Facile Route for Fabrication of Ferrimagnetic Mn 3 O 4 Spinel Material for Supercapacitors with Enhanced Capacitance

: The purpose of this investigation was the development of a new colloidal route for the fabrication of Mn 3 O 4 electrodes for supercapacitors with enhanced charge storage performance. Mn 3 O 4 -carbon nanotube electrodes were fabricated with record-high capacitances of 6.67 F cm − 2 obtained from cyclic voltammetry tests at a scan rate of 2 mV s − 1 and 7.55 F cm − 2 obtained from the galvanostatic charge–discharge tests at a current density of 3 mA cm − 2 in 0.5 M Na 2 SO 4 electrolyte in a potential window of 0.9 V. The approach involves the use of murexide as a capping agent for the synthesis of Mn 3 O 4 and a co-dispersant for Mn 3 O 4 and carbon nanotubes. Good electrochemical performance of the electrode material was achieved at a high active mass loading of 40 mg cm − 2 and was linked to a reduced agglomeration of Mn 3 O 4 nanoparticles and efﬁcient co-dispersion of Mn 3 O 4 with carbon nanotubes. The mechanisms of murexide adsorption on Mn 3 O 4 and carbon nanotube are discussed. With the proposed method, the time-consuming electrode activation procedure for Mn 3 O 4 electrodes can be avoided. The approach developed in this investigation paves the way for the fabrication of advanced cathodes for asymmetric supercapacitors and multifunctional devices, combining capacitive, magnetic, and other functional properties. formal analysis, W.Y.; investigation, W.Y.; resources, M.N. and I.Z.; data curation, W.Y.; writing—original draft preparation, W.Y., M.N. and I.Z.; writing—review and editing, W.Y. and I.Z.; visualization, W.Y.; supervision, I.Z.; project administration, I.Z.; I.Z.


Introduction
In recent years, advanced materials have emerged for energy storage in supercapacitors, including metal oxides, conductive polymers, graphene, and other carbon materials, MXenes, complex hydroxides, and composites [1][2][3][4][5][6][7]. Oxide materials such as MnO 2 , Fe 3 O 4 , BiMn 2 O 5 , and V 2 O 3 are increasingly being explored due to their large potential windows and high capacitance [6]. Spinel-type oxide materials have generated significant interest due to their promising performance and beneficial materials science aspects [8][9][10][11]. Atoms of transition metal elements with different valence states were incorporated into the spinel structure and exhibited redox behavior, imparting advanced pseudocapacitive properties to the spinel oxides [8,[12][13][14]. A large pool of spinel oxides provides a basis for the fabrication of spinel solid solutions with advanced properties [15]. Solid solutions allow a significant improvement in functional properties of materials by a controlled variation in their composition. Solid solutions are widely utilized in energy storage and other fields and often outperform individual spinel oxides for various applications [16][17][18][19]. Spinel materials are of particular interest because they exhibit advanced magnetic, catalytic, and other properties and can be used for the fabrication of multifunctional materials [14,[20][21][22][23][24]. Mn 3 O 4 is a spinel-type ferrimagnetic material that is widely used for the fabrication of advanced spinel solid solutions with enhanced magnetization, catalytic, and energy storage properties in batteries [25][26][27]. However, the potential of Mn 3 O 4 for supercapacitor technology is only beginning to be recognized [28,29]. Mn 3 O 4 can potentially outperform MnO 2 , which is currently one of the best materials for cathodes of asymmetric supercapacitors. solution in DI water before pH adjustment as a capping agent for synthesis, and MWCNTs were added after the synthesis. In both methods, Mn 3 O 4 and MWCNTs were co-dispersed using murexide. The mass ratio of Mn 3 O 4 :CNT:murexide was 4:1:1. The obtained mixtures of Mn 3 O 4 with MWCNTs, containing murexide, were ultrasonicated to achieve improved dispersion and mixing, washed, and dried. Obtained powders were used for the fabrication of electrodes using slurries of Mn 3 O 4 and MWCNTs in ethanol with a PVB binder. The binder content was 3% of the total mass of Mn 3 O 4 and MWCNTs. The slurries were used for impregnation of commercial Ni foam (Vale, Canada) current collectors. The total mass of impregnated material after drying was 40 mg cm −2 .

