polymer composite by covalent modification of graphene oxide surface

polymer composite by covalent modification of graphene oxide surface

Accepted Manuscript Achieving significantly enhanced dielectric performance of reduced graphene oxide/polymer composite by covalent modification of gr...

1MB Sizes 28 Downloads 63 Views

Accepted Manuscript Achieving significantly enhanced dielectric performance of reduced graphene oxide/polymer composite by covalent modification of graphene oxide surface Wangshu Tong, Yihe Zhang, Qian Zhang, Xinglong Luan, Yang Duan, Shaofeng Pan, Fengzhu Lv, Qi An PII: DOI: Reference:

S0008-6223(15)30040-3 http://dx.doi.org/10.1016/j.carbon.2015.07.005 CARBON 10077

To appear in:


Received Date: Accepted Date:

18 March 2015 2 July 2015

Please cite this article as: Tong, W., Zhang, Y., Zhang, Q., Luan, X., Duan, Y., Pan, S., Lv, F., An, Q., Achieving significantly enhanced dielectric performance of reduced graphene oxide/polymer composite by covalent modification of graphene oxide surface, Carbon (2015), doi: http://dx.doi.org/10.1016/j.carbon.2015.07.005

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Achieving significantly enhanced dielectric performance of reduced graphene oxide/polymer composite by covalent modification of graphene oxide surface

Wangshu Tong, Yihe Zhang*, Qian Zhang, Xinglong Luan, Yang Duan, Shaofeng Pan, Fengzhu Lv, Qi An* National Laboratory of Mineral Materials, School of Materials Sciences and Technology, China University of Geosciences, Beijing 100083, China






nanocomposite (PVDF-HFP)

films incorporating


poly(vinylidene polyethylenimine

(PEI)-covalently modified graphene sheets (rGO-PEI) was prepared by a solution-cast method, and the dielectric properties of the films were studied. Infrared spectroscopy, atom force microscope, X-ray photoelectron spectroscopy, solid state NMR spectroscopy, Raman, thermogravimetry analysis and additional verification experiment indicated that PEI chains were successfully grafted on the surface of graphene oxide (GO). A dielectric constant of 67 (100 Hz), which was five times that of pure PVDF-HFP film, was obtained for the composite films when the concentration of rGO-PEI was 8wt% and the dielectric loss was as low as 0.12. A microcapacitor model explained the behavior of the dielectric composites. The effect of the covalent modification was further studied by introducing control experiments using non-covalently prepared N-GO-PEI as the fillers to create the composite films. The results demonstrated that covalent linkages between GO and PEI were indispensable in obtaining

*Corresponding author. E-mail: [email protected] (Yihe Zhang), [email protected] (Qi An).

composite films with high dielectric constants and low losses.

1. Introduction Polymeric flexible composite films with high dielectric constants and low dielectric losses have attracted intense research attention due to their potential applications in artificial muscles and skins, charge-storage devices, and flexible electronics [1-5]. The fabrication of flexible dielectric films using polymers and electric conductive fillers has proven to be a promising strategy in producing films with advantageous dielectric properties [5-12]. Research has especially been conducted on the usage of graphene or carbon nanotubes as the conductive filler due to their high electronic conductivity and high aspect ratios. As a result, films with high dielectric constants have been successfully achieved [1,2,11]. Most of these efforts followed the theory of percolation, and the high dielectric constants close to the percolation threshold have been highlighted in numerous studies [13-19]. For example, by using multi-walled carbon nanotubes as the filler in a Poly(vinylidene fluoride) (PVDF) matrix, a giant dielectric constant of 8000 has been achieved [13]. Composite films with a high dielectric constant and low percolation threshold were obtained by using exfoliated graphite nanoplates dispersed in a PVDF matrix [14]. And reduced-graphene oxide dispersed in a PVDF matrix also led to a high dielectric constant [15]. However, two disadvantages remain for these films. One is the abrupt variations of the dielectric performance around the percolation threshold, including the dielectric constant, dielectric loss, and conductivity, which make the dielectric performance of these films difficult to control. The other disadvantage is that the dielectric losses of these films are very high. These two

disadvantages have hindered the realistic applications of these films where stable performances of the materials are required and low losses are crucial. In order to obtain dielectric films with stable high dielectric constants and low losses, a variety of strategies have been developed. These efforts include enhancing the aspect ratios of the fillers [11], increasing the conductivity of the fillers [15,16], inducing orientation of the fillers [20,21], and increasing interfacial compatibilities between the fillers and the matrix [14,22-24]. Preliminary successes have been achieved by using one or combinations of these strategies. Of particular interest is that, through increased experimentation, the micro-capacitor model was suggested to be the mechanism causing the phenomena observed for the dielectric films with relative high dielectric constants and low losses [24-30]. Despite the inherent challenges, research investigating the preparation of dielectric films with a high dielectric constant and low loss and the establishment of theoretical models to explain the phenomena and to guide practical applications remain highly desirable. In this study, we have prepared flexible films that simultaneously possess high dielectric constants and low losses by using the chemically converted reduced graphene oxide (rGO) as the fillers and poly(vinylidene fluoride-co-hexafluoropropylene) (PVDF-HFP) as the matrix. Branched polyethylenimine (PEI) was covalently grafted onto the surface of rGO (rGO-PEI) using the reaction between amines and epoxides in order to enhance interfacial compatibility between rGO and PVDF-HFP shown in Fig. 1. The prepared films presented a high dielectric constant of 67 and a dielectric loss of 0.12, which was almost as low as the pristine PVDF-HFP film, at the frequency of 100 Hz. We believe the model of “pseudothreshold” explained our experimental phenomena more appropriately when compared with the

percolation model. To learn the covalent functionalization and the effect on dielectric properties, the noncovalently functionalized GO (N-GO-PEI) was prepared to compare. The indispensable role of the covalent surface modification of rGO in obtaining the low dielectric losses was studied and analyzed. This report provides a facile method for the preparation of flexible films with a high dielectric constant and low loss. We believe that this study also provides a perspective for the improvement of the dielectric performance of flexible films.

