当前位置:首页 >> 能源/化工 >>

Mesoporous carbon–manganese oxide composite as negative electrode material for supercapacitors

Available online at www.sciencedirect.com

Microporous and Mesoporous Materials 110 (2008) 167–176 www.elsevier.com/locate/micromeso

Mesoporous carbon–manganese oxide com

posite as negative electrode material for supercapacitors
? ? Yannick Lei, Claire Fournier, Jean-Louis Pascal, Frederic Favier
Received 31 May 2007; received in revised form 26 October 2007; accepted 30 October 2007 Available online 5 November 2007


` ? AIME, Institut Charles Gerhardt, CNRS-UMR 5253, Universite Montpellier 2, Place Eugene Bataillon, 34095 Montpellier, Cedex 5, France

Abstract 3D-assembles of silica spheres were used as hard template to synthesize porous carbon materials with large mesopores to be included as current collectors in supercapacitors and developing large surface areas reaching up to 900 m2/g. Birnessite-type MnO2 was deposited by a co-precipitation method in the porous network. Electrochemical performances of resulting MnO2/C nanocomposites were evaluated by cyclic voltammetry and showed an initial capacitance and a retained capacitance after 500 cycles for the nanocomposite at 6 wt% MnO2 in C of about 660 and 490 F/g, respectively. ? 2007 Elsevier Inc. All rights reserved.
Keywords: Supercapacitor; Mesoporous carbon; Manganese oxide; Nanocomposite; Capacitance

1. Introduction With the unrelenting depletion of available stock of fossil fuel and because of increasingly serious environmental alerts associated to greenhouse e?ect, alternatives power or energy sources have been researched with these underlying objectives: less pollutant and, at least, as e?cient as combustion engines. Foreseen energy production means vary depending on the targeted use, mobile or domestic, portable or industrial: fuel cells, solar or wind power, biomass, etc. are potential, more or less viable, candidates for a ‘‘cleaner’’ generation of energy, mainly in the form of electricity. Along with this strategy of deported, delocalised and/or discontinuous (alternative) production, electricity storage is becoming a critical issue justifying the tremendous R&D e?orts focused on electrochemical storage devices: batteries and supercapacitors [1–6]. Supercapacitors have slightly lower energy density than batteries but thanks to their high power density, they are especially appropriate for applications where energy must


Corresponding author. Tel.: +33 4 67 14 33 32; fax: +33 4 67 14 33 04. E-mail address: fredf@univ-montp2.fr (F. Favier).

be delivered in pulsed mode with discharge duration ranging from few milliseconds to few minutes. For examples, they are currently used for aircraft emergency doors and slides actuation systems, catenary-less tramways and are perfect companions of regular batteries for an electricity powered vehicle at start-up or speed-up. In contrast with batteries, supercapacitors o?er extremely long-life services over 100,000 to 1,000,000 cycles of charge–discharge and can now be designed on the basis of environmentally friendly components [7–11]. In the course of high performance supercapacitors and because of their optimised electrolyte–active material and active material–current collector interfaces, thin ?lms of active materials have shown the best electrode design [12– 15]. However, their main drawbacks are low energy and low power coming from the small amount of active material at the ?at current collector with low surface area. Thus, increasing the surface area of the current collector appears as an obvious strategy to overcome this limit. For supercapacitors, the use of active microporous carbons with surface areas of several thousands of m2/g [16] is the most remarkable validation of this approach, although others have chosen nanostructured metal current collectors for the preparation of composite electrodes [17]. In another

1387-1811/$ - see front matter ? 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.micromeso.2007.10.048


Y. Lei et al. / Microporous and Mesoporous Materials 110 (2008) 167–176

approach, Beguin and co-workers have investigated carbon nanotubes loaded with manganese oxide [18]. Despite an excellent electronic conductivity, performances are limited since only the external surface of the tubes is decorated by the electro-active material. Recently, Dong et al. have studied the performances of a composite electrode supported by mesoporous carbons with small mesopores [19]. Performances de?nitively looked attractive but authors still have been confronted to the limited accessibility of the electrolyte to the active material into the composite porous structure. We are convinced electrochemical performances can be improved by the deposition of a controlled thickness of MnO2 at the surface of carbon with interconnected large mesopores. Electrochemical performances are expected to be balanced by several material design parameters including the thickness of the layer for an optimised quality of the electro-active material–current collector interface while maximizing corresponding energy and power, and the structure of the porous current collector, especially the size of the pores allowing both electro-active material loading and electrolyte accessibility. In this paper the electrochemical performances of mesoporous carbon–birnessite-type MnO2 nanocomposites used as electrode materials for supercapacitors are presented. Ordered mesoporous interconnected carbons have been prepared by using assembled silica spheres as hard templates. Various amounts of MnO2 have been precipitated in situ for the preparation of the composite electrodes. Performances have been compared for these prepared composite electrodes on the basis of initial capacitance and capacitance retention upon cycling. 2. Experimental section 2.1. Synthesis of the ‘‘3D-assemble of silica spheres’’ hard template Silica spheres were synthesized by Stober method [20]. It ¨ proceeds by hydrolysis of tetraethoxysilane (TEOS, 98%, ACROS Organics), as silica precursor, in dehydrated ethyl alcohol (EtOH, denatured with 5% MEK, ACROS Organics), in presence of NH4OH (28%) and in an appropriate [TEOS]/[NH4OH] ratio. In a typical synthesis, silica spheres of 150–200 nm in diameter were produced through the hydrolysis of 24 g of TEOS in 720 ml EtOH solution by 4 g of H2O in presence of 48 g of NH4OH ([TEOS]/ [NH4OH] = 1/6). Then, the spheres were kept in suspension in EtOH and assembled by slow evaporation of the solvent. Finally, a thermal treatment at 800 °C for 6 h was processed to enhance the contact between the spheres in the 3D-assemble. Final powders were immersed for equilibration at pH 3.5 in sulfuric acid solutions. 2.2. Synthesis of mesoporous carbon Silica 3D-assemble powder (1.2 g) was impregnated with 1 g furfuryl alcohol (FA, 98% GC, Aldrich) for a