Characterization Techniques
Electron microscopy studies were performed using a JEOL SEM (scanning electron microscope, JEOL, JSM-7000F). X-ray diffraction (XRD) analysis (diffractometer Bruker D8, UK) was performed using Cu-Kα radiation at the rate of 0.01 degrees per second. Electrochemical studies were performed in an aqueous 0.5 M Na 2 SO 4 electrolyte using PARSTAT 2273 (Ametek) potentiostat for cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS), and BioLogic VMP 300 potentiostat for the galvanostatic charge-discharge (GCD) investigations. Testing was performed using a 3-electrode electrochemical cell containing a working electrode (impregnated Ni foam), counter-electrode (Pt mesh), and reference electrode (SCE, saturated calomel electrode). The capacitive properties of electrode material were presented in gravimetric (C m , F g −1 ) and areal (C S , F cm −2 ) capacitance forms. Capacitances C m and C S were calculated from the CV, EIS, and GCD data as described in reference [6]. The capacitances calculated from the CV and GCD data represented the integral capacitances measured in a voltage window of 0-0.9 V versus SCE. The capacitances calculated from the EIS data represented differential capacitances measured at an open circuit potential at a voltage amplitude of 5 mV. CV testing procedures (TP) involved obtaining CV at scan rates of 2, 5, 10, 20, 50, and 100 mV s −1 . EIS measurements were performed after each TP. GCD measurements were performed after the last TP. The slurries were used for impregnation of commercial Ni foam collectors. The total mass of impregnated material after drying w

Characterization Techniques
Electron microscopy studies were performed using a JEOL microscope, JEOL, JSM-7000F). X-ray diffraction (XRD) analysi D8, UK) was performed using Cu-Kα radiation at the rate of 0.01 d trochemical studies were performed in an aqueous 0.5 M Na2SO STAT 2273 (Ametek) potentiostat for cyclic voltammetry (CV) and ance spectroscopy (EIS), and BioLogic VMP 300 potentiostat for t discharge (GCD) investigations. Testing was performed using a ical cell containing a working electrode (impregnated Ni foam mesh), and reference electrode (SCE, saturated calomel electrode ties of electrode material were presented in gravimetric (Cm, F g capacitance forms. Capacitances Cm and CS were calculated from data as described in reference [6]. The capacitances calculated fro represented the integral capacitances measured in a voltage wi SCE. The capacitances calculated from the EIS data represented measured at an open circuit potential at a voltage amplitude of dures (TP) involved obtaining CV at scan rates of 2, 5, 10, 20, 50, a urements were performed after each TP. GCD measurements w last TP.    The approach developed in this investigation was based on colloidal processing, which offers benefits for the fabrication of materials with advanced microstructures [36][37][38][39]. In colloidal processing methods, advanced capping agents and dispersants are necessary for the synthesis of nanomaterials and the fabrication of advanced composites. Capping agents and dispersants must be adsorbed on the particles. A non-adsorbed ionic species can stimulate particle agglomeration. Previous investigations [40] highlighted the benefits of chelating dispersants, which are strongly adsorbed on inorganic particles by bidentate bonding to the surface metal atoms. Figure 2A shows a chemical structure of murexide used in this investigation as a capping and dispersing agent. The approach developed in this investigation was based on colloidal processing, which offers benefits for the fabrication of materials with advanced microstructures [36][37][38][39]. In colloidal processing methods, advanced capping agents and dispersants are necessary for the synthesis of nanomaterials and the fabrication of advanced composites. Capping agents and dispersants must be adsorbed on the particles. A non-adsorbed ionic species can stimulate particle agglomeration. Previous investigations [40] highlighted the benefits of chelating dispersants, which are strongly adsorbed on inorganic particles by bidentate bonding to the surface metal atoms. Figure 2A shows a chemical structure of murexide used in this investigation as a capping and dispersing agent. Murexide exhibits chelating properties [41,42], which are related to its strong tridentate bonding to different metal atoms. Dissociated murexide acquires a negative charge in solutions ( Figure 2). It was hypothesized that murexide was adsorbed on the Mn3O4 surface by creating a tridentate bonding to Mn atoms on the surface ( Figure 2B). The adsorption of murexide on MWCNTs involved interactions [43] of two barbiturate rings of murexide with carbon rings of MWCNTs. The adsorbed murexide imparted a negative charge to the Mn3O4 particles and MWCNTs for their electrostatic co-dispersion. It should be noted that many commercial dispersants allow for the dispersion of only inorganic particles or carbon materials. In contrast, murexide allows for the dispersion of both Mn3O4 and MWCNTs, facilitating their efficient co-dispersion and mixing.

Results and Discussion
In this investigation, MWCNTs were used as conductive additives. Previous investigations of as-received MWCNT powders showed that MWCNTs formed large agglomerates with a typical size of 0.5 mm [44]. Therefore, efficient dispersion of MWCNTs was critically important for the fabrication of nanocomposites. Figure 3 shows SEM images of the electrodes prepared by Methods 1 and 2. The SEM images at low magnification show the porous structure of electrodes ( Figure 3A,B). The images at higher magnification (Figure 3C,D) show that the size of primary Mn3O4 particles was below 100 nm. MWCNTs were distributed between the Mn3O4 particles, which was beneficial for the enhancement of electronic conductivity of the composite.