Fig. 1 - Schematic representation of the functionalization of graphene oxide (GO) by PEI and rGO-PEI/PVDF-HFP. (a) Epoxide groups on the surface of GO, (b) the mixture of GO and PEI, (c) the nucleophilic ring-opening reaction of GO and PEI, and (d) composite film composed of PVDF-HFP and rGO-PEI.

2. Experimental 2.1. Materials Natural flake graphite (300 mesh) was provided by Shuangxing graphite processing plant, China and PVDF-HFP (density 1.78 g cm-3, 5-20% molar of hexafluoropropene) was purchased









3-glycidoxypropyltrimethoxysilane was purchased from Aladdin. Potassium persulfate (KPS, ≥99.5%), phosphorus pentoxide (PPO, ≥98%), potassium permanganate (KMnO4, ≥99.5%), sulfuric acid (H2SO4, 95-98%), perhydrol (H2O2, 30%), and N,N-dimethyl formamide (DMF, ≥99.5%) were obtained from local commercial sources and used as received. 2.2. Preparation of rGO-PEI and N-GO-PEI GO was prepared with a modified Hummers method [31]. To get completely oxidized, an additional graphite oxidation procedure was needed and the pre-oxidized graphite was prepared by KPS and PPO in concentrated H2SO4 [32]. This pre-oxidized graphite was then subjected to oxidation by Hummers’ method. GO of 0.05 g was loaded in a 250 ml round-bottom flask, to which distilled water (100 ml), 0.1g KOH and 2g PEI (50% aqueous solution) were added. The large amount of PEI avoided cross-linking of adjacent GO sheets. After sonication for10 min, the solution was heated up to 80 oC and held for 10 h. Then, the mixture of GO-PEI and PEI were washed by distilled water to remove extra PEI. Finally, the GO-PEI was obtained and dried at 40oC. The sample of non-covalently modified GO with PEI (N-GO-PEI) was prepared with the same method except that the mixture was not heated up to 80 oC but kept in room temperature for 10 h. GO was partially reduced during the covalent modification process. Further reduction of GO, in the covalently- or non-covalent-

modified samples, were achieved during the PVDF-HFP composite film preparation processes (kept in an oven at 220oC for 1 h) as stated below. 2.3. Preparation of GO-modified quartz surface ([Si]-PEI-GO) in both covalent and non-covalent manners As the reported procedure by Kim et al [33], the PEI modified silicon wafers [Si]-PEI was prepared. Silicon wafers covered by epoxy silanes were obtained by immersing in a solution of 3-glycidoxypropyltrimethoxysilane (3-GPTS) in toluene and heated at 90 ºC for 3 h. Then, [Si]-3-GPTS reacted with branched polyethyleneimine (Mn ~ 70000, 0.2 wt%) in ethanol at 80 °C for 6 h to form [Si]-PEI. (1) The covalently attached GO was achieved as follows: [Si]-PEI was immersed in GO solution with catalyst of KOH for 80°C for 10 h, and the surface was cleaned ultrasonically in PEI solvent for several times and then washed with water to form [Si]-PEI-GO (S). (2) The non-covalently attached (GO [Si]-PEI-N-GO (S)) was obtained in a similar manner but the process was conducted at room temperature. (3) [Si]-PEI-N-GO was obtained in a similar manner as (2) but without ultrasonic cleaning process. 2.4. Preparation of composite film A measured amount of rGO-PEI (N-GO-PEI) was loaded in a 100 ml round-bottom flask and DMF (20 ml) was added. The dispersion was sonicated using a SB-5200 DTDN ultrasonic bath cleaner (160W, Ningbo Scientz Biotechnology Co., Ltd.) for 3h to form a homogeneous suspension. After mixing with PVDF-HFP particles (3g) and stirring at 80oC for 2h, the rGO-PEI/PVDF-HFP (N-GO-PEI/PVDF-HFP) solution was obtained. The as-prepared rGO-PEI/PVDF-HFP (N-GO-PEI/PVDF-HFP) solution was drop-casted on a glass plate and

kept in an oven at 80 oC for 3 h to evaporate the solvent and then kept in an oven at 220oC for 1 h to obtain the rGO-PEI/PVDF-HFP (N-rGO-PEI/PVDF-HFP) composite films. 2.5. Characterization of rGO-PEI and rGO-PEI/PVDF-HFP The morphology and structure of the GO and rGO-PEI were examined by atomic force microscopy (AFM) obtained by scanning probe microscope in the tapping mode. Scanning electron microscopy (SEM) measurements and elemental scanning were carried out on an EVO MA25 instrument at 20.0 kV. X-ray photoelectron spectroscopy (XPS) measurements were conducted using a Kratos Axis Ultra system with a monochromatized Al Kα radiation at 1486.6 eV as the X-ray source. Solid-state Magic Angle Spinning Nuclear Magnetic Resonance analyses (NMR) were performed by the spectrometer from Bruker BioSpin GmbH (AV300). Powder X-ray diffraction (XRD) patterns were obtained on a D/MAX-RC diffractometer (Rigaku, Japan) at a scanning rate of 8°/min using Cu Kα radiation (40 kV, 100 mA,). Fourier transform infrared spectroscopy (FTIR, PerkinElmer Spectrum 100). Thermogravimetric (TG) analysis was conducted on a TA Q500 at a heating rate of 10 ºC/min in a nitrogen atmosphere. The dielectric property was determined by an impedance analyzer (Agilent 4294 A) at frequencies ranging from 102 Hz to 106 Hz. The film strips were cut accurately from the samples. Prior to the measurement, silver electrodes were fabricated on the sides of these strips using conductive silver paint (Agar no. 0443). 3. Results and discussion In order to improve interfacial compatibility of rGO and the matrix PVDF-HFP, surface modification of GO was conducted by the covalent reaction between amines of PEI and epoxide groups on the GO surfaces. The reaction is a ring-open reaction catalyzed by

