Aqueous electrostatic dispersion and heterocoagulation of multiwalled carbon nanotubes and manganese dioxide for the fabrication of supercapacitor electrodes and devices

Abstract

A conceptually new approach has been developed for the fabrication of MnO₂-multiwalled carbon nanotube (MWCNT) composites for electrodes of electrochemical supercapacitors (ES). Benzyldimethylhexadecylammonium chloride (BAC) surfactant and caffeic acid (CA) selectively adsorbed on MWCNT and MnO₂, respectively, and allowed the formation of stable aqueous suspensions of positively charged MWCNT and negatively charged MnO₂. The comparison of the electrophoretic deposition yield data for BAC, CA and other molecules provided an insight into the influence of the molecular structure on adsorption of the molecules and dispersion of MWCNT and MnO₂. Advanced composite materials with good mixing of the individual components were obtained by heterocoagulation, based on an ion-pairing assembly of BAC and CA. The composite electrodes, prepared by the new method showed superior electrochemical performance. It was found that high capacitance and good capacitance retention at high charge–discharge rates can be achieved at high active mass loadings. A practical outcome of this study was the fabrication of an asymmetric SC device, containing a positive MnO₂–MWCNT electrode and negative activated carbon–carbon black composite electrode with a voltage window of 1.8 V in an aqueous electrolyte. The asymmetric device showed high capacitance, high power-energy characteristics, good capacitance retention at high charge–discharge rates and cyclic stability.

---

1. Introduction

The growing interest in application of MnO₂ for electrodes of electrochemical supercapacitors (ES) is attributed to high specific capacitance (SC), a relatively large voltage window and the low cost of this material.^1,2 There are excellent reviews, describing recent advances in the development of MnO₂ electrodes for ES.^3–5 The charge–discharge behavior of MnO₂ is given by the following reaction:⁶

MnO₂ + A⁺ + e⁻ ↔ MnO₂A (1)

where A⁺ = Li⁺, Na⁺, K⁺, H⁺. Eqn (1) indicates that high electronic and ionic conductivities are necessary in order to utilize capacitive properties of MnO₂ in ES electrodes. However, the electronic conductivity of MnO₂ is low. This problem was addressed by the fabrication of MnO₂–carbon nanotube (CNT) composites.^7–11 It was found that MnO₂ and composite MnO₂–CNT electrodes show capacitive behavior in a voltage window of 1 V in aqueous electrolytes. It should be noted that the voltage windows of symmetric capacitors, containing two similar MnO₂ based electrodes in aqueous electrolytes is smaller than the voltage windows of symmetric RuO₂/RuO₂ (1.6 V (ref. 12)) or Ru/Ru (1.8 V (ref. 13)) devices. The development of asymmetric cells,^5,14 containing MnO₂ positive electrodes and activated carbon (AC) negative electrodes, allowed enlarged voltage window and improved power-energy characteristics of the ES devices.

Many investigations were focused on the optimization of microstructure and performance of MnO₂–CNT electrodes. The interest in CNT as a conductive additive is attributed to high conductivity and low percolation limit. However, the specific capacitance of CNT is low. Therefore, CNT must be well dispersed in the MnO₂ matrix and the CNT concentration in the composites must be optimized. In order to improve the electrical contact of CNT and MnO₂, composite materials were prepared by reduction of Mn⁷⁺ species in KMnO₄ solutions, containing CNT, to form MnO₂ coated CNT:^15–17

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

Further development of this method allowed the fabrication of flexible supercapacitor devices.¹⁸ New composites with various MnO₂ concentration were obtained by this method.¹⁷ However, it was demonstrated¹⁹ that reaction (2) results in a chemical degradation and reduced electrical conductivity of CNT. In order to reduce the chemical degradation of CNT, ethanol was used as a reducing agent¹⁹ for KMnO₄. In another strategy, cathodic electrosynthesis of MnO₂ was performed from KMnO₄ solutions on CNT coated substrates, prepared by a chemical vapor deposition.²⁰ However, relatively low active mass loading was achieved in this method.²⁰

Recent studies highlighted the importance of the fabrication of efficient ES electrodes with high active mass loading and high active material to current collector mass ratio.²¹ It is challenging to achieve high capacitance, good capacitance retention at high charge–discharge rates and cyclic stability for electrodes with mass loadings of 10–20 mg cm⁻², which are required for many practical applications.^21,22 The use of composites, prepared by mixing of MnO₂ nanofibers and multiwalled CNT (MWCNT) allowed²³ the fabrication of electrodes with mass loading of 40 mg cm⁻². However, due to the poor dispersion of the individual components, the electrodes showed low capacitance retention at high charge–discharge rates. The need in improved dispersion of MnO₂ and CNT has generated interest in the development of new colloidal methods for the fabrication of MnO₂–CNT composites.

