Controllable fabrication of superhierarchical carbon nanonetworks from 2D molecular brushes and their use in electrodes of flexible supercapacitors

LU Yu-heng, TANG You-chen, TANG Ke-han, WU Ding-cai,*, MA Qian

（1. PCFM Laboratory, School of Chemistry, Sun Yat-sen University, Guangzhou 510006, China;2. Department of Orthopedics, the Eighth Affiliated Hospital, Sun Yat-sen University, Shenzhen 518000, China;3. Research Center of Medical Sciences, Guangdong Provincial People’s Hospital, Guangdong Academy of Medical Sciences, Guangzhou 510080, China）

Abstract： Three-dimensional carbon nanonetworks (3D CNNs) have interconnected conductive skeletons and accessible pore structures, which provide multi-level transport channels and thus have promising applications in many areas. However, the physical stacking of these network units to form long-range conductive paths is hard to accomplish, and the introduction of micropores and small mesopores is usually difficult. We report a simple yet efficient strategy to construct CNNs with a nitrogen-doped micro-mesomacroporous carbon nanonetwork using Schiff-base gelation followed by carbonization. Using a polyacrolein-grafted graphene oxide molecular brush as the building block and tetrakis (4-aminophenyl) methane as the crosslinking agent, the obtained molecular brush nanonetworks have a high carbon yield and largely retain the original morphology, leading to the formation of a 3D continuous nanonetwork after carbonization. The materials have a micro-meso-macroporous structure with a high surface area and a highly conductive N-doped carbon backbone. This unique structure has a large number of exposed active sites and excellent charge/mass transfer ability. When loaded on carbon cloth and used as the electrodes of a flexible supercapacitor, the CNN has a specific capacitance of 180 F g-1 at 1 A g-1 and a high capacitance retention of 91.4% after 10 000 cycles at 8 A g-1 .

Key words: Molecular brushes；Carbon nanonetworks；Superhierarchical carbon；Flexible supercapacitors

1 Introduction

Owing to unique interconnected conductive skeletons and accessible porous structures, three-dimensional carbon nanonetworks (3D CNNs) have attracted extensive scientific and technological attention in the past decades[1]. With the rapid development of nanotechnology, a series of strategies by using building blocks with different dimensions (e.g., nanospheres[2],nanotubes[3-5]and nanosheets[6-8]) are used to construct 3D CNNs. These bottom-up strategies can not only maintain the intrinsic properties of building brocks but also achieve multi-level transport channels for electrons and ions, resulting in promising potentials in various fields, including energy storage[9,10]and conversion[11,12], separation[13,14]and catalysis[15].Among them, 3D CNNs formed by nanosheets with high aspect ratios, such as graphene[16,17], graphitic carbon nitride[18], and 2D transition metal carbide[19,20],possess abundant exposed active sites[21]. However,stacking of nanosheets mainly forms large-sized mesopores and macropores instead of micropores and small-sized mesopores[22], which greatly restricts their performance in the applications relying on accessible active surface areas. In addition, 3D CNNs are usually formed by physical stacking of nanosheets rather than chemical covalent connection of nanosheets, thus lacking stable long-range 3D conductive pathways[23-25].

In this context, various strategies, such as postsynthesis activation[26]and additional spacer components[27,28], are proposed to create nanoporosity in 3D CNNs. However, for the post-activation process, 3D CNNs are subjected to thermal treatment in the presence of an active gas/vapor or with the incorporation of an activating agent, which is not only costly and time-consuming but also easy to cause the collapse of the 3D CNN structure. For the spacer-involved procedures, spacer components tend to aggregate together or disperse inhomogeneously between nanosheet layers, resulting in uncontrolled nano-scale and mesoscale architectures of 3D CNNs. Therefore, it remains a challenge to develop a facile yet efficient strategy for constructing the micro-meso-macroporous 3D CNNs by nanosheet building blocks.