potassium hydroxide (KOH) [33-35]. XPS, FTIR, NMR, Raman, TGA characterizations proved that the covalent reaction between amino and epoxy occurred successfully. The photo images of the samples dispersed in DMF and XPS survey spectra are shown in Fig. 2a-c. After surface modification, the color of the sample turned darker, and the product dispersed evenly in the solvent of DMF. XPS survey spectra indicate that PEI was grafted on GO surfaces: compared with GO samples, the peak of N1s appeared after the PEI functionalization. High-resolution XPS spectra confirmed that epoxy groups were consumed during PEI modification, and at the same time GO was reduced. For samples before modification, C1s spectrum of the GO could be quantitatively differentiated into five different carbon species: the sp2-hybridized carbon atoms of graphene at 284.4 eV (C-C sp2), the sp3-hybridized carbon atoms at 285.2 eV (C-C sp3), the carbon in hydroxyl and epoxide groups (C-O) at 286.5 eV, the ones in carbonyl groups (C=O) at 288.0 eV, and those in carboxyl groups (O-C=O) at 289.2 eV shown in Fig. 2d [18,36]. Analysis of the peak areas indicated that the carbons in the hydroxyl and epoxide groups on GO surfaces took up to 52% of all the carbon species. After PEI functionalization, an additional component at 285.9 corresponding to C-N was observed shown in Fig. 2e [34,37], indicating that PEI was grafted successfully. The dramatic decrease in signals characteristic of C-O group indicated that C-O was consumed by PEI and GO was reduced. The N1s core-level spectra in Fig. 2f consists of two components with binding energies of 399.0 eV [38] and 400 eV [39], respectively, corresponding to -NH- and -NH2 groups in PEI.

Fig. 2 - (a)-(c)The photo images about GO and rGO-PEI dispersibility and XPS survey spectra, (d) and (e) high-resolution XPS C1s core-level spectra of GO and rGO-PEI respectively, and (f) high-resolution XPS N1s of rGO-PEI. FTIR clearly demonstrated the successful modification of the GO surfaces by PEI molecules. Fig. 3a displays the FTIR spectra of GO, rGO-PEI and PEI. The spectra of GO illustrates the presence of C-O-C corresponding to cyclic ether moieties (vC-O-C at 1225 and 1060 cm-1), C-OH (vC-OH at 3400cm-1), C=O in carboxylic acid (vC=O at 1730cm-1) and O-H vibration in carboxyl groups (vO-H at 1381 cm-1) [40-42]. These results indicated that there were epoxide groups on the surface of GO. After the covalent reaction, the characteristic absorbance of PEI appeared. The broad peak around 3425 cm-1 was attributed to nitrogen-containing surface groups (-NH or –NH2). The doublet absorption bands at 2841 cm-1 and 2943 cm-1 corresponded to symmetric vH-C-H and asymmetric vH-C-H of the PEI chains [43,44]. The bending vibration of N-H was detected at 1571 cm-1, as well as the C-N stretching vibrations at 1476, 1303 and 1114 cm-1 [43]. In addition, the peaks corresponding

to C-O-C disappeared after the reaction, indicating that the C-O-C groups had been consumed by the nucleophilic ring-opening reaction with PEI. These results indicated that the covalent reaction between amino moieties of PEI and the epoxide groups on GO sheets successfully occurred. At the same time, the peaks corresponding to C=O in carboxylic acid (1730cm-1) and O-H (1381 cm-1) also disappeared, indicating that GO was partially reduced during the reaction [45]. Thus, the product after the covalent modification was termed reduced GO-PEI (rGO-PEI).

Fig. 3 - (a) FTIR spectra of GO, rGO-PEI and PEI, (b) Raman spectra of GO and rGO-PEI and (c) TGA curves for pristine GO, PEI, and rGO-PEI at a heating rate of 10 o

C/min in nitrogen. The Raman spectra for GO sheets and the rGO-PEI both displayed typical D and G

bands for graphitic materials at 1334 and 1596 cm-1 respectively (Fig. 3b). The D band was believed to stem from the multiple photo scattering of defects or amorphous carbon and the G band to the stretching of the conjugated double bonds corresponding to sp2 hybridization [23,25]. The slightly increased ratio of ID/IG for rGO-PEI herein was presumed to stem not only from the intercalation-related strain, but also from the functionalization of the graphene basal plane [26]. The amount of PEI grafted onto rGO surfaces was calculated using TGA measurement, as shown in Fig. 3c. The curve for GO displayed little weight loss below 120