The goal of this investigation was the development of composite MnO₂–MWCNT electrodes by a new aqueous colloidal method, based on the electrostatic heterocoagulation of negatively charged MnO₂ nanoparticles and positively charged MWCNT. New strategy involved the use of anionic and cationic dispersants, which selectively adsorbed on MnO₂ and MWCNT, respectively. Proof-of-concept studies resulted in the formation of MnO₂–MWCNT composites with high active mass loadings, which showed significant improvement in capacitance at high charge–discharge rates. The use of colloidal methods for the fabrication of MnO₂–MWCNT nanocomposites offers advantages, because the problem of MWCNT degradation (eqn (2)) can be avoided. The results presented below indicated that the method allowed the fabrication of efficient asymmetric capacitors with a voltage window of 1.8 V and good electrochemical performance.

2. Experimental procedures

2.1. Materials

Benzyldimethylhexadecylammonium chloride (BAC), hexadecyltrimethylammonium bromide (CTAB), caffeic acid (CA), trans-cinnamic acid (TCA), p-coumaric acid (PCA), 2,4-dihydroxycinnamic acid (DCA), poly(vinyl butyral-co-vinyl alcohol-co-vinyl acetate) (PVB, average M_w = 50 000–80 000), KMnO₄ and Na₂SO₄ were purchased from Aldrich company. Manganese dioxide nanoparticles with an average particle size of 30 nm and an oxidation state of Mn of 3.6 were prepared by the reduction of aqueous KMnO₄ solutions using the method described in a previous investigation.²⁴ Multiwalled carbon nanotubes (MWCNT) were purchased from Bayer company. Ni foams with a porosity of 95% were supplied by Vale company.

2.2. Fabrication of suspensions and electrophoresis

Suspensions of MnO₂ and MWCNT in water, containing dispersants, were ultrasonically agitated during 30 min before the analysis and fabrication of MnO₂–MWCNT composites. The limitations of the zeta potential concept for the analysis of electrokinetic properties of nanoparticles, containing adsorbed organic molecules, were described in the literature.^25–27 Therefore, the charging mechanism and electrokinetic behavior of the particles were investigated by deposition yield measurements in electrophoretic deposition (EPD) experiments. The EPD cells contained a stainless steel working electrode and a Pt counter electrode. The distance between the electrodes was 15 mm. The cell voltage was 20 V. Deposition yield was studied for the deposits formed on stainless steel substrates.

2.3. Fabrication of electrodes and cells

The MnO₂ suspensions, containing CA dispersant, and MWCNT suspensions, containing BAC dispersant, were mixed, filtered, washed with water and dried. For comparison, MnO₂ and MWCNT mixtures were prepared using the same procedure without dispersing agents. The MnO₂–MWCNT mixtures, containing 5% PVB binder, were impregnated into Ni foam current collectors and then pressed. The mass of the impregnated material was 40 mg cm⁻² (based on projected area). The fabrication of activated carbon (AC) electrodes, containing carbon black (CB) as a conductive additive, is described in the ESI.† Composite MnO₂–MWCNT electrodes and AC–CB electrodes (described in the ESI†), separated by a porous polyethylene membrane (mean pore size 0.4 μm) were combined for the fabrication of coin cells (CR2032 type, MTI corporation, USA), which were sealed using a hydraulic crimping machine (MSK-110, MTI Corporation, USA). The electrode area was 1.1 cm². Envelope cells contained two electrodes with an area of 6 cm², separated by a porous membrane, and aqueous 0.5 M Na₂SO₄ solution as an electrolyte in a sealed plastic envelope.

2.4. Characterization techniques

Electron microscopy studies were performed using a JEOL JSM-7000F scanning electron microscope (SEM). Electrochemical studies of individual electrodes and devices were performed using a potentiostat (PARSTAT 2273, Princeton Applied Research). Surface area of the working electrode in the three electrode cells was 1 cm². The counter electrode was a platinum gauze, and the reference electrode was a standard calomel electrode (SCE). Capacitive behavior and electrochemical impedance of the electrodes and cells were investigated in 0.5 M Na₂SO₄ aqueous solutions. Cyclic voltammetry (CV) studies were performed at scan rates of 2–100 mV s⁻¹. The total capacitance C = Q/ΔV was calculated using half the integrated area of the CV curve to obtain the charge (Q), and subsequently dividing the charge by the width of the potential window (ΔV). The mass-normalized specific capacitance C_m = C/m (m-sample mass) and area-normalized specific capacitance C_s = C/S (S-sample area), calculated from the CV data, were investigated. The electrochemical impedance spectroscopy (EIS) studies of complex impedance Z* = Z′ − iZ′′ were conducted in the frequency range of 10 mHz–100 kHz at the amplitude of the alternating current signal of 5 mV. The complex capacitance C* = C′ − iC′′ (ref. 28) was calculated from the impedance data as C′ = Z′′/ω|Z|² and C′′ = Z′/ω|Z|², where ω = 2πf, f – frequency. Charge–discharge behaviour of the cells was investigated in a voltage window of 1.8 V using battery analyzers (BST8-MA and BST8-3, MTI Corporation, USA).

