Nanostructured carbon–metal oxide composite electrodes for supercapacitors: a review

Abstract

This paper presents a review of the research progress in the carbon–metal oxide composites for supercapacitor electrodes. In the past decade, various carbon–metal oxide composite electrodes have been developed by integrating metal oxides into different carbon nanostructures including zero-dimensional carbon nanoparticles, one-dimensional nanostructures (carbon nanotubes and carbon nanofibers), two-dimensional nanosheets (graphene and reduced graphene oxides) as well as three-dimensional porous carbon nano-architectures. This paper has described the constituent, the structure and the properties of the carbon–metal oxide composites. An emphasis is placed on the synergistic effects of the composite on the performance of supercapacitors in terms of specific capacitance, energy density, power density, rate capability and cyclic stability. This paper has also discussed the physico-chemical processes such as charge transport, ion diffusion and redox reactions involved in supercapacitors.

---

Mingjia Zhi

Mingjia Zhi received his B.S. and M.S. degrees from Department of Materials Science and Engineering at Zhejiang University in 2004 and 2006, respectively. He received his Ph.D. degree in Mechanical Engineering at West Virginia University in USA. He currently works as a Postdoctoral Fellow, co-supervised by Dr Nianqiang (Nick) Wu at West Virginia University and Dr A. Manivannan at National Energy Technology Laboratory, DOE. His research interests include nanomaterials, electrochemical energy conversion and storage devices.

Chengcheng Xiang

Chengcheng Xiang received her MS degrees in Material Physics at Central South University in China in 2007 and in Industrial Engineering at West Virginia University in 2011. Currently she is a Ph.D. student in Industrial and Management System Engineering at West Virginia University in USA. Her research interest is focused on experimental design and statistical analysis, nanomaterials and their application in supercapacitors.

Jiangtian Li

Jiangtian Li received his Ph.D. degree in Materials Physics and Chemistry at Shanghai Institute of Ceramics, Chinese Academy of Science in 2008. He then worked as an Assistant Professor with Prof. Fuqiang Huang. Afterwards, he worked as a Postdoctoral Fellow at University of Cologne, Germany. In 2010, he joined Prof. Nianqiang (Nick) Wu's group at West Virginia University. His current research interest is focused on nanostructured materials, solar energy conversion and storage.

Ming Li

Ming Li received his B.Sc. degree in Fine Chemicals from Dalian University of Technology in China in 2005 and his M.S. degree from Department of Materials Science and Engineering at Zhejiang University in China. He then joined Prof. Nianqiang (Nick) Wu's group as a Ph.D. student. He received his Ph.D. degree from Department of Mechanical and Aerospace Engineering at West Virginia University in USA in 2012. His research interest lies in functional nanomaterials, biosensors as well as energy conversion and storage devices.

Nianqiang Wu

Nianqiang Wu received his Ph.D. degree from Zhejiang University, China. After two years of postdoctoral training in University of Pittsburgh, he joined Keck Interdisciplinary Surface Science Center at Northwestern University. In 2005, he took a faculty position in West Virginia University. He is currently Associate Professor in Department of Mechanical and Aerospace Engineering at West Virginia University in USA. His research areas include (i) nanomaterials and nanolithography, (ii) electrochemical devices for energy conversion and storage, (iii) chemical sensors and biosensors, and (iv) photocatalysts and photoelectrochemical cells.

---

1 Introduction

Supercapacitors, also called ultracapacitors, are finding increasing applications in hybrid electric vehicles, large industrial equipment, memory back-up devices and renewable energy power plants.^1–5 Supercapacitors provide high power density, fast charge–discharge, and long service life as compared to batteries.^6–10 The specific capacitance (C) of a supercapacitor is defined aswhere Q is the charge stored on the electrode per unit mass and V is the operating voltage window.

Supercapacitors store the energy via two operating mechanisms:

(i) Electrochemical double-layer capacitance (EDLC),^11,12 which results from the electrical double-layer surrounding the surface of electrode. That is, the depletion of the oppositely charged species stores the energy at the interface of the electrode and the electrolyte, respectively. The accumulation of electrons at the electrode is a non-Faradaic process. The specific capacitance of an EDLC is measured as

where ε_r is the relative permittivity of the medium in the electrical double-layer, ε₀ is the permittivity of vacuum, A is the specific surface area of the electrode, and d is the effective thickness of the electrical double-layer.

(ii) Pseudo-capacitance, which is originated from the redox reaction of the electrode material with the electrolyte.^13–15 The accumulation of electrons at the electrode is a Faradaic process where the electrons produced by the redox reaction are transferred across the electrolyte–electrode interface. The theoretical pseudo-capacitance of metal oxide can be calculated as

where n is the mean number of the electrons transferred in the redox reaction, F is the Faraday constant, M is the molar mass of the metal oxide and V is the operating voltage window.

The energy density (E) of a supercapacitor is expressed as

The maximum power density of a supercapacitor is determined by

where R is the equivalent series resistance of all the components in the device. Both EDLC and pseudo-capacitance can be simultaneously generated in a single supercapacitor to form a hybrid supercapacitor. By utilizing both Faradaic and non-Faradaic processes to store charges, hybrid supercapacitors can achieve higher energy and power densities while keeping good cycling stability. For example, the redox reaction takes place on one of the electrodes and the non-Faradaic charge–discharge process occurs on another electrode in a single supercapacitor. Such a supercapacitor is called an asymmetric hybrid supercapacitor. Other two types of hybrid supercapacitors are battery-like hybrids and composite hybrids. In practice, various types of supercapacitors have been developed to combine different electrode materials in order to improve the device performance. Table 1 gives a summary of different types of supercapacitors.

Table 1 Summary of the different types of supercapacitor devices

Type of supercapacitor		Electrode material	Charge storage mechanism	Merits/shortcomings
Electrochemical double layer capacitor (EDLC)		Carbon	Electrochemical double layer (EDL), non-Faradaic process	Good cycling stability, good rate capability, low specific capacitance, low energy density
Pseudocapacitor		Redox metal oxide or redox polymer	Redox reaction, Faradaic process	High specific capacitance, relatively high energy density, relatively high power density, relatively low rate capability
Hybrid capacitor	Asymmetric hybrid	Anode: pseudocapacitance materials, cathode: carbon	Anode: redox reaction, cathode: EDL	High energy density, high power density and good cyclability
	Symmetric composite hybrid	Redox metal oxide/carbon or redox polymer/carbon	Redox reaction plus EDL	High energy density, moderate cost and moderate stability
	Battery-like hybrid	Anode: Li-insertion material, cathode: carbon	Anode: Lithiation/delithiation cathode: EDL	High energy density, high cost and requires electrode material capacity match

---

The key performance parameters of supercapacitors include specific capacitance (normalized by electrode mass, volume, or area), energy density, power density, rate capability (retained capacitance at a high current loading) and cycling stability.¹⁶ To increase the energy density and power density of a supercapacitor, it is desirable to increase the specific capacitance (C) and the operating voltage window (V) as well as reduce the equivalent series resistance (R). For EDLC supercapacitors, the maximum operating voltage window (V_m) is mainly dependent on the electrolyte used, which is limited by the stability of the electrolyte. Normally V_m is 1 V for the supercapacitors based on the aqueous electrolyte. One of the current research trends in supercapacitors is to develop non-aqueous electrolytes with high V_m. For example, an ionic liquid-based electrolyte is able to be operated at V_m as high as 3.5 V.¹⁷ For a review on the electrolytes refer to the previous paper.¹⁸ This paper is focused on the electrode materials. Ideal electrode materials are required to possess:

• high specific surface area, which governs the specific capacitance,

• controlled porosity, which affects the specific capacitance and the rate capability,

• high electronic conductivity, which is crucial to the rate capability and the power density,

• desirable electroactive sites, which enable pseudo-capacitance,

• high thermal stability and chemical stability, which affect the cyclic stability, and

• low costs of raw materials and manufacturing.

Currently the electrodes of most commercial supercapacitors are made of carbon that is inexpensive and has good resistance to corrosion. Carbon-based supercapacitors are operated under the EDLC. The carbon-based EDLC supercapacitors have excellent cyclic stability and long service lifetime since the electrode undergoes no chemical change during the charge/discharge processes. However, their maximum capacitance is restricted by the active electrode surface area and the pore size distribution (typically 0.15–0.4 F m⁻², or ∼150 F g⁻¹ for carbon).^19,20 Overall, the energy density of currently commercial carbon-based EDLC supercapacitors is typically 3–5 W h kg⁻¹, which is much lower than that of an electrochemical battery (30–40 W h kg⁻¹ for a lead acid battery and 10–250 W h kg⁻¹ for a lithium-ion battery).²¹ Such low energy-density cannot fulfill the need of energy storage devices for vehicles, wind-farms and solar power plants.

To improve the specific capacitance and the energy density, transition metal oxides are being investigated as the alternative materials for supercapacitor electrodes. The investigated metal oxides include RuO₂, MnO₂, NiO, Co₃O₄, SnO₂, ZnO, TiO₂, V₂O₅, CuO, Fe₂O₃, WO₃, etc.^22–33 In particular, some metal oxides show excellent pseudo-capacitance. Unlike a lithium battery in which the ions are deeply intercalated in the materials lattice, the pseudo-capacitance is originated from the weakly attached surface ions. The surface functional groups, the defects and the grain boundaries can serve as the excellent redox centers for the charge storage reactions. Table 2 lists the theoretical capacitance of typical metal oxides as well as the charge storage reactions. It can be seen that the transition metal oxide electrodes have one order higher specific capacitance than the carbon electrodes. Several reports have shown that bare metal oxide electrodes can deliver large specific capacitance and high energy density at a slow scan rate or at a low current density. For example, the electrodeposited NiO thin film electrode showed a specific capacitance of 1776 F g⁻¹ in the 1 M KOH aqueous electrolyte at a scan rate of 1 mV s⁻¹. However, only 23% of capacitance was retained when the scan rate increased from 1 mV s⁻¹ to 100 mV s⁻¹.⁴⁴ Therefore metal oxides may not be employed alone as the supercapacitor electrodes for practical purpose due to the following drawbacks.

Table 2 Pseudo-capacitance and conductivity of selected metal oxides

Oxide	Electrolyte	Charge storage reaction	Theoretical capacitance (F g^−1)	Conductivity (S cm^−1)
MnO2	Na2SO4	MnO2 + M^+ + e^− = MMnO2 (M could be H^+, Li^+, Na^+, K^+)	1380 (0.9 V)^34	10^−5 to 10^−6(ref. 35)
NiO	KOH, NaOH	NiO + OH^− = NiOOH + e^−	2584 (0.5 V)^36	0.01–0.32 (ref. 37)
Co3O4	KOH, NaOH	Co3O4 + OH^− + H2O = 3CoOOH + e^− CoOOH + OH^− = CoO2 + H2O + e^−	3560 (0.45 V)^38,39	10^−4 to 10^−2(ref. 40)
V2O5	NaCl, Na2SO4	V2O5 + 4M^+ + 4e^− = M2V2O5 (M could be H^+, Li^+, Na^+, K^+)	2120 (1 V)	10^−4 to 10^−2(ref. 41)
RuO2, xH2O	H2SO4, Na2SO4	RuO2 + xH^+ + xe^− = RuO2−x, (OH)x (0 < x < 2)	1200–2200 (1.23 V)^42	10^3 for polycrystalline, ∼1 for amorphous^43

---

• The conductivity of most metal oxides except for RuO₂ is very low. The high resistivity of metal oxides increases both the sheet resistance and the charge transfer resistance of the electrode and especially causes a large IR loss at a high current density. Thus the power density and the rate capability are poor, which hinders the practical application.