ºC, corresponding to the loss of water. The substantial loss in the temperature range from 150 ºC to 200 ºC was attributed to the removal of oxygen-containing functional groups [46]. The curve for PEI presented significant weight loss in the temperature range from 300 ºC to 380 ºC, and reached zero when the temperature was 380 ºC, indicating the complete consumption of PEI. However, the significant weight loss of rGO-PEI started at 250 ºC, which was 100 ºC higher than that of GO. The increased onset temperature of the decomposition of rGO-PEI indicated that the oxygen-containing functional groups on GO were consumed and new chemical bonds were formed. Therefore, we reached the conclusion that PEI was chemically grafted onto GO surfaces. In addition, quantitative analysis indicated that the weight percentage of PEI grafted onto the GO surfaces was 32 wt %, based on the overall weight loss values of GO (62%), PEI (100%), and rGO-PEI (74%). The XRD data as shown in Fig. S1 indicated that after surface modification, the rGO-PEI did not aggregate into a crystalized structure but remained as well-dispersed rGO sheets which showed a broad peak around 26.5°. AFM images (Fig. S2) confirmed that the rGO-PEI was still dispersed as two-dimensional sheets with thicknesses several nanometers larger than the GO sheets. The uniform distribution and complete coverage of the grafted PEI molecules on the rGO surfaces was indicated by the elemental mapping measurement. Fig. 4 shows that the area covered by the N element, provided by the PEI molecules, was identical to the area covered by the C and O elements, indicating that the entire rGO surface was homogeneously covered by the PEI molecules. The EDX analysis from the samples indicated that the content of carbon, oxygen and nitrogen elements were 62.3%, 22.1% and 15.6%, respectively, which was consistent with that of typical rGO-PEI [44,47].

Fig. 4 - Elemental mapping images of rGO-PEI: (a) a typical SEM image and the corresponding elemental mapping images of (b) carbon, (c) oxygen, (d) nitrogen, indicating the homogeneous dispersion of C, O and N in rGO-PEI. After surface modification by PEI molecules, GO was mixed with PVDF-HFP in the solvent of DMF, and the mixture was drop-casted and heated to obtain the composite dielectric film. XRD data (in Fig. 5) verified that rGO-PEI did not aggregate in the matrix but remained in the dispersed status. In the diffraction patterns for the composite films, peaks appeared at 18.4, 19.9 and 26.5 degrees, corresponding to the (020), (110), and (021) crystalline planes of the α-form PVDF-HFP.

Fig. 5 - XRD patterns of rGO-PEI, PVDF-HFP and rGO-PEI/PVDF-HFP composites with a rGO-PEI content of 8 wt% .

The dielectric properties of the composite films that contained rGO-PEI as the filler (rGO-PEI/PVDF-HFP) was advantageous compared with that of the pristine PVDF-HFP films and GO/PVDF-HFP films in its simultaneous possession of two characteristics: (1) the film presented high dielectric constants, and (2) it gave low dielectric losses. Fig. 6a presents the frequency dependence of the dielectric constant of the PVDF-HFP film and the composite rGO-PEI/PVDF-HFP films. Over the entire tested frequency range from 100 Hz to 1 MHz, the dielectric constant increased with the increase of the fraction of rGO-PEI, till the weight percentage of 8%. At the frequency of 100 Hz, a high dielectric constant of 67, which was almost 5 times that of the pure PVDF-HFP film, was obtained when the fraction of rGO-PEI was 8 wt%, and the dielectric loss was as low as 0.12, which was only 67% higher that of the PVDF-HFP film (Fig. 6b). When the fraction of rGO-PEI was increased further to 10 wt%, the dielectric constant decreased to the value of 56, and, at the same time, the dielectric loss was substantially increased to 0.35. With the increase of frequency, the dielectric constant slightly decreased till the frequency reached 100 kHz, and when the frequency was larger than 100 kHz, the dielectric constant dropped rapidly. The dielectric loss slightly decreased at first with the increase of the frequency, and then increased significantly when the frequency was higher than 100 kHz. Most interestingly, the typical dielectric performance of the polymeric composite films using graphitic fillers, which was the simultaneous display of high dielectric constants and high dielectric losses, was not observed in our prepared films. These findings correspond to results by a few other researchers who have reported polymeric composite films with relatively high dielectric constant and low dielectric losses [24-30].

Fig. 6 - Frequency response of (a) the dielectric constants and (b) dielectric losses of the rGO-PEI/PVDF-HFP composites. Compared with those reports of graphene composites, the dielectric performance of the films reported in this study is good. Some other dielectric films that presented a dielectric performance substantially superior to the films reported here exists. That study fabricated the composite film using the cyanoethyl pullulan polymer as the matrix and chlorinated graphene as the filler, and the film presented a dielectric constant of ~100 and a dielectric loss of 0.035 at 100 Hz [29]. However, the mechanical properties and long-term stability of the cyanoethyl pullulan polymer might raise concerns in realistic applications. Comparable to the dielectric performance of the film in our study, another reported film was fabricated using a PVDF-based polymer as the matrix and chemically grafted rGO as the fillers. The chemical bonding between the matrix and the fillers ensured an optimized dielectric constant of 74 and

a dielectric loss of 0.08 (as opposed to 60 and 0.048, respectively, in this study) at 103 Hz [23]. But the matrix therein was a PVDF-based polymer that contained double bonds and was synthesized by a relatively tedious process. In contrast, the dielectric film reported in this study was fabricated by using commercially available raw materials. In addition, the chemical modification process of rGO was a one-step reaction which was facile to conduct. In the report of the composite film using carbon nanotube encapsulated in graphene as the fillers in the polyurethane matrix, a dielectric constant of about 56 and a dielectric loss of 0.12 (as opposed to 60 and 0.048, respectively in this study) at 103 Hz was achieved [4]. Thus, we believe the reported film herein possesses advantageous comprehensive properties and is an addition to the current choices of dielectric films with superior performances. The substantial improvement of the dielectric constant for the rGO-PEI/PVDF-HFP composite











