50–100 μm-thick pseudocapacitive electrodes of MnO2 nanoparticles uniformly electrodeposited in carbon nanotube papers

To overcome the tradeoff between the gravimetric capacitance and loading density of pseudocapacitive MnO2, we electrodeposited MnO2 nanoparticles on the carbon nanotube (CNT) surfaces in 18–37 mmthick self-supporting CNT papers. We examined the electrodeposition conditions including constant potential, constant current, and potential pulses, and obtained MnO2–CNT hybrid electrodes containing MnO2 nanoparticles uniformly deposited at 60–90 wt% with an expanded CNT matrix. The MnO2–CNT hybrid electrode with a thickness of 62 mm, density of 1.09 g cm , areal mass of 6.75 mg cm , and 82 wt% MnO2 load showed a total gravimetric capacitance of 120 and 51 Ftotal gelectrode , volumetric capacitance of 131 and 56 Ftotal cm 3 and areal capacitance of 0.81 and 0.34 Ftotal cm 2 at scan rates of 2 and 200 mV s , respectively. The large thickness, moderately high mass density, and fairly conductive CNT matrix realized such high values of gravimetric, areal and volumetric capacitances that are important for practical devices.


Introduction
In recent years, owing to the growing demand for normalizing power uctuations of solar and wind power generation and energy recovery in automobiles, electrochemical capacitors have attracted increasing attention as high power electrochemical energy storage devices. Activated carbon (AC), which has a huge specic surface area, is generally used as an active material for electric double layer capacitors, but there is limited room to enhance the capacitance by controlling the pore size distribution. To enhance the energy densities, pseudocapacitors using redox reactions of active materials such as metal oxides and conductive polymers, have been extensively studied. Ru compounds have been studied most extensively as active materials because they have high conductivity, proton mobility, and high capacitance in a wide potential window. [1][2][3] It was reported that amorphous ruthenium oxide material calcined at 150 C showed a maximum capacitance of 720 F g À1 in sulfuric acid electrolyte, 1 and crystalline ruthenium oxide calcined at 200 C showed a maximum capacitance of 710 F g À1 in KOH electrolyte. 4 However, because ruthenium is a rare metal with a high cost and toxicity, less expensive materials have been studied. 5 MnO 2 , with a high theoretical capacity of 1100 C g À1 , is a promising candidate as an active material for commercial use because it is abundant, inexpensive, and environmentally friendly. 6 However, it is necessary to use ne MnO 2 particles with a high specic surface area because the conductivity of MnO 2 is low and the redox reaction occurs preferentially on or near the surface during charge and discharge (eqn (1)). 7 MnOOA % MnO 2 + A + + e À A very high gravimetric capacitance of 1200 F g MnO 2 À1 at 3 mV s À1 was conrmed, 8 but for a very thin layer on indium tin oxide-coated glass substrate, resulting in a very small areal capacitance of 6.8 Â 10 À4 F cm À2 . Conventional capacitor electrodes are built on current collectors of metal foils on which active materials are attached by binders with conductive llers (Fig. 1a). Although high gravimetric capacitance values have been reported for thin MnO 2 electrodes (Fig. 1b), the capacitance per device is small because of the much larger mass of the current collector and separator compared with the active materials. In contrast, the gravimetric capacitance values of thick MnO 2 electrodes are small (Fig. 1c). To utilize ne MnO 2 particles at higher loads, their composites with carbon materials including carbon nanotubes (CNTs) or conductive polymers have also been studied. [9][10][11][12][13][14] CNTs are a conductive material with high aspect ratios and specic surface areas, high tensile strength with exibility, and high thermal and chemical stability, and are expected to be applied in various electric and electronic devices. We have developed a uidized bed chemical vapour deposition method that enables semi-continuous production of 200-400 mm-long few-wall CNTs with an average diameter of 6.5-8 nm, carbon purity of over 99 wt%, and a specic surface area of 400-440 m 2 g À1 . 