3. Results and discussion

The fabrication of stable suspensions, containing well dispersed MnO₂ nanoparticles and MWCNT, is extremely important for the development of MnO₂–MWCNT nanocomposites with controlled structure and properties. The chemical modification of MnO₂ and MWCNT with adsorbed dispersing agents allows their efficient dispersion in the suspension and uniform distribution in the composite. A critical property of a dispersant is its adsorption on the particle surface.²⁹ An adsorbed ionic dispersant imparts a charge to the particles and provides their electrostatic repulsion and dispersion. However, a non-adsorbed ionic dispersant can be considered as an electrolyte. According to the DLVO theory of colloidal stability, it increases the ionic strength of the suspension, decreases the electrostatic repulsion of particles and facilitates agglomeration.²⁷ In our investigation, BAC was used as a new dispersing agent for MWCNT. BAC is a long chain cationic surfactant (Fig. 1A), containing an alkyl chain, a charged group and a benzene ring moiety. The sedimentation test results showed that BAC provided efficient dispersion of MWCNT in water. EPD experiments indicated that MWCNT were positively charged in the suspensions, containing BAC, and formed films on the cathodic substrates (Fig. S1(A)†). The deposit mass increased with increasing BAC concentration in the suspension and deposition time (Fig. 2). It is suggested that BAC adsorbed on the MWCNT surface and imparted a positive charge to MWCNT. The increase in BAC concentration in the suspension resulted in increasing BAC adsorption on MWCNT, which in turn resulted in increasing charge and deposition yield (Fig. 2). It is known³⁰ that surfactants disperse CNT in water mainly through hydrophobic/hydrophilic interactions, in which the alkyl chain of the surfactant molecule adsorbs on the CNT surface, while the hydrophilic head associates with water and promotes dispersion. Other investigations indicated that dispersion efficiency increased with increasing length of the alkyl chains and it is energetically favorable for the chains to lie along the length of the nanotubes.³¹

Fig. 1 Molecular structures of (A) benzyldimethylhexadelcylammonium chloride (BAC) and (B) caffeic acid (CA).

Fig. 2 Deposit mass for 1 g L⁻¹ MWCNT suspension in water at a deposition voltage of 20 V: (A) versus BAC concentration at a deposition time of 3 min and (B) versus deposition time at BAC concentration of 1 g L⁻¹.

Recent studies of anionic surfactants³² showed that the benzene ring of sodium dodecylbenzenesulfonate (NaDDBS) is one of the main reasons of superior performance of this material, compared to other anionic surfactants, such as sodium dodecylsulfate (SDS). The NaDDBS, containing a benzene ring moiety, enhanced the stability of CNTs in water by a factor on the order of 10 to 100, compared to other commonly employed surfactants and polymers.³¹

In our investigation we compared cationic BAC, containing a benzene ring, with CTAB, which is another cationic surfactant with a similar structure, but without a benzene ring (Fig. S2†). CTAB is commonly used for the dispersion of CNT in water. Our sedimentation test for 1 g L⁻¹ MWCNT suspensions, containing 0.25–1.0 g L⁻¹ BAC or CTAB showed that the use of BAC as a dispersant allowed improved suspension stability. The comparison of the deposition yield data (Fig. S3†) showed that BAC allowed significantly higher deposition yield of MWCNT, compared to the deposition yield of MWCNT, achieved using CTAB dispersant at similar deposition conditions. The difference can be attributed to improved adsorption of charged BAC on MWCNT, which resulted in improved dispersion and higher charge of MWCNT. The results obtained for cationic BAC, correlate with literature data, comparing dispersion efficiency of the anionic NaDDBS, containing a benzene ring, with other anionic surfactants without a benzene ring.³¹ The influence of the benzene ring on surfactant adsorption is not well understood.³³ We suggest that electrostatic repulsion of charged groups of the surfactants, exposed to the aqueous phase,³⁴ is detrimental for the adsorption. Such repulsions can be partially reduced by the π–π attractive interactions of the benzene rings of adjacent surfactant molecules, adsorbed on the MWCNT surface.

