Exposing residual catalyst in a carbon nanotube sponge

Abstract

Carbon nanotubes (CNTs) are grown from metal catalysts; after growth, residual catalyst particles are usually encapsulated within the tube cavities and are difficult to remove. Here, we directly expose controlled amounts of Fe catalyst from a bulk CNT sponge by thermal annealing at mild temperature, and produce uniformly dispersed Fe₂O₃ nanoparticles grafted onto CNTs throughout the sponge. Those exposed catalyst particles remain active and can be used to synthesize CNTs again. The resulting CNT–Fe₂O₃ composite sponges, which possess a highly porous and conductive network, also serve as freestanding supercapacitor electrodes with significantly enhanced specific capacitance than original CNT sponges. Our results indicate that residual metal catalysts, widely present in CNT-based materials, can be reactivated and utilized for applications in catalysis and energy areas.

---

Introduction

Carbon nanotubes (CNTs) have excellent properties and many promising applications in electronics, energy and environmental areas.^1–6 CNTs are synthesized by a number of methods, among those, catalytic chemical vapor deposition (CVD) is commonly used as it produces large-area samples (e.g. films, arrays) on specific substrates, with high flexibility in structural control (e.g. patterning, doping) and the potential to be scaled up.^7–12 Those CVD processes demand catalyst loading for the growth of CNTs, usually via a vapor–liquid–solid mechanism, with transitional metals such as Fe, Co and Ni being most efficient. As a result, as-grown CNT samples contain an appreciable amount of residual catalyst, which exist as small particles adhered onto the CNT tips or encapsulated within the tube cavities. These catalyst particles are covered by graphitic layers thus very difficult to remove. Methods that have been proposed to purify CNTs include oxidation, bromination, chloration, steam treatment, heat treatment and acid treatment.^13–19 Even after vigorous processing, there remains a certain amount of metallic impurities, and high-degree purification is usually accompanied by disturbance of CNT structure and degradation of properties.²⁰

On the other hand, there were also efforts in possible utilization of catalysts contained in CNTs. For example, catalyst particles pinned to the supporting layer were recycled to grow either multi-walled or single-walled CNTs for multiple times.^21–25 Although residual catalysts (e.g. Fe, Co and Ni) are often sheathed by graphitic layers, several studies have demonstrated that those metal impurities are responsible for the electrochemical (redox) activity in CNT samples, even at a small amount (middle-ppm range).^26–29 But for applications such as electrode materials, further enhancement in the electrocatalytic activity is necessary, by making those trapped catalyst particles directly accessible to outside. Emmett, et al. utilized voltammetric cycling to rupture the CNT walls and make the internal Fe particles accessible to electrolyte, and their buckypaper samples showed improved capacitance due to the Faradaic process of iron catalyst.³⁰ Thus the CNT materials containing metallic impurities could be a promising candidate for making high-performance supercapacitors, by combining the high surface area of CNTs and the redox activity from partially or fully exposed catalyst particles.

Here, we show that thermal annealing at mild temperatures can directly expose encapsulated catalyst particles in a bulk CNT sponge, while retaining the highly porous network structure. We found that the exposed Fe nanoparticles were simultaneously transformed into crystalline oxide particles during annealing. The resulting CNT-nanoparticle composite sponges acted as a porous catalyst substrate for further growth of new CNTs, and also served as freestanding conductive supercapacitor electrodes with 9-fold capacitance increase than original CNT sponges. Our work demonstrated a simple and low-cost way to reuse residual catalyst in CNT-based materials and to fabricate functional composite structures for various applications.

Results and discussion

The CNT sponges have a three-dimensional (3D) porous structure, as reported by our team.³¹ During synthesis, catalyst and carbon precursors (ferrocene and dichlorobenzene) were continuously injected into the CVD system, and as-grown sponges contain a large amount of trapped Fe catalyst. Here, the annealing process was carried out in a temperature range of 450 to 800 °C in air or inertia atmosphere (see Experimental for details). As illustrated in Fig. 1a, Fe particles encapsulated at the CNT tips or inside the tube cavities were exposed and simultaneously oxidized into Fe₂O₃ crystals. These oxide crystals still attached to the CNT network, resulting in the formation of a CNT–Fe₂O₃ composite sponge. Two applications were explored later based on this composite sponge, as a catalyst substrate to re-grow CNTs or pseudo-electrode for supercapacitors.