Herein, we develop a novel strategy to construct a class of superhierarchical CNNs (SHCNNs,i.e.,nitrogen-doped micro-meso-macroporous carbon nanonetworks) by a facile Schiff-base gelation, using 2D molecular brushes (i.e., polyacrolein-grafted graphene oxide, GO-g-PA) as the building block and tetrakis (4-aminophenyl) methane (TAPM) as the crosslinking agent, and subsequent carbonization (Fig. 1). Due to intra-brush and inter-brush Schiff-base reactions with aldehyde groups (-CHO) in hairy PA side-chains,TAPM molecules act as junctions for crosslinking neighboring GO-g-PA brushes in various directions to form the molecular brush nanonetworks (MBNNs).During carbonization, the MBNNs can bein-situtransformed into SHCNNs with N-doped microporous carbon shells, carbonaceous junctions, and highly long-range conductive frameworks. This superhierarchical structure combines the following advantageous characteristics: (i)In-situintroduction of micropore-rich carbon shell into the nanonetwork unit provides large areas contributing to electric double layer capacitance. (ii) Interconnected meso-macroporous nanonetwork structure serves as fast mass transport pathway. (iii) The long-range network with highly conductive reduced graphene oxide (rGO)cores enables efficient charge transport. (iv) N-doping enhances wettability and electronic conductivity.Therefore, SHCNNs can provide an unusual opportunity to enhance the properties in many applications.For example, when used as electrodes of flexible supercapacitors, the SHCNNs show large specific capacitances (i.e., 180 F g-1at 1 A g-1) and good cycling stability (i. e., 91.4% after 10 000 cycles at 8 A g-1).

2 Experimental

2.1 Materials

GO aqueous dispersions (GO-1, 10 mg g-1) were purchased from Hangzhou Gaoxi Technology Co.,Ltd. Tetrakis(4-aminophenyl)methane (TAPM, 95%),potassium persulfate (99.5%), N, N-dimethylformamide (AR), and ethylacetate (99%) were purchased from Shanghai Macklin Biochemical Co., Ltd. Acrolein and Nifion117 were purchased from Shanghai Aladdin Biochemical Co., Ltd. Carbon cloth(W0S1011) was purchased from Suzhou Keshenghe company. A commercial activated carbon (YP50) was purchased from Kuraray, Japan.

2.2 Preparation of SHCNNs

In a typical synthesis, 5 g of graphene aqueous dispersion (concentration of 10 mg g-1) was added into 25 mL of deionized water under ultrasonication for 30 min at room temperature. The mixture was further stirred for 10 min to obtain a homogeneous dispersion. Afterwards, 3 mL of acrolein monomer was added under ultrasonication for 15 min, and 60 mg of potassium persulfate initiator was added under quickly stirring for 1 min. Subsequently, the air inside the container was removed by pumping and backfilled with nitrogen. The mixture was stirred at 70 °C for 24 h. The obtained product was centrifuged and washed with deionized water for 3 times. The resulting GO-g-PA was further dispersed in 1, 4-dioxane to form a dispersion (20 mg mL-1). TAPM was dissolved in 1, 4-dioxane to form a solution (20 mg mL-1, 0.25 mL) and then the solution was added into the GO-g-PA dispersion. After shaking for 30 s,the mixture was immediately added into 0.25 mL of acetic acid. The mixture was continuously shaken for 30 s and reacted at 60 °C for 24 h. After freeze-drying, the MBNN was prepared. In the above synthesis condition, the molar ratio of -NH2to -CHO was 0.22. The SHCNN was prepared by carbonization of the MBNN at 900 °C for 20 h at a heating rate of 2 °C min-1in a tube furnace with flowing N2. Likely,the other SHCNN products (SHCNN-x-y) were obtained via carbonizing the MBNN under other carbonization conditions (x and y represent carbonization temperature and carbonization time, respectively).Other MBNN products with different molar ratios of—NH2to —CHO (i.e., 0.45, 0.89 and 1.78) were synthesized by different feedings of TAPM, and then carbonized at 900 °C for 3 h to obtain the corresponding SHCNN products. GO-C and GO-g-PA-C were also prepared as control samples by carbonizing GO and GO-g-PA at 900 °C for 3 h, respectively.

2.3 Fabrication of SHCNN/CC electrodes

The active materials, conductive carbon black(Super-P), and Nifion117 were ultrasonically dispersed in an aqueous ethanol solution (75%) in a mass ratio of 90∶5∶5, and the concentration was configured to 3 mg mL-1. Subsequently, the above mixture was added dropwise to the carbon cloth and then dried at 100 °C for 1 h. The electrode possesses a working area of 0.5 cm × 1 cm and a SHCNN loading of 1.8 mg cm-2.

2.4 Assembly of SHCNN/CC//YP50/CC device

The SHCNN/CC and YP50/CC electrodes were fabricated by loading SHCNN (1.8 mg cm-2) and YP50 (3.0 mg cm-2) on 0.5 cm × 2.8 cm carbon cloth,respectively. The loading mass ratios of the electrodes were set based on the following equation[29]:

wheremis the mass of active electrode material (g),Cis the specific capacitance (F g-1), and ΔUis the potential window (V).