Our results indicated that BAC can be used for the efficient dispersion of MWCNT in water, however the dispersion of MnO₂ presents difficulties. We have found a solution of this problem by utilizing important properties of surfactants, which were not paid enough attention earlier. It is known that cationic surfactants form ion-pairs with tungestosilicic acid,³⁵ anionic dyes,^36–38 caffeic acid³⁹ and other ion-pairing agents.⁴⁰ The electrostatic ion-pair interactions between the cationic surfactants and anionic organic molecules are involved in many self-organized systems and biological systems.⁴¹ Of special interest are the investigations of the formation of ion-pairs between surfactants and chelating agents,^37,40 which were utilized for the extraction of ions, surface modification of inorganic particles, modification of absorption properties of organic dyes and other applications, utilizing fascinating properties of the ion-pair assemblies. The interaction of cationic surfactants with caffeic acid (CA)³⁹ is of special interest for nanotechnology of composites.

CA belongs to the catechol family of materials (Fig. 1B). The anionic properties of CA are attributed to the carboxylic group. It is known that materials from the catechol family show strong adsorption on different inorganic surfaces.²⁹ The growing interest in the catecholate type adsorption is motivated by recent advances in the investigation of the mechanism of mussel adhesion to metal and mineral surfaces in water, which showed that strong adhesion is provided by the catechol ligand of the natural catecholic amino acid, L-3,4-dihydroxyphenylalanine (DOPA).^42–44 CA is a closest molecular analogue of DOPA. Previous investigations showed strong adsorption of CA on MnO₂ in ethanol.⁴⁵ It should be noted that the adsorption of organic molecules on inorganic particles is influenced by particle–solvent interactions, solvation effects and dielectric constant of the solvent.⁴⁶ It is known that water and other liquids with high dielectric constants lead to the deterioration of performance of various adhesives.⁴⁶ In this investigation we analysed the adsorption of CA on MnO₂ in water. The EPD experiments showed that MnO₂ particles were negatively charged in water. However, the deposition rate of MnO₂ was relatively low (Fig. 3A) and the suspension was unstable. The natural negative charge of MnO₂ particles in the aqueous suspension without CA can result from preferred adsorption of OH⁻ ions.²⁷ The addition of CA resulted in improved suspension stability and significant increase of the deposition yield (Fig. 3). The deposit mass increased with increasing deposition time, indicating film growth. SEM observations confirmed the formation of anodic deposits, containing MnO₂ nanoparticles (Fig. S1(B)†). It is suggested that strong adsorption of anionic CA on MnO₂ particles resulted in increased particle charge, improved suspension stability and increased deposition yield. We found that the addition of other aromatic molecules, such as TCA, PCA and DCA, to MnO₂ suspensions resulted in reduced suspension stability and reduced deposition yield. EPD of MnO₂ has not been observed at TCA, PCA and DCA concentrations of 0.4 g L⁻¹. The comparison of the chemical structures of CA (Fig. 1B), TCA, PCA and DCA (Fig. S4†) provided an insight into the adsorption mechanism of CA in water. It is important to note that CA, TCA, PCA and DCA have a similar structure. However, the TCA structure does not include OH groups, bonded to the aromatic ring. The structure of PCA includes one phenolic OH group. The structure of DCA includes two phenolic OH groups, however, they are bonded to non-adjacent carbon atoms of the aromatic ring. In contrast, similar to other molecules of the catechol family, the structure of CA includes two OH groups, bonded to the adjacent carbon atoms of the aromatic ring. The dissociation of CA, TCA, PCA and DCA in water resulted in the formation of anionic CA⁻, TCA⁻, PCA⁻ and DCA⁻ species and H⁺. The charge of the MnO₂ particles was governed by competitive adsorption of the anionic species and H⁺. The sedimentation tests and EPD deposition yield measurements indicated that adsorption of TCA⁻, PCA⁻ and DCA⁻ species was weak and the preferred adsorption of H⁺ resulted in the reduction of the negative charge of the MnO₂ particles and reduction of the deposition yield. In contrast, preferred adsorption of CA⁻ resulted in increasing negative charge of MnO₂ particles in water and increasing deposition yield. The results indicated that similar to DOPA, the adsorption of CA involves two OH groups of the catechol ligand.⁴⁷ In contrast, the interactions of individual OH groups of PCA and DCA or carboxylic groups of CA, TCA, PCA and DCA with MnO₂ particles were relatively weak.

Fig. 3 Deposit mass for 4 g L⁻¹ MnO₂ suspension in water at a deposition voltage of 20 V: (A) versus CA concentration at a deposition time of 3 min and (B) versus deposition time for CA concentration of 0.4 g L⁻¹.