Fig. 1 Fabrication of CNT–Fe₂O₃ sponges, contact angle and mechanical measurement. (a) Schematic illustration of the synthetic method towards exposed residual catalysts in CNTs and the applications of CNT–Fe₂O₃ composite sponges for porous catalyst substrate and supercapacitor electrode. (b) Photos of an original bulk CNT sponge and annealed CNT–Fe₂O₃ sponges (for 60 min and 150 min, 450 °C in air). Red color of the third sample indicates severe aggregation of Fe₂O₃. (c) Contact angle measurement of an original and annealed CNT sponge. A water droplet injected onto the flat surface of a CNT sponge with a contact angle of 142 ± 2°, while the water droplet has been absorbed completely upon touching the annealed sponge. (d) Compressive stress–strain curves of an original CNT and annealed CNT–Fe₂O₃ sponge (60 min, 450 °C in air) at a set strain of 30%, respectively.

An original bulk, freestanding sponge appeared black. After annealing at 450 °C for 60 minutes, there was no obvious change in the macroscopic morphology (some samples showed a slight red tinge on the surface). With longer annealing period or higher temperature, the sponges shrank considerably with red color particles covering on the surface (Fig. 1b). It showed that as more catalyst particles were exposed and aggregated, the damage on the CNT network became severe, causing structural collapse. The annealing process also changed the surface properties of CNT sponges from initially hydrophobic (contact angle of 142 ± 2°) to hydrophilic (the water droplet was absorbed completely when it touched the annealed sponge) (Fig. 1c). Hence, an original sponge was floating on the water surface while the annealed sponge was soaked into water quickly. In addition, the mechanical property also changed. The compressive strength decreased from about 22 to 14 kPa after annealing, and while the original sponge was compressed elastically up to a strain of 30%, the annealed sponge tended to develop plastic deformation after repeated compression (Fig. 1d).

We have characterized the microstructure evolution in the CNT sponges by scanning electron microscopy (SEM). The original sponge consists of randomly overlapped multi-walled nanotubes forming a 3D porous network (Fig. 2a). The surface of CNTs appears rather clean, free of foreign particles. In comparison, we observe many small particles uniformly dispersed through the CNT network in the annealed sponge (450 °C in air, 60 min), indicating that the Fe catalyst has been exposed (Fig. 2b). The exposed Fe particles are present both on the surface and inside the sponge (more particles on the surface) (ESI, Fig. S1†). On the other hand, annealing at 800 °C in Ar (with residual air) produced cubic or polyhedral shape particles terminated by sharp facets, indicating the growth of single-crystals due to higher temperature treatment (Fig. 2c–f). Many of these crystalline particles attach to the ends of CNTs, as they came from the Fe particles originally encapsulated within the CNT tips as characterized by transmission electron microscopy (TEM) (Fig. S2†). The crystal sizes range from 20 to 250 nm; this wide size distribution is caused by the disparity in the volume of individual Fe particles trapped inside the CNTs. The particle size is further increased upon oxidation and formation of crystalline oxides. There is occasionally a particle sitting between the broken regions in the middle of a CNT (Fig. 2e). We suppose that this particle may come from the tube cavities along the CNT axis. Most of the particles are exposed from the CNT tips, indicating the tips are more defective and prone to oxidation. Because of the limited amount of air during annealing in Ar, oxidation mainly occurs near the surface region of the sponge, and no crystalline particles observed inside.

Fig. 2 Structural characterization of CNT and CNT–Fe₂O₃ sponges. (a) SEM image of CNT sponges (inside) showing randomly overlapped nanotubes with clean surface. (b) SEM image of CNT–Fe₂O₃ sponges annealed at 450 °C (air, 60 min) with uniform dispersion of nanoparticles through the CNT network. (c) SEM image of CNT–Fe₂O₃ sponges annealed at 800 °C (Ar, 30 min) with single-crystals due to higher temperature treatment. (d) Enlarged view showing sharp facets of crystals. (e) Crystalline particles attached to the ends of CNTs. (f) A cubic crystal suspended between the broken region in the middle of a CNT.