• The strain developed in the pure metal oxide during the charge–discharge processes causes the cracking of the electrode, leading to poor long-term stability.

• The surface area, the pore distribution as well as the porosity are difficult to tailor in metal oxides.

It is logical to develop a composite electrode containing both carbon and transition metal oxide as the supercapacitor electrode, which combines the merits and mitigate the shortcomings of both the components. In such carbon nanostructure–metal oxide composite electrodes, the carbon nanostructures not only serve as the physical support of metal oxides but also provide the channels for charge transport. The high electronic conductivity of carbon nanostructures benefits to the rate capability and power density at a large charge/discharge current. The metal oxides are the main sources that store the charge and the energy. The electro-activities of metal oxides contribute to high specific capacitance and high energy density of the carbon nanostructure–metal oxide composite electrodes. A synergistic effect could be expected and the materials cost can be reduced. The compositional constituent, microstructure and physical properties of metal oxide–carbon composites govern the performance of the supercapacitor electrodes. The electrode porosity, electronic conductivity, pore size distribution, specific surface area affect the cell performance. Table 3 gives a comparison of carbon, metal oxide and metal oxide–carbon composite electrodes.

Table 3 Comparison of carbon, metal oxide and metal oxide–carbon composite as supercapacitor electrode

Electrode materials	Surface area	Pore size distribution	Specific capacitance	Conductivity	Rate capability	Stability	Cost
Carbon	High	Various ways to tailor	Low	High	High	Good	Low
Metal oxide	Normally low	Difficult to tailor	High	Low	Low	Poor	High
Metal oxide–carbon composite	Normally controlled by a carbon support	Various ways to tailor depending on the carbon support	High	Tunable, depending on the carbon support	Good	Good	Moderate

---

The aim of research on the carbon–metal oxide composite electrodes is to develop electrodes that simultaneously possess high power density, high energy density as well as good rate capability and cyclic stability. In the past decade, carbon materials were engineered at different dimensions as the building blocks to couple with metal oxides to form composites for supercapacitor electrodes. Therefore, this review article will be organized according to the roles of carbon nanostructures with different dimensions^45,46 in the composite electrodes.

1. Zero-dimensional (0-D) carbon nanoparticles: 0-D carbon particles refer to round-shaped particles with an aspect ratio close to 1. 0-D carbon particles mainly include ultrafine activated carbon (AC), carbon nanospheres, and mesoporous carbon. They have high specific area (∼3000 m² g⁻¹ for AC^47,48). More importantly, the pore content and size distribution can be tailored, which makes them suitable for supporting metal oxides for the supercapacitor electrodes.

2. One-dimensional (1-D) carbon nanostructures: 1-D nanostructures are defined as the fiber-shaped materials with a high aspect ratio. Carbon nanotubes (CNT), carbon nanofibers (CNF) and carbon nanocoils fall into the category of 1-D carbon nanostructures. They have high aspect ratio and good electronic transport properties, which is expected to facilitate the kinetics of the electrochemical reactions.

3. Two-dimensional (2-D) nanosheets: 2-D nanostructures are defined as the sheet-shaped materials with a high aspect ratio. Graphene, graphene oxide (GO) or reduced graphene oxide (rGO) are the representatives of the 2-D carbon nanosheets. Graphene and GO exhibit great mechanical strength, excellent electronic conductivity as well as high specific surface area, which are promising candidates for supercapacitor electrodes. For example, the theoretical surface area of single layer graphene is 2756 m² g⁻¹ with a charge mobility of 200 000 cm² V⁻¹ s⁻¹.⁴⁹

4. Three-dimensional (3-D) porous architectures: 3-D architectures are made of low dimensional building blocks. 3-D porous carbon nano-architectures used in the supercapacitor electrodes are mainly carbon nanofoams or sponges. When used in supercapacitor electrodes, such a foam has high specific surface area, large area of electrolyte–electrode interface and continuous electron transport path.

2 Zero-dimensional carbon-supported metal oxide electrodes

Activated carbon (AC): AC is the most widely used carbon support material in supercapacitors. AC features moderate electronic conductivity and high specific surface area, which is suitable for enlarging the effective surface area of the metal oxide while providing electronic conductance.⁵⁰ An example is the NiO–AC composite electrode.⁵¹ The surface areas for NiO and AC alone were 150 m² g⁻¹ and 776 m² g⁻¹, respectively. After a NiO–AC composite was formed, the specific surface area was maintained from 575 m² g⁻¹ to 700 m² g⁻¹ and the NiO agglomeration was prevented. As a result, a specific capacitance of ∼100 F g⁻¹ was obtained for the NiO component at 7 wt% of NiO loading in the 6 M KOH electrolyte, which was measured by cyclic voltammetry (CV) at a scan rate of 5 mV s⁻¹, which was 5 times higher than that of pristine NiO (∼20 F g⁻¹). The reason was that the dispersed NiO had larger active surface area as compared with the sintered pristine NiO. Another example is the carbon–RuO₂ composite obtained by dispersing 10 nm sized Ru metallic nanoparticles on the carbon aero-gel.⁵² Such a carbon–RuO₂ composite had a specific surface area of 520 m² g⁻¹ and showed a specific capacitance of 256 F g⁻¹ at a current density of 1 A g⁻¹ in 30 wt% H₂SO₄ electrolyte. The high capacitance was estimated based on the total electrode mass, which corresponded to 50% utilization of the theoretical capacitance of RuO₂. It was also reported that a hybrid supercapacitor electrode was constructed by mixing MnO₂ nanopowder with active carbon black. This resulted in a specific capacitance of 60 F g⁻¹ in 1 M LiOH electrolyte at a current density of 0.1 A g⁻¹ and a working window of 1.5 V.⁵³

Mesoporous carbon: Besides high specific surface area, 0-D carbon can bring another advantage to the composite electrode which is its tunable pore size and fraction. The pore size of carbon is one of the critical factors that govern the specific capacitance and the rate capability. Desirable carbon electrodes have a hierarchical pore structure that contains macropores as the ion-buffering reservoirs, mesopores (2–50 nm) for promoting the ion transport, and nanopores (<2 nm) for enhancing the charge storage.⁵⁴

The accessibility of redox molecules and ions into the nanopores (<2 nm) is dependent on both the solvated molecular size and the pore diameter.⁵⁵ When the pores are smaller than the solvated ions, the ions will have difficulty in accessing to the pore inwall.⁵⁶ Only when the pore diameter is comparable to the ion size (that is, typically 0.5–1 nm for the aqueous electrolyte), the nanopores will make maximum contribution to the EDLC since the strong interaction of ions with the pore inwall will occur, leading to high electro-adsorption. Partial desolvation of hydrated ions has been found to occur in the nanopores (<1 nm), which contributes to overall capacitance.⁵⁷ However, the nanopores have negligible contribution to the ion transport in the electrolyte. In contrast, the mesopores in a diameter of 2–50 nm favor the rapid mass transport of ions to the electrode surface, which facilitates the charging/discharging of the electric double-layer.⁵⁸ Especially, the ion transport time constant can be expressed as τ = L²/D, where L is the ion transport length and D is the ion diffusion constant. Thus suitable mesopores with large pore size and short ion diffusion length can reduce the ion transport barrier. Another effect that needs to be taken into account is the regularity of the pores. A narrow pore size distribution can reduce the ion scattering, which improves the electrode kinetics.⁵⁴ It has been found that the differential capacitance in an aqueous solution increases linearly with the surface area of the mesopores.⁵⁶ The existence of interconnected mesopores will allow the efficient propagation of solvated ions to the nanopores throughout the electrode, which is the key to high rate capability. In short, the optimal performance (high specific capacitance and good rate capability) is expected for the porous carbon electrodes with a mixture of a large fraction of nanopores with the narrow size distribution and a suitable amount of mesopores.⁵⁴

An ideal pore structure for a supercapacitor should fulfill the following criteria: large pore volume for storage of energy, suitable ion transport channel for fast delivery of ions into the electrode and active pore walls for increasing the specific capacitance. Similar laws are also applicable to the metal oxide–carbon composite electrodes since the pseudo-capacitance of metal oxide is dependent on the fast surface redox reaction. Hence the access and transport of the ions in the composite electrode play a critical role. The carbon support offers a tunable platform for tailoring the fine pore microstructure and the specific surface area in the metal oxide–carbon electrodes. It is easier to adjust the property of carbon than the metal oxide. The unique features of the carbon support are desirable to be preserved in the composite.⁵⁴ For example, mesocarbon microbeads (MCMBs) were activated in the KOH solution at high temperature to yield an active carbon with a high mesopore content of 64.5% and a specific surface area of ∼3000 m² g⁻¹. The physical mixture of the activated MCMBs and MnO₂ nanosphere was determined to be suitable for the organic electrolyte (LiPF₆) due to the optimized pore structure in the activated MCMBs in contrast to the commercial carbon–MnO₂ composite electrode. The activated MCMB–MnO₂ composite electrode was operated at a window of 3 V, and exhibited an energy density of 128 W h kg⁻¹.⁵⁹

Moreover, carbon can be tuned to have ordered mesoporous structure, which is also an attractive feature for supercapacitors. Since microporous (>10 μm) ACs have a long diffusion distance (>5 μm) and a high ion-transport resistance, a large IR drop and limited ion-accessible surface area are consequently expected at a high current density. In contrast, ordered mesoporous carbons exhibit a particle size of 1–2 μm, which results in a short diffusion distance of 0.5–1 μm. Ordered mesoporous carbons also have mesoporous channels (4–6 nm), which lead to a low ion-transport resistance.⁶⁰ A recent review paper has summarized the progress of mesoporous/metal oxide as a supercapacitor electrode.¹⁴ For instance, MnO₂ was deposited on the wall of ordered mesoporous carbon (denoted as CMK-3).⁶¹ The microstructure of the materials can be found in Fig. 1. It has been found that the CMK-3–MnO₂ composite retained its high specific surface area (700–900 m² g⁻¹), which was comparable to bare CMK-3 carbon (968 m² g⁻¹). The pore size was also kept constant (3.6 nm) after different MnO₂ loadings. The specific capacitance of such a nanostructured electrode reached 605 F g⁻¹ for MnO₂ at a scan rate of 5 mV s⁻¹ in the aqueous Na₂SO₄ electrolyte. Mesoporous carbon was also used as the template for preparing nanostructured MnO₂. As compared with MnO₂ prepared with a CNT template, the mesoporous carbon-derived MnO₂ not only showed higher specific surface area but also narrower pore size distribution (3–7 nm vs. 20–80 nm). Such a unique microstructure enhanced the capacitive behavior of MnO₂. It exhibited a specific capacitance of 221 F g⁻¹ at 5 mA cm⁻² and a good rate capability (87.4% capacitance was retained when the current density increased from 2.5 to 10 mA cm⁻²). In contrast, 98 F g⁻¹ of specific capacitance and 71.7% of capacitance remained at the same condition for the CNT-derived MnO₂.⁶² Recently peapod-like Ni nanoparticles have been deposited on the mesoporous carbon with a specific surface area of 178 m² g⁻¹ and an average pore size of 3.6 nm. Such an electrode showed a specific capacitance of 972 F g⁻¹ for Ni at a current density of 0.5 A g⁻¹, which was higher than that of the NiO nanospheres and the NiO nanotubes.⁶³

Fig. 1 TEM image of 26 wt% MnO₂ loaded mesoporous carbon, inset is the high resolution TEM image, and the arrows indicate the embedded MnO₂ (reprinted with permission from ref. 61. Copyright (2006) American Chemical Society).