The possibility of dispersion of MWCNT using cationic BAC and dispersion of MnO₂ using anionic CA paved the way for the formation of composite materials (Fig. 4). This approach is based on the ion-pairing of BAC and CA, which resulted in attraction of positively charged MWCNT and negatively charged MnO₂. It is known that ion-pair assemblies of aromatic ion-pairing agents and surfactants, containing a benzene ring, are governed by the electrostatic interactions and π–π interactions.⁴⁸ The attraction of MWCNT and CA allowed the formation of a composite with improved mixing of the individual components. SEM observation of the composite material showed MnO₂ nanoparticles attached to the non-agglomerated MWCNT (Fig. 5), which formed a 3-D conductive network. In contrast, the composites, prepared without dispersants, showed large agglomerates of MWCNT and MnO₂ (Fig. S5†).

Fig. 4 Fabrication of MnO₂–MWCNT composite by dispersion of MWCNT using cationic BAC, dispersion of MnO₂ using anionic CA and electrostatic heterocoagulation.

Fig. 5 SEM image of MnO₂–MWCNT powder.

The improved mixing of MnO₂ and MWCNT allowed good electrochemical performance of the composite electrodes. Fig. 6 compares CVs at different scan rates for electrodes with different MWCNT contents, prepared without and with dispersing agents. The electrodes, prepared without dispersants showed tilted CVs with low area, indicating poor capacitive performance. The electrodes, prepared using BAC and CA, showed significantly larger CV areas, indicating higher capacitance. The electrodes, prepared using 20% MWCNT showed nearly ideal box shape CVs. The CV areas increased with increasing scan rate, indicating good capacitance retention. It is important to note that good capacitive behavior was achieved at active mass loadings of 40 mg cm⁻², which meet requirements^21,22 for practical applications in electrodes of ES. The specific capacitances, calculated from the CV data are presented in Fig. 7.

Fig. 6 CVs for electrodes with mass loading of 40 mg cm⁻², prepared (A–C) without dispersants and (D–F) with BAC and CA dispersants, containing (A and D) 85% MnO₂ and 15% MWCNT, (B and E) 80% MnO₂ and 20% MWCNT and (C and F) 75% MnO₂ and 25% MWCNT at scan rates of (a) 2, (b) 5, (c) 10 and (d) 20 mV s⁻¹ in 0.5 M aqueous Na₂SO₄ electrolyte.

Fig. 7 C_s and C_m, calculated from the CV data, for different electrodes with mass loading of 40 mg cm⁻²: (a) pure MnO₂ and (b–g) different composite electrodes, prepared (b–d) without dispersants, (e–g) using BAC and CA dispersants, containing (b and e) 85% MnO₂ and 15% MWCNT, (c and f) 80% MnO₂ and 20% MWCNT and (d and g) 75% MnO₂ and 25% MWCNT.

Pure MnO₂ electrodes, prepared without MWCNT showed a specific capacitance of 1.19 F cm⁻² (29.8 F g⁻¹) at a scan rate of 2 mV s⁻¹. The capacitance decreased with increasing scan rate. The composite electrodes, prepared without BAC and CA showed capacitances of 4.29 F cm⁻² (107.2 F g⁻¹), 4.30 F cm⁻² (107.5 F g⁻¹), 3.98 F cm⁻² (99.5 F g⁻¹) for MWCNT content of 15, 20 and 25%, respectively, at a scan rate of 2 mV s⁻¹. However, the capacitance decreased rapidly with increasing scan rate. The electrodes, prepared using BAC and CA showed specific capacitances of 5.16 F cm⁻² (129.1 F g⁻¹), 5.26 F cm⁻² (131.6 F g⁻¹), 4.95 F cm⁻² (123.7 F g⁻¹) for MWCNT content of 15, 20 and 25%, respectively at a scan rate of 2 mV s⁻¹. It is important to note that the addition of MWCNT results in improved electronic conductivity of the composite and higher capacitance (eqn (1)). However the specific capacitance of MWCNT is significantly lower than that of MnO₂. The increase in MWCNT content in the composite above the optimum amount can result in decreasing capacitance. Therefore, the lower capacitance of the electrode, containing 25% MWCNT, compared to the capacitance of the electrode, containing 20% MWCNT, at a scan rate of 2 mV s⁻¹ can be attributed to lower capacitance of MWCNT. The composite electrodes, prepared using BAC and CA showed capacitance retention of 31.2, 34.8 and 47.8%, for MWCNT content of 15, 20 and 25%, respectively, at a scan rate of 100 mV s⁻¹. It is important to note that at scan rates above 20 mV s⁻¹ the composite electrodes, prepared using BAC and CA showed 2–3 times higher capacitance, compared to the capacitance of the electrodes prepared without dispersants. The electrodes, containing BAC and CA, showed significant improvement in area normalized capacitance at high scan rates, compared to other investigations.^23,49 It is important to note that good electrochemical performance at high scan rates was achieved at relatively high mass loadings, which are important for practical applications. The high capacitance, achieved at high scan rates, is especially important for the development of high power devices.