We have characterized the crystal structure, composition and functional groups in the CNT sponges before and after annealing, by X-ray diffraction (XRD), Raman, X-ray photoelectron spectroscopy (XPS), FTIR and TGA. From the XRD pattern of the annealed sponge, there are 10 characteristic peaks that can be indexed to α-Fe₂O₃ (JCPDS no. 33-0664) (Fig. 3a). Raman spectra also show A_1g modes (peaks at 215, and 490 cm⁻¹) and E_g modes (peaks at 278, 388, and 591 cm⁻¹) of α-Fe₂O₃ in the annealed sample (Fig. 3b).^32,33 The XPS survey spectra reveal coexistence of C 1s peak (284.85 eV), O 1s peak (532.84 eV) and Fe 2p peaks in the annealed sponge, while the original sponge only shows C and O peaks (Fig. 3c). High-resolution spectrum of Fe 2p shows two peaks at 711.2 and 724.5 eV, corresponding to the Fe 2p_1/2 and Fe 2p_3/2 spin–orbit peaks of Fe₂O₃, respectively (Fig. 3d). Furthermore, a broad satellite peak around 719.3 eV matches well with the previous report of Fe³⁺.^34,35 After annealing, the bands appearing at 628, 1618 and 3415 cm⁻¹ in FTIR spectra are attributed to the Fe–O, C=O and –OH peaks, respectively (Fig. 3e). The increase of oxygen containing functional groups explains the enhanced hydrophilicity of annealed sponge (as seen in Fig. 1c). The Fe–O peak also confirms the presence of iron oxide. We have performed TGA on the CNT sponge following the annealing procedure, in which the sample was held at 450 °C for 60 minutes during the heating process in air. We can see a ca. 5% weight loss occurring at that constant temperature period, due to the combined effects between the combustion of CNTs (weight loss) and the oxidation of Fe particles (weight gain) (Fig. 3f). It confirms that some of the CNTs have been oxidized accompanied by the formation of Fe₂O₃ particles during annealing, as seen in SEM images. Vejpravova, et al. have shown that thermal annealing at different temperatures results in the formation of hematite particles on CNTs with magnetic response.³⁶ After complete combustion (at ∼570 °C) by rising temperature up to 800 °C, the remaining weight (∼15%) accounts for the total content of oxidized residual catalyst encapsulated in the as-grown CNT sponge. The above results indicate that we have fabricated CNT–Fe₂O₃ composite sponges composed of well-crystalline α-Fe₂O₃ nanoparticles grafted on the CNT network.

Fig. 3 Spectroscopy analysis and thermal stability of CNT and CNT–Fe₂O₃ sponges. XRD patterns (a), Raman spectra (b), XPS wide-scan spectra (c) of CNT and CNT–Fe₂O₃ sponges (air, 450 °C, 60–90 min). (d) High-resolution XPS spectra of Fe 2p peak in CNT–Fe₂O₃ sponges. (e) FTIR spectra of the CNT and CNT–Fe₂O₃ sponge. (f) TGA curve following the annealing procedure (heated in air, held at 450 °C for 60 min) and then rising temperature up to 800 °C. Inset, enlarged view of TGA curve around 450 °C showing the combined effects between the combustion of CNTs (weight loss) and the oxidation of Fe particles (weight gain).

