Performance of Manganese Oxide/CNTs Composites as Electrode Materials for Electrochemical Capacitors

Figure 1 shows the adsorption isotherm of the prepared manganese oxide. The isotherm is typical for a mesoporous material with a hysteresis loop at high partial pressures. The slight adsorption at low relative pressure indicates some moderate amount of micropores in this material. A surface area of 220 m²/g was calculated by applying the BET equation to the adsorption data, that fits with previous reports.^{16 17 21} The material obtained has a relatively high surface area, considering that it is a metallic oxide, what is interesting for its further application as the electrode in electrochemical capacitors.

Zoom In Zoom Out Reset image size

The XRD Cu Kα pattern of manganese oxide did not show any discernible reflection, which confirms its amorphous nature. The amorphous structure will favor the fast proton/ions diffusion rates. Thus, an important characteristic needed for a metallic oxide to be used as electrode materials is fulfilled.⁴

Thermal analysis was also performed on the oxide in order to characterize the chemical nature of water. Figure 2 shows the weight loss during a heat-treatment from room temperature up to 800°C in a nitrogen atmosphere. The gases evolved during the treatment were analyzed by mass spectroscopy and only water was detected (Fig. 2). The weight loss of the material, ca. 11%, occurs in a single step between 80 and 400°C. Since water evolution is continuous and mostly above 100°C, that indicates that it is not adsorbed water, but water in the oxide structure. Hence, the prepared oxide is hydrated, which will favor a pseudocapacitive behavior.^{5 7} These different analyses confirm that the material prepared is amorphous and hydrated manganese oxide with a high surface area.

Zoom In Zoom Out Reset image size

The resistance of amorphous was estimated from the ohmic drop between galvanostatic charge and discharge^{36 37} of a symmetric capacitor built with two manganese oxide electrodes (film Teflon) of 1 cm² surface in 1 mol L⁻¹ as the electrolyte. The resistance value found was around 2000 Ω cm². Thus, the electric conductivity of this material is not high enough to be used as a pure electrode component of an electrochemical capacitor. Generally, carbon black is used in electrochemical cells to enhance the conductivity of the electrodes. In the present work, we want to determine if multiwalled CNTs could be more efficient than carbon black for this purpose. Indeed, the SEM image in Fig. 3a shows long and flexible nanotubes, highly entangled. Moreover, as shown by the TEM picture in Fig. 3b, the CNTs are formed by continuous layers surrounding the central canal, that is favorable for a good electrical conductivity of each nanotube. Therefore, we can predict that this kind of network of conducting CNTs should be able to perfectly connect the particles in the /CNTs composite.

Zoom In Zoom Out Reset image size

Figure 1 presents the adsorption isotherm for the pristine carbon nanotubes. The shape of the curve corresponds to a mesoporous material and, considering that the tips of these nanotubes are closed, the mesopores are due to the open network of entangled nanotubes. The calculated BET specific surface area is 320 m²/g. Figure 1 also includes the isotherm for the /CNTs composite prepared with 15 wt % of nanotubes. The BET specific surface area of the composite is similar to that of pure manganese oxide, ca. 210 m²/g. However, although the carbon nanotubes content is only 15 wt %, the isotherm of the composite is not similar to that of raw manganese oxide (Fig. 1). Actually, the shapes of both adsorption isotherms for the composite and the pristine nanotubes are identical, indicating that the open mesoporous network formed by the nanotubes is maintained after the addition of manganese oxide.

Figure 4 shows the pore size distribution obtained from the isotherms presented in Fig. 1 by applying the Barret-Joyner-Halendo (BJH) method. This analysis confirms that the mesopore size distribution of the /CNTs composite and CNTs are comparable. Moreover, compared with the /CNTs composite presents an additional mesoporosity. We can postulate that, due to this open network, the ions accessibility from the electrolyte to the active mass of the /CNTs composite should be favored, while keeping the advantage of the excellent electrical conductivity provided by the nanotubes. Moreover, the flexibility and entanglement of the nanotubes should ensure good mechanical properties of the /CNTs composite electrodes. In the following part of this manuscript, we show that all these assumptions are proved to a large extent.