Carbon nanospheres: Another type of 0-D carbon nanomaterial is carbon nanospheres. Carbon nanospheres are prepared using the SiO₂ sphere as the template and ferrocene as the carbon source.⁶⁴ The as-grown hollow carbon spheres are highly graphitic, which ensures good electronic conductivity. MnO₂ nanofibers were then grown by reduction of KMnO₄ solution and a mild reflux method. Such carbon sphere–MnO₂ nanofiber electrode showed a specific capacitance of 190 F g⁻¹ at a current density of 0.1 A g⁻¹ with 67 wt% MnO₂. The electrode demonstrated good rate capability (55% capacitance retained at 10 A g⁻¹). In addition, an asymmetric hybrid supercapacitor was fabricated using the carbon spheres and carbon sphere–MnO₂ as the electrode and operated at a voltage window of 0–2 V, which increased the energy density to 22.1 W h kg⁻¹. The MnO₂ nanoribbons were grown on the onion-like carbon nanostructures to form the MnO₂–C nanospheres composite, such a structure ensured the active materials to accept electrons and ions in all the directions, which yielded good rate capability (177.5 F g⁻¹ at 2 A g⁻¹, 56.3% of capacitance was retained when the current increased from 0.2 to 2 A g⁻¹).⁶⁵ Carbon-coated Li₄Ti₅O₁₂ nanoparticles were used as the anode for a metal oxide–C/AC hybrid supercapacitor. The specific capacity of Li₄Ti₅O₁₂ was as high as 271 F g⁻¹ at a current loading of 980 mA g⁻¹. The full cell can deliver a power density of 440 W kg⁻¹ at an energy density of 16 W h kg⁻¹.⁶⁶

3 One-dimensional carbon–metal oxide electrodes

A collection of round-shaped 0-D nanoparticles are widely used for electrodes and current collectors, in which electrons are transported either via hopping between trap states on the neighboring nanoparticles or via diffusion within the extended states slowed down by the trapping–detrapping processes.⁶⁷ In contrast, in the 1-D nanostructure-based electrodes, the long 1-D nanostructures form a continuous network for charge transport. Owing to the reduced contact resistance between neighboring nanoparticles, 1-D nanostructures show better charge transport ability than their 0-D nanoparticle counterparts.^68–70 In particular, single-crystalline 1-D nanostructures can act as a “highway” for charge transport along the longitude direction.^71,72 It has been proved that the conductivity of the electrode material is crucial to a high power density supercapacitor. Therefore, 1-D nanostructures are the promising candidates for supercapacitor electrodes due to their super charge transport property, which will greatly improve the rate capability of supercapacitors. In addition, 1-D nanostructures can form a percolating network to avoid high mass loading as compared with 0-D nanoparticles.⁶⁹ Also, 1-D nanostructures, which have relatively high surface area, can form a porous network that acts as the scaffold/support of metal oxides.^68,69 The porous network can facilitate the mass transport of the solvated ions in the electrolyte to the electrode material surface. Furthermore, the 1-D carbon nanostructure–metal oxide composites have better mechanical toughness than pure metal oxides, which will benefit to the long-term cyclic stability. Many approaches including infiltration, electrophoretic deposition (EPD), hydrothermal, co-precipitation and vapor growth have been successfully developed to deposit metal oxides to form the 1-D carbon nanostructure–metal oxide composites.^73–83

Carbon nanotubes (CNTs): CNTs are the mostly investigated material among the 1-D carbon nanostructures for supercapacitor electrode applications. CNTs have a much higher electronic conductivity than amorphous AC. It has been reported that pure CNT electrodes are characteristic of the EDLC and the pseudocapacitance. Broad Faradaic peaks are observed in the MWCNT (multi-walled carbon nanotube) electrode during CV scan in various aqueous electrolytes, which is due to the fact that the surface functional groups and/or the impurities act as the redox centers. The highest specific capacitance of pure CNT electrode is found to be ∼100 F g⁻¹.⁷³

CNTs have been coupled with NiO,⁸⁴ V₂O₅,⁸⁵ MnO₂^86,87 and CuO⁸⁸ to form the CNT–metal oxide composite electrodes. The NiO–CNT composite was prepared by in situ growing the NiO nanoparticles in the CNT suspension during a hydrothermal process.⁸⁷ A more rectangular shape of CV was observed as compared with the pristine NiO powder electrode, indicating a more ideal capacitor character. The IR loss of the NiO–CNT composite electrode was reduced as the CNT content increased, which improved the rate capability of the electrode. After forming the composite, the specific capacitance of the active NiO component increased from 122 F g⁻¹ to ∼1600 F g⁻¹ at a scan rate of 2 mV s⁻¹ in the alkaline electrolyte. The increase in the specific capacitance was also observed when the V₂O₅ nanowires were integrated with MWCNTs.⁸⁵ The V₂O₅ component in the inter-weaved V₂O₅ nanowire–MWCNT composite delivered as high as 450 F g⁻¹ of specific capacitance at a current density of 1 A g⁻¹, which far exceeded that (100 F g⁻¹) of the pure V₂O₅ nanowire electrode. The CV of the V₂O₅–MWCNT composite showed much better defined redox peaks associated with the redox reaction of V₂O₅, which suggested that charge transfer properties were improved when the MWCNTs acted as the electron transport channel. Thus a high power density of 3.75 kW kg⁻¹ was achieved at an energy density of 5.5 W h kg⁻¹. The advantages of the composite were further demonstrated by high performance of the MnO₂–CNT electrode.⁸⁶ To prepare this composite electrode, the CNTs were immersed in the KMnO₄ solution to in situ form MnO₂ in the presence of carbon. The high electron mobility in the CNTs facilitated the following reaction

4MnO₄⁻ + 3C + H₂O = 4 MnO₂ + 2HCO₃⁻ + CO₃²⁻ (6)

The above reaction resulted in a MnO₂–CNT composite. After casting ∼35 mg cm⁻² of the 65 wt% MnO₂–CNT composite on the graphite electrode, a high area-capacitance of 5.07 F cm⁻² was achieved, which is the highest value for the carbon–metal oxide composites that have been investigated. Moreover, the location of MnO₂ on CNT had significant influence on the capacitor performance. It was found that MnO₂ in the inner wall of CNT showed much higher capacitance than those decorated on the outer shell (1250 F g⁻¹versus 790 F g⁻¹).⁸⁷ One of the possible reasons was due to the gap between the inner MnO₂ and the CNT, which served as the additional EDCL sites, thus brought extra capacitance by trapping ions inside the gap. Another possible reason was that the CNT contact site was not affected by MnO₂ confined inside the nanotubes. Hence a continuous charge transport pathway was retained. Further improvement of the metal oxide–CNT composite materials was realized by introducing a conductive polymer. Hou et al. have fabricated ternary MnO₂–CNT–PEDOT-PSS composite. In this composite, the CNT offered conductive pathway and supported the porous MnO₂ nanospheres. The introduction of PEDOT-PSS stabilized the MnO₂–CNT composite to facilitate the electrode fabrication, improved the conductive interconnection between metal oxide particles and CNT, and offered additional capacitance. A good stability (99% of retention after 1000 cycles) and increased specific capacitance (200 F g⁻¹ at 5 mA cm⁻²) were observed.⁸⁹

The metal oxide–CNT composite is also used to construct a symmetric composite hybrid supercapacitor such that the working window in an aqueous electrolyte can be extended, leading to an enhancement in the energy density. It has been reported that a Co–Al hydroxide–MWCNT symmetric hybrid supercapacitor shows a higher power density (6.4 kW kg⁻¹) as compared with the MWCNT–MWCNT capacitor (2.6 kW kg⁻¹). The addition of CNT into the hydroxide can effectively balance the electrode matching and increase the working window up to 1.6 V without decomposition of the electrolyte (6 M KOH).⁹⁰

It is also worth noting that the metal oxide–CNT composite paste can be injection-printed, which makes it possible to realize a printable wear-resistant solid supercapacitor device. For example, a mixed RuO₂ nanowire–SWCNT paste was printed on various substrates by the ink injection method.⁹¹ The conductivity of pristine CNT film reached 1562 S cm⁻¹. Specific capacitance of 138 F g⁻¹ and energy density of 18.8 W h kg⁻¹ were achieved in the solid polymer–gel electrolyte (polyvinyl alcohol in phosphoric acid, PVA–H₃PO₄).

Carbon nanofibers (CNFs): CNFs are inexpensive since they can be massively fabricated by various methods such as electrospinning or vapor growth. Activated electrospun CNF (E-CNF) has been employed as the alternative of activated carbon for supercapacitor electrodes. The pore size and the specific surface area can be adjusted via gas phase activation (H₂O vapor or CO₂) or chemical activation (KOH and ZnCl₂).^92,93 The specific surface area of the CNFs could reach 940 m² g⁻¹ with the optimized pore distribution.⁹³ The activated E-CNF electrode showed a high specific capacitance of 140 F g⁻¹,⁹² which was comparable to the AC-based electrode.

CNFs are very attractive as the support for metal oxides in the composite electrodes. Typically the metal oxides are coated on the CNF surface to form a core–shell structure in which the CNF not only serves as the physical backbone support but also offers the channel for charge transport, and the metal oxide acts as the redox center that contributes to specific capacitance and energy density. Various metal oxides such as V₂O₅,⁹⁴ SnO₂,⁹⁵ RuO₂,⁹⁶ and MnO₂^97,98 have been deposited on the E-CNFs. Amorphous V₂O₅ was coated on the activated CNF surface using the electrodeposition method to form a coaxial nanostructure.⁹⁴ The 4 nm thick V₂O₅ shell on the E-CNF exhibited a specific capacitance of 1308 F g⁻¹ in the KCl electrolyte at a scan rate of 2 mV s⁻¹. The specific capacitance was gradually reduced when the shell became thicker. When the shell thickness increased to 20 nm, the specific capacitance dropped to 560 F g⁻¹. It was ascribed to the surface-limited redox sites and the reduction of the electronic conductivity of the composite. The V₂O₅–E-CNF composite electrode showed 3 times enhancement in the specific capacitance as compared with pure electrospun V₂O₅ nanofibers.

MnO₂ is one of the best candidates for pseudo-capacitor electrodes because it has high theoretical specific capacitance (∼1380 F g⁻¹) and is environmentally benign and earth-abundant. Unfortunately the poor electrical conductivity of MnO₂ limits the full utilization of its high pseudo-capacitance. Recently the carbon–MnO₂ coaxial nanofibers have also shown a specific capacitance of 539 F g⁻¹ at 0.38 mg cm⁻² of MnO₂ loading.⁹⁷ However, such high capacitance values can be obtained only at a very slow scan rate (2 mV s⁻¹) and at a low metal oxide loading (<10 wt%). Severe loss occurred at a high scan rate or at a high metal oxide loading, which hinders their practical application when both high power and energy densities are needed.