Since the residual catalyst has been removed from CNT cavities in the annealed sponge, the next question is whether those exposed (and oxidized) particles still possess catalytic activity. To study this, we used the CNT–Fe₂O₃ composite sponge (annealed at 450 °C in air, sample shown in Fig. 2b) as a porous catalyst substrate to grow new CNTs on the exposed Fe₂O₃ particles. A CVD process was carried out by flowing H₂ at 800 °C for 20 minutes to reduce Fe₂O₃, and ethanol was fed as carbon source to grow CNTs. We find that the growth process includes the following 3 steps: (1) covering of graphitic layers around the reduced Fe particles (C–Fe vesicles), (2) formation of protruding antennas (CNT caps), and 3) elongation into new nanotubes (as illustrated in Fig. 4a). SEM image of the sample after a short growth period (e.g. 10 minutes) clearly reveals many vesicular structures consisting of an inner Fe particle (high contrast core) enclosed by an outer carbon layer (low contrast skin) (Fig. 4b). Compared with faceted oxide particles as shown in Fig. 2, the particle morphology change was caused by H₂ reduction and carbon decomposition onto the particle surface. Close view shows that several half-spherical caps emerge from one particle along different directions, which represent the early stage (nucleation) of CNT growth (Fig. 4c). We suppose that different terminal facets in the catalyst can act as the nucleation sites of CNTs, resulting in the formation of multiple antenna-like caps around one particle. Based on the vapor–liquid–solid mechanism, these caps acted as the growth seeds which were elongated continuously into long nanotubes. As a result, we observe multiple CNTs (short segments and long tubes) protruding from individual particles after longer CVD period; these are newly grown CNTs from reactivated residual catalyst, which still have a multi-walled structure (Fig. 4d and e, Fig. S3†). TEM images following different re-growth stages clearly show the structures including the C–Fe vesicles (Fe particles wrapped by carbon layer), vesicles with small CNT caps emerging from the particle surface, and long CNTs grown from the Fe particles (Fig. S3†). The morphology and structure of catalyst particles have been monitored by TEM and selected area electron diffraction (SAED) at different re-growth stages (Fig. 4f). Specifically, exposed particles in annealed sponges are indexed as Fe₂O₃ single-crystals by SAED with sharp facets under TEM, and after re-growth, they have been reduced to nearly spherical Fe₃C particles covered by graphitic layers (corresponding to the vesicular structure observed in Fig. 4b).The above results indicate that exposed (versus encapsulated) catalyst particles can be reactivated to grow CNTs. Also, since the Fe particles are located on a highly porous CNT network, carbon source can infiltrate into the sponge to grow CNTs inside, and the new CNTs growing within the porous space extend to arbitrary directions. This explains the phenomenon that multiple CNT segments have grown in different directions from a single particle. Previous work has demonstrated repeated growth of aligned CNT arrays from rigid substrates (e.g. silicon),^21–25 here our CNT sponge has served as a flexible and porous substrate for the re-growth. Controlling the size of exposed catalyst particles and optimization of CVD conditions may further improve the structure and morphology of the secondary CNTs, and for example, make higher density porous CNT structures.

Fig. 4 Application of CNT–Fe₂O₃ sponges as porous catalyst substrate. (a) Illustration of the process including H₂ reduction of Fe₂O₃ particles and regrowth of CNTs. (b) SEM image showing vesicular structures consisting of an inner Fe particle (high contrast core) enclosed by an outer carbon layer (low contrast shell) (see arrows). (c) SEM image showing several half-spherical caps emerging from one particle along different directions (see arrows), which represent the early stage (nucleation) of CNT growth. (d) SEM image showing short segments protruding from individual particles after longer CVD period. (e) SEM image showing several long nanotubes which are newly grown from reactivated residual catalyst. (f) TEM images of the exposed Fe₂O₃ particle after annealing and the C–Fe vesicle after re-growth and their corresponding SAED pattern.

Iron and iron oxide are considered as promising pseudocapacitive materials for electrochemical devices owing to their variable oxidation states, natural abundance, low cost, and high theoretical capacity.³⁷ Previously, Fe₂O₃ nanoparticles have been mixed with CNT powders or grafted on CNT sponges to serve as hybrid supercapacitor electrodes with improved performance.^34,35,38 Here, our CNT–Fe₂O₃ composite sponges have a number of advantages compared with previous structures. First, we do not introduce foreign pseudocapacitive-materials to combine with the CNT sponges; rather, residual Fe catalyst is directly exposed from the CNT cavities with a uniform distribution throughout the porous sponge. Second, many exposed Fe₂O₃ nanoparticles have strong connection to the CNT tips (as shown in Fig. 2), leading to efficient electron transport through the electrode and high structural stability (avoid detaching Fe₂O₃ particles into the electrolyte).

