In situ polymerization of graphene-polyaniline@polyimide composite films with high EMI shielding and electrical properties

With the increasing demands of the electronics industry, electromagnetic interference (EMI) shielding has become a critical issue that severely restricts the application of devices. In this work, we have proposed a “non-covalent welding” method to fabricate graphene-polyaniline (Gr-PANI) composite fillers. The Gr sheets are welded with PANI via π–π non-covalent interactions. Furthermore, a flexible polyimide (PI) composite film with superior EMI shielding effectiveness is prepared by in situ polymerization. The 40% content of Gr-PANI10:1 (the mass ratio of Gr to PANI is 10 : 1) shows a superior electrical conductivity (σ) as high as 2.1 ± 0.1 S cm−1, 1.45 times higher than that of Gr@PI film at the same loading. Moreover, the total shielding effectiveness (SET) of EMI of the Gr-PANI10:1@PI reaches ∼21.3 dB and an extremely high specific shielding effectiveness value (SSE) of 4096.2 dB cm2 g−1 is achieved. Such a “non-covalent welding” approach provides a facile strategy to prepare high-performance PI-based materials for efficient EMI shielding.


Introduction
With the rapid development of communication systems and electronic devices, electromagnetic interference (EMI) has become a serious problem as it not only interrupts the operation of electronic items, but also threatens people's health. 1,2 The traditional EMI shielding materials made by metal show several limitations including high density, poor exibility, and undesirable susceptibility to corrosion. 3 Therefore, it is essential to develop novel materials with low density, high exibility and high shielding effectiveness for EMI shielding.
PI, as a kind of high-performance engineering polymer, has been widely used in electronic industries as a promising polymer matrix for EMI shielding because of its superior exibility, excellent thermal stability as well as high wear resistance. 4,5 Great efforts have been taken to introduce Gr sheets into PI matrix to fabricate the composite lm with high electrical conductivity and excellent EMI shielding property. Recently, carbon materials, such as carbon nanotubes (CNT), 6,7 electrical carbon black, 8 carbon bers, 9 and graphene 10 have been explored and mixed with polymers as conductive llers to fabricate lightweight and high shielding effectiveness polymerbased materials.
Among these potential carbon-based llers, graphene, due to its outstanding electrical conductivity, has become a promising candidate as electrical ller in polymer. 11 Feng and coworkers fabricated rGO@PANI/PI lms via in situ polymerization. And the dielectric properties and thermal stability increased signicantly. 12 Ma et al. prepared rGOpoly(ethylene glycol) (PEG)/PI with chemically reduced method. 13 However, it is very difficult to uniformly disperse rGO sheets in organic solvent owing to van der Waals force and the p-p stacking interaction between rGO sheets. Meanwhile, the interaction between the rGO sheets and PI matrix is relatively weak, leading to the inferior mechanical properties of PI/rGO lm. In this regard, the thermal reduction of GO has advantages in mild reaction conditions and less deterioration of GO compared to chemical reduction of GO. A lightweight porous PI/rGO lm was obtained using nonsolvent induced phase separation (NIPS) process. 14 They found that the PI/rGO lm exhibited electrical conductivity of 1.5 Â 10 À4 S cm À1 , corresponding to a SSE value of 693 dB cm 2 g À1 . The GO reduced through thermal reduction still maintains some oxygen functional groups, contributing to construct strong interfacial interaction with PI matrix. Therefore, these rGO@PI lms exhibit the better mechanical properties. Nevertheless, these composite lms reduced at high temperature have poor electrical conductivity limiting the improvement of electrical property and EMI shielding performance. An effective way to improve the dispersibility of Gr is needed to enhance the performance for EMI shielding and electrical conductivity.
In this work, we propose a "non-covalent welding" method to fabricate Gr-PANI ller to improve the electrical conductivity, mechanical and EMI shielding properties of the PI composite lm. PANI, as a solder, not only successfully connects adjacent Gr sheets to provide the pathway for electron transportation, but also effectively prevents the stacking of Gr sheets via p-p non-covalent bonding. The Gr-PANI@PI lm prepared through in situ polymerization, contributing to disperse Gr sheets homogenously in the PI matrix. The composite lm displays superior mechanical properties ($56.8 MPa), electrical conductivity (2.1 AE 0.1 S cm À1 ), EMI shielding performance ($21.3 dB) and an extremely high SSE value of 4096.2 dB cm 2 g À1 with a thickness of only 0.04 mm. The reported "non-covalent welding" conception is also an effective technique to prepare polymer materials with Gr sheets in other elds.