Zoom In Zoom Out Reset image size

Figure 5 shows the cyclic voltammograms (CVs), with a scan rate of potential of 2 mV/s, for pure (pressing a mixture of 90% active PVDF) and for the CNTs (pressing a mixture of 90% active PVDF) in 1 mol L⁻¹ electrolyte. It can be observed that presents a very high resistance and the specific capacitance obtained is only 0.1 F/g, because the electrolyte probably does not penetrate in the bulk of the material. On the contrary, carbon nanotubes present the characteristic box-like shape of an ideal capacitor, but the specific capacitance obtained is not higher than 20 F/g. Therefore, carbon nanotubes cannot be considered as an electrochemical active mass giving a significant additional capacitance in the composites.

Zoom In Zoom Out Reset image size

Table I displays the specific capacitance of composites prepared with different nanotube loadings during the synthesis of and the resistance estimated from the galvanostatic charge/discharge cycling experiments. It is clearly seen that the addition of carbon nanotubes improves the behavior of as a capacitor electrode. The specific capacitance values referred to the oxide increase with the amount of CNTs, indicating that more energy can be extracted from manganese oxide, in other words that the active mass of is more accessible with a large enough amount of CNTs. In addition, the resistance decreases when increasing the amount of CNTs, proving their efficiency as a conductivity agent. However, when the specific capacitance is referred to the total mass of electrode material, to be realistic, it can be noticed that carbon nanotubes loadings higher than 10-15 wt % are not improving the performance of the electrodes because they are not electrochemically active enough by themselves. Therefore, 10-15 wt % of CNTs as additive are sufficient to increase the capacitance of from 0.1 to 140 F/g. Hence, carbon nanotubes are found to be useful as additives for manganese oxide, but it is necessary to test if they present any advantage over another conventional additive, as carbon black.

Table I. 

Specific capacitance in 1 mol L−1 medium for loaded with different weight percentages of carbon nanotubes and capacitor resistance.
Carbon nanotubesloading(% wt)	Specificcapacitance	Specificcapacitance	Resistance(Ω cm2)
0	0.1	0.1	2000
5	19	20	625
10	137	154	5.2
15	137	161	4.0
20	139	174	3.8
25	141	188	3.5
30	138	197	3.0
40	140	234	2.7
50	138	277	2.5
100	20	⋯	2.0

The comparative voltamperometric analyses presented in Fig. 6 for electrodes loaded either with 15 wt % of CNTs or with 15 wt % of carbon black demonstrate that the electrochemical behavior of both composites is completely different. The capacitor built from the material prepared with carbon black is highly resistive, presenting a CV curve very far from the ideal box-like shape. The obtained capacitance with this composite material is not higher than ∼70 F/g. On the contrary, when carbon nanotubes are used as an additive to the resistance of the electrodes is much lower than in the case of using the conventional carbon black, finding a performance closer to an ideal capacitor. Hence, the conductivity of was very effectively enhanced by the presence of carbon nanotubes. Additionally, another highly valuable effect of the use of CNTs is the possibility of extracting more energy from the electrodes, obtaining specific capacitance values per gram of electrode material of ∼140 F/g.

Zoom In Zoom Out Reset image size

The scanning electron microscopy (SEM) analyses helped to better understand the different behaviors of both composite materials. The SEM pictures of the /carbon black and /CNTs composites, both with 15 wt % of carbon additive, are shown in Fig. 7. Figure 7a shows that the carbon black and manganese oxide particles are homogeneously mixed. In the case of CNTs (Fig. 7b), the manganese oxide particles are integrated and surrounded in the entangled network of nanotubes in agreement with the adsorption results presented in Fig. 4. In both cases the dispersion of the carbon additive is very good. However, remarkable differences between both materials are observed on the surface of the film electrodes prepared with 10 wt % of binder. Figure 7c shows that when using carbon black as an additive, the surface of the film electrode is not very homogeneous and the connectivity of manganese oxide particles is rather poor. On the contrary, a film electrode prepared in similar conditions with the same proportion of carbon nanotubes presents a more homogeneous surface (Fig. 7d). This better homogeneity is attributed to the entangled network of CNTs which improves the connectivity of the particles. As a consequence of this good connectivity between the particles, the conductivity is improved, as observed in the voltammograms of Fig. 6, by comparison with a traditional additive as carbon black. Additionally, the carbon nanotubes generate an open mesoporous network which facilitates the contact between the electrolyte and the active mass of the nanocomposite, allowing more energy to be extracted from manganese oxide.