In order to improve the rate capability of the CNF–metal oxide composite electrode, the electron transfer in the composite and the ion accessibility must be enhanced. A recent paper has demonstrated the highly conductive CNF–MnO₂ coaxial cables⁹⁸ (Fig. 2(a)). In the hierarchical MnO₂ structure, an around 4 nm thick sheath surrounds the carbon nanofiber (CNF) in a diameter of 200 nm, and nano-whiskers grew radically outward from the sheath in view of the cross-section of the coaxial cables, which yielded a high specific surface area of MnO₂ (Fig. 2(b)). The pseudo-capacitance of CNF–MnO₂ coaxial cable electrode was mainly contributed by the redox reactions, which took place only on the outermost surface layer of MnO₂ in contact with the aqueous Na₂SO₄ electrolyte:

MnO_{2 surface} + Na⁺ + e⁻ = MnO₂⁻Na⁺_surface (surface absorption) (7)

MnO_{2 bulk} + Na⁺ + e⁻ = MnOONa_bulk (bulk ion intercalation) (8)

Fig. 2 (a) SEM image of the AAI–CNF–MnO₂ electrode with 39 wt% MnO₂ loading and (b) corresponding TEM image. (c) Specific capacitance of CNF, AAI–CNF, AAI–CNF–MnO₂ and regular CNF–MnO₂ electrodes at different scan rates, the MnO₂ loading was 39 wt% and (d) the Bode plot of the CNF, AAI–CNF, AAI–CNF@MnO₂ and CNF–MnO₂ electrodes, the time constant was labeled. Adopted with permission from ref. 98. (Copyright (2012) Elsevier).

According to the Nyquist plots of impedance, the capacitance retardation in the CNF@MnO₂ samples was ascribed to high series resistance attributed to insufficient conductivity of the electrode and to large charge transfer resistance at the CNF–MnO₂ interface that inhibits the Na⁺ insertion/adsorption process in MnO₂, leading to low rate capability. In other words, rapid transfer of Na⁺ cation and electrons to MnO₂ simultaneously is the key to completing the charge storage reaction. Reaction (7) can be achieved by increasing the MnO₂ surface area exposed to the electrolyte, which is realized by the high specific surface area of the AAI–CNF support. Also, a thin and porous MnO₂ sheath and the nano-whisker structures shorten the path of ion diffusion. Moreover, owing to high surface area of the AAI–CNF backbone and ultra-small thickness of the MnO₂ shell, the CNF–MnO₂ contact area is large and the interfacial resistance between carbon and MnO₂ is very small. This greatly facilitates the transfer of the electron and the ions involved in the redox reactions. Reaction (8) requires the fast electron transport inside the CNF backbone to effectively transport electrons to the MnO₂ phase. In this study, iron(III) acetylacetonate (AAI), an iron organic compound, was incorporated in the electrospinning precursor to improve the electronic conductivity when preparing the E-CNFs. The resulting AAI–CNF cable showed much higher electronic conductivity and surface area than the regular CNFs (10 S cm⁻¹versus 1 S cm⁻¹, and 128 m² g⁻¹versus 28 m² g⁻¹). The improvement in the electronic conductivity was originated from the larger amount of the ordered graphite phase in the AAI–CNF. As a result, the AAI–CNF–MnO₂ electrode showed an impressive specific capacitance of 311 F g⁻¹ for the whole composite electrode and 900 F g⁻¹ for the MnO₂ shell at a scan rate of 2 mV s⁻¹ as shown in Fig. 2(c). In contrast, the regular CNF–MnO₂ showed only 122 F g⁻¹ for the whole composite and ∼320 F g⁻¹ for the MnO₂ under the same test conditions. An improved rate capability was observed for the AAI–CNF@MnO₂ hybrid. That is, ∼50% of the AAI–CNF–MnO₂ capacitance was retained from 2 mV s⁻¹ to 200 mV s⁻¹ as compared with 11% for the regular CNF–MnO₂. The optimal AAI–CNF@MnO₂ electrode showed good cycling stability, high energy density (80.2 W h kg⁻¹) and high power density (57.7 kW kg⁻¹) for the active MnO₂.

Complex Bode plots of impedance were also used to study the rate capability of the electrodes. The impedance of a supercapacitor device can be expressed as

where C is the device capacitance and ω is the radius speed; the equation then yields

Herein the total capacitance of the device can be written as

C = C′ − jC′′ (11)

whereand

The real part C′ is defined as the effective capacitance that the device can deliver, and the imaginary part C′′ is related to the irreversible resistivity loss in the device. The frequency f is the character frequency at which C′′ reaches the maximum and τ = 1/f is the time constant of the device, which are characteristic of the rate capability. High power density supercapacitors generally possess high character frequency and smaller τ. A comparison of the Bode plot for the different CNF–MnO₂ electrodes is shown in Fig. 2(d). It could be seen that character frequencies for AAI–CNF and AAI–CNF–MnO₂ with 39 wt% MnO₂ loading were at 0.48 Hz and 0.1 Hz, respectively. It should be noted that the character frequency of AAI–CNF–MnO₂ was comparable with that of the activated carbon electrode in which no charge transfer process happened (0.1 Hz, 10 s). However, for the CNF–MnO₂ with similar 39 wt% MnO₂ loading, the character frequency was far below the lowest testing frequency of 0.01 Hz. This implied that the AAI–CNF–MnO₂ electrode was able to deliver energy much faster than its CNF@MnO₂ counterparts. This work indicated that high electrical conductivity of the electrode material was crucial to achieving high power and energy density for pseudo-supercapacitors.

The addition of CNTs or graphene into the precursor of the E-CNFs is an alternative and effective method to improve the electronic conductivity.⁹⁹ The CNF and graphene can be self-aligned inside the fiber along the longitude direction to promote the electron transport. For instance, the addition of 3 wt% of CNTs into the E-CNF increased the electronic conductivity of the E-CNF from 0.42 S cm⁻¹ to 0.98 S cm⁻¹. The specific surface area was also enlarged two times. As a result, for the RuO₂–E-CNF composite electrode, the specific capacitance increased from 390 F g⁻¹ to 590 F g⁻¹ based on the RuO₂ content.⁹⁹

Carbon nanofibers network was also used to improve the performance of the battery-like hybrid supercapacitor. The Li₄Ti₅O₁₂–carbon hybrid nanofibers were prepared by an electrospinning method following by conductive polypyrrole coating. Such an electrode showed 122 mA h g⁻¹ of capacity at 4000 mA g⁻¹. The hybrid supercapacitor exhibited comparable energy density to a lithium ion battery and better power density than an AC–AC supercapacitor, and preserved the good cycling property of Li₄Ti₅O₁₂.¹⁰⁰

1-D carbon nanostructure-based papers: Due to the strong van der Waals force and rich functional groups, sheet- or paper-like CNT films can be obtained using simple filtration and vacuum suction methods. Such resulting papers show excellent mechanical properties. Hence they can be used as flexible supercapacitor electrodes without need of a binder or a current collector. The CuO nanobelt-CNT free-standing mat has been prepared by mixing the CuO nanobelt with SWCNTs as shown in Fig. 3(a) and (b).⁸⁸ The SEM image in Fig. 3(c) shows that the CuO nanobelt and SWCNT form a rigid and porous structure. This bendable CuO nanobelt–SWCNT supercapacitor electrode was operated at a voltage window of 0–2.5 V in the LiPF₆ electrolyte. Such an electrode exhibited a higher specific capacitance (150 F g⁻¹) at a current density of 1 A g⁻¹ than the CuO nanoparticle–carbon black electrode (100 F g⁻¹) and the pure CuO nanobelt electrode (78 F g⁻¹). The enhancement was attributed to the large surface area of the CuO nanobelts and to the conductive CNT matrix, which facilitated the charge transfer and ion diffusion. In addition, a layer-by-layer (LBL) self-assembly method was developed to prepare a CNT film. In such a process, the CNT surface was functionalized with different chemical groups for different surface charge polarities. Alternative change in the surface polarity of the CNT allowed the firm assembly of a multilayer of CNT films on the indium tin oxide (ITO) coated glass. Such a CNT film was used as the support for MnO₂ as the electrode.¹⁰¹ The MnO₂ component in the CNT–MnO₂ composite electrode delivered a specific capacitance of 965 F g⁻¹ at a current density of 0.25 A g⁻¹ in the Na₂SO₄ electrolyte. The volume capacitance of the composite electrode can reach 200 F cm⁻³. A linear relationship was observed of the area capacitance versus the film thickness, indicating the potential of scale-up fabrication.

Fig. 3 (a) Image of the flexible CuO–CNT electrode, (b) electrode subjected to bending and (c) SEM image of the CuO nanobelt–CNT electrode. Reprinted with permission from ref. 88. Copyright (2011) American Chemical Society.

Carbon fibers can also form a paper. Liu et al. have recently used carbon fiber paper (CFP) as the backbone to fabricate a CFP/Co₃O₄ paper electrode.¹⁰² The Co₃O₄ nanonet in the CFP/Co₃O₄ paper electrode exhibited a specific capacitance of 1190 F g⁻¹ at a current density of 0.25 A cm⁻² in the NaOH electrolyte. The CFP/Co₃O₄ nanonet paper electrode retained 94.4% of its capacitance when the current density increased to 25 A g⁻¹. Such a good rate capability is better than any Co₃O₄ based electrodes reported so far. One of the reasons for the remarkable specific capacitance and rate capability was attributed to the unique 1-D nanonet structure of the electrodes. Another reason was that the CNF paper played provided excellent electronic conductive channel. The third reason was that CFP/Co₃O₄ nanonet paper electrode favored the ion diffusion. Li et al. have grown Zn₂SnO₄ nanoneedles on the CNF paper using the vapor deposition method.¹⁰³ A carbon layer was further coated on Zn₂SnO₄ nanoneedles, followed by growth of MnO₂. The electrons and ions were effectively transported via the network of MnO₂–carbon–Zn₂SnO₄–CNF, which resulted in a specific capacitance of 565 F g⁻¹ for the MnO₂ component at 2 mV s⁻¹. Additionally, the mechanical strength of the CNF paper allowed the electrode to be operated at the bending condition with little performance degradation over 1000 cycles. A similar device was built by growing the WO_3−x@Au@MnO₂ core–shell nanowires on a CFP. The Au interlayer accelerated the electron transport and improved the charge transfer capability of the electrode. As a result, the IR loss was reduced and a remarkable specific capacitance of 1195 F g⁻¹ was recorded for the active MnO₂ mass at a current density of 0.75 A g⁻¹.¹⁰⁴

Vertically aligned 1-D nanostructures: 1-D nanostructures can be directly grown on the electrode or the current collector and vertically aligned. Such aligned nanostructures have been proven to greatly promote the charge transport and mass transport in various electrochemical cells. In addition, the aligned 1-D nanostructures are in intimate contact with the current collector, leading to reduced contact resistance at the current collector–CNT interface, which will reduce the IR loss, especially at a high current loading. Recently the vertically aligned NiO nanoplatelet array has been grown on the conductive F-doped tin oxide (FTO) substrate as the supercapacitor electrode¹⁰⁵ (Fig. 4(a)). The CV curves exhibited approximately rectangular shape in the 1 M KOH aqueous solution with a three-electrode cell (Fig. 4(b)). The NiO nanoarray electrode exhibited a high specific area capacitance (620 F m⁻² equivalent to 312 F g⁻¹) at a current density of 4 mA cm⁻². The cyclic stability of such an electrode was excellent and the specific capacitance exhibited no degradation after continuous charge/discharge measurement over 1000 cycles at a current density of 4 mA cm⁻² and a voltage range of −0.2 to 0.3 V. Generally, the kinetics of a redox reaction at an electrode is governed by the charge transfer process in the electrode, the surface adsorption of ions on the electrode surface, and the diffusion of ions in the electrolyte. The redox peak current of NiO in Fig. 4(c), which was derived from the CV curves, shows a linear relationship with the scan rate. This indicated that surface adsorption of ions was the rate-limiting step. The single-crystalline nature of the NiO nanoplatelet array facilitated electron transport, leading to small charge transfer resistance, which was further confirmed by the impedance analysis. In addition, the individual vertically aligned nanoplatelets act as ultra-small electrodes, which shortened the distance of ion diffusion to the electrodes in the solution and reduced the IR loss.