The above electrodes were immersed in 6 mol L-1KOH overnight, assembled with a 0.6 cm × 3.5 cm cellulose separator, and then sealed in polyolefin film.

2.5 Material characterization

The nanostructures of the samples were investigated by a Hitachi S-4800 scanning electron microscope (SEM). N2adsorption measurements were carried out on a Quantachrome Autosorb-IQ3 analyzer at 77 K. The BET surface areas (SBET) were determined by Brunauer-Emmett-Teller (BET) theory. The micropore surface area (Smic) were determined by t-plot method. The pore size distributions were analyzed by density functional theory (DFT). X-ray diffraction(XRD) patterns were recorded on a D-MAX 2200 VPC diffractometer using CuKα radiation (40 kV,26 mA). X-ray photoelectron spectroscopy (XPS)measurements were carried out on an ESCALAB 250 spectrometer. Fourier-transform infrared (FT-IR)spectra were recorded on a Bruker Equinox 55 FT-IR spectrometer. Raman spectra were collected on a HORIBA JY with 532 nm laser.

The compaction density measurement was evaluated by pressing the SHCNN in a round mold with a diameter of 1.8 cm at a pressure of 18 MPa for 3 min, and the compaction density (g cm-3) was calculated using the following equation:

wheremSHCNNis the mass of the SHCNN after pressing (g),Amoldis the area of the mold (cm2), andTHKSHCNNis the thickness of the SHCNN after pressing (cm).

The conductivity was measured by a four-point probe method, using a 4-point probe resistivity measurement system (PROBES TECH, Guangzhou). The electrodes for measurements were fabricated by loading active materials (e.g., SHCNN and GO-g-PA-C,1.8 mg cm-2) on 0.5 cm × 2 cm carbon cloth.

2.6 Electrochemical test

All electrochemical measurements except the conductivity measurement were carried out on a CHI 660E electrochemical workstation (CH Instruments,Inc., Shanghai). For a standard three-electrode system,the SHCNN/CC electrode, Pt foil, and saturated Hg/HgO electrode (filled with saturated KCl) were used as the working electrode, counter electrode, and reference electrode, respectively, in an aqueous 6 mol L-1KOH. The cyclic voltammograms (CVs)were tested from -1 to 0 V at scan rates from 1 to 50 mV s-1. The galvanostatic charge/discharge (GCD)measurements were performed at the current densities of 0.1-8 A g-1. Electrochemical impedance spectra(EIS) were recorded by applying the open-circuit potential with an amplitude of 5 mV over the frequency range from 100 kHz to 0.01 Hz. Cycling stability was characterized using GCD measurements over 10 000 cycles at a current density of 8 A g-1. The specific capacitances (F g-1) of the electrodes in three-electrode system were calculated using the following equation[30]:

whereIis the discharge current (A), Δtis the discharge time (s), ΔUis the potential window (V), andmis the mass of the active materials on carbon cloth current collectors (g).

The specific capacitances of the flexible supercapacitor were calculated using the following equation[30]:

whereIis the discharge current (A), Δtis the discharge time (s), ΔUis the potential window (V), andMis the total mass of the active materials on SHCNN/CC and YP50/CC electrodes (g).

The energy density (Wh kg-1) of SHCNN/CC//YP50/CC device was calculated using the following equation[30]:

whereCMis the specific capacitance of a supercapacitor (F g-1), and ΔUis the potential window (V). The power density (W kg-1) of SHCNN/CC//YP50/CC device was calculated using the following equation[30]:

whereEis the energy density (Wh kg-1), and Δtis the discharging time (s).