Zoom In Zoom Out Reset image size

Another important aspect for a material to be used as a capacitor electrode is the electrochemical stability through repeated charge/discharge cycling. Figure 8 presents the specific discharge capacitance vs. the cycle number for (with 15 wt % of CNTs) in 1 mol L⁻¹ at two different pHs. At pH 6.4, the specific capacitance loss reaches 20% after 200 cycles. However, if the pH of the electrolyte is fixed at 10 by adding NaOH, the stability is remarkably increased, with a loss of only 4% of the initial specific capacitance after 200 cycles.

Zoom In Zoom Out Reset image size

In order to explain this behavior, it has to be considered that the pseudocapacitance of hydrous oxides like is attributed to reversible redox transitions involving protons and/or cations exchange with the electrolyte following the equation²⁰

where and indicate interfacial at high and low oxidation states, respectively. However, due to the solubility of either Mn(II) or Mn(VII) in the electrolytic solution, the irreversible reactions Mn(IV) to Mn(II) and Mn(IV) to Mn(VII) should be avoided during charging and discharging the capacitor. If one of these reactions takes place, it will result in a partial dissolution of the electrode and a subsequent decrease of capacitance. From the Pourbaix potential-pH diagram for manganese,³⁸ it is possible to determine that the reaction Mn(IV) to Mn(II) occurs at potentials of 0.47 and 0.05 V (vs. NHE) for pH 6.4 and 10, respectively. In the case of the transformation of Mn(IV) into Mn(VII) the equilibrium potentials are 1.19 and 0.98 V (vs. NHE) for pH 6.4 and 10, respectively. Therefore, in order to confirm that the behavior found in Fig. 8 is related to some dissolution of we have measured the potential of both electrodes when the cell voltage varies between 0 and 0.6 V during the galvanostatic charge-discharge cycling in the two aqueous solutions of different pH.

For this purpose, we have used a three-electrode Swagelok-type cell consisting of two electrodes from the /CNTs composite, as the working and the auxiliary ones, and a reference electrode. Figure 9 presents, for the two electrolytic media, the variation of the cell voltage during charge-discharge cycling at a constant current density of 100 mA/g (taking into account the total mass of the electrode) and the potential profile for both positive and negative electrodes. Figure 9a shows that with the electrolyte of pH 6.4, the initial potential for both electrodes in the discharged state of the capacitor is 0.76 V vs. NHE. After charging the capacitor to 0.6 V, the potential of the positive electrode increases up to 1.04 V, while for the negative one it decreases from 0.76 to 0.44 V vs. NHE. Since the irreversible reduction of Mn(IV) into Mn(II) starts to occur at 0.47 V for a pH 6.4,³⁸ it demonstrates that the negative electrode is actually working below this potential, which results in a partial dissolution of the active material in the electrolyte. Hence, the capacitance decrease observed in Fig. 8 during repeated cycling of a capacitor in this medium is due to the deterioration of the negative electrode. On the contrary, from Fig. 9b it is clear that at pH 10 both electrodes work in the electrochemical stability window where there is not any irreversible reaction producing either Mn(II) or Mn(VII). In this electrolyte, the initial potential for both electrodes of the discharged capacitor is 0.66 V vs. NHE and, after charging the capacitor up to 0.6 V, the potential of the negative electrode decreases to 0.38 V, very far from the value of 0.05 V needed for the destructive reaction Mn(IV) to Mn(II) at pH 10.³⁸ This fact explains the good cycle life observed in Fig. 8 for a capacitor based on and an electrolyte of pH 10.

Zoom In Zoom Out Reset image size

The results presented above demonstrate that irreversible electrode reactions leading to a specific capacitance loss with time can be avoided by controlling the pH of the electrolytic solution. However, Fig. 9b shows that using the electrolyte with pH 10, the positive electrode is just working at the limit of the irreversible transformation of manganese oxide to Mn(VII). As a consequence, the operating voltage of this capacitor cannot be increased to a value higher than 0.6 V.