We have fabricated two CNT–Fe₂O₃ electrodes with different Fe₂O₃ contents by annealing the CNT sponges at 450 °C for 30 and 60 minutes, respectively, resulting in exposed nanoparticle average areal densities of about 8 (number of particles enclosed) and 41 particles per μm², as characterized by SEM. As the amount of active sites (exposed catalyst particles) is different, we have performed TGA in air (burn CNTs but leave metal particles) on those samples to estimate the loading of exposed particles. We find that the remaining weight percentage of Fe₂O₃ in annealed sponges (19.9 to 25.6 wt%) is higher than that of an as-grown sponge (16.5 wt%) (Fig. S4†), because there is already a portion of catalyst exposed in annealed sponges before TGA test. Therefore, the loading of exposed Fe particles could be related to this difference, which is about 3.4 wt% and 9.1 wt% for the sponges annealed for different periods (corresponding to 8 and 41 particles per μm², respectively). The results show that the specific capacitance is enhanced with increasing loading of exposed catalyst. With increasing particle areal density, the Brunauer–Emmett–Teller (BET) specific surface area has increased consistently from 60–100 m² g⁻¹ of original CNT sponges to ∼123 m² g⁻¹ (8 particles per μm²) and 183 m² g⁻¹ (41 particles per μm²), respectively (Fig. 5a). All three samples exhibit type IV characteristics with increased meso-pore volume in the range of 4 to 12 nm of annealed sponges according to the pore size distribution (Fig. S5†). Enhanced surface area is due to the formation of defects, fracture of CNTs into segments, exposure of inner tube cavities during the annealing process as shown in SEM characterization and the weight loss in TGA.

Fig. 5 Application of CNT and CNT–Fe₂O₃ sponges as supercapacitor electrode. Nitrogen adsorption and desorption isotherm (a), CV curves at a scan rate of 200 mV s⁻¹ (b), GCD curves at a current density of 1 A g⁻¹ (c), specific capacitance at different scan rates from 2 to 200 mV s⁻¹ (d), EIS spectra (e), cycling tests at a scan rate of 100 mV s⁻¹ for 1000 cycles (f) of CNT and CNT–Fe₂O₃ sponges with exposed nanoparticle average areal densities of about 8 (number of particles enclosed) and 41 particles per μm². Inset in (f) shows CV curves of the 1st, and 1000th cycles of CNT–Fe₂O₃ sponges (41 particles per μm²).

To evaluate the electrochemical behavior of the CNT–Fe₂O₃ sponge electrodes, we performed cyclic voltammetry (CV), galvanostatic charge discharge (GCD) measurements within a potential window between −0.8 and 0 V (vs. Hg/HgO) in 1 M KOH aqueous electrolyte, in a three-electrode testing system. The annealed CNT sponges maintain rectangular CV curves with larger enclosed area than the original sponge, indicating enhanced electrochemical property after annealing, especially at a higher nanoparticle density (41 particles per μm²) (Fig. 5b). The rectangular CV shape suggests that charge propagation at the electrode surface follows the electric double-layer supercapacitor (EDLC) mechanism. The annealed sponges have larger surface area and more defects, providing more available sites for charge adsorption in the bulk porous electrode. Improved hydrophilicity after annealing also facilitates the infiltration of aqueous electrolyte into the porous sponge. In addition, the presence of oxygen-containing functional groups and exposed Fe₂O₃ nanoparticles produces pseudo-capacitive effect as seen from the redox peaks at around −0.6 V. GCD measurements under a constant current density of 1 A g⁻¹ show nearly triangular curves, in which the annealed sponges possess longer discharge duration attributed to the combination of EDLC and faradaic capacitance (Fig. 5c). The specific capacitances have been calculated from the CV curves at scan rates of 2 to 200 mV s⁻¹ (Fig. 5d). The CNT–Fe₂O₃ sponge (41 particles per μm²) shows the highest specific capacitance of ∼198 F g⁻¹ at 2 mV s⁻¹ (with 56% retention at 200 mV s⁻¹), compared with the original CNT sponge (22 F g⁻¹) and annealed sponge with less oxide particles (42 F g⁻¹). Here, the number of exposed Fe₂O₃ nanoparticles and loading are limiting factors, and higher capacitance may be achieved by combining the exposure of residual catalyst with the deposition of other inorganic and organic pseudocapacitive materials.^39,40 The Nyquist plots of electrochemical impedance spectroscopy (EIS) contain a semicircle in the high frequency region and almost vertical capacitive lines in the low frequency region (Fig. 5e). All three electrodes have similar internal resistances (R_s ≈ 1.4 Ω, which is the intercept at the real axis), suggesting that the presence of Fe₂O₃ does not influence the network conductivity. The CNT–Fe₂O₃ sponge (41 particles per μm²) also shows the largest charge-transfer resistance (R_ct, semicircle diameter) than other samples, due to the relatively slow redox reaction of Fe₂O₃ and oxygen-containing groups. All three electrodes show excellent cycling stability, with 91.2%, 91.8% and 92.8% capacitance retention after 1000 cycles at 100 mV s⁻¹ in the CNT, CNT–Fe₂O₃ sponges containing 8 and 41 particles per μm², respectively (Fig. 5f). The above results demonstrate that exposing the residual catalyst in CNT sponges, accompanied by simultaneous increase of specific surface area and introduction of oxygen-containing groups, could make supercapacitor electrodes with enhanced performance.