SiO2/FA molar ratio of 2/1, and a polymerisation step was processed at 80 °C for about 1 day, followed by a calcination under N2 ?ow at 800 °C for 16 h. To open the interconnected porosity, the hard template was removed from the resulting silica spheres/carbon composite by etching in a 3 M NaOH solution for a week. Finally, the carbon was washed with distilled water 3 times and once with ethyl alcohol. 2.3. Preparation of the carbon/birnessite-type MnO2 by a co-precipitation method Birnessite-type MnO2 was co-precipitated in the porous network of the carbon. Solutions of MnSO4 and KMnO4 were used as precursors. Carbon was impregnated for 3 h in a 0.5 M MnSO4 solution. Powder was ?ltered on Whatman paper before to be kept in 5.10?4–5.10?2 M KMnO4 solutions for 5 min. MnO2 loading ratio was adjusted by varying the concentration of KMnO4 in the 5.10?4– 5.10?2 M concentration range. A thermal treatment at 200 °C for 6 h was then processed on MnO2/C composite powders to slightly enhance MnO2-birnessite crystallinity. All chemicals used for the various synthetic steps have been used as purchased without any further puri?cation. KMnO4 is a very oxidative reagent which must be handle with special care. 2.4. Preparation of the electrode for cyclic voltammetry (shaping) The composite electrode material was shaped by using a suspension of PTFE (60 wt%) in ethyl alcohol [21]. A paste containing 92 wt% of electrode material and 8 wt% of PTFE was obtained and roll-pressed. Squares (1 cm2) of the composite active material paste were cut and placed between two stainless steel grids to be pressed under 10 tons for 2 min. 2.5. Physical characterisations Various analytical methods including thermal gravimetry analysis (TGA/DSC), X-ray di?raction (XRD), scanning electron microscopy (SEM) and surface area calculation by Brunauer–Emmett–Teller (BET) method were performed for the physical characterisation of the prepared powders at the various synthetic stages. TGA/ DSC measurements, performed using a NETZSCH STA409PC equipment, allowed the calculation of the birnessite loading ratio in the ?nal composite materials. XRD patterns were measured using a Phillips X’Pert diffractometer (Cu Ka1) in Bragg–Brentano con?guration. SBET calculation extracted from experimental adsorption isotherms of Krypton using a Micromeritics ASAP equipment, provided an estimation of the surface areas of the prepared powders. Samples were also imaged using a JEOL scanning electron microscope (SEM, JSM-6300F).

Y. Lei et al. / Microporous and Mesoporous Materials 110 (2008) 167–176


2.6. Electrochemical characterisation Cyclic voltammetry was performed using an Autolab electrochemical system PGSTAT12. The electrochemical cell used is a conventional three-electrode system composed with a Pt counter electrode, an Ag/AgCl reference and the composite electrode as working electrode. Measurements were done in an aqueous 0.1 M K2SO4 electrolyte solution. The electrolyte solution was deoxygenated by bubbling N2 for 5–10 min before measurement. The determination of initial capacitance was done at a scan rate of 5 mV s?1, while retained capacitance upon cycling was measured at 10 mV s?1. To discriminate between the relative contributions of C and MnO2 to the total capacitance of the composite materials, measurements have also been done with N(Et)4BF4 0.1 M in anhydrous acetonitrile. Measurements have then been done under controlled dry Ar atmosphere in a sealed electrochemical cell. 3. Results 3.1. Synthesis Stober route remains a method of choice for the prepara¨ tion of nano to meso SiO2 particles. The particle size, ranging from few tens to few hundreds of nanometers with an excellent size dispersity, is controlled by [TEOS]/[NH4OH] ratio. The lower this ratio, the larger the size of the prepared particles. Spherical SiO2 particles can easily be assembled from colloidal solutions in 2D or hexagonal 3D-assembles by various methods processing by slow solvent evaporation [24]. Since the ?nal porous material is an inverted replica of the assemble used as a mould, the porosity characteristics, pore size, shape and interconnectivity depend, respectively, on the particle size and shape, and on the contacts in between particles within the assemble. SiO2 sphere assembles are thermally treated at 800 °C for 6 h. This thermal treatment has two e?ects: (i) the contact areas in between spheres are slightly increased and (ii) the mechanical properties of the assemble are enhanced. The polymerisation of furfuryl alcohol proceeds by condensation between the alcohol function and the proton in a of the furane group. This polymerisation is e?ective at 80 °C in acidic medium. This acidity is provided by the acidic species at the silica surface issued from the equilibration of the assemble at pH 3.5 prior to furfuryl alcohol impregnation. Thermal decomposition of polyfurfuryl alcohol starts at 200 °C (N2 1 atm). In presence of Ar or N2 it carbonises at higher temperature. After 16 h at 800 °C, the carbonisation is complete. Hard template removal allowing the opening of the material porosity is processed by chemical etching of the silica spheres. For safety reasons, NaOH etching is preferred rather than HF etching. During these various synthetic steps, polymerisation, calcination and ?nally etching, the particle assemble under-