The analysis of the impedance data (Fig. 8) indicated that the addition of MWCNT to MnO₂ resulted in reduced resistance R = Z′. The electrodes prepared using BAC and CA (Fig. 8B(e–g)) showed lower resistance, compared to the electrodes prepared without dispersants (Fig. 8B(b and c)). The difference can be attributed to better mixing of the components in the composite materials. Fig. 8C and D show frequency dependencies of the components of complex capacitance C* = C′ − iC′′, calculated from the impedance data. The composite MnO₂–MWCNT electrodes showed significantly higher capacitance C′, compared to the capacitance of pure MnO₂ at low frequencies. The frequency dependences of capacitances for composite electrodes showed typical relaxation type dispersions,⁵⁰ as indicated by the reduction of C′ with increasing frequency and corresponding maxima in the frequency dependencies of C′′.

Fig. 8 (A and B) Nyquist plots of complex impedance and (C) C^′_s and (D) C^′′_s components of complex capacitance versus frequency for electrodes with mass loading of 40 mg cm⁻²: (a) pure MnO₂ and (b–g) composites, prepared (b–d) without dispersants and (e–g) using CA and BAC dispersants, containing (b and e) 85% MnO₂ and 15% MWCNT, (c and f) 80% MnO₂ and 20% MWCNT and (d and g) 75% MnO₂ and 25% MWCNT.

The electrodes, prepared using BAC and CA, showed relaxation type dispersion at higher frequencies, as illustrated by the reduction in C′ at higher frequencies and corresponding shift of relaxation maxima of C′′. This result is in agreement with the data, presented in Fig. 7, which indicated improved capacitance retention at high scan rates for the electrodes, prepared in the presence of dispersants.

The composite MnO₂–MWCNT positive electrodes, containing 20% MWCNT, and activated carbon–carbon black (AC–CB, 10% CB) negative electrodes were used for the fabrication of asymmetric devices. The method of preparation of AC–CB electrodes, detailed electrochemical and impedance characterization data for different compositions are presented in ESI (Fig. S6–S9).† The asymmetric device shoved a box shape CV (Fig. 9) in a voltage window of 1.8 V. The hydrogen and oxygen evolution at electrodes at higher voltages limits the voltage windows of the asymmetric devices for practical applications.⁵¹ The increase in scan rate resulted in increasing current, indicating good capacitive behavior. The capacitance of the asymmetric cells, calculated from the CV data decreased from 2.3 to 0.9 F cm⁻² with increasing scan rate from 2 to 100 mV s⁻¹. Such decrease can be attributed to electrolyte diffusion limitations in pores of individual electrodes. The components of complex AC capacitance of the asymmetric devices were calculated from the impedance data at different frequencies (Fig. 10). The capacitance C′ of 1.6 F cm⁻² was obtained at a frequency of 10 mHz. The frequency dependence of capacitance showed typical relaxation type dispersion, as indicated by decrease in C′ with increasing frequency and corresponding maximum in the frequency dependence of C′′. It is important to note that capacitance, calculated from the CV data depends on scan rate, whereas AC capacitance depends on frequency. The specific capacitance was also calculated from the chronopotentiometry data. Fig. 11 shows typical charge–discharge curves for the asymmetric coin cells. The charge–discharge curves at different current densities were of symmetric triangular shape, indicating good Coulombic efficiency. The capacitance calculated from the chronopotentiometry data decreased from 1.36 to 1.09 F cm⁻² with increasing discharge current in the range of 1–10 mA (Fig. 11).

Fig. 9 CVs for the two electrode asymmetric cell, containing a positive MnO₂–MWCNT electrode (80% MnO₂ and 20% MWCNT) and a negative AC–CB electrode (90% AC and 10% CB) at scan rates of (a) 2, (b) 5, (c) 10 and (d) 20 mV s⁻¹ in 0.5 M Na₂SO₄ electrolyte.

Fig. 10 Frequency dependencies of components of complex capacitance (A) C^′_s, C^′_m and (B) C^′′_s, C^′′_m, calculated from the impedance data for the two electrode asymmetric cell, containing a positive MnO₂–MWCNT electrode (80% MnO₂ and 20% MWCNT) and a negative AC–CB electrode (90% AC and 10% CB).

Fig. 11 (A) Galvanostatic charge–discharge behavior of an asymmetric coin cell at currents of (a) 1, (b) 2, (c) 3, (d) 4, (e) 5, (f) 6, (g) 8, and (h) 10 mA, (B) cell capacitance C_s and C_m versus current, inset shows coin cells.

Fig. 12A shows Ragone plot for the coin cell asymmetric device. The maximum energy density of 10.2 W h kg⁻¹ and power density of 2.67 kW kg⁻¹ were achieved. The investigation of the cyclic stability of the coin cells showed capacitance retention of 95.1% after 1000 cycles (Fig. 12B). The method, developed in this investigation was used for the fabrication of larger envelope cells with area of 6 cm². Two envelope cells were used for powering of 19 LED bulbs with a nominal current of 20 mA (Fig. 13). The results indicated that the asymmetric ES cells, developed in this investigation, are promising for practical applications.