Fig. 4 (a) SEM image of the aligned NiO nanoplates array, (b) CV scan of the NiO array electrode at different scan rates, and (c) redox peak current normalized to scan rate. Reproduced from ref. 105 with permission from The Royal Society of Chemistry.

The vertically aligned CNT forest was grown and collected by a directional chemical vapor deposition (CVD) method. The CNT matrix not only served as the metal oxide support but also as the current collector, which was suitable for the supercapacitor electrode.¹⁰⁶ Vertically aligned CNT electrodes normally had specific capacitance of less than 100 F g⁻¹ but good rate capability.¹⁰⁷ The specific capacitance can be improved by depositing metal oxide on the CNTs. Different metal oxides such as Co₃O₄, NiO and Mn₂O₃ were deposited on the CNT forest by thermal decomposition of metal nitrates.¹⁰⁶ The resulting Mn₂O₃–CNT composite electrode showed not only high specific capacitance (508 F g⁻¹ for the whole composite and 800 F g⁻¹ for active Mn₂O₃ at 10 mV s⁻¹) but also good rate capability (245 F g⁻¹ for the composite at a current density of 155 A g⁻¹). As a result, high energy density (30 W h kg⁻¹) and power density (60 kW kg⁻¹) were achieved. The super performance was attributed to the unique aligned structure, which shortened the charge transfer path of electrons to the metal oxide, as well as the high in-plane conductivity of the CNT sheet. In addition, a hybrid CNT@MnO₂ coaxial nanotube array has been prepared.¹⁰⁸ The MnO₂ nanotubes were first prepared by vacuum filtration into a porous anodic aluminum oxide (AAO) template. The CNTs were then grown inside the MnO₂ nanotubes as the conductive pathway. An intimate connection between the working electrode and the current collector was then formed, thus the ohmic drop in the charge/discharge process was negligible. The symmetric MnO₂–CNT supercapacitor showed a specific capacitance of 60 F g⁻¹ at a current density of 6 A g⁻¹ in the Na₂SO₄ electrolyte with long-term stability. Similar vertical aligned CNT–MnO₂ electrode showed a specific capacitance of 642 F g⁻¹ for active MnO₂ at a scan rate of 10 mV s⁻¹.¹⁰⁹ Recently, flower-like MnO₂ has been successfully grown on the CNTs. 84% of capacitance was retained when the scan rate increased from 10 mV s⁻¹ to 100 mV s⁻¹, which showed much better rate capability due to smaller charge transfer and diffusion resistance as compared with the electrode made of pure MnO₂ nano-flower.¹¹⁰

4 Two-dimensional carbon–metal oxide electrodes

The 2-D carbon nanosheets such as graphene, graphene oxide (GO) and reduced graphene oxide (rGO) have been used for the supercapacitors. Graphene has a large specific surface area and extremely high electronic conductivity, which make it attractable for supercapacitor electrode. Besides graphene, chemically rGO is also attractive as the supercapacitor electrode because of its hydrophilic feature, which is not only good for further compositing with metal oxides but also for wettability in the aqueous electrolyte. In addition, rGO still has high electronic conductivity. Moreover, it is rich in surface functioned groups such as alcohol, epoxide, carbonyl and carboxylic acid. It is also possible to introduce additional functional groups on rGO by further modification.¹¹¹ The functional groups may serve as the redox centers, which can contribute to pseudo-capacitance. As a result, GO may have higher specific capacitance than graphene.¹¹² Ruoff et al. have produced the activated rGO with a specific surface area of 3100 m² g⁻¹. The electrode made of such activated rGO exhibited a specific capacitance of 166 F g⁻¹ at a current density of 5.7 A g⁻¹.¹¹³

The unique planar structure of the graphene family materials makes them good supports for metal oxides. As compared with 1-D carbon nanostructures, 2-D carbon nanostructures are even easier and more flexible to integrate with metal oxides. The formation of rGO (or graphene)–metal oxide composite can be realized via various methods including electrodeposition,^114–116 hydrothermal processing,^117–119 co-precipitation,^120,121 spray pyrolyzing¹²² and so on. Yu's group has developed a general strategy to prepare rGO–metal oxide core–shell structures, which could be also used to fabricate supercapacitor electrodes.¹²³ So far rGO has been coupled with various metal oxides such as TiO₂,²⁷ NiO,^114,115,120 Co₃O₄¹¹⁹ and MnO₂^116,124,125 as the supercapacitor electrodes. Table 4 lists the representatives of the metal oxide–rGO composite electrodes.

Table 4 Representatives of the metal oxide–rGO composite electrodes

Synthesis method	Metal oxide–rGO composite	Specific capacitance	References
Electrodepositing	NiO porous film on rGO sheet	432 F g^−1 at 2 A g^−1 for NiO, 636 mF cm^−2 at 0.5 mA cm^−2	114
	NiO nanoparticles–rGO	569 F g^−1 at 5 A g^−1 for NiO	115
	MnO2 nanoparticle–rGO	476 F g^−1 at 1 A g^−1 for MnO2	117
Hydrothermal method	MnO2 nanowire–rGO	211.2 F g^−1 at 0.15 A g^−1 for composite	117
	Flower-like NiO–rGO sheet	346 F g^−1 for composite 778.7 F g^−1 for NiO at 1.5 A g^−1	118
	Co3O4 nanoparticle–rGO	415 F g^−1 at 3 A g^−1	119
Co-precipitation	Monolayer NiO–rGO	525 F g^−1 at 0.2 A g^−1	120
	MnO2 nanowire–rGO	176 F g^−1 at 5 mV s^−1	121

---

A synergistic enhancement effect was observed when coupling rGO with metal oxides. For example, the RuO₂ nanoparticles were anchored on the rGO surface by the precipitation method.¹²⁶ The specific capacitance of the RuO₂–rGO composite was higher than that of either pure RuO₂ or rGO. One reason was that the RuO₂ nanoparticles separated the rGO sheets, which increased the EDLC of the rGO. Also, the dispersed rGO prevented the agglomeration of the RuO₂ nanoparticles, and offered conductive pathway. Similar results were also observed in the MnO₂ nanowires–rGO and Co₃O₄–rGO composite.^127,128 More detailed description of such a synergistic effect can be found in a recent review.¹²⁹

An additional study of coupling rGO with various kinds of metal oxides as the supercapacitor electrodes was carried out by Ramaprabhu and Mishra.¹³⁰ It has been found that the rGO-supported metal oxide electrode had higher capacitance as compared with other carbon supports due to the large contact area between rGO and the active metal oxide materials, which suggested the importance of rGO for electrochemical energy storage. It has also been reported that a rGO–MnO₂/carbon nanofiber asymmetric supercapacitor exhibited a specific capacitance of 70 F g⁻¹ with an operating window of 2.5 V at 2 mV s⁻¹ scan, which gave a high energy density of 51.5 kW h kg⁻¹.¹³¹ Ni(OH)₂ nano-flowers were also self-assembled on the rGO to form the rGO–Ni(OH)₂ composite electrode, which delivered a power density of 15.2 kW kg⁻¹ at the energy density of 13.5 W h kg⁻¹.¹³² When rGO was used as the metal oxide support, the specific surface area of rGO can be further increased by an activation process. Zhao et al. prepared a series of KOH-activated microwave-expanded graphite oxide and then coupled such graphite oxide with MnO₂ to form a composite. The specific surface area of the activated graphite oxide reached 2690 m² g⁻¹. After deposition of MnO₂, the pore size showed a negligible change (constant 1.8 nm). The asymmetric hybrid supercapacitor with the activated graphite oxide as the positive electrode and the composite as the negative electrode showed a specific capacitance of 175 F g⁻¹ at 0.25 A g⁻¹ with a working window of 2 V.¹³³

Recently the single-crystalline TiO₂ nanobelts have been deposited on the rGO sheets to form the rGO–TiO₂ nanobelt composite as the supercapacitor electrode (Fig. 5).¹³⁴ A mixture of GO and TiO₂ nanobelts (NBs) was subject to mild hydrothermal processing in the ethanol solution, which reduced GO to rGO and incorporated the TiO₂ nanobelts into the rGO sheets simultaneously. For a comparative study, the rGO–TiO₂ nanoparticle (NP) composite was also prepared. The results have shown that the electrochemical performance of the rGO–TiO₂ composite was much better than that of any individual components, indicating the synergistic effect. The monolithic rGO electrode had a specific capacitance of 40 F g⁻¹. The TiO₂ nanobelt and nanosphere electrodes showed capacitances of 17 F g⁻¹ and 11 F g⁻¹ in the Na₂SO₄ electrolyte at a scan rate of 2 mV s⁻¹, respectively. After rGO was coupled with TiO₂ nanobelts or nanoparticles, the specific capacitance significantly increased. The optimized rGO/TiO₂ mass ratio was 7 : 3 for both the rGO–TiO₂ nanoblets and the rGO–TiO₂ nanoparticles, which resulted in 200 F g⁻¹ and 60 F g⁻¹ of specific capacitance, respectively. Interestingly, the shape of the TiO₂ nanomaterial had significant influence on the performance of the composite electrode. The rGO–TiO₂ nanobelt composite electrode showed much higher specific capacitance, energy density and power density than its rGO–TiO₂ nanosphere counterpart (Fig. 5). One of the reasons was that the TiO₂ nanobelts favored the charge mobility as compared with the TiO₂ nanoparticles. In addition, the TiO₂ nanobelts were ∼10 nm thick, which was close to the theoretical ion intercalation depth of TiO₂. Therefore the whole TiO₂ material in the nanobelts was actively involved in the charge storage process while only the outermost surface layer (∼10 nm thick) was active in the TiO₂ nanoparticles in a diameter of 150 nm. Furthermore, the TiO₂ nanoparticles had a point contact with the rGO sheet while The TiO₂ nanobelts had planar contact with the rGO sheet. This resulted in a much higher contact area at the rGO–TiO₂ interface, which improved the charge transfer and the charge storage ability of the electrode. Impedance spectra analysis also confirmed the rGO–TiO₂ nanobelt composite electrode had fast time constant as compared with the rGO–TiO₂ nanosphere electrode.

Fig. 5 Schematic of the rGO–TiO₂ composite with different TiO₂ morphologies and corresponding electrochemical performance. Reproduced from ref. 134 with permission from The Royal Society of Chemistry.