The electrodes prepared by Methods 1 and 2 were tested in 0.5 M Na2SO4 electrolyte. Figure 4 shows CVs for the electrodes prepared by Method 1 for different TPs. The CVs obtained at low sweep rates were nearly rectangular. The comparison of the CVs obtained at the same sweep rates for different TPs showed that the area of CV increased with increasing TP number. Murexide exhibits chelating properties [41,42], which are related to its strong tridentate bonding to different metal atoms. Dissociated murexide acquires a negative charge in solutions ( Figure 2). It was hypothesized that murexide was adsorbed on the Mn 3 O 4 surface by creating a tridentate bonding to Mn atoms on the surface ( Figure 2B). The adsorption of murexide on MWCNTs involved interactions [43] of two barbiturate rings of murexide with carbon rings of MWCNTs. The adsorbed murexide imparted a negative charge to the Mn 3 O 4 particles and MWCNTs for their electrostatic co-dispersion. It should be noted that many commercial dispersants allow for the dispersion of only inorganic particles or carbon materials. In contrast, murexide allows for the dispersion of both Mn 3 O 4 and MWCNTs, facilitating their efficient co-dispersion and mixing.
In this investigation, MWCNTs were used as conductive additives. Previous investigations of as-received MWCNT powders showed that MWCNTs formed large agglomerates with a typical size of 0.5 mm [44]. Therefore, efficient dispersion of MWCNTs was critically important for the fabrication of nanocomposites. Figure 3 shows SEM images of the electrodes prepared by Methods 1 and 2. The SEM images at low magnification show the porous structure of electrodes ( Figure 3A,B). The images at higher magnification ( Figure 3C,D) show that the size of primary Mn 3 O 4 particles was below 100 nm. MWCNTs were distributed between the Mn 3 O 4 particles, which was beneficial for the enhancement of electronic conductivity of the composite.
The electrodes prepared by Methods 1 and 2 were tested in 0.5 M Na 2 SO 4 electrolyte. Figure 4 shows CVs for the electrodes prepared by Method 1 for different TPs. The CVs obtained at low sweep rates were nearly rectangular. The comparison of the CVs obtained at the same sweep rates for different TPs showed that the area of CV increased with increasing TP number. This is in agreement with previous investigations, which showed a capacitance increase during cycling [30][31][32][33]. Several previous XPS investigations showed an oxidation of Mn 2+ and Mn 3+ ions with an increased content of Mn 4+ ions on the Mn 3 O 4 particle surface during cycling and linked this process to the capacitance increase [31,[33][34][35]. Moreover, previous investigations showed that the application of Mn 3 O 4 for supercapacitors requires the use of time-consuming activation procedures [31,33,35]. Such procedures are detrimental for practical applications of Mn 3 O 4 in supercapacitors.  This is in agreement with previous investigations, which showed a capacitance increase during cycling [30][31][32][33]. Several previous XPS investigations showed an oxidation of Mn 2+ and Mn 3+ ions with an increased content of Mn 4+ ions on the Mn3O4 particle surface during cycling and linked this process to the capacitance increase [31,[33][34][35]. Moreover, previous investigations showed that the application of Mn3O4 for supercapacitors requires the use of time-consuming activation procedures [31,33,35]. Such procedures are detrimental for practical applications of Mn3O4 in supercapacitors.
It should be noted that we investigated electrodes with high active mass loadings of 40 mg cm −2 . In this investigation, commercial Ni foam current collectors were used, which were designed for batteries and supercapacitors, based on inorganic active materials with a typical electrode mass of 30-50 mg cm −2 . High active mass loading is important for practical applications for reducing the contribution of current collectors and other passive components to the total electrode mass. An active mass of about 10 mg cm −2 is required for commercial activated carbon electrodes [6,45]. Inorganic materials, such as Mn3O4, have a significantly higher density than the density of activated carbon. Therefore, larger mass loadings can be achieved at the same electrode volume.