Fig. 12 (A) Ragone plot and (B) capacitance retention versus cycle number for an asymmetric coin cell, inset shows multiple charge–discharge cycles at a constant current of 10 mA.

Fig. 13 Two envelope asymmetric cells power 19 LED bulbs with a nominal current of 20 mA for each bulb.

4. Conclusions

The adsorption of BAC on MWCNT in water allowed efficient dispersion of MWCNT, imparted a positive charge to MWCNT and allowed cathodic EPD of MWCNT films. The adsorption of CA on MWCNT in water allowed efficient dispersion of MnO₂, imparted a negative charge to MnO₂ and allowed anodic EPD of MnO₂ films. Composite MnO₂–MWCNT electrodes can be obtained by heterocoagulation, based on ion-pairing assembly of BAC and CA. The method allowed improved mixing of the individual components, which resulted in superior electrochemical performance of the MnO₂–MWCNT electrodes. High capacitance and good capacitance retention can be achieved at high active mass loadings. The asymmetric devices, based on MnO₂–MWCNT positive electrodes and AC–CB negative electrodes showed good capacitive behavior in a voltage window of 1.8 V. The specific power of 2.67 kW kg⁻¹ and specific energy 10.2 W h kg⁻¹ were achieved. The devices showed good capacitance retention at high charge–discharge rates and good cyclic stability. The ES devices are promising for practical applications.

Acknowledgements

The authors gratefully acknowledge the Natural Sciences and Engineering Research Council of Canada for the financial support and Vale Canada for the donation of Ni foam.