goes severe mechanical stresses but thanks to the thermal treatment and the resulting strong particle–particle interactions, it keeps its structural integrity as demonstrated by the ordered porous morphology of the resulting carboneous material. The next synthetic step consists in the controlled deposition of MnO2 material at the pore surface for the preparation of MnO2/mesoporous carbon nanocomposites. This can be achieved by in situ precipitation of MnO2 from a solubilized precursor dispersed in the porous structure. This preparation route deriving from the so-called co-precipitation method developed by Lee and Goodenough [7] and recently re-actualised by Toupin and co-workers is a very fast and suitable method processing at room temperature [8]. Prepared MnO2 is usually mainly amorphous showing birnessite-like limited long range crystalline ordering depending on the synthetic parameters. Our method proceeds by impregnation of MnSO4 solution into the carbon porous structure. A special care is devoted to this critical synthetic step to maximize MnSO4 loading inside the porous structure while preventing crystallisation at the carbon grain surface. MnO2 is precipitated by adding KMnO4 oxidative solution. MnO2 loading can easily be controlled in the 0–13 wt% range by the concentration of the KMnO4 solution. Precipitation is followed by a thermal treatment at moderate temperature (200 °C) since it slightly enhances both material crystallinity and electrochemical performances [22]. 3.2. Physical characterisation Fig. 1 shows SEM pictures of resulting materials at the successive synthetic steps from SiO2 sphere assembles to ?nal MnO2/C nanocomposite material. Fig. 1a con?rms the 3D hexagonal packing of silica spheres obtained by slow solvent evaporation and heating at 800 °C for 6 h. If thermal treatment exceeds this duration, spheres start to fuse and void in between spheres collapses. SEM micrography in Fig. 1b was taken after few hours of contact with the NaOH etching solution since the simultaneous presence of silica spheres embedded in the carbon matrix was then more easily observed after partial dissolution of silica spheres at the material surface than on the raw SiO2/C material. After a week in contact with the etching solution, the material porosity is fully opened con?rming the interconnectivity of the porous structure allowing the di?usion of NaOH solution. Fig. 1c reveals the ordered porous arrangement of the resulting carbon as a perfect replica of the initial silica sphere 3D-assemble. From Fig. 1c to d, the presence of the deposited MnO2 at the carbon pore surface can hardly be observed by SEM. However, thanks to TGA measurements (Fig. 2A), it was con?rmed that MnO2 has been deposited during the in situ co-precipitation step according to the following weight MnO2 to C ratios: 0, 3, 4, 6, 10, 11, 12, 13, and 100. Fig. 2A depicts the TGA/DSC curves MnO2/C composite sample with 3 wt% MnO2. Up to 100 °C, TGA curve


Y. Lei et al. / Microporous and Mesoporous Materials 110 (2008) 167–176

Fig. 1. SEM micrographies of (a) assembled silica spheres prepared by Stober method, (b) SiO2/C composite after partial etching, (c) mesoporous carbon ¨ after complete template removal, and (d) ?nal MnO2/C composite.

Fig. 2. (A) TGA/DSC measurements of a MnO2/C composite material with 3 wt% MnO2 (B) XRD patterns of prepared mesoporous carbon (a), precipitated birnessite-MnO2 (b) and MnO2/C composite with 3 wt% MnO2 (c). Bragg positions are given for pure birnessite (JCPDF #18-0802) (d).

shows the rapid loss of water from hydrated MnO2. Carbon decomposition starts at about 300 °C. This exothermic reaction is completed at 500 °C while only carbon-free MnO2 remains.