15,16 Such CNTs can form self-supporting papers with arbitrary thicknesses through a simple dispersion-ltration process without any binding additives. The CNT papers have a much larger surface area (2000-6000 cm 2 for 20-60 mm-thick, 1 cm 2 CNT paper) than metal foils (2 cm 2 surface area for 1 cm 2 foil) and can capture various active materials inside their nanometre-sized pores at loads much larger than their own mass. Positive electrodes of oxidized CNT papers for lithium batteries, 17 positive electrodes of polymerized pyrene derivatives held in CNT papers for pseudocapacitors, 18 and biomassderived carbonaceous positive electrodes for lithium cells 19 have been realized. We have also realized a self-supporting paper of AC (90 wt%) held by CNTs (10 wt%), which showed a three-times higher specic capacitance than the pure singlewall CNT electrodes 20 and worked as a capacitor electrode not only in full contact but also in line contact to a metal mesh. 21 The practical use of MnO 2 as an active material in electrochemical capacitors requires: (1) deposition of ne MnO 2 particles at a high density, (2) building of conductive paths to all MnO 2 particles, and (3) electrodes that are sufficiently thick (several tens of micrometres). In this study, we aimed to fabricate electrodes using MnO 2 -CNT hybrids that have a high capacitance per mass and area of the electrode (Fig. 1d). Various methods such as hydrothermal synthesis, 22 sol-gel method, 23 electrostatic spray deposition, 24 electrophoretic deposition, 25 anodic oxidation, 26 and cathodic reduction 27 have been reported for making MnO 2 . Among them, electrodeposition 26,27 is attractive because it forms conductive paths to every MnO 2 particle. Uniform deposition of nickel oxide particles in vertically aligned CNT forests have been reported although the volumetric capacitance was small (1.26 F cm À3 ). 28 Therefore, we electrodeposited MnO 2 directly on CNT papers by anodically oxidizing Mn 2+ ions using the CNT papers as a working electrode in a MnSO 4 aqueous solution. A high potential is needed to deposit small MnO 2 particles at a high density by enhancing their nucleation. But at the same time, high potential leads to a high deposition rate and diffusion-limited deposition, resulting in the preferential deposition of MnO 2 on the outer surface of thick CNT papers. We examined electrodeposition at a constant potential (CP), at a constant current (CC), or by applying high potential pulses (hereaer "Pulse"), and analysed micro-and macroscopic structural changes and chargedischarge characteristics. The thick MnO 2 -CNT hybrids reported here have a higher areal capacitance than any previous report on MnO 2 (Fig. 1e), expect for the very high value of 2.8 F cm À2 recorded at a very small scan rate of 0.05 mV s À1 for MnO 2 -CNT-textile with a high MnO 2 load of 8.3 mg cm À2 . 14 Experimental Preparation of the MnO 2 -CNT hybrid papers Sub-millimetre-long few-wall CNTs (3-10 mg) synthesized by uidized bed chemical vapour deposition 16 were mixed with a 0.5 wt% sodium dodecylbenzene sulfonate aqueous solution (30-100 mL) and dispersed by ultrasonication (bath-type, 100 W, 30-100 min). The dispersed CNT solution was vacuum-ltrated on a membrane lter (polytetrauoroethylene, pore size of 0.5 mm), and the CNT lm on the membrane lter was washed by hot distilled water (80 C), dried at 90 C for 2 h under air, and then separated from the membrane lter using tweezers. The areal mass of the CNTs was controlled at 0.6-1.4 mg cm À2 , yielding self-supporting CNT papers with a mass density of 0.25-0.48 g cm À3 and a thickness of 18-37 mm. MnO 2 -CNT hybrids were then prepared by electrodeposition using the condition reported by Jin et al. 29 MnO 2 were deposited on/in the CNT paper immersed in 0.6 M MnSO 4 /0.8 M H 2 SO 4 aqueous electrolyte using the CNT paper as a working electrode, a graphite sheet as a counter electrode, and Ag/AgCl electrode (in 3 M NaCl aqueous solution) as a reference electrode. The electrodeposition was conducted either at CP, CC, or Pulse. The obtained electrodes were dried at 90 C for 2 h under air.