The value of specific capacitance obtained in a two-electrode cell for amorphous manganese oxide prepared with 15 wt % of carbon nanotubes, ca. 140 F/g (of total electrode mass), seems to be lower than some of the values presented in the literature for amorphous manganese oxide prepared with 25 wt % of acetylene black, ca. 200 F/g.^{6 21} These promising results were obtained in three-electrode cell devices, as all the other studies on the use of manganese oxide as an electrode material for capacitors.^{16 17 18 19 20 21 22 23 24 25} The following part demonstrates that the values of specific capacitance are overestimated when they are obtained from a three-electrode configuration.

In a two-electrode cell, each electrode works in a different potential range. Taking into account the experiment presented in Fig. 9a for pH 6.4, it is very easy to observe that after charging the capacitor to 0.6 V, the polarization of the negative electrode V) is larger than the polarization of the positive electrode V). The specific capacitance calculated per gram of electrode material is 133 and 144 F/g for the negative and the positive electrode, respectively. The same effect is observed in Fig. 9b for pH 10, where the negative electrode is working in a potential range of 0.28 V giving a specific capacitance of 150 F/g, while the range is 0.32 V for the positive electrode, giving a specific capacitance of 134 F/g. Hence, it is clear that the specific capacitance of each electrode depends on its working potential range. In addition, according to the known relationship

the capacitance measured in a two-electrode cell is mainly imposed by the electrode with the smallest capacitance.

Completely different capacitance values are determined when the /CNTs composite is tested in a three-electrode cell being used only as a working electrode, with a Pt wire as an auxiliary electrode and as the reference electrode. Figure 10 presents the cyclic voltammograms of the /CNTs composite obtained at pH 6.4 and 10 with a 2 mV/s scan rate of potential. First, Fig. 10 confirms that the possible working potential of depends on the pH of electrolyte. At pH 10, can be polarized with smaller potential values than at pH 6.4. In the latter case, some redox process starts to take place at ca. 0.4 V, as shown by the peak appearing from this potential, due to the Mn(IV) to Mn(II) transformation. An opposite behavior is rather found under positive polarization at pH 6.4, for which the electrode can be reversibly polarized at potentials as high as 1.2 V, whereas at pH 10, the potential corresponding to the Mn(IV) to Mn(VII) transformation should not exceed 1.0 V.

Zoom In Zoom Out Reset image size

When the specific capacitance is calculated from the CV curves presented in Fig. 10 for the three-electrode cell, the values obtained are 230 F/g (total mass of electrode) and 200 F/g for pH 6.4 and 10, respectively, that are similar to the maximum values presented by other authors.^{16 21} The reasons for such noticeable differences depending on whether using a two- or a three-electrode cell can be interpreted by comparing the data presented in Fig. 9 and 10. In the three-electrode cell, the total potential range of the working electrode is much higher than the operating potential range of each electrode in the two-electrode cell (ca. 0.3 V). In an ideal capacitor, the capacitance should not depend on the potential window. However, when the electrode is from a material with pseudo-capacitance properties such as the capacitance is the result of the overlapping of several redox reactions, each of them occurring in a given potential range. Thus, taking into account that operating with a higher potential range, additional redox processes can take place, this explains why high values of specific capacitance are only determined in the three-electrode configuration. Hence, when a material with pseudo-capacitance properties, e.g., is used for both electrodes of a symmetric capacitor, the data found with a three-electrode configuration cannot be extrapolated.

Although our results state clearly that symmetric capacitors based on can operate reversibly in a voltage of 0.6 V, Fig. 10 shows that at pH 6.4 the /CNTs composite remains a very promising material for a positive electrode, allowing the polarization to be extended up to 1.2 V. Taking into account that is irreversibly dissolved at low potentials, an interesting alternative to circumvent this problem in a neutral medium would be to use only for the positive electrode and another material for the negative electrode. Such so-called asymmetric systems have been already successfully tested with conducting polymers^{39 40 41} and activated carbon as the positive and negative electrodes, respectively. Recently, it has been confirmed that a relatively high voltage can be reached in an asymmetric system based on manganese oxide as the positive electrode, and activated carbon as the negative one.^{42 43} We are presently investigating this system in order to better understand the respective roles of both electrodes in the charge-storage mechanism.