The XRD patterns depicted in Fig. 2B show the crystalline nature and quality of pure mesoporous carbon from carbonisation of polyfurfuryl alcohol followed by etching of the silica sphere assemble (a), and composite material

Y. Lei et al. / Microporous and Mesoporous Materials 110 (2008) 167–176


from in situ co-precipitation of MnO2 in the mesoporous matrix followed by a 6 h thermal treatment at 200 °C (c). As reference, experimental XRD pattern of pure MnO2birnessite as prepared by co-precipitation method and annealed at 200 °C for 6 h (b) as well as Bragg positions (d) from literature data (JCPDF #18-0802 and Ref. [23]) are also included. Only weak and broad peaks could be extracted from the background at 36.5°, 42.6°, 54.9°, and 65.7°(2h) in pattern (b) corresponding, however, to the main di?raction peaks of pure hexagonal MnO2-birnessite [23]. Despite its poor quality, (c) XRD pattern shows characteristic features from both pure carbon and MnO2-birnessite, validating the composite formation. Despite an annealing step, MnO2 remains mostly XRD amorphous within the composite material. The absence of long range order can be explained by the small size of the MnO2 particles deposited at the surface of the pores while carbon does not favour any crystalline growth of MnO2-birnessite. The resemblance between patterns from pure co-precipitated and carbon-supported MnO2 con?rms the limited impact of material con?nement within carbon of such large mesopores on MnO2 powder morphology. The hysteresis loop in the adsorption isotherm of Kr on 10 wt% MnO2 composite sample depicted in Fig. 3 demonstrates the mesoporous characteristics of the prepared material. Measurements by N2-adsorption have not been possible with satisfying repeatability and accuracy neither on pure carbons nor on composite materials. Numerical data extracted from shape-less N2-adsorption have been rejected and Kr-adsorption performed on prepared materials. In the whole series, surface characteristics show the same trend as depicted in Fig. 3 and BET surfaces areas

Table 1 Summary of the surface areas determined by BET method for prepared samples MnO2/C sample (wt%) 0 (pure C) 2 4 6 10 11 100 (pure MnO2) Measured BET surface (m2/g) 933 ± 29 827 ± 16 552 ± 12 532 ± 13 458 ± 9 475 ± 11 16.9 ± 0.2 Calculated BET surface (m2/g) 915 896 878 841 832

Calculated surface areas have been obtained using the following formula: Scalc = RwiSi with wi the weight percentage for C and MnO2-birnessite, respectively, within the corresponding sample and Si the measured BET surface area for pure C (933 m2/g) and pure MnO2-birnessite (17 m2/g).

have been measured for each prepared mesoporous carbon without and with deposited MnO2 (Table 1). Surface areas developed by prepared carbons are quite large, up to 930 cm2/g. Although, BET surface is expected to mainly originate from the geometrical surface area developed by the ordered spherical mesoporous structure, the shape of the sorption isotherm does not dismiss a contribution from the porous carbon walls. After co-precipitation of MnO2, the corresponding surface slightly decreases: the larger the MnO2 content the smaller the developed surface area. For each composite sample, the measured surface area is signi?cantly smaller than calculated by the sum of those for pure carbon (933 m2/g) and pure co-precipitated birnessite (16 m2/g) weighted by the corresponding content ratio. This latter point con?rms that MnO2 is precipitated within the porous structure of the composite material rather than as separated powder.

Fig. 3. Adsorption Isotherm of Kr on MnO2/C composite electrode material with 10 wt% MnO2.

Fig. 4. Cyclic voltammograms for pure carbon (a), pure co-precipitated MnO2-birnessite (b) and 10 wt% MnO2 in carbon/MnO2 composite (c). Measured currents are relative to weight content of electro-active material within the sample.


Y. Lei et al. / Microporous and Mesoporous Materials 110 (2008) 167–176

3.3. Electrochemical characterisation Fig. 4 shows the cyclic voltammograms measured for as prepared mesoporous carbon (a), birnessite (b) obtained by co-precipitation method and MnO2/C composite electrode with 10 wt% MnO2(c). By comparing the integrals of each curve, the incorporation of birnessite into the porous carbon structure obviously results in a strong improvement of the electrode capacitance. A more precise calculation of the corresponding capacitance also demonstrates that the composite electrode capacitance is larger than the direct sum of weighted capacitances of carbon and birnessite relative to the composite content. This clearly supports the anticipated composite e?ect on electrode capacitance. Electrochemical experiments have been set-up for the performance comparisons of the prepared composite electrodes through the evaluation of two electrochemical parameters: the initial capacitance and the capacitance retention upon electrode cycling. For the whole series, both parameters have been evaluated in the same conditions: same electrode shaping, electrode material weight and size in the same ranges, and single cell set-up. Although experimental conditions have been kept similar in the whole evaluation series, these have not been optimised and neither have been the electrochemical performances. These data have been measured and analysed for a comparative purpose only and could probably be strongly improved. Initial capacitances of prepared composite electrodes have been measured. Reported values (Fig. 5) have been calculated relative to MnO2/mesoporous carbon weight content (+) as well as relative to MnO2 exclusive weight content (s). Curves (+) and (s) obviously show the same trends demonstrating the limited contribution of the car-