3 Results and discussion

As shown in Fig. 2a, the obtained MBNN clearly exhibits a gelation phenomenon, indicating that the imine crosslinking bridges formed by the Schiff-base reaction can promote the formation of a gel. The bond formation processes during gelation were measured by FT-IR as shown in Fig. 2b. Compared with GO, a strong absorption peak around 1 720 cm-1can be obviously observed in GO-g-PA, indicating the existence of aldehyde groups in hairy PA side-chains[31].The peak at 3 394 cm-1assigned to the stretching vibration of the amino group disappears and a new peak centred at 1 608 cm-1assigned to the imine group appears in the MBNN[32], indicating the Schiff-base reaction occurs between GO-g-PA and TAPM. The morphological and structural features of the corresponding samples were characterized by SEM. After PA side-chains were grafted on GO nanosheets (Fig. S1a),GO-g-PA still possesses well-defined 2D structure and smooth surface (Fig. 2c and Fig. S1b), indicating that the uniform distribution of hairy PA on the surface of GO. After gelation between GO-g-PA and TAPM, the obtained MBNN shows apparent continuous macropores (Fig. 2d and Fig. S1c), exhibiting the distinct structure of 3D network. Obviously, the surface of MBNN becomes rougher after the crosslinking. Furthermore, with increasing the molar ratio of-NH2to -CHO, the surface roughness of the corresponding MBNNs is increased (Fig. S1d-f), demonstrating that the degree of crosslinking on intra-brush and inter-brush could be controlled by the ratio of-NH2to -CHO. When the molar ratio of -NH2to-CHO is within 0.89, the resulting SHCNNs present a 3D nanonetwork structure with nanosheet network units (Fig. S2a-d), which is similar to that of its precursor, MBNN. Under the certain molar ratio of-NH2to -CHO (e.g.,0.22), when the carbonization time is extended from 3 to 20 h, the 2D morphology of network units and interconnected macropores are still well retained in the obtained SHCNNs and the nanosheets tend to pack more tightly (Fig. 2e and Fig. S3). Evidently, the SHCNNs can withstand molecular overflow and rearrangement during high temperature and longtime treatment without apparent destruction, presenting outstanding nanostructural inheritability.

To explore the effect of crosslinking bridges on the thermal stability of MBNNs, thermal gravity analysis (TGA) was carried out. According to the TGA curves, the PA content in GO-g-PA is about 66%, indicating that the 2D molecular brushes have a high polymer grafting rate and a large amount of aldehyde group reactive sites (Fig. 3a). It is worth noting that the weight loss of MBNN is about 70% at 700 °C,which is lower than those of PA (100%), GO-g-PA(83%) and TAPM (94%). Moreover, carbonization yields of products obtained from different MBNN precursors are higher than those of GO-g-PA and TAPM precursors (Table S1). Especially, carbonization yields of SHCNNs gradually increase with increasing the amounts of TAPM in MBNN. These results prove that abundant crosslinking bridges effectively resist the loss of carbon at high temperature and improve the carbonization yields. As shown in Fig. 3b, there is no sharp peak in the XRD pattern of MBNN, indicating that the hairy polymer can inhibit the stacking of GO plane[33]. The XRD pattern of SHCNN shows broad peak at 25°, suggesting the formation of amorphous carbon. The Raman spectra show that SHCNN exhibit higher intensity ratio ofDband toGband (ID/IG)than MBNN, demonstrating a lower degree of graphitization, which could be attributed to the transformation of polymer into amorphous carbon during carbonization (Fig. 3c). This is consistent with the XRD result. The XPS analysis was performed to investigate the N states in MBNN and SHCNN. As shown in Fig. 3d, the high-resolution N1s spectrum of MBNN is deconvoluted into three peaks located at 399.3(C-N), 400.1 (C＝N), and 402.2 eV (N-H)[34,35],suggesting the formation of imine groups by Schiffbase reaction and retaining of amino group in TAPM.After carbonization at 900 °C for 20 h, SHCNN still keep a N content of 2.69% (Fig. S4). Elemental mapping demonstrates the presence and homogeneous distribution of N element in the SHCNN (Fig. S5). The peaks of pyridinic N (398.5 eV), pyrrolic N (399.7 eV), graphitic N (400.7 eV) and oxidized N (401.6 eV) in the high-resolution N1s spectrum of SHCNN indicate the formation of N-doped carbon frameworks by the rearrangement of molecular during the high-temperature pyrolysis (Fig. 3e)[36].

N2adsorption-desorption isotherms and DFT pore size distribution curves reveal the hierarchical porous structure of the SHCNN. As shown in Fig. 3f,there is a high nitrogen uptake at low relative pressure in SHCNN (obtained at 900 °C, 20 h) in contrast of MBNN, indicating the introduction of numerous micropores in SHCNN by carbonization. In addition,the isotherm of SHCNN shows a hysteresis loop at medium relative pressure, revealing the presence of mesopores, and has no limiting adsorption at high relative pressure, indicating the existence of macropores[37]. The BET surface area (SBET) of SHCNN is 1 187 m2g-1, which is dominated by micropores (Smic=985 m2g-1, Table S2) and is also much higher than that of MBNN (24 m2g-1). The DFT pore size distribution also suggests the hierarchical porous structure of the SHCNN, which is obviously different from nearly nonporous structure for MBNN. Similar to the sample GO-C (Fig. S6a and b), the meso- and macropores in the SHCNN may be attributed to interspace between aggregated nanosheets. The compaction density of above SHCNN is about 0.417 g cm-3.Furthermore, the microporous structure can be adjusted by the carbonization time and temperature. For example, as shown in Fig. S6c-h and Table S2, theSBETslightly decreases to 926 m2g-1and theSmicsignificantly decreases to 543 m2g-1for SHCNN-900-3 when the carbonization time is decreased to 3 h at 900 °C.TheSBETandSmicdecrease to 368 and 151 m2g-1for SHCNN-700-20, respectively, when the carbonization temperature is decreased to 700 °C for 20 h.These results indicate the development of micropores are more sensitive to temperature than the carbonization time.