Preparation of Gr-PANI composite ller
The preparation of Gr-PANI@PI composite lms was illustrated in Scheme 1. Concretely, 1 g Gr and 0.1 g AN were dispersed in 100 mL of 1 mol L À1 HCl under ultrasonication. Then, the solution of FeCl 3 $6H 2 O (0.1458 g, 0.54 mmol) was rapidly transferred to the above solution containing AN and Gr. Next, the mixture was magnetically stirred for 4 h in an ice-water bath. Aer that, the Gr-PANI composite ller was obtained by ltration, washed with deionized water and methanol until the washing solution was completely colorless. In this process, free PANI particles and unreacted substance were separated and removed. Finally, the mixture was freeze-dried for 24 h. The asobtained was named as Gr-PANI 10:1 (mass ratio of Gr to AN was 10 : 1). The Gr-PANI 15:1 and Gr-PANI 5:1 composite ller were prepared in the same manner.

2.3.
In situ synthesis of Gr-PANI 10:1 @PI composite lm The fabrication process of Gr-PANI 10:1 @PI composite lm involved three steps. First, 1.25 g Gr-PANI 10:1 was dispersed in anhydrous NMP (15.33 mL) by ultrasonication (1 h) at ambient temperature in order to homogeneously disperse Gr-PANI 10:1 in the solution. Then, ODA (2 g, 10 mmol) and PMDA (2.167 g, 9.94 mmol) were added into the solution under a nitrogen atmosphere. Aer continuous stirring at 25 C for 2 h, the homogeneous and viscous Gr-PANI 10:1 @PAA (polyamic acid, the precursor of polyimide) solution was obtained. The mixture was bar-coated on a clean glass dish and treated at 140 C for 1 h, and 350 C for 2 h. The as-obtained composite lm was coded as x%Gr-PANI 10:1 @PI, where x was the weight percentage of the Gr-PANI ller to the matrix. The 30%Gr-PANI 10:1 @PI obtained had a thickness of 40 mm. The 35%Gr-PANI 10:1 @PI and 40%Gr-PANI 10:1 @PI lms were prepared in the same manner. In the same manner, we fabricated the x%Gr@PI (without PANI) composite lm, where x represented the weight percentage of Gr ller to the matrix.

Characterization
Fourier transform infrared spectroscopy (FTIR, P.E. Spectrum 100, USA) was collected on a spectrum from 400 to 4000 cm À1 at a resolution of 2 cm À1 and for an accumulation of 8 times. Raman spectra (Horiba, LabRAM HR Evolution, France) were collected from 200 to 3000 cm À1 with a laser length at 532 nm. X-ray photoelectron spectra (XPS, Physical Electronics PHI 5000C ESCA) were performed to acquire the chemical bonding and the valence states of Gr and Gr-PANI with a monochromatized Al Ka X-ray source (1486.71 eV). X-ray diffraction (XRD, Bruker D8 ADVANCE, USA) patterns were conducted using nickel-ltered Cu Ka (k ¼ 0.154 nm) radiation with a generator voltage of 40 kV and a current of 40 mA. The scanning speed was 5 min À1 and the step size was 0.02 with the range from 10 to 60 . The surface and cross-section morphology of samples were observed by scanning electron microscopy (SEM, FEI Quanta FEG, USA) with an accelerating voltage of 20 kV. The elemental distribution mapping of Gr-PANI 10:1 was measured by energy dispersive spectrometer (EDS, Oxford Instruments, United Kingdom). The electrical conductivity of lms was measured by in-line four-point probe method (RTS-9, Scientic Equipment and Services) at room temperature, and three specimens with the thicknesses of 0.04 mm were used for each sample. The tensile strength and the elongation at break of the Gr-PANI@PI and Gr@PI lms were measured by a universal stretching machine (UTM, 5567A, Instron, USA). Specimens were tested at a speed of 2 mm min À1 and a preload of 0.1 N. The size of each sample was 10 mm in length and 4 mm in width. Five specimens were used for each sample in the tensile test. The EMI shielding properties of the composite lms were performed in the frequency range of 8.2-12.4 GHz (X band) using the waveguide method via the vector network analyzer (Agilent N5222B). All samples were cut into rectangle plates with a size of 22.8 Â 10.1 mm 2 to t the waveguide sample holder. The scattering parameters (S 11 and S 21 ) of the Gr-PANI@PI composite lms were collected to calculate the EMI shielding effectiveness. Thermogravimetric analysis (TGA, Q5000, TA Instruments, USA) was performed to determine the thermal stability of lms. Thermal gravimetric analysis (TGA) was conducted under the N 2 atmosphere from 50 to 900 C at a heating rate of 10 C min À1 .