Conclusion

We demonstrated direct exposure of residual catalyst and fabrication of CNT–Fe₂O₃ composite sponges, through a simple thermal annealing method. The resulting CNT–Fe₂O₃ sponges were used as a porous catalyst substrate to re-grow CNTs, and also acted as supercapacitor electrodes with improved electrochemical performance. In the future, exposure of residual catalyst in other CNT-based materials, for example, aligned CNT arrays, may produce a layer of metal nanoparticles supported by the CNT tips. Our CNT sponges containing exposed catalyst particles have potential applications in energy and environmental areas such as Li-ion batteries, catalysis and adsorption of heavy elements.

Experimental methods

1. Fabrication of CNT–Fe₂O₃ sponges

CNT sponges were synthesized by CVD, as reported by our team previously, by using ferrocene and dichlorobenzene as catalyst and carbon precursors, respectively.²⁵ CNT–Fe₂O₃ sponges were prepared via a simple annealing process, which can be easily scaled up. We adopted two conditions to expose residual metal catalysts and transform them into Fe₂O₃ nanoparticles. First, an as-synthesized CNT sponge block with edge size of several mm was annealed in an ambient atmosphere at 450 °C for different durations. By controlling the oxidation time, we can obtain CNT–Fe₂O₃ sponges with different content of Fe₂O₃ nanoparticles. Second, raw CNT sponges were annealed at 700–800 °C in Ar atmosphere for 60 min. Oxidation also occurs because of remnant oxygen in the quartz tube.

2. Material characterization

The sample morphology and structure were characterized using field emission scanning electron microscopy (FESEM) (Hitachi S4800), high resolution transmission electron microscopy (HRTEM) (Tecnai F20) and X-ray diffraction (XRD) (PANalytical X-Pert3 Powder). Contact angle measurement was tested on instrument (Dataphysics OCA 20). Mechanical tests were carried out in a single-column static instrument (Instron 5843) equipped with two flat-surface compression stages and a load cell of 10 N. Raman spectra were recorded on Renishaw with a 514 nm laser. Fourier transform infrared (FTIR) spectra were tested on a FTIR system (Perkin Elmer Spectrum II). Thermal gravimetric analysis (TGA) was performed on a TA Instruments SDT Q600 analyzer and the heat procedure follows the annealing process of samples, heating from room temperature to 450 °C in air at a heating rate of 10 °C min⁻¹ and stay 60 min. Then, we continue to heat up from 450 °C to 800 °C. X-ray photoelectron spectroscopy (XPS) measurements were performed using a Imaging Photoelectron Spectrometer (Kratos Analytical Axis Ultra). Nitrogen sorption isotherms were measured at 77 K on a Quantachrome Autosorb-IQ gas adsorption analyzer. Before the sorption test, the samples were degassed at 200 °C for 5 hours under dynamic vacuum. The Brunauer–Emmett–Teller (BET) specific surface area of sponges was measured. The pore volume and pore size distribution were obtained using the density functional theory (DFT) method.