Characterization of the MnO 2 -CNT hybrids
The microstructure of the MnO 2 -CNT hybrids was characterized using a scanning electron microscope (SEM, Hitachi S-4800, Tokyo, Japan). The composition of the MnO 2 -CNT hybrids was evaluated using energy-dispersive X-ray spectroscopy (EDAX Genesis, AMETEK, Elancourt, France) equipped to SEM and mass change of the CNT papers. Their crystal structure was analysed using X-ray diffraction (Ultima III system, Rigaku, Akishima, Japan). Three-electrode cells were used to run cyclic voltammetry (CV), electrochemical impedance spectroscopy, and galvanostatic charge-discharge tests with a VMP3 potentiostat (Bio-Logic, Grenoble, France). The three-electrode cell consisted of a hybrid electrode as the working electrode, an AC-CNT hybrid lm (90 wt% AC with 10 wt% CNTs, $200 mm in thickness) 21 as the counter electrode and an Ag/AgCl electrode (in 3 M NaCl aqueous solution) as the reference electrode with 1 M Na 2 SO 4 aqueous solution as the electrolyte. Structure and electrochemical performance are summarized for all and representative samples in ESI Tables S1 and S2, respectively. †

Results and discussion
Rate process of electrodeposition and microstructure of the resulting MnO 2 in CNT papers ited MnO 2 ), respectively. The deposition rate of MnO 2 is controlled by the potential of the CNT paper. We will discuss each aspect separately in detail below. Fig. 2a and b shows the current density proles during electrodeposition at CP of 1.15, 1.20, and 1.30 V against the time of electrodeposition and the integrated electric charge, respectively. At all three potentials, a large current density was initially observed just aer applying the electric potential, which quickly decayed in the rst several seconds (Fig. 2a). The current density at 1.15 V became very small. Conversely, the current density increased for $5 and $2 min, became saturated, and decreased aer $16 and $5 min at 1.20 and 1.30 V, respectively. Fig. 2c-i shows the plan-view SEM images of the surfaces of the CNT papers before (Fig. 2c) and aer ( Fig. 2d-i) electrodeposition. Flower-shaped particles formed inside the CNT papers (Fig. 2d, e and g) with increasing size and number density with time ( Fig. 2d vs. 2e), and dense layers formed on the surface of the CNT papers aer the decrease in the current density (Fig. 2f, h and i). These results suggest that the initial large current density aer several seconds corresponds to the charge accumulation by the electric double layer on the CNT surface. The subsequent increase in current density corresponds to the increasing surface area of MnO 2 by nucleation and growth of MnO 2 particles and the nal decrease in current density corresponds to the decreasing surface area and increasing electric resistance of the dense MnO 2 layer. The increase in current density with the nucleation and growth of MnO 2 particles indicates that the MnO 2 deposition is preferred on the MnO 2 surface rather than on the CNT surface. The interior of the CNT paper was lled with MnO 2 particles uniformly from top to the bottom of the paper (Fig. 2j-m and S2a †), indicating that the Mn 2+ diffuses into the CNT paper until the surface of the CNT paper is covered with a dense MnO 2 layer. The MnO 2 deposition accompanied the expansion of the CNT paper matrix, from 30 to 61-63 mm in thickness with the areal mass increase from 1.23 to 6.75 mg cm À2 and density increase from 0.41 to 1.09 g cm À3 during 10 min deposition at 1.20 V (Fig. 2j). As the deposition proceeds, the tortuosity of the CNT paper increases whereas the porosity decreases, thereby inhibiting the diffusion of Mn 2+ and making the deposition of MnO 2 preferential at the exterior of the paper and resulting in a dense MnO 2 layer. Fig. 2b shows the change of current density with the integrated electric charge (corresponding to the amount of deposited MnO 2 ). The current density decreased at a smaller integrated electric charge of $10 C cm À2 at 1.30 V than at $17 C cm À2 at 1.20 V, showing that less MnO 2 nanoparticles can be deposited inside the CNT paper at higher potential owing to the accelerated formation of a dense layer. Comparison of Fig. 2e and g shows that smaller MnO 2 particles were deposited at a higher number density at potentials of 1.30 V than at 1.20 V, indicating that a higher potential enhances the nucleation more than the growth of MnO 2 particles. To realize high areal capacitance of the hybrid electrode, ne particles must be deposited at a high number density. Here arises a difficult tradeoff; a higher potential is needed to enhance the nucleation density of MnO 2 particles, which at the same time causes concentration distribution of Mn 2+ ions within the CNT papers, resulting in a dense MnO 2 layer on the exterior of the CNT papers.