Characteristics and morphology
The optical photos of PI and Gr-PANI 10:1 @PI lm with different loading of 30%, 35% and 40% are displayed in Fig. 1(a). With the increasing amount of the ller, the Gr-PANI 10:1 @PI lms become darker compared to the transparent yellow PI lm. Furthermore, the 40%Gr-PANI 10:1 @PI still remains excellent exibility. It can be twisted and folded into a paper plane ( Fig. 1(b)). The size of 40%Gr-PANI 10:1 @PI lm reaches 22 Â 33 cm 2 as displayed in Fig. S1(a) in ESI. † The dispersibility of ller in a polymer matrix is crucial for the preparation of high performance composites. 15 In our experiment, the PANI is in situ synthesized on the surface of Gr sheets to connect adjacent Gr sheets through p-p non-covalent bonding. When the Gr-PANI composites are added into the NMP, the PANI can effectively prevent Gr sheets from agglomerating. 16,17 As displayed in Fig. S1(b) in ESI, † aer 8 hours of standing, Gr sheets aggregate partly in NMP, but, Gr-PANI 10:1 composite ller shows excellent dispersibility due to the p-p non-covalent interactions between Gr and PANI, indicating that the PANI can avoid the agglomeration of Gr sheets in the organic solvent. 18 The SEM morphologies of Gr, Gr-PANI 10:1 , PI and 40%Gr-PANI 10:1 @PI are shown in Fig. 2. The pure Gr ( Fig. 2(a)) displays a relatively smooth and at surface. However, in Fig. 2(b), there are some particles attached on the surface of Gr sheets. It demonstrates that the PANI is aligned on the Gr surface homogeneously, which is ascribed to the p-p interaction between PANI and Gr. 19,20 The size of the PANI absorbed on the Gr sheets is smaller than that of pure PANI ( Fig. S3(a) in ESI †), which is benecial for shortening the diffusion path of electrons and for enhancing the conductivity of the composite lm. 21 Additionally, the EDS mapping of Gr-PANI 10:1 , as demonstrated in Fig. S2, † shows that the C and N element is uniformly distributed on the Gr-PANI 10:1 ller. The atomic percentage of N reaches 22.19 at%, indicating the successful attachment of PANI in the ller (Table S1 †). While in the Gr- Fig. 1 (a) The photos of pure PI and PI films with different Gr-PANI 10:1 contents; (b) the flexible 40%Gr-PANI 10:1 @PI film. PANI 5:1 (Fig. S3(b) in ESI †), the agglomeration of PANI can be observed on the surface of the Gr sheets, hampering the formation of the conductive network. The results indicate that the excessive PANI will lead to agglomeration in the composites. The cross sections of the pure PI and the 40%Gr-PANI 10:1 @PI are further investigated, as shown in Fig. 2(c) and (d). The PI lm displays a relatively smooth structure, while the crosssection of Gr-PANI 10:1 @PI exhibits a typical stake-up morphology, demonstrating that Gr sheets are uniformly distributed in the PI matrix and form coarser fractured surfaces compared with pure PI. It can be attributed to the excellent compatibility and interfacial interaction between the Gr-PANI 10:1 composites and the PI matrix. 22 The PANI acts as the solder to connect adjacent Gr sheets into an integrated conductive structure with less gaps, thereby improving the electrical performance. Fig. 3(a) shows the FTIR spectra of Gr, PANI, and Gr-PANI 10:1 ller. The typical adsorption peaks of PANI are located at 1572 cm À1 , 1483 cm À1 , 1307 cm À1 , 1243 cm À1 and 1140 cm À1 , which represent C]C vibration in the quinoid ring, C]C stretching in the benzene ring, C-N stretching in the secondary aromatic amines, and C-H bonding in the benzenoid ring and the quinoid ring, 23 respectively. In the Gr-PANI 10:1 , these characteristic peaks are shied to 1566 cm À1 , 1476 cm À1 , 1296 cm À1 , 1232 cm À1 , and 1086 cm À1 , respectively, indicating that PANI is distributed intimately on the surface of the Gr sheets, 24 due to the electrostatic interactions of p-p non-covalent bonding between the Gr and PANI. 25 The peak shi can be also observed in Gr-PANI 15:1 and Gr-PANI 5:1 (Fig. S4(a) (ESI †)).