bon matrix to the composite electrode capacitance. Carbon contribution remains constant for about 46 F/g in the whole series of composite electrodes. This contribution has been extrapolated from measurements done in anhydrous acetonitrile electrolyte. For pure MnO2, capacitance in acetonitrile is negligible. From pure mesoporous carbon to composite electrodes the corresponding capacitances decrease from 52 F/g to 46 F/g whatever the MnO2/C content ratio. Curve (s) shows a sharper pro?le than curve (+). These curves are clearly not ‘‘parallel’’ suggesting performance improvements as not simply related nor proportional to an increase of MnO2 electro-active content in the composite electrode. Adding MnO2 into the carbon mesoporous matrix obviously increases the initial capacitance of the composite electrode compared to pure carbon. However, this increase does not linearly correlate MnO2 content: with 3 wt% or 4 wt% of MnO2 in the composite electrode, the weight content of electro-active material is limited as is the resulting electrode capacitance. For highest percentage values, corresponding electrode capacitances decrease as MnO2 contents increase up to 100 wt% MnO2. For curves (+) and (s), maxima do not correspond and the highest capacity value at about 90 F/g is reached for the composite electrode formulated with 10 wt% MnO2 content (+) while maximum capacitance relative to MnO2 content at 700 F/g is for 6 wt% MnO2 (s). Noteworthy are the very close capacitance values at about 50–62 F/g for both pure MnO2 and mesoporous carbon, respectively. Fig. 6 depicts the behaviour upon cycling of the various electrodes involved in the present study. Capacitances relative to MnO2 contents have been measured for the ?rst 500

Fig. 5. Initial capacitances (F/g) for various MnO2/C ratios. Capacitances have been calculated relative to MnO2/C weight content (+) and relative to MnO2 weight content (s). For the latter, carbon contribution has been excluded.

Fig. 6. Capacitance upon cycling relative to MnO2 content over 500 cycles for various MnO2/C composite electrodes: 0 wt%-pure carbon (open circles), 4 wt% (open squares), 6 wt% (open diamonds), 10 wt% (·), 12 wt% (crosses), and 100 wt% -pure MnO2 (open triangles).

Y. Lei et al. / Microporous and Mesoporous Materials 110 (2008) 167–176


cycles for the whole series of prepared electrodes. Pure mesoporous carbon and pure MnO2 have been included as reference materials. Two distinct behaviours can be observed: (i) for electrodes composed of pure carbon, and low loading composites with 4 or 6 wt% of MnO2, starting from high initial capacitance values, a more or less progressive capacitance fading is observed upon cycling. (ii) For electrodes from pure MnO2-birnessite, and composite electrodes with 10 or 12 wt% of MnO2, the capacitance increases during the ?rst 200 ca cycles before to reach a steady state showing capacitance values higher than the initial capacitance. Cycling over 500 cycles induces some drastic changes in the capacitance of prepared electrodes. As an example, after 500 cycles the highest capacitance remains for the 6 wt%-MnO2 composite electrode but shows a decrease of more than 25% of the initial value (C500 = 490 F/g for Cinit = 660 F/g). In contrast, in the meantime the 12 wt%MnO2 composite electrode has seen its capacitance increased by 20% (C500 = 360 F/g for Cinit = 300 F/g). These electrodes are not the only two to be a?ected by electrode cycling and the initial capacitance ranking as follows: C(pure C) < C(MnO2-birnessite) < C(10 wt%) < C(12 wt%) < C(4 wt%) < C(6 wt%) has been inverted after 500 cycles: C(pure C) < C(MnO2-birnessite) $ C(10 wt%) < C(4 wt%) < C(12 wt%) < C(6 wt%). 4. Discussion 4.1. Synthesis Prepared MnO2/mesoporous carbon composites are complex materials and electrochemical performances of the corresponding electrodes rely on various structural as well as interfacial parameters. Starting from the ordered assembling of silica spheres to the in situ precipitation of MnO2 into the carbon matrix, a ?ne control of each successive synthetic step is necessary for an optimization of the electrode performances. Ordering of the porous structure is de?nitively not a critical parameter for the material electrochemical behaviour since similar electrode performances could certainly be achieved from a disordered structure. There is then no special need for inverse opal-like material for this speci?c application, however, ordering of the silica spheres, by hexagonal close-packing in the present case, is certainly among the most appropriate ways to ensure interconnectivity between neighbouring pores. Contacts in between spheres originate for the pore to pore connectivity which guarantees essential liquid di?usion through the porous structure during composite in situ synthesis as well as for use. The pore to pore channel size is relative to the area of the sphere to sphere contact surface. For perfect hard spheres, the ‘‘contact surface’’ is theoretically limited to a single point but, hopefully, surface roughness of imperfect silica spheres, strongly increase the contact surface area. More-