To electrodeposit MnO 2 particles in CNT papers uniformly at a high density, we investigated the control of the reaction rate by electrodeposition with CC. Fig. 3a and b shows potential proles during electrodeposition at CC of 2 and 20 mA cm À2 , respectively. At 2 mA cm À2 , the potential changed slightly between 1.15 and 1.19 V for 180 min; it initially increased and then decreased and remained at a constant value. At 20 mA cm À2 , the potential showed a similar change at $1.2 V for 10 min (less than 1/10 of that for 2 mA cm À2 ) and then increased abruptly to $1.5 V at $12 min. Fig. 3c shows the change in current density with integrated electric charge. It clearly shows that the MnO 2 deposition proceeded at a higher potential with an abrupt potential increase with the larger current density of 20 mA cm À2 . Fig. 3d-g shows plan-view SEM images of the surface of the MnO 2 -CNT hybrids aer electrodeposition with CC of 2 mA cm À2 . Particles grew continuously without forming a layer on the CNT paper surface even aer a long electrodeposition time at 2 mA cm À2 . Conversely, at 20 mA cm À2 , small MnO 2 particles formed at a higher density (Fig. 3h) and a dense layer formed on the surface of the CNT paper aer deposition for 18 min (Fig. 3i). The initial increase, decrease, plateau, and abrupt increase (only with 20 mA cm À2 ) in the potential correspond to the charge accumulation by double-layer capacitance, the increased reaction area as a result of the formation of the MnO 2 particles, the sufficient surface area for reaction and interspace for diffusion of the growing MnO 2 particles, and the inhibited diffusion owing to the dense layer formation, respectively. It can be said that Mn 2+ ions can diffuse uniformly even with a large amount of deposited MnO 2 (22 C cm À2 and 87 wt%) when MnO 2 is deposited slowly at 2 mA cm À2 . Fig. 3j-m shows the cross-section of the lm aer electrodeposition at 20 mA cm À2 for 9 min (the same lm as Fig. 3h). It indicates that MnO 2 particles were uniformly electrodeposited inside the CNT paper. Increased current density (20 mA cm À2 ) and, thus, the overpotential ($1.2 V) enhances the nucleation of MnO 2 particles, initially making small MnO 2 particles at high density ( Fig. 3h and j). However, the potential increases at $12 min owing to the formation of a surface layer (Fig. 3i); the accessible surface area of MnO 2 for Mn 2+ ions decreases and the overpotential increases to balance the constant current.
To electrodeposit ne MnO 2 particles at a high density, next we investigated intermittently applying high potential pulses. We expected the nucleation of MnO 2 particles to be accelerated by the high potential, and a uniform deposition of ne MnO 2 particles to be achieved by allowing the Mn 2+ ions to diffuse into the CNT paper uniformly during the rests between the pulses. Fig. 4a and b shows typical time proles of the applied potential and resulting current density. Fig. 4c-f shows the SEM images of the electrode surface aer pulse-deposition by repeatedly applying potential pulses of 2.0 V for 0.5 s at 10 s intervals (15-60 times). Small MnO 2 particles covered the CNT bundles aer 15 pulses, and CNT-MnO 2 core-shell wires became thicker aer more pulses. SEM images and EDS elemental mapping of the cross-section (Fig. 4g-j and S2c †) show that MnO 2 was uniformly electrodeposited inside the CNT paper. In the case of CC, the small current and thus the low potential nucleated MnO 2 particles at a low density resulting in large particles (Fig. 3). Conversely, the high potential pulse-deposition enhanced the nucleation of MnO 2 particles owing to a high overpotential, resulting in small MnO 2 particles at a very high density covering the CNT surfaces.