References

1. J. Zhao, Z. Lu, M. Shao, D. Yan, M. Wei, D. G. Evans and X. Duan, RSC Adv., 2013, 3, 1045–1049.
2. S. Roshny, R. Ranjusha, M. S. Deepak, N. Sanoj Rejinold, R. Jayakumar, S. V. Nair and A. Balakrishnan, RSC Adv., 2014, 4, 15863–15869.
3. W. Wei, X. Cui, W. Chen and D. G. Ivey, Chem. Soc. Rev., 2011, 40, 1697–1721.
4. C. Xu, F. Kang, B. Li and H. Du, J. Mater. Res., 2010, 25, 1421.
5. F. Wang, S. Xiao, Y. Hou, C. Hu, L. Liu and Y. Wu, RSC Adv., 2013, 3, 13059–13084.
6. S. Devaraj and N. Munichandraiah, Electrochem. Solid-State Lett., 2005, 8, A373–A377.
7. R. Jiang, T. Huang, Y. Tang, J. Liu, L. Xue, J. Zhuang and A. Yu, Electrochim. Acta, 2009, 54, 7173–7179.
8. J. Yan, Z. Fan, T. Wei, J. Cheng, B. Shao, K. Wang, L. Song and M. Zhang, J. Power Sources, 2009, 194, 1202–1207.
9. J.-G. Wang, Y. Yang, Z.-H. Huang and F. Kang, J. Power Sources, 2013, 224, 86–92.
10. S. Hassan, M. Suzuki, S. Mori and A. A. El-Moneim, RSC Adv., 2014, 4, 20479–20488.
11. Y. Gao, Y. S. Zhou, M. Qian, H. M. Li, J. Redepenning, L. S. Fan, X. N. He, W. Xiong, X. Huang, M. Majhouri-Samani, L. Jiang and Y. F. Lu, RSC Adv., 2013, 3, 20613–20618.
12. H. Xia, Y. S. Meng, G. Yuan, C. Cui and L. Lu, Electrochem. Solid-State Lett., 2012, 15, A60–A63.
13. H. Xia, B. Li and L. Lu, RSC Adv., 2014, 4, 11111–11114.
14. V. Khomenko, E. Raymundo-Piñero and F. Béguin, J. Power Sources, 2006, 153, 183–190.
15. H. Xia, Y. Wang, J. Lin and L. Lu, Nanoscale Res. Lett., 2012, 7, 1–10.
16. H. Xia, M. Lai and L. Lu, J. Mater. Chem., 2010, 20, 6896–6902.
17. S.-B. Yoon, J.-P. Jegal, K. C. Roh and K.-B. Kim, J. Electrochem. Soc., 2014, 161, H207–H213.
18. W. Jiang, K. Zhang, L. Wei, D. Yu, J. Wei and Y. Chen, Nanoscale, 2013, 5, 11108–11117.
19. C.-W. Lee, S.-B. Yoon, S.-M. Bak, J. Han, K. C. Roh and K.-B. Kim, J. Mater. Chem. A, 2014, 2, 3641–3647.
20. Y. Wang, H. Liu, X. Sun and I. Zhitomirsky, Scr. Mater., 2009, 61, 1079–1082.
21. Y. Gogotsi and P. Simon, Science, 2011, 334, 917–918.
22. L. Wei and G. Yushin, Nano Energy, 2012, 1, 552–565.
23. J. Li, Q. M. Yang and I. Zhitomirsky, J. Power Sources, 2008, 185, 1569–1574.
24. M. Cheong and I. Zhitomirsky, Surf. Eng., 2009, 25, 346–352.
25. H. Ohshima, Colloids Surf., A, 1995, 103, 249–255.
26. H. Ohshima, J. Colloid Interface Sci., 2001, 233, 142–152.
27. I. Zhitomirsky, Adv. Colloid Interface Sci., 2002, 97, 277–315.
28. P. Taberna, P. Simon and J. F. Fauvarque, J. Electrochem. Soc., 2003, 150, A292–A300.
29. M. Ata, Y. Liu and I. Zhitomirsky, RSC Adv., 2014, 4, 22716–22732.
30. H. Wang, Curr. Opin. Colloid Interface Sci., 2009, 14, 364–371.
31. M. Islam, E. Rojas, D. Bergey, A. Johnson and A. Yodh, Nano Lett., 2003, 3, 269–273.
32. V. Datsyuk, P. r. Landois, J. Fitremann, A. Peigney, A. M. Galibert, B. Soula and E. Flahaut, J. Mater. Chem., 2009, 19, 2729–2736.
33. O. Matarredona, H. Rhoads, Z. Li, J. H. Harwell, L. Balzano and D. E. Resasco, J. Phys. Chem. B, 2003, 107, 13357–13367.
34. Q. Wang, Y. Han, Y. Wang, Y. Qin and Z.-X. Guo, J. Phys. Chem. B, 2008, 112, 7227–7233.
35. R. Hosseinzadeh, R. Tahmasebi, K. Farhadi, A. A. Moosavi-Movahedi, A. Jouyban and J. Badraghi, Colloids Surf., B, 2011, 84, 13–17.
36. H. Akbas and T. Taner, Spectrochim. Acta, Part A, 2009, 73, 150–153.
37. A. C. R. Gomes, L. D. F. Cafer, P. Homem-de-Mello, M. D. Coutinho-Neto and I. Gaubeur, J. Colloid Interface Sci., 2012, 367, 370–377.
38. A. Ali, S. Uzair, N. A. Malik and M. Ali, J. Mol. Liq., 2014, 196, 395–403.
39. S. Rudiuk, M. Delample, S. Franceschi-Messant, N. Chouini-Lalanne, E. Perez, J.-C. Garrigues and I. Rico-Lattes, J. Dispersion Sci. Technol., 2010, 31, 384–391.
40. S. R. Yousefi, E. Zolfonoun, M. R. Pourjavid and S. J. Ahmadi, Anal. Methods, 2014, 6, 3694–3699.
41. E. L. Tan, J.-C. Liu and Y. W. Chien, J. Dispersion Sci. Technol., 1993, 14, 609–624.
42. J. H. Waite, Nat. Mater., 2008, 7, 8–9.
43. L. Haeshin, S. M. Dellatore, W. M. Miller and P. B. Messersmith, Science, 2007, 318, 426–430.
44. X. Fan, L. Lin and P. B. Messersmith, Compos. Sci. Technol., 2006, 66, 1198–1204.
45. Y. Wang and I. Zhitomirsky, J. Colloid Interface Sci., 2012, 380, 8–15.
46. B. P. Lee, P. B. Messersmith, J. N. Israelachvili and J. H. Waite, Annu. Rev. Mater. Res., 2011, 41, 99–132.
47. I. A. Jankovic, Z. V. Saponjic, M. I. Comor and J. M. Nedeljkovic, J. Phys. Chem. C, 2009, 113, 12645–12652.
48. A. A. Shahir, M. Rashidi-Alavijeh, S. Javadian, J. Kakemam and A. Yousefi, Fluid Phase Equilib., 2011, 305, 219–226.
49. Y. Wang, Y. Liu and I. Zhitomirsky, J. Mater. Chem. A, 2013, 1, 12519–12526.
50. K. Shi and I. Zhitomirsky, J. Mater. Chem. A, 2013, 1, 11614–11622.
51. T. Brousse, M. Toupin and D. Belanger, J. Electrochem. Soc., 2004, 151, A614–A622.

---

Footnote

† Electronic supplementary information (ESI) available: Chemical structures of molecules, deposition yield data, SEM data, fabrication and testing of AC–CB electrodes. See DOI: 10.1039/c4ra05879h

---