Maxwell−Wagner−Sillars (MWS) polarization, between the rGO and polymer (PEI and PVDF-HFP) [29]. In general, MWS polarization leads to the enhancement of the dielectric constants at low frequencies, and induces a rapid decrease of the dielectric constants at high frequency ranges. At high frequency ranges, the polarization of the chemical bonds can no longer follow the rapid changes in the external electric field, leading to the dissipation of energy, causing high dielectric losses and a decrease in dielectric constants. One factor that contributed to the enhancement of the MWS polarization in our study (and thus the dielectric constant) was believed to be the reduction of the chemically grafted rGO in the composite film with the chemical and heating treatment, which led to increased conductivity of the rGO nanosheet (Table S1). The enhanced conductivity of the rGO nanosheets have increased the

density of the accumulated charge carriers at the interfaces between the rGO nanosheets and the matrix, leading to increased MWS polarization. Another factor that possibly contributed was the positive compatibility of the rGO-PEI and the PVDF-HFP matrix. PEI interacted with PVDF through polar interactions and hydrogen bonds. Thus the chemically grafted rGO dispersed homogeneously in the matrix and held a thin layer of the dielectric layer (PEI and PVDF-HFP) in between two adjacent rGO-PEI layers. This layer, in turn, formed internal barrier layer capacitors, which contributed a large capacitance. It is noteworthy that when the fraction of the rGO-PEI was 8 wt% and the dielectric constant of the film was 67, the composite film did not show any detectable direct-current conductivity. Despite a large number of previous studies reporting electric conductivity for composite film with advantageous dielectric properties, a few researchers did observe similar phenomena as reported here that the film was non-conductive to a direct current [26,29]. Based on the above analysis, we believe the covalent modification of the rGO surface improved the dielectric behaviors of the composite film. Though the film presented maximal dielectric performance at the fraction of rGO-PEI at 8 wt% and decreased with further increase of the filler, we tend to consider the value of 8 wt% a “pseudothreshold” instead of a threshold [26]. The threshold theory states that, at the fraction of the threshold, the conductive











electric-conductivity. This, however, was not the case in the current study. We suggest, therefore, that the “pseudothreshold” theory explains the decrease of the dielectric constants when the filler fraction is above a certain value as follows. At the point of the “pseudothreshold,” the dielectric layers in between the conductive planes were the thinnest

possible. Further increases of the amount of the conductive fillers would lead to partial aggregation of the fillers inside the matrix. This partial aggregation of the filler led to local direct-current conductive clusters, which caused decreases of dielectric constants and increases of dielectric losses. Therefore, positive compatibility between the surface modifiers of the conductive fillers and the matrix is the prerequisite for the emergence of the “pseudothreshold.” Further studies demonstrated that the covalent linkages between PEI and rGO were indispensable in obtaining the advantageous dielectric properties of the composite films. We employed the non-covalently modified rGO-PEI as the control sample (N-GO-PEI). N-GO-PEI was prepared by mixing GO and PEI together in the same ratios as the covalently modified rGO-PEI. But the heating treatment of the mixture, which was necessary for the reaction between amines and epoxies to take place, was removed. Under such experimental conditions, amines would adsorb onto the surface of the rGO through non-covalent interactions. Solid-state


C NMR spectra indicated that epoxides on GO surface were still

present in large quantities without the heating treatment, thus confirmed that in this circumstances PEI absorbed on GO surface in non-covalent manners (Fig. 7). For the N-GO-PEI, the peaks at 60 and 70 ppm represent epoxide and hydroxyl groups. The peak at 133 ppm belongs to the un-oxidized sp2 carbon atoms [48]. The flat peak of PEI (38-55 ppm) demonstrated the presence of PEI on the surface of GO [49,50]. The intensity corresponding to PEI was low for the N-GO-PEI sample because the NMR samples were subjected to fierce washing procedures that partially disassembled the non-covalently attached PEI, as to be discussed below. In contrast to the N-GO-PEI sample, the peaks of oxygenated and carbonyl

carbons were absent for the GO-PEI sample and the peaks at 133 ppm shifted to 123 ppm due to the chemical environment variations of carbon atom induced by reduction [37]. The strong peak corresponding to carbons in PEI indicated that plenty of PEI existed on GO surface even after the fierce washing treatment conducted for the NMR samples, highlighting the strong stability provided by covalent modifications.

Fig. 7 - Solid-state 13C NMR spectra of rGO-PEI and N-GO-PEI. The dielectric properties of the composite film using N-GO-PEI as fillers were studied and appear in Fig. 8 and S3. The dielectric constants of the film increased with an increasing amount of the filler till the filler fraction reached 5%, where a maximum dielectric constant of 35 was obtained. We noted that despite the lack of the polar C-N bonds in these films due to the non-covalent nature of the modification method, these films presented comparable but somewhat larger dielectric constants as compared to the rGO-PEI/PVDF-HFP films with the same filler fractions. The high dielectric constants for the N-rGO-PEI/PVDF-HFP films might arise from the higher conductivity of the filler of N-rGO-PEI, compared with rGO-PEI (Table S1). When the filler fractions exceeded 3%, aggregation of the fillers took place, and the uneven distribution of the filler was clearly observed in the optical images shown in Fig.

9a. The uneven distributions of the filler lead to fluctuations of local dielectric properties in different area of the film. The dielectric losses of the N-rGO-PEI/PVDF-HFP films were larger than that of the rGO-PEI/PVDF-HFP films. At the filler fraction of 5%, the dielectric loss of the N-rGO-PEI/PVDF-HFP films was almost three times the value for the rGO-PEI/PVDF-HFP films. The inhomogeneity and high dielectric loss of the N-rGO-PEI/PVDF-HFP films were mainly explained by the weakened interfacial interaction between rGO and PEI, as verified below.

Fig. 8 - Dependence of the dielectric constants (squares) and dielectric losses (circles) of the rGO-PEI/PVDF-HFP and N-rGO-PEI/PVDF-HFP composite films on the content of the rGO-PEI and N-rGO-PEI respectively.

