their high dielectric constant, the fabricated composites bear the potential to be used as energy storage materials, but their dissipation factors has to be minimized for better performance.