Electrochemical performances of MnO 2 -CNT hybrid electrodes
The electrochemical performance of the MnO 2 -CNT hybrid electrodes was examined using a three-electrode cell with 1 M Na 2 SO 4 aqueous solution at a potential range of 0-0.8 V. Fig. S4 † shows typical CV curves of the MnO 2 -CNT electrode electrodeposited by CP at 1.20 V for 10 min. The CV curves show rectangular shapes at low scan rates, which are typical of capacitor electrodes and suggest that sufficient electric conduction pathways are provided by the CNT matrix. At higher scan rates $100 mV s À1 , their shapes change to parallelogram, resulting in decreasing capacitances with scan rates. Fig. 5 shows the rate performances of the MnO 2 -CNT hybrid electrodes electrodeposited by CP (Fig. 5a-c), CC (Fig. 5d-f), and Pulse ( Fig. 5g-i). The MnO 2 -based capacitances (Fig. 5c, f and i) were estimated by subtracting the capacitance of the CNT paper from the total capacitance of the electrodes (Fig. 5b, e and h). Because the CNT papers lose their exposed surface as MnO 2 is deposited, the contribution of the CNTs is overestimated and the actual MnO 2 -based capacitance should be higher.
The total areal capacitance at 1 mV s À1 of the MnO 2 -CNT hybrids prepared by CP at 1.20 V increased signicantly with deposition time, from 0.02 F cm À2 for pristine CNTs to 1.08 F cm À2 aer 30 min deposition (Fig. 5a). The total gravimetric capacitance at 1 mV s À1 increased from 22 F total g electrode À1 for pristine CNTs to 124 F total g electrode À1 aer 10 min deposition, but it changed little at a low scan rate and decreased at a high scan rate aer a 30 min deposition (Fig. 5b). The MnO 2 -based capacitance was largest aer a 2 min deposition (170 and 131 F MnO 2 g MnO 2 À1 at 1 and 200 mV s À1 , respectively) and decreased monotonically aer deposition for 2 to 30 min (Fig. 5c) owing to the increased particle size of MnO 2 (Fig. 2). Previous reports mainly discuss the gravimetric capacitance of MnO 2 , for which the largest value can be obtained for a small MnO 2 load. However, the total capacitance of an electrode is more important for practical applications, and the largest gravimetric and areal capacitances can be obtained for a moderate (10 min) and large (30 min) load of MnO 2 , respectively. The MnO 2 -CNT hybrid deposited at 1.20 V for 10 min had a large thickness of 62 mm, high areal mass of 6.75 mg cm À2 and high mass density of 1.09 g cm À3 , yielding a moderate total gravimetric capacitance of 120 and 51 F total g electrode À1 , high volumetric capacitance of 131 and 56 F total cm À3 and high areal capacitance of 0.81 and 0.34 F total cm À2 at scan rates of 2 and 200 mV s À1 , respectively. The changes in the capacitance values for the hybrids prepared by CP at 1.30 V were different from those prepared by CP at 1.20 V. The total areal capacitances were largest (0.48 F cm À2 at 1 mV s À1 ) aer a 2 min deposition and decreased aer further deposition for 10 and 30 min (Fig. 5a). Additional deposition of over 10 min resulted in continuous lm formation on the electrode surface ( Fig. 2h and i), which inhibited the use of MnO 2 particles inside the CNT papers. A similar change was observed for the total and MnO 2 -based gravimetric capacitances (Fig. 5b and c).