over, after 6 h at 800 °C, silica spheres do not melt but strongly aggregate, promoting larger contact surface areas resulting in larger pore to pore channels. This thermal treatment processed after sphere assembling, also enhances the mechanical strength of the ‘‘silica mould’’ such as strains and stresses from capillary forces and thermal modi?cations induced by polymerisation and during carbonisation do not destroy the initial 3D structure. Furfuryl alcohol shows a pretty low viscosity which practically allows its facile di?usion into the open void in between the sphere assemble for a complete template impregnation. After polymerisation, the resulting composite material is an homogeneous intimate mixture of the silica assemble and polyfurfuryl alcohol. This homogeneity is also a guaranty for the structural uniformity of the ?nal porous carbon material. The polymer decomposition is processed at 800 °C under N2 ?ow. Under these conditions, graphitisation of the resulting carbon is low as demonstrated by XRD measurements: only features from X-ray amorphous carbon are observed in the corresponding pattern (Fig. 2B-(a)). Porosity opening from hard template dissolution can be achieved either ways by HF or NaOH etching. Although dissolution kinetics are more favourable using HF, NaOH was preferred for obvious safety reasons. Moreover, NaOH etching e?ciency has been con?rmed since, after one week of contact, silica has been completely dissolved and washed o? after rinsing with water as demonstrated by energy dispersive X-ray analysis (EDX – data not shown). In situ preparation of solids within a porous structure from precursors dissolved in liquid phases is possible only if kinetics of both precursor di?usions and reactivities are favourable. Impregnation of a reactant such as MnSO4 in such large mesoporous and interconnected carbon matrix is easily achieved by immersion of a carbon powder into a concentrated solution of MnSO4: in such open medium, solubilised MnSO4 can di?use almost freely within the porous matrix. Allowed to dry, the resulting composite consists then in solid MnSO4 dispersed within the carbon porous volume. Any speci?c interaction is expected between carbon and MnSO4 and if an MnSO4-free aqueous solution is added, MnSO4 dissolves and starts to di?use to equilibrate the whole solution volume inducing an unfavourable dilution of the reactant inside the porous material. Oxidation of MnSO4 by KMnO4 is however so fast that, at KMnO4 solution addition, any prejudicial dilution does not occur before MnO2 precipitation without any material loss. 4.2. BET results Geometrical surface area developed by the porous material can be calculated from the diameter of the silica particles used as hard template according to the following formula: S ? N spheres ? S sphere ? 1=qcarbon ?1?


Y. Lei et al. / Microporous and Mesoporous Materials 110 (2008) 167–176

and N spheres ? V =V sphere ?2?

with Nspheres = number of silica spheres which can be assemble in an hexagonal compact arrangement in an 1 cm3 volume; V = 0.74 · 10?6 m3: volume occupied by an hexagonal arrangement in a cube of 1 cm3; Vsphere = p · d3 · 10?27/6 m3: volume of a single particle of d diameter; Ssphere = surface developed by a single particle sphere; qcarbon = 2.25 g/cm3; from (1) and (2), S = (0.74 · 6 · 103)/(d · qcarbon). Starting from SiO2 spheres of 75 nm in diameter, this gives S75 = 26.31 m2 geometrical surface area developed by 1 g of carbon and S120 = 16.44 m2/g and S200 = 9.87 m2/g starting from 120 nm and 200 nm particle diameters, respectively. Compared to measured BET surface areas, up to 933 m2/g for the mesoporous C samples prepared from silica particles of about 120 nm in diameter, these limited geometrical surface areas represent less than 2% of the overall developed surface. In these materials, most of the surface developed comes from the mesopores in the structure walls. These carboneous ordered materials are thus built on a hierarchical porous structure including large mesopores perfectly suited for liquid di?usion and mass transfer within the structure while smaller mesopores in the thin carbon walls remain accessible for electro-active material deposition. The loss of BET surface area from MnO2-free carbon material to composite originates from the decrease in the developed surface while porous volume is fed with MnO2 deposit. These high values however demonstrate that a large part of material surface areas remain accessible (to Kr and certainly to electrolytic species) after in situ co-precipitation of MnO2 and composite preparation. Prepared mesoporous carbons and MnO2/C nanocomposites can then be assimilated to hierarchically structured materials based on a carbon structure showing large interconnected mesopores opening access to smaller mesopores within the structure walls. MnO2 is precipitated within the whole porous volume, including larger as well as smaller mesopores for an optimised accessibility of the electrolyte to the electro-active material. 4.3. Electrochemical performances Electrochemical performances of the composite electrodes, especially on a weight basis, appear as based on a compromise between the electrochemical e?ciency, including both charge storage and charge transfer capabilities of the electro-active material and the presence of active carbon considered as a more or less dead weight. Two distinct points of view can be used to discuss performance evaluations of the prepared composite electrodes: Engineers will consider electrodes as a whole including electro-active material, conductive additive, binder, and grid support while researchers will also extract the speci?c contribution of the electro-active material from the overall