Fig. 9 - (a) The photo images of

PVDF-HFP composites with rGO-PEI and

N-rGO-PEI respectively, (b) SEM image of the cross section of 5 wt% N-rGO-PEI/PVDF-HFP, and (c) and (d) SEM image of the cross section of 5 wt% rGO-PEI/PVDF-HFP. By bridging the rGO and the matrix, the covalently grafted PEI provided a strong interaction with rGO by covalent bonds. In addition, the covalently grafted PEI presented strong non-covalent interactions with the PVDF-HFP matrix due to the existence of the polar amine functional groups. In so doing, the covalently grafted PEI increased the highest possible filler fractions in the homogeneous films and also decreased local aggregations at relatively high filler fractions. In comparison, in the N-rGO-PEI/PVDF-HFP films, PEI would be easily detached from GO surface in the viscous PVDF matrix, and as a result, interfacial compatibility provided by the PEI linker would be diminished. This loss of interfacial compatibility due to the detachment of PEI counteracted the larger conductivity of rGO (during the heating treatment that took place after the stirring-assistant mixing of GO filler and PVDF matrix, GO was reduced) provided by non-covalent modification. To support this hypothesis and verify the difference in stabilities of PEI modification in the N-rGO-PEI

and rGO-PEI sample, an experiment of PEI assembly on quartz surface was designed as shown in Fig. 10a. In this experiment, quartz sheets were first modified covalently with epoxy groups, resembling the functional groups on GO surfaces. Then PEI was attached covalently. Subsequently, GO was assembled in both covalent and non-covalent manners on different substrates. The substrates were then subjected to thorough washing treatment. Afterwards, only the covalently modified substrate (by GO) displayed obvious absorbance characteristic of partially reduced GO at around 250 nm shown in Fig. 10b [48,51]. In clear comparison, the non-covalently modified substrate, although displayed strong absorbance corresponding to GO after the assembly process, lost GO signal after the washing treatment. This experiment demonstrated that remarkable differences in stabilities existed for the covalent and non-covalent manner, and the covalent linkage between GO and PEI was appreciably stronger.

Fig. 10 - (a) Schematic representation of fabrication of the GO-covalently-modified surface








GO-non-covalently-modified surface washed by ultrasonic cleaning ([Si]-PEI-N-GO (S)) and (b) UV-Vis spectrum of (1) [Si]-PEI-GO (S), (2) [Si]-PEI-N-GO (S) and (3) GO-non-covalently-modified surface without ultrasonic cleaning ([Si]-PEI-N-GO). This stronger linkage of PEI-GO then lead to increased compatibility between the filler

and the PVDF matrix, as verified by SEM images shown in Fig. 9b. The cross-section images of the rGO-PEI/PVDF-HFP composite (5%) displayed homogeneous structures with particulate features. In clear contrast, the N-rGO-PEI/PVDF-HFP composite (5%) displayed layered structures, stemmed from insufficient compatibilities between GO fillers and the matrix [23]. The differences in interfacial compatibilities further lead to different distribution status in the films. The rGO-PEI/PVDF-HFP composite (up to filler concentration 8%) presented homogeneous dark color. While the N-rGO-PEI/PVDF-HFP composite displayed aggregates of fillers even observable with bare eyes (Fig. 9a). All in all, the correspondence between the dielectric properties of the composite films and the structural features highlighted the importance of strong interfacial interactions in obtaining composite films with high dielectric constants and low dielectric losses. Consistent with the previous report, these results showed that covalent linkages between rGO and the matrix would significantly enhance the dielectric performance of the film [23]. 4. Summary In conclusion, we have prepared flexible composite films consisting of PEI-grafted-rGO and PVDF-HFP. PEI was covalently grafted onto rGO surfaces by using the reaction between amines and epoxides. After the reaction, PEI molecules homogeneously covered the surface of rGO. The films simultaneously presented high dielectric constants and low dielectric losses. At the filler fraction of 8wt%, a dielectric constant of 67 and a dielectric loss of 0.12 were obtained at the frequency of 100 Hz. The dielectric constant was 5 times the value of the pure PVDF-HFP films, and the dielectric loss was only 67% higher than that of the PVDF-HFP films. The substantial improvement of the dielectric performance of the prepared

composite films was possibly caused by the MWS polarization. By comparing with the non-covalently modified rGO (by PEI), the indispensable role of covalent modification was verified as follows: covalent modification provided stable molecular interface between PVDF-HFP and rGO fillers, leading to improved interfacial compatibility between fillers and matrix, results in enhanced dielectric constant and remained the dielectric loss at low values. Acknowledgements This work was supported by the NSFC (21303169), the Fundamental Research Funds for the








(Z141103001814064), Beijing Specific Project to Foster Elitist (2013D009015000001) and the Open Project of State Key Laboratory of Chemical Resource Engineering (CRE-2013-C-201). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at xxxx. References: [1] Nan CW, Shen Y, Ma J. Physical properties of composites near percolation. Annu Rev Mater Res 2010;40:131-51. [2] Dang ZM, Yuan JK, Zha JW, Zhou T, Li ST, Hud GH. Processes and applications of high-permittivity polymer-matrix composites. Prog Mater Sci 2012;57(4):660-723. [3] Kelarakis A, Hayrapetyan S, Ansari S, Fang J, Estevez L, Giannelis EP. Clay nanocomposites based on poly(vinylidene fluoride-co-hexafluoropropylene): structure and properties. Polymer 2010;51(2):469-74. [4] Wu C, Huang XY, Wu XF, Xie LY, Yang K, Jiang PK. Graphene oxide-encapsulated carbon nanotube hybrids for high dielectric performance nanocomposites with enhanced energy storage density. Nanoscale 2013;5(9):3847-55. [5] Kohlmeyer RR, Javadi A, Pradhan B, Pilla S, Setyowati K, Chen J, et al. Electrical and dielectric properties of hydroxylated carbon nanotube-elastomer composites. J Phys Chem C 2009;113(41):17626-9. [6] He F, Lam K, Ma D, Fan J T, Chan LH, Zhang LM. Fabrication of graphene nanosheet (GNS)–Fe3O4 hybrids and GNS–Fe3O4/syndiotactic polystyrene composites with high dielectric permittivity. Carbon 2013;58:175-84. [7] Zhang QM, Bharti V, Zhao X. Giant electrostriction and relaxor ferroelectric behavior in