For the MnO 2 -CNT hybrids prepared by CC at 2 mA cm À2 , the total areal capacitance increased with the deposition time (and thus MnO 2 load) (Fig. 5d). The total gravimetric capacitance increased with deposition time, showing a maximum at 90 min and decreasing at 180 min (Fig. 5e). The MnO 2 -based gravimetric capacitance decreased monotonically with increasing deposition time owing to the monotonic increase in the particle size of MnO 2 (Fig. 3d-g). The capacitances of the materials obtained by CC at 20 mA cm À2 were signicantly larger than those obtained by CC at 2 mA cm À2 (Fig. 5d-f). The total electric charge for electrodeposition was the same between 2 mA cm À2 for 90 min and 20 mA cm À2 for 9 min and between 2 mA cm À2 for 180 min and 20 mA cm À2 for 18 min, yielding  Fig. S2c. † similar content of MnO 2 in the hybrid lms (75 and 81 wt% for the former and 87 and 87 wt% for the latter). The signicantly high capacitance of the hybrids obtained by CC at 20 mA cm À2 was attributed to the smaller MnO 2 particles at higher density (Fig. 3h) although a continuous lm covers the electrode surface aer deposition for 18 min (Fig. 3i). When we carefully compare the two hybrids at 20 mA cm À2 , both the total and MnO 2 -based capacitances were higher for the hybrid deposited for 9 min than that for 18 min at high scan rates ($50 mV s À1 ). The capacitance tends to be limited by the ionic diffusion in electrolyte because more ions need to diffuse through the pores of smaller volume and/or larger tortuosity in the hybrids with higher MnO 2 content. The best conditions change for the targeted capacitances (i.e., areal or gravimetric), and in view of gravimetric capacitance, the best hybrid obtained by CC at 20 mA cm À2 for 9 min (Fig. 5e) showed a similar performance with the best hybrid obtained by CP at 1.20 V for 10 min (Fig. 5b). This result is reasonable if we consider the similar electrodeposition conditions between the former (81 wt% MnO 2 at 20 mA cm À2 and $1.2 V for 9 min, Fig. 3b and h) and the latter (82 wt% MnO 2 at 18-19 mA cm À2 and 1.20 V for 10 min, Fig. 2a and e).
For the MnO 2 -CNT hybrids prepared by applying Pulse (Fig. 5g), the total areal capacitance also increased with the number of pulses and thus the MnO 2 load, but did not increase as much compared with the hybrids prepared by CP and CC. Their total gravimetric capacitances were fairly high for all of the conditions between 15 and 60 pulses (Fig. 5h); 96-105 and 44-65 F total g electrode À1 at 1 and 200 mV s À1 , respectively. Their MnO 2 -based gravimetric capacitance showed some decrease with increasing number of pulses; from 154 to 109 and from 100 to 49 F MnO 2 g MnO 2 À1 at scan rates of 1 and 200 mV s À1 for 15 and 60 pulses, respectively. Because of the smaller MnO 2 particles electrodeposited by Pulse, the MnO 2 -based gravimetric capacitance did not change much (Fig. 5i), resulting in almost constant total gravimetric capacitances (Fig. 5h) and increasing total areal capacitances ( Fig. 5g) with increasing number of pulses and thus MnO 2 load. But the inferior crystallinity of the MnO 2 particles prepared by Pulse compared with those prepared by CP and CC (Fig. S3 †) may have cancelled the advantage of small particle size, resulting in moderate MnO 2based gravimetric capacitance values. Further enhancement of the capacitances could be expected by improving the crystallinity of MnO 2 by adjusting the potential more carefully. Fig. 6 summarizes the capacitance of our MnO 2 -CNT hybrid electrodes electrodeposited under various conditions. Fig. 6a clearly shows that the total areal capacitance increased with MnO 2 load under most conditions, and largest values were achieved for the lms by CP at 1.20 V and CC at 20 mA cm À2 . Fig. 6b shows that the total gravimetric capacitance increased with MnO 2 load of #3 mg cm À2 but became saturated or even decreased for higher MnO 2 load. The capacitance enhancement is more signicant than the mass increase for small MnO 2 load but the mass increase becomes more signicant than capacitance enhancement for large MnO 2 load. Fig. 6c shows that the MnO 2 -based gravimetric capacitance was very high for very small MnO 2 load (346 F MnO 2 g MnO 2 À1 at 2 mV s À1 for a hybrid with MnO 2 load of 0.12 mg cm À2 and 8.5 wt%), which decreased monotonically to 30-150 F MnO 2 g MnO 2 À1 with increasing MnO 2 load. The utility ratio of MnO 2 decreased with increasing MnO 2 load owing to the increasing particle size, increasing tortuosity, decreasing porosity, and/or continuous lm formation ( Fig. 2-4). Fig. 6d shows the lm performance plotted against the capacitance values that