response of the formulated electrodes. Since the purpose of this paper is also to con?rm our approach as suitable for a maximisation of the charge storage performances of MnO2 using composite electrode, the following discussion will focus on both series of data. Previous results on electrochemical performances of thin (or thick) MnO2 ?lms have shown that the thinner the electro-active ?lm, the higher the capacitance relative to the MnO2 content [13]. Data gained in the present study from composite electrodes with low MnO2 contents do not fully correlate these previous results. Composite e?ect is demonstrated by the capacitances measured for composite electrodes that are, with the exception of MnO2/C electrode with 3 wt% MnO2, higher than the balanced sum of the individual capacitance values at about 52–60 F/g for both pure MnO2 and mesoporous carbon. At MnO2/C with 3 wt% MnO2, the electrode capacitance is lower than those for pure C: the presence of MnO2 particles at the carbon surface, blocking electrolyte access, is prejudicial to carbon speci?c capacitance. At low MnO2 content such as 3 wt%, capacitance provided by MnO2 does not compensate the loss of carbon capacitance. In contrast, compared to pure MnO2 capacitance value (60 F/g), those for MnO2/C with 10 wt% MnO2 at 86 F/g is 25% higher. At this content ratio, capacitance is the largest in the prepared electrode series. We have not reached so far any de?nitive explanation for this speci?c value at 10 wt% MnO2 but it clearly sets the limit between two composition domains for which performances rely on two distinct electrochemical behaviours: below 10 wt% MnO2 the higher the MnO2 loading within the porous structure, the higher the resulting electrode capacitance. In this composition domain, resulting materials show optimised electrochemical e?ciency in terms of carbon surface available for MnO2 deposition and particle size or layer thickness. Below this 10 wt% MnO2 threshold, any precipitated MnO2 particle sets some e?cient electrical contact with the carbon surface for a good quality electro-active material/substrate solid–solid interface, while MnO2 particles remain small enough for a maximisation of the electroactive material amount in direct contact with the electrolyte (and a maximisation of the corresponding electro-active material/electrolyte solid–liquid interface). Above 10 wt% MnO2, adding MnO2 within the carbon porous structure induces a decrease in the resulting composite electrode capacitance. It seems reasonable to state that at 10 wt% MnO2 in MnO2/C, the whole carbon surface is covered with MnO2 and that any additional MnO2 will precipitate at MnO2 surface already present at the pore surface for an increase of the electro-active material thickness. Above with 10 wt% MnO2 in MnO2/C, the thickness of precipitated MnO2 layer is too large to allow any optimised contact of the electro-active material with the electrolyte. Unaccessible in depth MnO2 layer is then less electrochemically active which sink-down the overall elec-

Y. Lei et al. / Microporous and Mesoporous Materials 110 (2008) 167–176


trode performances. Moreover, as a non-conductive barrier it also limits the charge transfer from/to the current collector for the disadvantage of the electrochemical storage process too. Individual MnO2 capacitance values (open circle curve in Fig. 5) have been extracted from measured electrode capacity values according to the following formula: C MnO2 ? ?melect =mMnO2 ? ? ?C elect ? ?mc =melect ? ? C c ? with melect: weight of electro-active material; mMnO2 : weight of MnO2 in composite electrode; mc: weight of carbon in composite; Celect: capacitance of electro-active material (MnO2 + C); Cc: capacitance of carbon in composite electrode. To discriminate between the carbon and MnO2 contributions to the total capacitances of composite materials and determine Cc, electrochemical measurements have been done in an acetonitrile electrolytic solution. Since charge storage in MnO2 is known to proceed by intercalation of protons [13] or hydrolysed alkaline metal cations [7], MnO2 does not show any capacitive behaviour in aprotic solvents such as acetonitrile. Therefore, capacitances measured for composite materials in acetonitrile originate only from the mesoporous carbon substrate. These measurements have demonstrated that whatever the MnO2/C content ratio, Cc remains constant at 46 F/g. This shows that at these low MnO2 loadings, accessible carbon surface hardly changes despite the presence of precipitated MnO2 at the pore surface. Capacitances for mesoporous carbon-supported MnO2, calculated up to 700 F/g, can however reach higher values than those obtained from raw powders usually ranging from 80 to 150 F/g [25]. They are in better agreement with those measured from thin ?lms [13,14] demonstrating the consistency of our approach for the design of composite electrodes for supercapacitors. Low loadings allow a maximisation of the electrochemical activity of MnO2 with an optimal relative content at 6 wt%. At higher MnO2 loadings, performances decrease for the reasons exposed above in terms of layer thickness, electrolyte accessibility and charge transfer e?ciency. Supercapacitor devices are expected to run for few hundreds of thousands cycles and capacitance retention is de?nitively one of the most critical performance parameter to control while designing new electrode materials for supercapacitors. As for the measurement of the initial capacitances, the electrochemical set-up used for long duration measurements over a week, was not optimised in the present work and actual performances could certainly be strongly improved. However, measurements have been performed in the same conditions for the whole series of prepared composite materials. From curves on Fig. 6, charge–discharge cycling a?ects the electrode performances and some drastic capacity fading is observed after 500 cycles in operation. A loss of electro-active material upon electrode cycling is suspected to originate for the observed capacity loss. The co-precipita-