It is worth noting that a MnO₂–rGO composite electrode has been prepared by immersing the rGO sheets in the KMnO₄ solution.¹²⁸ Nanoscale MnO₂ was produced on the rGO surface by a self-limited electroless deposition process. With 78 wt% of MnO₂ loading, the specific capacitance of the composite electrode reached 310 F g⁻¹ at a scan rate of 2 mV s⁻¹. Moreover, 74% of the capacitance was retained even when the scan rate increased to 500 mV s⁻¹, which implied excellent rate capability. Such impressive performance was beneficial from the high specific surface area of the composite electrode (287 m² g⁻¹) and the high electronic conductivity of the composite electrodes. As the metal oxide support, the specific surface area of rGO can be further increased by an activation process. A recent report shows that the addition of CNT into the rGO–MnO₂ composite can further enhance the conductivity of electrode, leading to the improvement in the rate capability.¹³⁵

Another advantage of graphene and rGO based supercapacitor electrodes is their high mechanical strength. rGO can form a robust film or a free-standing paper using a simple filtration and vacuum suction process. The Young's modulus of graphene can reach as high as 1.0 TPa, which is more resilient than other carbon materials based papers.⁴⁹ The electrode volume change of metal oxide supported on such a flexible matrix during the charge–discharge process can be migrated, which can further improve the long-term stability. Fig. 6 shows the fabrication process of the supercapacitor electrode based on the flexible rGO–MnO₂ paper. Such an electrode exhibited a specific capacitance of 256 F g⁻¹ at a discharge current of 0.5 A g⁻¹.¹³²

Fig. 6 Schematic of the fabrication process of the rGO–MnO₂ paper electrode. Reproduced from ref. 132 with permission from The Royal Society of Chemistry.

5 Three-dimensional porous carbon–metal oxide electrodes

High specific capacitance, energy density and power density are always desirable for a supercapacitor. On the other hand, the total energy and power stored in the whole device are sought to be maximized without increasing the overall footprint of device. Hence, each electrode is required to load active materials as much as possible to store a total amount of capacitance, energy and power as high as possible. It is a challenge to load a large amount of material on each electrode without undermining the electrochemical performance in terms of specific capacitance, rate capability, energy density and power density. 3-D porous carbon–metal oxide nano-architectures are being developed to address this challenge. Thus many carbon nanostructures are assembled with the building blocks of lower dimensional carbon nanostructures. Such building blocks form a hierarchical structure in the 3-D space but the individual building blocks can still keep their original merits.

3-D carbon nanofoams are promising candidates for supporting metal oxides. One of the fabrication methods is the infiltration of carbon fiber paper with polymer resorcinol and subsequent pyrolysis at high temperatures. 3-D carbon nanofoams had a “through-connected” pore network and high electronic conductivity (∼1 S cm⁻¹).^136,137 The 3-D nanofoam structures had large specific surface areas to allow high metal oxide loading (∼50 wt%). FeO_x and MnO₂ were then electrolessly deposited on the carbon nanofoam via in situ reduction of high-valence Mn and Fe ions by carbon, respectively. The carbon–MnO₂ and carbon–FeO_x nanofoam electrodes showed high area specific capacitance of 1.5 F cm⁻² and 0.85 F cm⁻², respectively.

It was reported that graphene was directly grown on a Ni foam by the CVD process, in which the Ni foam served not only as the template but also as the catalyst. Subsequently the Ni foam template was dissolved to form a porous, highly conductive 3-D graphene network.¹³⁸ Such a highly interconnected graphene foam had conductivity of over 10 S cm⁻¹ and the specific surface was close to the theoretical value for single-layer graphene. The graphene nanofoam alone can act as good supercapacitor electrode material, which exhibited a specific capacitance of ∼100 F g⁻¹ with good rate capability. Incorporation of metal oxides such as Co₃O₄ and ZnO into this graphene foam can further improve the specific capacitance as the metal oxides serve as the redox centers. For example, the Co₃O₄ nanowires were deposited to the graphene foam to form a Co₃O₄–graphene nanofoam electrode (Fig. 7),¹³⁹ which showed a specific capacitance of 765 F g⁻¹. 80% of capacitance was retained when the scan rate increased from 2 mV s⁻¹ to 100 mV s⁻¹. Recently ZnO has been hydrothermally grown on the graphene foam and used as the supercapacitor electrode.¹⁴⁰ Clear redox peaks of ZnO were observed in the KOH electrolyte. A 211 F g⁻¹ of specific capacitance was obtained at a current density of 33.3 A g⁻¹. The results indicated that the enlarged surface area as well as the fast charge transfer ability brought by the graphene foam was the key to high performance of supercapacitor electrodes.

Fig. 7 (a) 3-D graphene nanofoam, (b) Co₃O₄ nanowires incorporated into graphene nanofoam, (c) the charge–discharge plots of the electrode at different current density and (d) the long term stability of the Co₃O₄ nanowires-3-D graphene foam electrode. Reprinted with permission from ref. 139. Copyright (2012) American Chemical Society.

Cui's group has developed a facile method to immobilize rGO on a flexible substrate such as the polymer texture.¹⁴¹ Such a mesh-like rGO network was flexible and conductive, and can survive after mechanical bending. MnO₂ was then grown on such a scaffold with the hydrothermal method. The resulting MnO₂–rGO-mesh composite electrode exhibited a specific capacitance of 315 F g⁻¹ for the active MnO₂ at a scan rate of 2 mV s⁻¹. In addition, a 3-D rGO framework has been prepared by using polystyrene spheres as the sacrificial templates.¹⁴² Such 3-D nanostructures not only offered high surface area but also preserved good electronic conductivity (1204 S m⁻¹). The 3-D MnO₂–rGO framework composite electrode exhibited a specific capacitance of 389 F g⁻¹ at a current density of 1 A g⁻¹. Moreover, 97.7% of capacitance was retained when the current density increased to 35 A g⁻¹.

Besides graphene and rGO foams, CNT foams (or sponges) have also been fabricated as the supercapacitor electrodes in a similar way. The porous structure allows for the accessibility of the ions to the CNT electrode surface.¹⁴³ The interconnected network provides continuous channels for electron transport. The CNT sponge electrode has been demonstrated to be capable of operation at a scan rate as high as 200 V s⁻¹. After coating MnO₂ on the CNT sponge, the power density and the energy density were further improved and reached 63 kW kg⁻¹ and 31 W h kg⁻¹, respectively. The specific capacitance of the active MnO₂ component in the composite electrode reached 1270 F g⁻¹, which was close to the theoretical value.

6 Summary and future direction

This paper has reviewed the research progress in the carbon–metal oxide composites for supercapacitor electrodes. In such a composite electrode, the metal oxide provides the source of high specific capacitance and high energy density; and the carbon nanostructure ensures good rate capability and high power density at a large current (or at a high scan rate). Moreover, the carbon nanostructures at different dimensions have their own unique features, which provide the flexibility to tailor the microstructure and properties of carbon–metal oxide composite electrodes. Each type of carbon nanostructure has its advantages as follows:

• 0-D carbon nanoparticles provide the flexibility for tuning the porosity, which can be optimized to be suitable for different electrolytes.

• 1-D carbon nanostructures can facilitate the charge transport in the supercapacitor electrode. Although they have relatively smaller specific surface area as compared with 0-D carbon nanomaterials, the high conductivity of 1-D carbon nanostructures is greatly beneficial to the redox activities of the metal oxides, which enhances the pseudo-capacitance and improves the rate capability.

• 2-D carbon nanostructures combine the high surface area, high electronic conductivity and high mechanical strength. Such properties are very attractive for flexible energy-storage devices and for improvement in the charge/discharge reaction kinetics of supercapacitor electrodes.

• 3-D carbon nanostructures result from the assembly of low-dimensional carbon nanomaterials while retaining the advantages of the building blocks, which enlarges the total active volume of the supercapacitor electrodes without increasing the overall footprint of the device. This will result in compact energy storage devices.

The global market for supercapacitors is $470 million in the world in 2010.¹⁴⁴ The market is projected to grow at an annual rate of 21.4%¹⁴⁴ and reach $1.2 billion by 2015.¹⁴⁵ Taking 5% of the battery market, the supercapacitor market is projected to be $3.5 billion in 2020.¹⁴⁶ The increasing market share will further stimulate the research and development of supercapacitors. It is essential to further improve the performance including specific capacitance, energy density, power density, rate capability, long-term cyclic stability and performance/price ratio in order to meet the commercial need of supercapacitors. In particular, it remains a significant challenge to overcome the barrier of insufficient rate capability of carbon–metal oxide composite electrodes. The critical factors governing the performance of carbon–metal oxide composite supercapacitor electrodes should be considered in the future research:

Optimizing the dimension, shape, orientation and porosity of carbon nanostructures: since the carbon nanostructure serves as the physical support of metal oxides, its structure determines the architecture of the whole carbon–metal oxide composite. High specific surface area of the carbon support is the prerequisite to maximize the loading of metal oxides and to achieve large area of the carbon–metal interface. Small dimension and optimized porosity will shorten the distance of the ion diffusion to the electrode surface.

Improving the electronic conductivity of carbon nanostructures: the rate capability of a supercapacitor is dependent heavily on the electronic conductivity of electrodes. Since metal oxides generally have very low electronic conductivity, the carbon nanostructure is the only option of tuning the electronic conductivity. The intimate contact between the carbon nanostructure and the current collector is also the key to minimizing the interfacial resistance.

Enlarging the interfacial carbon–oxide area and controlling the thickness of metal oxide: intimate contact at the carbon–oxide interface and large carbon–oxide interfacial area are the important factors that affect the charge transfer ability of the composite electrode. Since the redox reactions only occur on the outermost surface layer of metal oxide for pseudo-supercapacitors, the thickness of metal oxides supported on the carbon nanostructure should be thin enough (comparable to the ion insertion depth) to allow the full utilization of electro-active materials. On the other hand, since the electrons produced in the Faradaic process have to be transferred across the carbon–metal oxide interface, the thickness of metal oxide should also be comparable to the electron diffusion length in the metal oxide so that the electrons involved in the redox reactions can be transferred to the current collector.

Developing hybrid supercapacitors: current symmetric supercapacitor has a low energy density. To improve the energy density, one of the effective routes is to develop hybrid supercapacitors. For example, a hybrid supercapacitor with the nanostructured NiO cathode and the carbon anode exhibited a high energy density of 10 W h kg⁻¹ at a high power density of 10 kW kg⁻¹.¹⁴⁷ Since the energy density is directly proportional to the square of the operating voltage window, the non-aqueous electrolyte is desirable for supercapacitors. It is recommended that the metal oxide–carbon composite electrode to be used in hybrid supercapacitors with a large operating voltage window.

Exploring sustainable materials for supercapacitors: we must respond to an unprecedented rise in the demand for reduction of raw material consumption and minimizing the environmental impact of manufacturing process. Therefore there is strong incentive to develop “green materials”, which are earth-abundant, renewable, recyclable and non-toxic.

Establishing the testing standards: it is necessary to establish the standards for testing supercapacitors in order to compare the performance data taken from different laboratories.^16,148 In many previous papers, the specific capacitance data were only taken from the metal oxide component. In practical application, it is important that the overall performance data are given for composite electrodes.