are important for practical application; total areal capacitance vs. total gravimetric capacitance. The hybrids deposited by CP at 1.20 V for 10 and 30 min and those deposited by CC at 20 mA cm À2 for 9 and 18 min showed high gravimetric and areal capacitances at low scan rates of 2-20 mV s À1 owing to the high load. In contrast, they showed a decreased capacitance at a high scan rate of 200 mV s À1 owing to the large particle size and/or continuous surface layer of MnO 2 ( Fig. 3 and 4). The hybrids deposited by applying Pulse with 45 and 60 pulses showed moderately high gravimetric and areal capacitances. Fig. 6e shows the lm performance plotted against the total areal capacitance and MnO 2based gravimetric capacitance. It is clear that the MnO 2 -based gravimetric capacitance, which is oen called "specic capacitance" and reported as a primary property, is in a tradeoff relationship with the areal capacitance that is important for practical applications. Fig. 1e shows a similar plot comparing the present work with previous reports in a logarithmic scale with a different denition of the gravimetric capacitance (i.e., MnO 2 -based capacitance divided by MnO 2 mass for Fig. 6e and total capacitance divided by MnO 2 mass for Fig. 1e). Encouraging values that were reported for MnO 2 -based gravimetric capacitance were realized by very small MnO 2 loads, resulting in very small total areal capacitance values. Our electrodes with ne MnO 2 particles uniformly electrodeposited in 18-37 mm thick CNT paper realized areal capacitance of $1 F cm À2 , which is much higher than most of previously reported values. Stability and performance changes of the MnO 2 -CNT hybrid electrodeposited by CP at 1.20 V for 10 min were tested by galvanostatic charge-discharge cycling between 0.0 and 0.8 V. The symmetric shape of the charge-discharge proles is very close to that expected for an ideal capacitor, with small IR (currentresistance) drops (Fig. S5a †). As for the cycle stability, the electrode showed some decay in capacitance to $90% aer 5000 cycles (Fig. S5b †).

Conclusions
MnO 2 -CNT hybrid electrodes were fabricated by electrodepositing MnO 2 on self-supporting CNT paper as a three dimensional current collector. For uniform electrodeposition of MnO 2 particles inside the 18-37 mm-thick CNT papers, we considered diffusion and reaction processes of Mn 2+ ions. A high potential enhanced the reaction, especially nucleation of MnO 2 particles, yielding small particles at high density. But the process was diffusioncontrolled, yielding a dense MnO 2 layer on the exterior surface of the electrodes. Such electrodes were not suitable for capacitor electrodes because of the inhibited ionic diffusion by the dense layer during capacitor operation. In contrast, a low potential made the reaction slow, making the process reaction controlled and realizing uniform MnO 2 deposition within the CNT matrix owing to the sufficient Mn 2+ diffusion. But the low potential made the nucleation of MnO 2 particles slow, yielding large particles at a low density. Such electrodes did not show high capacitance owing to the low electric conductivity of MnO 2 and slow ionic diffusion in MnO 2 . To overcome the tradeoff between the uniformity and particle size/density, we examined pulse electrodeposition, in which high potential pulses were intermittently applied to deposit ne MnO 2 particles at high density and allow the Mn 2+ ions to diffuse during the intervals. MnO 2 was deposited uniformly with amounts increasing roughly proportional to the charge, and we obtained electrodes with MnO 2 content of $80 wt% (and thus CNT # 20 wt%) without the formation of a dense MnO 2 layer on the electrode surface.
For the electrodes using 18-37 mm-thick CNT paper, the best performance was achieved when MnO 2 was deposited by CP of 1.20 V for 10 min. The resulting MnO 2 -CNT hybrid had a large thickness of 62 mm, high areal mass of 6.75 mg cm À2 and high mass density of 1.09 g cm À3 , yielding a moderate total gravimetric capacitance of 120 and 51 F total g electrode À1 , high volumetric capacitance of 131 and 56 F total cm À3 and high areal capacitance of 0.81 and 0.34 F total cm À2 at scan rates of 2 and 200 mV s À1 , respectively. The MnO 2 -CNT hybrid prepared by Pulse had small particles at high density uniformly dispersed within the CNT matrix, but showed moderate capacitance possibly owing to the inferior crystallinity resulting at the high applied potential of 2.0 V. We are trying to improve the crystallinity of the small MnO 2 particles deposited uniformly at high density by adjusting the potential used for the pulse deposition.