tion method used for the composite preparation does not allow any control on the MnO2 to C interaction while any strong adhesion of MnO2 is then expected at the carbon pore surface. Cycling under mass control is currently operated to con?rm this hypothesis and, if material loss is con?rmed, alternative deposition techniques including electrochemical deposition at pore surface will be explored. Electrode cycling does not however a?ect electrochemical performances the same way depending on the low or high MnO2 loading within the mesoporous carbon structure. For low loadings at 4 and 6 wt% MnO2 in MnO2/C, electrodes behave in a similar way as for pure carbon: the highest capacitance at ?rst cycles progressively fades down upon cycling. After 500 cycles, pure carbon loses 16% of its initial capacitance. This loss is of 40% for the 4 wt% MnO2 in MnO2/C sample. After 500 cycles, the 6 wt% MnO2 in MnO2/C composite electrode remains that having the most attractive performances, despite a 26% fading of initial capacitance. In contrast, high MnO2 loadings at 10 and 12 wt% in MnO2/C give electrodes behaving as from pure MnO2: the capacitance increases during the ?rst 200 ca cycles before to stabilise in long term use at higher values than the initial value. The capacitance after 500 cycles, for the 12 wt% MnO2 in MnO2/C sample for example, however remains lower than that of the 6 wt% MnO2 in MnO2/C composite electrode at the same cycling stage. This latter point would have to be con?rmed in longer term use. 5. Conclusion We have synthesized new negative electrode materials for supercapacitors using a multistep but easy preparation route. Prepared nanocomposite materials are based on mesoporous carbon with interconnected large mesopores as current collector and deposited manganese oxide as active material. Electrode materials were characterised by physical and electrochemical methods. They all show a surface area of several hundreds m2/g. Initial and retained capacitances were evaluated by cyclic voltammetry. Results have shown that a 6 wt% MnO2 in MnO2/C composite presents an optimal electrochemical reactivity, with an initial capacitance of 660 F/g and a retained capacitance after 500 cycles of 490 F/g. Capacity fading will be elucidated by in situ as well as ex situ physical characterisations of the electrode material in operation and after cycling. The next step of this study will be devoted to the design, fabrication and performance evaluation of monolithic selfsupported nanocomposite electrodes. Electrodeposition would then be used as an alternative method for the controlled deposition of MnO2 ?lms at the pore surface. Acknowledgments The Centre National pour la Recherche Scienti?que (CNRS) and the French ministry for education and research are gratefully acknowledged for ?nancial support


Y. Lei et al. / Microporous and Mesoporous Materials 110 (2008) 167–176 [13] S.C. Pang, M.A. Anderson, T.W. Chapman, J. Electrochem. Soc. 147 (2000) 444. [14] T. Shinomiya, V. Gupta, N. Miura, Electrochimica Acta 51 (2006) 4412. [15] C. Du, N. Pan, Nanotechnology 17 (2006) 5314. [16] W. Xing, S.Z. Qiao, R.G. Ding, F. Li, G.Q. Lu, Z.F. Yan, H.M. Cheng, Carbon 44 (2006) 216. [17] P.L. Taberna, S. Mitra, P. Poizot, P. Simon, J.M. Tarascon, Nat. Mater. 7 (2006) 567. [18] E. Raymundo-Pinero, V. Khomenko, E. Frackowiak, F. Beguin, J. Electrochem. Soc. 152 (1) (2005) A229. [19] X. Dong, W. Shen, J. Gu, L. Xiong, Y. Zhu, H. Li, Ji. Shi, J. Phys. Chem. B 110 (2006) 6015. [20] W. Stober, A. Fink, E. Bohn, J. Colloid Interf. Sci. 26 (1968) 62. ¨ [21] M. Toupin, T. Brousse, D. Belanger, Chem. Mater. 14 (2002) 3946. [22] T. Brousse, P.-L. Taberna, O. Crosnier, R. Dugas, P. Guillemet, Y. ? Scudeller, Y. Zhou, F. Favier, P. Simon, D. Belanger, J. Power Sources (2007), doi:10.1016/j.jpowsour.2007.04.074. [23] A.-C. Gaillot, D. Flot, V.A. Drits, A. Manceau, M. Burghammer, B. Lanson, Chem. Mater. 15 (2003) 4666. [24] W. Horst, Philos. Trans. Roy. Soc. A 361 (1803) (2003) 229. [25] T. Brousse, M. Toupin, R. Dugas, L. Athouel, O. Crosnier, D. ¨ ? Belanger, J. Electrochem. Soc. 153 (2006) A2171.

as is the European Commission for funding through INSIDE_POReS Network of Excellence programme. References
[1] E. Kardena, P. Shinn, P. Bostock, J. Cunningham, E. Schoultz, D. Kok, J. Power Sources 144 (2005) 505. [2] A.G. Pandolfo, A.F. Hollenkamp, J. Power Sources 157 (2006) 11. [3] M. Anderman, J. Power Sources 127 (2004) 2. [4] S.G. Chalk, J.F. Miller, J. Power Sources 159 (2006) 73. [5] R.A. Huggins, Solid State Ionics 134 (2000) 179. [6] J. Lee, J. Kim, T. Hyeon, Adv. Mater. 18 (2006) 2073. [7] H.Y. Lee, J.B. Goodenough, J. Solid State Chem. 144 (1999) 220. ? [8] T. Brousse, M. Toupin, D. Belanger, J. Electrochem. Soc. 151 (4) (2004) A614. [9] C. Chen, D. Zhao, D. Xu, X. Wang, Mater. Chem. Phys. 95 (2006) 84. [10] M. Wohlfahrt-Mehrens, J. Schenk, P.M. Wilde, E. Abdelmula, P. Axmann, J. Garche, J. Power Sources 105 (2002) 182. [11] V. Subramanian, H. Zhu, B. Wei, Electrochem. Com. 8 (2006) 827. [12] T.P. Gujar, V.R. Shinde, C.D. Lokhande, S. Han, J. Power Sources 161 (2006) 1479.