Notes and references

1. U. Eberle and R. von Helmolt, Energy Environ. Sci., 2010, 3, 689–699.
2. J. T. Zhang and X. S. Zhao, ChemSusChem, 2012, 5, 818–841.
3. G. P. Wang, L. Zhang and J. J. Zhang, Chem. Soc. Rev., 2012, 41, 797–828.
4. P. Simon and Y. Gogotsi, Nat. Mater., 2008, 7, 845–854.
5. M. Winter and R. J. Brodd, Chem. Rev., 2004, 104, 4245–4269.
6. M. Jayalakshmi and K. Balasubramanian, Int. J. Electrochem. Sci., 2008, 3, 1196–1217.
7. S. Mallika and R. S. Kumar, Int. J. Eng. Technol., 2011, 3, 37–43.
8. A. Kuperman and I. Aharon, Renewable Sustainable Energy Rev., 2011, 15, 981–992.
9. Y. Zhang, H. Feng, X. B. Wu, L. Z. Wang, A. Q. Zhang, T. C. Xia, H. C. Dong, X. F. Li and L. S. Zhang, Int. J. Hydrogen Energy, 2009, 34, 4889–4899.
10. X. Zhao, B. M. Sánchez, P. J. Dobson and P. S. Grant, Nanoscale, 2011, 3, 839–855.
11. S. Bose, T. Kuila, A. K. Mishra, R. Rajasekar, N. H. Kim and J. H. Lee, J. Mater. Chem., 2012, 22, 767–784.
12. L. L. Zhang and X. S. Zhao, Chem. Soc. Rev., 2009, 38, 2520–2531.
13. B. E. Conway, Electrochemical Supercapacitors: Scientific Fundamentals and Technological Applications, 1999, Plenum Pub Corp, p. 222.
14. H. Jiang, J. Ma and C. Z. Li, Adv. Mater., 2012, 24, 4197–4202.
15. C. D. Lokhande, D. P. Dubal and O. S. Joo, Curr. Appl. Phys., 2011, 11, 255–270.
16. M. D. Stoller and R. S. Ruoff, Energy Environ. Sci, 2010, 3, 1294–1301.
17. B. Andrea, D. Romain, T. Pierre-Louis, S. Patrice, P. Dominique, M. Marina and P. Stefano, J. Power Sources, 2007, 165, 922–927.
18. V. R. Koch, Recent Advances in Electrolyte for Electrochemical Double Layer Capacitors, 2005.
19. M. D. Stoller, C. W. Magnuson, Y. W. Zhu, S. Murali, J. W. Suk, R. Piner and R. S. Ruoff, Energ. Environ. Sci., 2011, 4, 4685–4689.
20. E. Frackowiak and F. Beguin, Carbon, 2001, 39, 937–950.
21. http://www.mpoweruk.com/performance.htm.
22. V. D. Patake, C. D. Lokhande and O. S. Joo, Appl. Surf. Sci., 2009, 255, 4192–4196.
23. D. L. Yan, Z. L. Guo, G. S. Zhu, Z. Z. Yu, H. R. Xu and A. B. Yu, J. Power Sources, 2012, 199, 409–412.
24. U. M. Patil, R. R. Salunkhe, K. V. Gurav and C. D. Lokhande, Appl. Surf. Sci., 2008, 255, 2603–2607.
25. B. R. Duan and Q. Cao, Electrochim. Acta, 2012, 64, 154–161; R. Z. Li, X. Ren, F. Zhang, C. Du and J. P. Liu, Chem. Commun., 2012, 48, 5010–5012.
26. T. Lu, Y. P. Zhang, H. B. Li, L. K. Pan, Y. L. Li and Z. Sun, Electrochim. Acta, 2010, 55, 4170–4173.
27. X. Sun, M. Xie, G. K. Wang, H. T. Sun, A. S. Cavanagh, J. J. Travis, S. M. George and J. Lian, J. Electrochem. Soc., 2012, 159, A364–A369.
28. X. Y. Chen, E. Pomerantseva, P. Banerjee, K. Gregorczyk, R. Ghodssi and G. Rubloff, Chem. Mater., 2012, 24, 1255–1261.
29. J. S. Shaikh, R. C. Pawar, R. S. Devan, Y. R. Ma, P. P. Salvi, S. S. Kolekar and P. S. Patil, Electrochim. Acta, 2011, 56, 2127–2134.
30. P. M. Kulal, D. P. Dubal, C. D. Lokhande and V. J. Fulari, J. Alloys Compd., 2011, 509, 2567–2571.
31. X. B. Ren, H. Y. Lu, H. B. Lin, Y. N. Liu and Y. Xing, Russ. J. Electrochem., 2010, 46, 77–80.
32. P. Lu, D. Xue, H. Yang and Y. Liu, Int. J. Smart Nano Mater., 2012, 1, 1–25.
33. X. Y. Lang, A. Hirata, T. Fujita and M. W. Chen, Nat. Nanotechnol., 2011, 6, 232–236.
34. M. Toupin, T. Brousse and D. Bélanger, Chem. Mater., 2004, 16, 3184–3190.
35. D. Bélanger, T. Brousse and J. W. Long, Electrochem. Soc. Interface, 2008, 17, 49–52.
36. D. S. Kong, J. M. Wang, H. B. Shao, J. Q. Zhang and C. N. Cao, J. Alloys Compd., 2011, 509, 5611–5616.
37. S. Nandy, U. N. Maiti, C. K. Ghosh and K. K. Chattopadhyay, J. Phys.: Condens. Matter, 2009, 21, 115804.
38. R. B. Rakhi, W. Chen, D. K. Cha and H. N. Alshareef, Nano Lett., 2012, 12, 2559–2567.
39. C. Z. Yuan, L. Yang, L. R. Hou, L. F. Shen, X. G. Zhang and X. W. Lou, Energy Environ. Sci., 2012, 5, 7883–7887.
40. M. Hamdani, R. N. Singh and P. Chartier, Int. J. Electrochem. Sci., 2010, 5, 556–577.
41. W. Pan and J. H. Gong, Key Eng. Mater., 2007, 336–338, 2134–2137.
42. C. C. Hu, W. C. Chen and K. H. Chang, J. Electrochem. Soc., 2004, 151, A281–A290.
43. A. Adeyemoa, G. Hunterb and P. K. Dutta, Sens. Actuators, B, 2011, 152, 307–315.
44. K. Liang, X. Z. Tang and W. C. Hu, J. Mater. Chem., 2012, 22, 11062–11067.
45. V. V. Pokropivny and V. V. Skorokhod, Mater. Sci. Eng., C, 2007, 27, 990–993.
46. J. N. Tiwari, R. N. Tiwari and K. S. Kim, Prog. Mater. Sci., 2012, 57, 724–803.
47. Y. Zou and B. X. Han, Energy Fuels, 2001, 15, 1383–1386.
48. G. P. Meisner and Q. Y. Hu, Nanotechnology, 2009, 20, 204023.
49. Y. Zhu, S. Murali, W. Cai, X. Li, J. W. Suk, J. R. Potts and R. S. Ruoff, Adv. Mater., 2010, 22, 3906–3924.
50. O. Barbieri, M. Hahn, A. Herzog and R. Kötz, Carbon, 2005, 43, 1303–1310.
51. K. Lota, A. Sierczynska and G. Lota, Int. J. Electrochem., 2011, 2011, 321473.
52. C. Lin, J. A. Ritter and B. N. Popov, J. Electrochem. Soc., 1999, 146, 3155–3160.
53. A. B. Yuan and Q. L. Zhang, Electrochem. Commun., 2006, 8, 1173–1178.
54. C. Liu, F. Li, L. P. Ma and H. M. Cheng, Adv. Mater, 2010, 22, E28–E62.
55. J. H. Bae, J. H. Han and T. D. Chung, Phys. Chem. Chem. Phys., 2012, 14, 448–463.
56. H. Yamada, H. Nakamura, F. Nakahara, I. Moriguchi and T. Kudo, J. Phys. Chem. C, 2007, 111, 227–233.
57. E. Raymundo-PiMero, K. Kierzek, J. Machnikowski and F. BKguin, Carbon, 2006, 44, 2498–2507.
58. G. Gryglewicza, J. Machnikowski, E. Lorenc-Grabowska, G. Lota and E. Frackowiak, Electrochim. Acta, 2005, 50, 1197–1206.
59. H. Q. Wang, Z. S. Li, Y. G. Huang, Q. Y. Li and X. Y. Wang, J. Mater. Chem., 2010, 20, 3883–3889.
60. D. W. Wang, F. Li, M. Liu, G. Q. Lu and H. M. Cheng, Angew. Chem., Int. Ed., 2008, 47, 373–376.
61. X. P. Dong, W. H. Shen, J. L. Gu, L. M. Xiong, Y. F. Zhu, H. Li and J. L. Shi, J. Phys. Chem. B, 2006, 110, 6015.
62. M. W. Xua, W. Jia, S. J. Bao, Z. Su and B. Dong, Electrochim. Acta, 2010, 55, 5117–5122.
63. H. Jiang, T. Sun, C. Z. Li and J. Ma, RSC Adv., 2011, 1, 954–957.
64. Z. B. Lei, J. T. Zhang and X. S. Zhao, J. Mater. Chem., 2012, 22, 153–160.
65. Y. Wang, S. F. Yu, C. Y. Sun, T. J. Zhu and H. Y. Yang, J. Mater. Chem., 2012, 22, 17584–17588.
66. J. F. Ni, L. X. Yang, H. B. Wang and L. J. Gao, J. Solid State Electrochem., 2012, 12, 2791–2796.
67. J. B. Baxter and E. S. Aydil, Appl. Phys. Lett., 2005, 86, 053114.
68. M. Zhi, N. Mariani, R. Gemmen, K. Gerdes and N. Q. Wu, Energy Environ. Sci., 2011, 4, 417–420.
69. M. Zhi, S. Lee, N. Miller, N. H. Menzler and N. Q. Wu, Energy Environ. Sci., 2012, 5, 7066–7071.
70. M. Yang, J. Wang, H. Li, J.-G. Zheng and N. Q. Wu, Nanotechnology, 2008, 19, 075502.
71. N. Q. Wu, J. Wang, D. Tafen, H. Wang, J.-G. Zheng, J. P. Lewis, X. Liu, S. S. Leonard and A. Manivannan, J. Am. Chem. Soc., 2010, 132, 6679–6685.
72. D. Wang, H. Zhao, N. Q. Wu, A. El Khakani and D. Ma, J. Phys. Chem. Lett., 2010, 1, 1030–1035.
73. H. Pan, J. Y. Li and Y. P. Feng, Nanoscale Res. Lett., 2010, 5, 654–668.
74. R. R. Bi, X. L. Wu, F. F. Cao, L. Y. Jiang, Y. G. Guo and L. J. Wan, J. Phys. Chem. C, 2010, 114, 2448–2451.
75. R. Vellacheri, V. K. Pillai and S. Kurungot, Nanoscale, 2012, 4, 890–896.
76. J. B. Mu, B. Chen, Z. C. Guo, M. Y. Zhang, Z. Y. Zhang, P. Zhang, C. L. Shao and Y. C. Liu, Nanoscale, 2011, 3, 5034–5040.
77. J. G. Wang, Y. Yang, Z. H. Huang and F. Y. Kang, Mater. Lett., 2012, 72, 18–21.
78. R. B. Rakhi, D. Y. Cha, W. Chen and H. N. Alshareef, J. Phys. Chem. C, 2011, 115, 14392–14399.
79. P. C. Chen, G. Shen, S. Sukcharoenchoke and C. Zhou, Appl. Phys. Lett., 2009, 94, 043113.