electron-irradiated poly(vinylidene fluoride-trifluoroethylene) copolymer. Science 1998;280(5372):2101-4. [8] Zhou X, Zhao X, Suo Z, Zou C, Runt J, Liu S, et al. Electrical breakdown and ultrahigh electrical energy density in poly(vinylidene fluoride-hexafluoropropylene) copolymer. Appl Phys Lett 2009;94(16):162901. [9] Ning NY, Bai X, Yang D, Zhang LQ, Lu YL, Nishi T, et al. Dramatically improved dielectric properties of polymer composites by controlling the alignment of carbon nanotubes in matrix. RSC Adv 2014;4(9):4543-51. [10] Fan P, Wang L, Yang JT, Chen F, Zhong MQ. Graphene/poly(vinylidene fluoride) composites with high dielectric constant and low percolation threshold. Nanotechnology 2012;23(36):365702. [11] Shehzad K, Dang ZM, Ahmad MN, Sagar RUR, Butt S, Farooq MU, et al. Effects of carbon nanotubes aspect ratio on the qualitative and quantitative aspects of frequency response of electrical conductivity and dielectric permittivity in the carbon nanotube/polymer composites. Carbon 2013;54:105-112. [12] Ameli A, Nofar M, Park CB, Tschke PPo, Rizvi G. Polypropylene/carbon nanotube nano/microcellular structures with high dielectric permittivity, low dielectric loss, and low percolation threshold. Carbon 2014;71:206-217. [13] Dang ZM, Wang L, Yin Y, Zhang Q, Lei QQ. Giant dielectric permittivities in functionalized carbon-nanotube/electroactive-polymer nanocomposites. Adv Mater. 2007;19(6):852-7. [14] He F, Lau S, Chan HL, Fan JT. High dielectric permittivity and low percolation threshold in nanocomposites based on poly(vinylidene fluoride) and exfoliated graphite nanoplates. Adv Mater 2009;21(6):710-5. [15] Tang HX, Ehlert GJ, Lin YR, Sodano HA. Highly efficient synthesis of graphene nanocomposites. Nano Lett 2012;12(1):84-90. [16] Kim JY, Kim TY, Suk JW, Chou H, Jang JH, Lee J H, et al. Enhanced dielectric performance in polymer composite films with carbon nanotube-reduced graphene oxide hybrid filler. Small 2014;10(16):3405-11. [17] Yuan JK, Yao SH, Dang ZM, Sylvestre A, Genestoux M, Bai JB. Giant dielectric permittivity nanocomposites: realizing true potential of pristine carbon nanotubes in polyvinylidene fluoride matrix through an enhanced interfacial interaction. J Phys Chem C 2011;115(13):5515-21. [18] Min C, Yu DM, Cao JY, Wang GL, Feng LH. A graphite nanoplatelet/epoxy composite with high dielectric constant and high thermal conductivity. Carbon 2013;55:116-25. [19] Wang DR, Zhang XM, Zha JW, Zhao J, Dang ZM, Hu GH. Dielectric properties of reduced graphene oxide/polypropylene composites with ultralow percolation threshold. Polymer 2013;54(7):1916-22. [20] Tong WS, Zhang YH, Yu L, Luan XL, An Q, Zhang Q, et al. Novel method for the fabrication of flexible film with oriented arrays of graphene in poly(vinylidene fluoride-cohexafluoropropylene) with low dielectric loss. J Phys Chem C 2014;118(20):10567-73. [21] Yao SH, Yuan JK, Zhou T, Dang ZM, Bai JB. Stretch-modulated carbon nanotube alignment in ferroelectric polymer composites: characterization of the orientation state