To investigate the interaction between Gr sheets and PANI, the Raman spectroscopy of PANI, Gr and Gr-PANI 10:1 is conducted, as displayed in Fig. 3(b). There are two prominent peaks corresponding to the G-band ($1570 cm À1 ) of the rst-order scattering of E 2g mode of sp 2 -hybridized carbon atoms and the D-band ($1345 cm À1 ) of the disordered amorphous carbon C-C bonds at the defects or edge boundaries, 26 respectively. Aer the in situ polymerization of PANI, the G-band and the Dband of the Gr-PANI 10:1 are shied to 1561.3 cm À1 (vs. 1574.1 cm À1 for Gr), and 1343.7 cm À1 (vs. 1345.0 cm À1 for Gr), respectively, suggesting the p-p interaction between PANI and Gr sheets. 27 Moreover, the I D /I G values of Gr and Gr-PANI 10:1 are 0.194 and 0.088, respectively. With the decreasing mass ratio of Gr to PANI, the decrease in I D /I G values can be also observed in Gr-PANI 15:1 and Gr-PANI 5:1 (Fig. S4(b) (ESI †)). The intensity ratio of D to G bands in Gr-PANI composites is attributed to the intimate interaction between the PANI and Gr sheets. 23,28,29 Together with the FTIR spectra, the Raman results conrm that the Gr sheets are welded by PANI through p-p bonding during the in situ polymerization, thereby obtaining the PI lm with high electrical property. Fig. 3(c) exhibits the XRD patterns of pure PI, 40%Gr@PI and 40%Gr-PANI 10:1 @PI. The XRD pattern of pure PI shows a broad This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 2368-2377 | 2371 peak centered at 2q ¼ 18.5 . 30,31 The sharp peaks centered at $26.50 in 40%Gr@PI and 40%Gr-PANI 10:1 @PI are attributed to the (002) of Gr. The 40%Gr@PI shows an obvious diffraction peak at 2q ¼ 26.54 corresponding to the d (002) of 0.3362 nm, revealing the incorporation of PANI enlarges the interlayer space of graphene. 30 The thermal stability is a key property of PI-based composite. Fig. S5 † shows that the TGA curves of pure PANI, Gr and Gr-PANI 10:1 composites under nitrogen atmosphere. The 15 wt% weight loss temperature of Gr-PANI 10:1 shis to higher temperature compared with pure PANI and Gr. The thermal stability of Gr-PANI 10:1 is remarkably enhanced due to the presence of PANI and the p-p non-covalent between the Gr sheets and PANI. 12 Meanwhile, TGA results demonstrate that the as-fabricated Gr-PANI 10:1 composites contain 12.5 wt% of PANI by calculation. Fig. 3(d) presents the TGA curves of pure PI, 40%Gr@PI, and 40%Gr-PANI 10:1 @PI at a heating rate of 10 C min À1 in nitrogen atmosphere. The pure PI lm shows an onset weight loss below 100 C due to the vaporization of water. 32 The main mass loss occurs at approximately 500 C, which is attributed to thermal decomposition of PI. 33 The 15% weight loss temperatures of pure PI, the 40%Gr@PI, and 40% Gr-PANI 10:1 @PI are 592 C, 599 C and 600 C, respectively. Meanwhile, the residual weight at 700 C for the 40%Gr-PANI 10:1 @PI is evidently improved from 66.4% (pure PI) to 72.7%, indicating that the Gr-PANI 10:1 @PI lm shows superior thermal stability. This phenomenon is owing to the excellent thermal stability of Gr-PANI 10:1 , its introduction reduces the TGA rate of the PI lm. 34 XPS spectroscopy is further characterized to analyze the elemental compositions and the valence states of Gr, PANI and Gr-PANI 10:1 . The C 1s spectrum of Gr ( Fig. 4(a)) consists of three peaks occurring at 284.6 eV, 285.5 eV and 286.6 eV, corresponding to the C]C, C-C and C-O, respectively. 