80. J. H. Park, J. M. Ko and O. O. Park, J. Electrochem. Soc., 2003, 150, A864–A867.
81. J. Yan, Z. J. Fan, T. Wei, J. Cheng, B. Shao, K. Wang, L. P. Song and M. L. Zhang, J. Power Sources, 2009, 194, 1202–1207.
82. R. Amade, E. Jover, B. Caglar, T. Mutlu and E. Bertran, J. Power Sources, 2011, 196, 5779–5783.
83. M. Jayalakshmi, M. M. Rao, N. Venugopal and K. B. Kim, J. Power Sources, 2007, 166, 578–583.
84. J. Y. Lee, K. Liang, K. H. An and Y. H. Lee, Synth. Met., 2005, 150, 153–157.
85. Z. Chen, Y. C. Qin, D. Weng, Q. F. Xiao, Y. T. Peng, X. L. Wang, H. X. Li, F. Wei and Y. F. Lu, Adv. Funct. Mater., 2009, 19, 3420–3426.
86. X. B. Jin, W. Z. Zhou, S. W. Zhang and G. Z. Chen, Small, 2007, 3, 1513–1517.
87. W. Chen, Z. L. Fan, L. Gu, X. H. Bao and C. L. Wang, Chem. Commun., 2010, 46, 3905–3907.
88. X. Zhang, W. Shi, J. Zhu, D. J. Kharistal, W. Zhao, B. S. Lalia, H. H. Hng and Q. Yan, ACS Nano, 2011, 5, 2013–2019.
89. Y. Hou, Y. W. Cheng, T. Hobson and J. Liu, ACS Nano, 2012, 6, 5404–5412.
90. L. H. Su, X. G. Zhang, C. Z. Yuan and B. Gao, J. Electrochem. Soc., 2008, 155, A110–A114.
91. P. C. Chen, H. T. Chen, J. Qiu and C. W. Zhou, Nano Res., 2010, 3, 594–603.
92. C. Kim, B. T. N. Ngoc, K. S. Yang, M. Kojima, Y. A. Kim, Y. J. Kim, M. Endo and S. C. Yang, Adv. Mater., 2007, 19, 2341–2346.
93. C. Kim, Y. I. Jeong, B. T. N. Ngoc, K. S. Yang, M. Kojima, Y. A. Kim, M. Endo and J. Lee, Small, 2007, 3, 91–95.
94. A. Ghosh, E. J. Ra, M. H. Jin, H. K. Jeong, T. H. Kim, C. Biswa and Y. H. Lee, Adv. Funct. Mater., 2011, 13, 2541–2547.
95. J. B. Mu, B. Chen, Z. C. Guo, M. Y. Zhang, Z. Y. Zhang, C. L. Shao and Y. C. Liu, J. Colloid Interface Sci., 2011, 356, 706–712.
96. C. M. Chuang, C. W. Huang, H. Teng and J. M. Ting, Compos. Sci. Technol., 2012, 72, 1524–1529.
97. J. G. Wang, Y. Yang, Z. H. Huang and F. Y. Kang, Electrochim. Acta, 2011, 56, 9240.
98. M. Zhi, A. Manivannan, F. Meng and N. Q. Wu, J. Power Sources, 2012, 208, 345–353.
99. K. S. Yang, Y. W. Joo, T. Kim, J. H. Kim and W. J. Lee, Proceedings of the 3^rd WSEAS/IASME Int. Conf. on Electroscience and Technology for Naval Engineering, Greece, July 14–16, 2006, pp. 6–10.
100. H. S. Choi, T. H. Kim, J. H. Im and C. R. Park, Nanotechnology, 2011, 22, 405402.
101. S. W. Lee, J. Y. Kim, S. Chen, P. T. Hammond and Y. Shao-Horn, ACS Nano, 2010, 4, 3889–3896.
102. L. Yang, S. Cheng, Y. Ding, X. B. Zhu, Z. L. Wang and M. L. Liu, Nano Lett., 2012, 12, 321–325.
103. L. H. Bao, J. F. Zang and X. D. Li, Nano Lett., 2011, 11, 1215–1220.
104. X. H. Lu, T. Zhai, X. H. Zhang, Y. Q. Shen, L. Y. Yuan, B. Hu, L. Gong, J. Chen, Y. H. Gao, J. Zhou, Y. X. Tong and Z. L. Wang, Adv. Mater., 2012, 24, 938–944.
105. J. Li, W. Zhao, F. Huang, A. Manivannan and N. Q. Wu, Nanoscale, 2011, 3, 5103–5109.
106. R. F. Zhou, C. Z. Meng, F. Zhu, Q. Q. Li, C. H. Liu, S. S. Fan and K. L. Jiang, Nanotechnology, 2010, 21, 345701.
107. L. Basirićo and G. Lanzara, Nanotechnology, 2012, 23, 305401.
108. A. L. M. Reddy, M. M. Shaijumon, S. R. Gowda and P. M. Ajayan, J. Phys. Chem. C, 2010, 114, 658–663.
109. R. Amade, E. Jover, B. Caglar, T. Mutlu and E. Bertran, J. Power Sources, 2011, 196, 5779–5783.
110. H. Xia, Y. Wang, J. Y. Lin and L. Lu, Nanoscale Res. Lett., 2012, 7, 33.
111. S. F. Pei and H. M. Cheng, Carbon, 2012, 50, 3210–3228.
112. B. Xu, S. F. Yue, Z. Y. Sui, X. T. Zhang, S. S. Hou, G. P. Cao and Y. P. Yang, Energy Environ. Sci., 2011, 4, 2826–2830.
113. Y. Zhu, S. Murali, M. D. Stoller, K. J. Ganesh, W. Cai, P. J. Ferreira, A. Pirkle, R. M. Wallace, K. A. Cychosz, M. Thommes, D. Su, E. A. Stach and R. S. Ruoff, Science, 2011, 332, 1537–1541.
114. X. H. Xia, J. P. Tu, Y. J. Mai, R. Chen, X. L. Wang, C. D. Gu and X. B. Zhao, Chem.–Eur. J., 2011, 17, 10898–10905.
115. M. S. Wu, Y. P. Lin, C. H. Lin and J. T. Lee, J. Mater. Chem., 2012, 22, 2442–2448.
116. Y. Q. Zhao, D. D. Zhao, P. Y. Tang, Y. W. Wang, C. L. Xu and H. L. Li, Mater. Lett., 2012, 76, 127–130.
117. S. Chen, J. W. Zhu, X. D. Wu, Q. F. Han and X. Wang, ACS Nano, 2010, 4, 2822–2830.
118. C. Y. Ge, Z. H. Hou, B. H. He, F. Y. Zeng, J. G. Cao, Y. M. Liu and Y. F. Kuang, J. Sol-Gel Sci. Technol., 2012, 63, 146–152.
119. G. Y. He, J. H. Li, H. Q. Chen, J. Shi, X. Q. Sun, S. Chen and X. Wang, Mater. Lett., 2012, 82, 61–63.
120. B. Zhao, J. S. Song, P. Liu, W. W. Xu, T. Fang, Z. Jiao, H. J. Zhang and Y. Jiang, J. Mater. Chem., 2011, 21, 18792–18798.
121. R. R. Jiang, T. Huang, J. L. Liu, J. H. Zhuang and A. S. Yu, Electrochim. Acta, 2009, 54, 3047–3052.
122. A. Chidembo, S. H. Aboutalebi, K. Konstantinov, M. Salari, B. Winton, S. A. Yamini, I. P. Nevirkovets and H. K. Liu, Energy Environ. Sci., 2012, 5, 5236–5240.
123. W. W. Zhou, J. X. Zhu, C. W. Cheng, J. P. Liu, H. P. Yang, C. X. Cong, C. Guan, X. T. Jia, H. J. Fan, Q. Y. Yan, C. M. Li and T. Yu, Energy Environ. Sci., 2011, 4, 4954–4961.
124. J. Yan, Z. J. Fan, T. Wei, W. Z. Qian, M. L. Zhang and F. Wei, Carbon, 2010, 48, 3825–3833.
125. H. J. Huang and X. Wang, Nanoscale, 2011, 3, 3185–3191.
126. Z. S. Wu, D. W. Wang, W. C. Ren, J. P. Zhao, G. M. Zhou, F. Li and H. M. Cheng, Adv. Funct. Mater., 2010, 20, 3595–3602.
127. Z. S. Wu, W. C. Ren, D. W. Wang, F. Li, B. L. Liu and H. M. Cheng, ACS Nano, 2010, 4, 5835–5842.
128. Z. S. Wu, W. C. Ren, L. Wen, L. B. Gao, J. P. Zhao, Z. P. Chen, G. M. Zhou, F. Li and H. M. Cheng, ACS Nano, 2010, 4, 3187–3194.
129. Z. S. Wu, G. M. Zhou, L. C. Yin, W. C. Ren, F. Li and H. M. Cheng, Nano Energy, 2012, 1, 107–131.
130. A. K. Mishra and S. Ramaprabhu, J. Phys. Chem. C, 2011, 115, 14006–14013.
131. Z. J. Fan, J. Yan, T. Wei, L. J. Zhi, G. Q. Ning, T. Y. Li and F. Wei, Adv. Funct. Mater., 2011, 21, 2366–2375.
132. Z. P. Li, Y. J. Mi, X. H. Liu, S. Liu, S. R. Yang and J. Q. Wang, J. Mater. Chem., 2011, 21, 14706–14711.
133. X. Zhao, L. L. Zhang, S. Murali, M. D. Stoller, Q. H. Zhang, Y. W. Zhu and R. S. Ruoff, ACS Nano, 2012, 6, 5404–5412.
134. C. Xiang, M. Li, M. Zhi, A. Manivannan and N. Q. Wu, J. Mater. Chem., 2012, 22, 19161–19167.
135. Y. W. Cheng, S. T. Lu, H. B. Zhang, C. V. Varanasi and J. Liu, Nano Lett., 2012, 12, 4206–4211.
136. M. B. Sassin, A. N. Mansour, K. A. Pettigrew, D. R. Rolison and J. W. Long, ACS Nano, 2010, 4, 4505–4514.
137. A. E. Fischer, K. A. Pettigrew, D. R. Rolison, R. M. Stroud and J. W. Long, Nano Lett., 2007, 7, 281–286.
138. Z. P. Chen, W. C. Ren, L. B. Gao, B. L. Liu, S. F. Pei and H.-M. Cheng, Nat. Mater., 2011, 10, 424–428.
139. X. C. Dong, H. Xu, X. W. Wang, Y. X. Huang, M. B. Chan-Park, H. Zhang, L. H. Wang, W. Huang and P. Chen, ACS Nano, 2012, 6, 3206–3213.
140. X. C. Dong, Y. F. Cao, J. Wang, M. B. Chan-Park, L. h. Wang, W. Huang and P. Chen, RSC Adv., 2012, 2, 4364–4369.
141. L. B. Hu, W. Chen, X. Xie, N. Liu, Y. Yang, H. Wu, Y. Yao, M. Pasta, H. N. Alshareef and Y. Cui, ACS Nano, 2011, 5, 8904–8913.
142. B. G. Choi, M. Yang, W. H. Hong, J. W. Choi and Y. S. Huh, ACS Nano, 2012, 6, 4020–4028.
143. W. Chen, R. B. Rakhi, L. B. Hu, X. Xie, Y. Cui and H. N. Alshareef, Nano Lett., 2011, 11, 5165–5172.
144. http://gigaom.com/cleantech/supercapacitor-market-to-surge-to-877m-by-2014/, accessed 24 July 2012.
145. http://www.marketresearchmedia.com/2012/03/16/ultracapacitor-market/, accessed 24 July 2012.
146. http://www.reportlinker.com/p0363454/Supercapacitors-Technology-Developments-and-Global-Markets.html, accessed 24 July 2012.
147. D. W. Wang, F. Li and H. M. Cheng, J. Power Sources, 2008, 185, 1563–1568.
148. Y. Gogotsi and P. Simon, Science, 2011, 18, 917–918.

---