and its influence on the dielectric properties. J Phys Chem C 2011;115(40):20011-7. [22] Yang K, Huang XY, Fang LJ, He JL, Jiang PK. Fluoro-polymer functionalized graphene for flexible ferroelectric polymer-based high-k nanocomposites with sduppressed dielectric loss and low percolation threshold. Nanoscale 2014;6(24):14740-53. [23] Wen F, Xu Z, Tan SB, Xia WM, Wei XY, Zhang ZC. Chemical bonding-induced low dielectric loss and low conductivity in high‑K poly(vinylidene fluoride-trifluorethylene)/ graphene nanosheets nanocomposites. ACS Appl Mater Interfaces 2013;5(19):9411-20. [24] Wang DR, Bao YR, Zha JW, Zhao J, Dang ZM, Hu GH. Improved dielectric properties of nanocomposites based on poly(vinylidene fluoride) and poly(vinyl alcohol)-functionalized graphene. ACS Appl Mater Interfaces 2012;4(11):6273-9. [25] Wu C, Huang XY, Wang GL, Wu XF, Yang K, Li ST, et al. Hyperbranched-polymer functionalization of graphene sheets for enhanced mechanical and dielectric properties of polyurethane composites. J Mater Chem 2012;22(14):7010-9. [26] Dimiev A, Zakhidov D, Genorio B, Oladimeji K, Crowgey B, Kempel L, et al. Permittivity of dielectric composite materials comprising graphene nanoribbons. ACS Appl Mater Interfaces 2013;5(15):7567-73. [27] Wu C, Huang XY, Xie LY, Wu XF, Yu JH, Jiang PK. Morphology-controllable graphene–TiO2 nanorod hybrid nanostructures for polymer composites with high dielectric performance. J Mater Chem 2011;21(44):17729-36. [28] Chu LY, Xue QZ, Sun J, Xia FJ, Xing W, Xia D, et al. Porous graphene sandwich/poly(vinylidene fluoride) composites with high dielectric properties. Compos Sci Technol 2013;86:70-5. [29] Kim JY, Lee WH, Suk JW, Potts JR, Chou H, Kholmanov IN, et al. Chlorination of reduced graphene oxide enhances the dielectric constant of reduced graphene oxide/polymer composites, Adv Mater 2013;25(16):2308-13. [30] Shang JW, Zhang YH, Yu L, Luan XL, Shen B, Zhang ZL, et al. Fabrication and enhanced dielectric properties of graphene polyvinylidene fluoride functional hybrid films with a polyaniline interlayer, J Mater Chem A 2013;1(3):884-90. [31] Kovtyukhova NI, Ollivier PJ, Martin BR, Mallouk TE, Chizhik SA, Buzaneva EV, et al. Layer-by-layer assembly of ultrathin composite films from micron-sized graphite oxide sheets and polycations. Chem Mater 1999;11(3):771-8. [32] HummersW, Offeman R. Preparation of graphitic oxide. J Am Chem Soc 1958;80(6):1339. [33] Yang HF, Shan CS, Li FH, Han DX, Zhang QX, Niu L. Covalent functionalization of polydisperse chemically-converted graphene sheets with amine-terminated ionic liquid. Chem Commun 2009;26:3880-2. [34] Liu JH, Chen GS, Jiang M. Supramolecular hybrid hydrogels from noncovalently functionalized graphene with block copolymers. Macromolecules 2011;44(19):7682-91. [35] Ahn Y, Jang Y, Selvapalam N, Yun G, Kim K. Supramolecular Velcro for reversible underwater adhesion. Angew Chem Int Edit 2013;52(11):3140-4. [36] Pumera M, Iwai H. Multicomponent metallic impurities and their influence upon the electrochemistry of carbon nanotubes, J Phys Chem C 2009;113(11):4401-5. [37] Stankovich S, Dikin DA, Piner RD, Kohlhaas KA, Kleinhammes A, Jia YY, et al.

Synthesis of graphene-based nanosheets via chemical reduction of exfoliated graphite oxide. Carbon 2007;45(7):1558-65. [38] Louette P, Bodino F, Pireaux JJ. Poly(ethylene imine) (PEI) XPS reference core level and energy loss spectra. Surf Sci Spectra 2005;12(5):54-8. [39] Jansen RJJ, Bekkum HV. XPS of nitrogen-containing functional groups on activated carbon. Carbon 1995;33(8):1021-7. [40] Zhu CZ, Guo SJ, Fang YX, Dong SJ, Reducing sugar: new functional molecules for the green synthesis of graphene nanosheets. ACS Nano 2010;4(4):2429-37. [41] Acik M, Mattevi C, Gong C, Lee G, Cho K, Chhowalla M, et al. The role of intercalated water in multilayered graphene oxide. ACS Nano 2010;4(10):5861-8. [42] Song M, Xu J. Preparation of polyethylenimine-functionalized graphene oxide composite and its application in electrochemical ammonia sensors. Electroanalysis 2013;25(2):523-30. [43] Ramanathan T, Fisher FT, Ruoff RS, Brinson LC. Amino-functionalized carbon nanotubes for binding to polymers and biological systems. Chem Mater 2005;17(6):1290-5. [44] Zhang LM, Lu ZX, Zhao QH, Huang J, Shen H, Zhang ZJ. Enhanced chemotherapy efficacy by sequential delivery of siRNA and anticancer drugs using PEI-grafted graphene oxide, Small 2011;7(4):460-4. [45] Che JF, Shen LY, Xiao YH. A new approach to fabricate graphene nanosheets in organic medium: combination of reduction and dispersion. J Mater Chem 2010;20(4):1722-7. [46] Fang M, Wang KG, Lu HB, Yang YL, Nutt S. Covalent polymer functionalization of graphene nanosheets and mechanical properties of composites, J Mater Chem 2009;19(38):7098-105. [47] Zhou X, Laroche F, Lamers GEM, Torraca V, Voskamp P, Lu T, et al. Ultra-small graphene oxide functionalized with polyethylenimine (PEI) for very efficient gene delivery in cell and zebrafish embryos. Nano Res 2012;5(10):703-9. [48] Sharmila TKB, Nair AB, Abraham BT, Beegum PMS, Thachil ET. Microwave exfoliated reduced graphene oxide epoxy nanocomposites for high performance applications. Polymer 2014;55(16):3614-27. [49] Holycross DR, Chai MH. Comprehensive NMR studies of the structures and properties of PEI polymers. Macromolecules 2013;46(17):6891-7. [50] Wong K, Sun GB, Zhang XQ, Dai H, Liu Y, He CB, et al. PEI-g-chitosan, a novel gene delivery system with transfection efficiency comparable to polyethylenimine in vitro and after liver administration in vivo. Bioconjugate Chem. 2006;17(1):152-8. [51] Lv ZZ, Yang X, Wang E. Highly concentrated polycations-functionalized graphene nanosheets with excellent solubility and stability, and its fast, facile and controllable assembly of multiple nanoparticles. Nanoscale 2013;5(2):663-70.