35 The content of each bonding is 87.5 atom%, 10.2 atom%, and 2.3 atom%, respectively. The three peaks of Gr-PANI 10:1 , as shown in the Fig. 4(b),are shied to 284.5 eV, 285.2 eV and 286.5 eV, which is ascribed to the p-p interaction between Gr and PANI. 36 Aer the in situ polymerization of PANI, the three peak areas show negligible changes 84.5 atom% (vs. 87.5 atom% for Gr), 10.0 atom% (vs. 10.2 atom% for Gr), 2.0 atom% (vs. 2.3 atom% for Gr), suggesting the interaction between Gr and PANI is not covalent bonding. Furthermore, an additional peak at 285.9 eV (3.5 atom%) corresponding to the C-N bonding is observed, suggesting that PANI has been successfully in situ synthesized on Gr sheets. The core-level N 1s spectrum of pure PANI presents three primary peaks -N] at 399.1 eV, -NH 2 at 399.8 eV and -N + at 401.0 eV in Fig. 4(c). 37 The peak shi is also observed in Gr-PANI 10:1 (Fig. 4(d)), and the three peaks move toward lower binding energy values of 399.1 eV, 399.5 eV and 400.6 eV, respectively. 38 Furthermore, there are no obvious changes in three peak areas 33.5 atom% (vs. 32.0 atom% for PANI), 33.1 atom% (vs. 37.0 atom% for PANI), 33.4 atom% (vs. 31.0 atom% for PANI).The above results conrm the gap between the Gr sheets could be coupled together by PANI via the p-p noncovalent bonding, which can signicantly enhance the dispersibility of Gr sheets.

Electrical conductivity
The electrical conductivity (s) of each ller is measured by fourpoint probe method (RTS-9, Scientic Equipment and Services). The pure Gr has the highest s of 618.1 AE 46.2 S cm À1 compared with Gr-PANI composites, as displayed in Fig. S6. † With the increasing amount of PANI, the s decreases to 541.0 AE 16.9 S cm À1 , 490.3 AE 39.0 S cm À1 , 392.4 AE 44.5 S cm À1 , corresponding to Gr-PANI 15:1 , Gr-PANI 10:1 and Gr-PANI 5:1 , respectively. Such a decrease in s is primary attributed to the factor that the electrical property of PANI is inferior to that of Gr. Furthermore, the s of the 40%Gr@PI is only 1.4 AE 0.1 S cm À1 due to the inevitable agglomeration of Gr sheets in PI matrix. But with the increasing mass ratio of Gr to PANI from 15 : 1 to 10 : 1 at a ller loading of 40%, the s increases to 1.8 AE 0.2 S cm À1 and 2.1 AE 0.1 S cm À1 , but when the ratio is up to 5 : 1, the s displays a dramatic decline to 1.3 AE 0.1 S cm À1 . It is noticeable that the maximum s reaches 2.1 AE 0.1 S cm À1 for 40%Gr-PANI 10:1 @PI, indicating the PANI welds the adjacent Gr sheets into conductive network so that it can increase the efficiency of electron transfer in PI lm via the "non-covalent welding" strategy. However, with the further addition of the PANI (Gr-PANI 5:1 ), there is an obvious decrease of s in 40%Gr-PANI 5:1 @PI. Taking the result of SEM images into consideration, the reason lies in that the surface of Gr sheets is wholly covered by excessive PANI. The excessive PANI solder would lead to agglomeration between the Gr sheets, impeding the electron transfer.