3. Regrowth of CNTs

The CNT–Fe₂O₃ sponge was put into a horizontal quartz tube furnace, followed by ramping up the temperature to 800 °C in 7 min and purging with Ar and H₂ at a flow rate of 400 sccm and 200 sccm, successively. Once the temperature was reached, it was held for 20 min to reduce the Fe₂O₃ nanoparticles in H₂ (200 sccm). Subsequently, CVD growth of CNTs proceeded at 800 °C with ethanol served as the carbon source for 10–60 min. After the growth, the sample was cooled down to room temperature under Ar atmosphere.

4. Electrochemical measurements

The electrochemical properties of samples including cyclic voltammetry (CV), galvanostatic charge/discharge (GCD), and electrochemical impedance spectroscopy (EIS) were carried out on a CHI660D electrochemical workstation in a three-electrode system. The sponge clamped by two polymeric blocks was used as working electrode. Platinum wire twisted around the polymeric clamp and connected to the sponge electrode was used as the current collector. A platinum wire and a Hg/HgO electrode were used as counter and reference electrodes, respectively. All electrochemical measurements were carried out in 1 M KOH aqueous solution as electrolyte at room temperature. The CV curves were measured under different scan rates of 2–200 mV s⁻¹ from −0.8 to 0 V. The EIS measurements were recorded under AC voltage amplitude of 5 mV, frequency range of 100 kHz to 0.01 Hz at open circuit potential. The specific capacitance of the sponge electrode (C_S) was calculated from the CV curves and discharging curves according to

C_S = (IΔt)/(mΔU)

where I is the response current (A), m is the total mass the sponge electrode (g), ΔV is the potential range in the CV (V), ν is the potential scan rate (mV s⁻¹). ΔU is potential window in the discharging process and Δt is the discharging time.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (NO.51325202).