Motivated by the unique superhierarchical structure, the SHCNNs are used as electrodes of flexible supercapacitors. As shown in Fig. 4a, the SHCNN loaded on carbon cloth (SHCNN/CC, the SHCNN was obtained at 900 °C for 20 h) show quasi-rectangular CV profiles at scan rates from 1 to 50 mV s-1, indicating the electric double layer capacitive behavior. The SHCNN/CC possesses high specific capacitances (calculated by the galvanostatic charge/discharge test,GCD) in a range from about 291 to 154 F g-1at current densities from 0.1 to 8 A g-1, indicating a good rate capability (Fig. S7 and Fig. 4b). Besides, symmetric GCD curves reveal a good coulombic efficiency of SHCNN/CC. Higher electrical conductivity plays an important role in the good rate capability and coulombic efficiency. SHCNN/CC and SHCNN-900-3/CC electrodes have higher electrical conductivities than GO-g-PA-C/CC electrode (Fig. S8). These enhanced electronic conductivities may be derived from the N-doped network units and long-range nanonetwork formation.

Notably, the specific capacitance of SHCNN/CC is 180 F g-1at 1 A g-1, which is superior to that of the commercial active carbon (YP50) electrode (109 F g-1at 1 A g-1). Considering the much largerSBET(1 740 m2g-1) of YP50, the higher specific capacitance of the SHCNN may be ascribed to the sufficient utilization of active sites (Fig. S9). According to the Nyquist plot, the SHCNN/CC show a low solution resistance of about 1.05 Ω, which is due to good wettability of N-doped surface (Fig. 4c)[38,39]. The almost vertical straight line in the low-frequency region indicates a good capacitive behavior of the SHCNN/CC,which can be ascribed to the hierarchical porous structure[40]. The cycling stability was investigated by GCD test at 8 A g-1for 10 000 cycles. As shown in Fig. 4d,the SHCNN/CC displays a capacitance retention of 91.4% after charging/discharging for 10 000 cycles,showing a good cycling stability. These results can be attributed to abundant and accessible active sites originated from 2D nanonetwork unit and effective mass/charge transfer channels provided by the superhierarchical porous structure and lone-range conductive framework[41,42].

An asymmetric flexible supercapacitor device was assembled by using SHCNN/CC and YP50/CC as electrodes (SHCNN/CC//YP50/CC) in 6 mol L-1KOH.As shown in Fig. 5a-d, the SHCNN/CC//YP50/CC device shows high flexibility and no obvious destruction under bending angles from 45° to 135°.Moreover, according to Fig. 5e, the SHCNN/CC//YP50/CC device displays similar shape of CV curves in different bending states, revealing its good flexibility. According to Fig. 5f, the energy density of the SHCNN/CC//YP50/CC device is 2.5 Wh kg-1at a power density of 93.2 W kg-1. These electrochemical performance presents the potential of SHCNN as electrodes of flexible supercapacitors.

4 Conclusion

In summary, we develop a novel strategy to construct SHCNNs with nitrogen-doped micro-mesomacroporous carbon nanonetwork through a Schiffbase gelation, followed by carbonization. Compared with the conventional CNN, the SHCNNs have Ndoped micropore-abundant carbon shells, carbonaceous junctions, and long-range conductive frameworks, due to an outstanding carbonizability and nanostructural inheritability benefitting from the intrabrush and inter-brush crosslinking bridges. Thus, the SHCNNs display well-developed active sites and excellent charge/mass transfer ability. When used as electrodes of flexible supercapacitor, the SHCNNs show large specific capacitances and high cycling stability. Therefore, this strategy may open a new opportunity to create novel high-performance 3D CNNs for energy storage.

Acknowledgements

The authors are grateful for the financial support from the projects of National Natural Science Foundation of China (51925308 and 51872336) and National Key Research and Development Program of China(2021YFF0500600).