We also make a comparison between the Gr@PI and the Gr-PANI 10:1 @PI with different ller contents, as shown in Fig. S6(b). † With the increasing amount of Gr from 30% to 40%, the s improves from 0.4 AE 0.1 S cm À1 to 1.4 AE 0.1 S cm À1 , respectively. In contrast, with the introduction of PANI, the s of PI is improved further. When the ller content is at 40%, the s of Gr-PANI 10:1 @PI is up to 2.1 AE 0.1 S cm À1 , which is approximately 45% higher than that of Gr@PI lm. The results demonstrate that the "non-covalent welding" strategy can successfully weld up the Gr sheets to improve the dispersibility of Gr and form conductive network in PI matrix, which can signicantly enhance the s of the composite lm.

Mechanical properties
The mechanical properties and stress-strain curves of the Gr@PI and Gr-PANI 10:1 @PI are shown in Fig. S7 and S8 (ESI †), This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 2368-2377 | 2373 respectively. The tensile strength decreases gradually with the increasing amount of ller from 65.9 AE 1.0 MPa (30%) to 54.3 AE 2.8 MPa (40%). Furthermore, the tensile strength of the Gr-PANI 10:1 @PI is higher than that of Gr@PI with identical ller loadings. Compared with 40%Gr@PI (54.3 AE 2.8 MPa), the tensile strength of 40%Gr-PANI 10:1 @PI increases to 56.8 AE 1.4 MPa. The elongation at break of Gr@PI and Gr-PANI 10:1 @PI presents a slight decline with increasing ller contents (Fig.-S7(b) †). The elongation at break of the 30%Gr@PI is only 3.4 AE 0.3%. However, the 30%Gr-PANI 10:1 @PI displays a slight increase in the elongation at break, which is about 1.29 times higher than that of 30%Gr@PI. The improvement in mechanical properties can be primarily ascribed to the homogeneous dispersibility of Gr-PANI 10:1 in PI matrix because of the "noncovalent welding" strategy. 39

EMI shielding effectiveness
The total EMI shielding effectiveness (SE T ) of the Gr@PI and Gr-PANI 10:1 @PI is measured with vector network analyzer at the Xband (Fig. S9 (ESI †)). The total EMI SE T of shielding materials can be dened as the logarithmic of the ratio of the incident wave P I to the transmitted wave P T , which is calculated by the following equation: 40 In general, the electromagnetic wave consists of three aspects: reection (SE R ), absorption (SE A ) and multiple reections (SE M ), respectively. The EMI performance is attributed to the interaction between the electromagnetic waves and free electrons on the surface of materials. The material with high electrical conductivity can provide electric dipoles. When the incident electromagnetic waves interact with the free electrons on the surface and some electromagnetic waves are reected from the surface of the shield. If electric dipoles interact with the electromagnetic elds in the radiation, the remaining electromagnetic waves would be absorbed by the dipoles. [41][42][43] In addition, the thickness also affects the value of SE M and SE A . The SE M can be neglected when the value of SE T is larger than 10 dB, therefore, SE T can be simplied as the equation: 44 In a vector network analyzer, we can record the parameters (S 11 and S 21 ) to calculate the EMI SE as shown in following equation: 45 shielding mechanism of Gr-PANI 10:1 @PI (P I : incident wave; P R : reflected wave; P T : transmitted wave).
As illustrated in the equation, T is transmittance coefficient, A is absorbance coefficient, R is reection coefficient.
The SE T of the PI lms with different amount of Gr and Gr-PANI 10:1 llers is shown in Fig. 5(a) and (b). The electrical property plays an important role in the SET of composite lms. As the amount ller from 30% to 40%, the SE T values of both Gr@PI and Gr-PANI 10:1 @PI increase signicantly due to the increasing electrical conductivity. As shown in Fig. 5(c), the SE T of Gr-PANI 10:1 @PI is $21.3 dB at 40%, which is approximately 125% enhancement compared with Gr@PI ($17.1 dB). The uniformly dispersed Gr sheets could attenuate electromagnetic radiation easily due to reection and scattering microwave in the Gr-PANI 10:1 @PI lm, and incident electromagnetic microwaves are transferred to heat by being absorbed. The Gr-PANI 10:1 @PI shows higher SE T value compared with Gr@PI, which can be ascribed to the structure of conductive network by the "non-covalent welding" approach.