References

1. S. Iijima, Nature, 1991, 354, 56.
2. P. M. Ajayan, Chem. Rev., 1999, 99, 1787.
3. S. Fan, M. G. Chapline, N. R. Franklin, T. W. Tombler, A. M. Cassell and H. Dai, Science, 1999, 283, 512.
4. Y. Hayamizu, T. Yamada, K. Mizuno, R. C. Davis, D. N. Futaba, M. Yumura and K. Hata, Nat. Nanotechnol., 2008, 3, 289.
5. H. Li, X. Gui, L. Zhang, S. Wang, C. Ji, J. Wei, K. Wang, H. Zhu, D. Wu and A. Cao, Chem. Commun., 2010, 46, 7966.
6. D. N. Futaba, K. Hata, T. Yamada, T. Hiraoka, Y. Hayamizu, Y. Kakudate, O. Tanaike, H. Hatori, M. Yumura and S. Iijima, Nat. Mater., 2006, 5, 987.
7. D. S. Bethune, C. H. Klang, M. S. de Vries, G. Gorman, R. Savoy, J. Vazquez and R. Beyers, Nature, 1993, 363, 605.
8. P. J. Harris, Carbon, 2007, 45, 229.
9. C. Bower, O. Zhou, W. Zhu, D. J. Werder and S. Jin, Appl. Phys. Lett., 2000, 77, 2767.
10. M. Kumar and Y. Ando, J. Nanosci. Nanotechnol., 2010, 10, 3739.
11. W. Ma, L. Song, R. Yang, T. Zhang, Y. Zhao, L. Sun, Y. Ren, D. Liu, J. Shen, Z. Zhang, Y. Xiang, W. Zhou and S. S. Xie, Nano Lett., 2007, 7, 2307.
12. K. Hata, D. N. Futaba, K. Mizuno, T. Namai, M. Yumura and S. Iijima, Science, 2004, 306, 1362.
13. A. C. Dillon, T. Gennett, K. M. Jones, J. L. Alleman, P. A. Parilla and M. J. Heben, Adv. Mater., 1999, 11, 1354.
14. Y. S. Park, Y. C. Choi, K. S. Kim, D. C. Chung, D. J. Bae, K. H. An, S. C. Lim, X. Y. Zhu and Y. H. Lee, Carbon, 2001, 39, 655.
15. P. X. Hou, S. Bai, Q. H. Yang, C. Liu and H. M. Cheng, Carbon, 2002, 40, 81.
16. W. Huang, Y. Wang, G. Luo and F. Wei, Carbon, 2003, 41, 2585.
17. Y. Q. Xu, H. Peng, R. H. Hauge and R. E. Smalley, Nano Lett., 2005, 5, 163.
18. G. Tobias, L. Shao, C. G. Salzmann, Y. Huh and M. L. H. Green, J. Phys. Chem. B, 2006, 110, 22318.
19. G. Mercier, C. Herold, J. Mareche, S. Cahen, J. Gleize, J. Ghanbaja, G. Lamura, C. Bellouard and B. Vigolo, New J. Chem., 2013, 37, 790.
20. M. Pumera, Langmuir, 2007, 23, 6453.
21. X. Li, A. Cao, Y. J. Jung, R. Vajtai and P. M. Ajayan, Nano Lett., 2005, 5, 1997.
22. T. Iwasaki, J. Robertson and H. Kawarada, Nano Lett., 2008, 8, 886.
23. D. Jung, M. Han and G. S. Lee, Appl. Phys. Express, 2014, 7, 025102.
24. C. L. Pint, N. Nicholas, J. G. Duque, A. N. G. Parra-Vasquez, M. Pasquali and R. Hauge, Chem. Mater., 2009, 21, 1550.
25. C. Chiu, M. Yoshimura, K. Ueda, Y. Kamizono, H. Shinohara, Y. Ohira and T. Tanji, J. Nanomater., 2010, 2010, 7.
26. C. E. Banks, A. Crossley, C. Salter, S. J. Wilkins and R. G. Compton, Angew. Chem., Int. Ed., 2006, 45, 2533.
27. M. Pumera and Y. Miyahara, Nanoscale, 2009, 1, 260.
28. M. Pumera and H. Iwai, Chem.–Asian J., 2009, 4, 554.
29. A. Ambrosi and M. Pumera, Chem.–Eur. J., 2012, 18, 3338.
30. R. K. Emmett, M. Karakaya, R. Podila, M. R. Arcila-Velez, J. Zhu, A. M. Rao and M. E. Roberts, J. Phys. Chem. C, 2014, 118, 26498.
31. X. Gui, J. Wei, K. Wang, A. Cao, H. Zhu, Y. Jia, Q. Shu and D. Wu, Adv. Mater., 2010, 22, 617.
32. H. Wang, Z. Xu, H. Yi, H. Wei, Z. Guo and X. Wang, Nano Energy, 2014, 7, 86.
33. L. F. Chen, Z. Y. Yu, X. Ma, Z. Y. Li and S. H. Yu, Nano Energy, 2014, 9, 345.
34. C. Guan, J. Liu, Y. Wang, L. Mao, Z. Fan, Z. Shen, H. Zhang and J. Wang, ACS Nano, 2015, 9, 5198.
35. X. Cheng, X. Gui, Z. Lin, Y. Zheng, M. Liu, R. Zhan, Y. Zhu and Z. Tang, J. Mater. Chem. A, 2015, 3, 20927.
36. B. P. Bittova, M. Kalbac, S. Kubickova, A. Mantlikova, S. Mangold and J. Vejpravova, Phys. Chem. Chem. Phys., 2013, 15, 5992.
37. G. Binitha, M. S. Soumya, A. A. Madhavan, P. Praveen, A. Balakrishnan, K. R. V. Subramanian, M. V. Reddy, S. V. Nair, A. S. Nair and N. Sivakumar, J. Mater. Chem. A, 2013, 1, 11698.
38. X. Zhao, C. Johnston and P. S. Grant, J. Mater. Chem., 2009, 19, 8755.
39. P. Li, E. Shi, Y. Yang, Y. Shang, Q. Peng, S. Wu, J. Wei, K. Wang, H. Zhu, Q. Yuan, A. Cao and D. Wu, Nano Res., 2014, 7, 209.
40. P. Li, Y. Yang, E. Shi, Q. Shen, Y. Shang, S. Wu, J. Wei, K. Wang, H. Zhu, Q. Yuan, A. Cao and D. Wu, ACS Appl. Mater. Interfaces, 2014, 6, 5228.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra05775f

---