The high electrical conductivity and the formation of conductive networks within the PI matrix play an important role in the EMI shielding performance of the lm. Fig. 5(d) shows the EMI shielding mechanism of Gr-PANI 10:1 @PI lm. A portion of electromagnetic wave P I is reected at the interface of the lm, because conductive networks with a large amount of charge carriers are benecial to EMI shielding effectiveness by reection. The remaining portion penetrating the surface is multiply reected or absorbed by the Gr layers. 46 Only a small amount of electromagnetic waves pass through the lm. The formation of an excellent conductive network between Gr and PANI in PI matrix ensures that the electromagnetic wave is greatly reduced, leading to an outstanding EMI shielding performance. On the other hand, the Gr-PANI@PI lm possesses the continuously cross-linked network structure Scheme 2 The mechanism of (a) Gr@PI without PANI and (b) Gr-PANI 10:1 @PI with PANI via "non-covalent welding".
This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 2368-2377 | 2375 entirely consisted of Gr-PANI layers, which could certainly enhance internal multiple reections more effectively than the agglomerated Gr layers in PI matrix. With the effect of the above two aspects, the EMI performance of Gr-PANI 10:1 @PI lm has been obviously improved. 47,48 Generally, the density and thickness of a material are also two signicant factors for evaluating its EMI shielding performance. Hence, the SSE on the basis of the EMI SE T , the density, and the thickness is an appropriate criterion to estimate the shielding performance of various EMI shielding materials. The SE T value of reported EMI shielding lms in the X-band region is listed in Table S2. † Compared with other carbon-based polymer lms, the 40%Gr-PANI@PI 10:1 has more excellent SE T (21.3 dB) and higher SSE (4096.2 dB cm 2 g À1 ) with the thickness only of 0.04 mm. This value is higher than that of the PS/ graphene ($17.3 dB) prepared by high-pressure compression molding 49 and PU/MWCNT ($20.0 dB) prepared ball mill method, 50 respectively. Moreover, it is also superior to that of the WPU/CNT composites by a facile freeze-drying method by 72.6% ($2143.0 dB cm 2 g À1 ) 51 and PI/rGO (17-21 dB) 52 with a thickness of 0.8 mm. The survey suggests that Gr-PANI@PI 10:1 lm has the potential advantages to meet the requirement of the commercial application (more than 20 dB).
3.5. The mechanism of "non-covalent welding" The mechanism of "non-covalent welding" between PANI and Gr is illustrated in Scheme 2. In Scheme 2(a), the Gr@PI lm is directly fabricated by in situ polymerization. Without the PANI, during stirring process, the Gr sheets tend to agglomerate due to Vander Waals force and p-p stacking interaction between Gr sheets resulting in an inhomogeneous and discrete network. 53 In Scheme 2(b), the PANI is in situ synthesized on the surface of Gr sheets to weld up adjacent Gr sheets through p-p noncovalent bonding, which effectively avoids the agglomeration of Gr sheets in organic solvent. Aer the thermal imidization, Gr-PANI forms a continuous conductive network in polymer matrix, making the electron transfer more effective. In this case, the combination of Gr and PANI only depends on p-p noncovalent bonding, and the structure of Gr is not destroyed, ensuring the excellent conductivity of Gr.

Conclusions
In summary, we have reported a novel "non-covalent welding" approach to prepare Gr-PANI 10:1 @PI lm with enhanced EMI shielding performance and electrical properties. The PANI, as a solder, can successfully weld up the graphene sheets via p-p non-covalent bonding and form a continuous conductive network in polymer matrix by in situ polymerization. The Gr-PANI 10:1 @PI exhibited a superior SE T of 21.3 dB and SSE of 4096.2 dB cm 2 g À1 at 40% content with a thickness of 40 mm compared to that of the Gr@PI (SE T 17.1 dB). The s of Gr-PANI 10:1 @PI was promoted to 2.1 AE 0.1 S cm À1 higher than Gr@PI (1.4 AE 0.1 S cm À1 ) at 40% content. Last but not least, we believe that such a "non-covalent welding" strategy shows a great potential to fabricate composites with inorganic llers.

Conflicts of interest
There are no conicts to declare.