The higher gravimetric capacitance of Mn3O4 and the higher active mass can potentially result in significantly higher capacitances of Mn3O4-based electrodes, compared to activated carbon electrodes of the same volume. However, it is challenging to achieve good electrode performance at a high active mass. It is known that gravimetric capacitance drops with active mass increase [6]. Moreover, the use of electrodes with high active mass loading aggravated the problem of Mn3O4 electrode activation, compared to the thin-film Mn3O4 electrodes. This is attributed to better electrolyte access to thin film electrodes, compared to the bulk electrodes with high active mass.   Figure 5A shows Cm and CS, derived from the CV data for different TPs. The depen ence of capacitance on scan rate for TP 1 (Figure 5A(a)) shows a maximum at 20 mV s −1 . is suggested that the electrode activation during cycling at lower scan rates resulted in t capacitance increase. It should be noted that we investigated electrodes with high active mass loadings of 40 mg cm −2 . In this investigation, commercial Ni foam current collectors were used, which were designed for batteries and supercapacitors, based on inorganic active materials with a typical electrode mass of 30-50 mg cm −2 . High active mass loading is important for practical applications for reducing the contribution of current collectors and other passive components to the total electrode mass. An active mass of about 10 mg cm −2 is required for commercial activated carbon electrodes [6,45]. Inorganic materials, such as Mn 3 O 4 , have a significantly higher density than the density of activated carbon. Therefore, larger mass loadings can be achieved at the same electrode volume.
The higher gravimetric capacitance of Mn 3 O 4 and the higher active mass can potentially result in significantly higher capacitances of Mn 3 O 4 -based electrodes, compared to activated carbon electrodes of the same volume. However, it is challenging to achieve good electrode performance at a high active mass. It is known that gravimetric capacitance drops with active mass increase [6]. Moreover, the use of electrodes with high active mass loading aggravated the problem of Mn 3 O 4 electrode activation, compared to the thin-film Mn 3 O 4 electrodes. This is attributed to better electrolyte access to thin film electrodes, compared to the bulk electrodes with high active mass. Figure 5A shows C m and C S , derived from the CV data for different TPs. The dependence of capacitance on scan rate for TP 1 (Figure 5A(a)) shows a maximum at 20 mV s −1 . It is suggested that the electrode activation during cycling at lower scan rates resulted in the capacitance increase.  Figure 5A shows Cm and CS, derived from the CV data for different TPs. The dependence of capacitance on scan rate for TP 1 (Figure 5A(a)) shows a maximum at 20 mV s −1 . It is suggested that the electrode activation during cycling at lower scan rates resulted in the capacitance increase.  However, the capacitance decreased at scan rates of 50 and 100 mV s −1 , resulting in a maximum ( Figure 5A(a)). The capacitance increased with increasing TP number from 1 to 5 ( Figure 5). The highest integral capacitance of 4.87 F cm −2 (121.8 F g −1 ) was achieved at 2 mV s −1 for TP 5. The components of the differential complex capacitance, obtained from the EIS data, showed significant variations for TP 1-5 ( Figure 5B,C). The frequency dependences of the capacitance components showed relaxation-type [46] dispersions. The real part of the differential complex capacitance increased with increasing TP number in agreement with the CV data. The GCD data obtained after TP 5 showed linear chargedischarge behavior, indicating good capacitive performance ( Figure 5D). The integral capacitance of 6.77 F cm −2 (169.3 F g −1 ) was obtained at a current density of 3 mA cm −2 . The capacitance slightly decreased with increasing current density and showed good capacitance retention.
Testing of the electrodes prepared by method 2 showed reduced capacitance variations during cycling, and significantly higher capacitances were obtained compared to method 1. Figure 6 compares CV data for TP 1 and TP 3. The areas of CVs increased from TP 1 to TP 3 only at low scan rates. At scan rates of 20 mV s −1 and higher, the CV areas were nearly similar for TP1 and TP3. The CV areas for TP 4 and TP 5 were practically the same as for TP 3 for all scan rates. tance retention.
Testing of the electrodes prepared by method 2 showed reduced capacitance variations during cycling, and significantly higher capacitances were obtained compared to method 1. Figure 6 compares CV data for TP 1 and TP 3. The areas of CVs increased from TP 1 to TP 3 only at low scan rates. At scan rates of 20 mV s −1 and higher, the CV areas were nearly similar for TP1 and TP3. The CV areas for TP 4 and TP 5 were practically the same as for TP 3 for all scan rates. Figure 7A shows integral capacitances, calculated from the CV data for the electrode prepared by Method 2. The capacitance for TP 1 (Figure 7A(a)) showed a maximum for a scan rate of 10 mV s −1 . A similar maximum was observed for TP 1 for the electrode prepared by method 1 (Figure 5A(a)). As pointed out above, such a maximum resulted from the activation of the electrode by cycling at low scan rates. Therefore, some activation occurred for the electrodes prepared by Method 2. However, it should be noted that the capacitance obtained for the first cycle at 2 mV s −1 for TP 1 for electrode prepared by Method 2 was 5.46 F cm −2 (136.4 F g −1 ), which is higher than the capacitance of 4.87 F cm −2 (121.8 F g −1 ) at 2 mV s −1 for TP 5 for the electrode prepared by Method 1. The highest capacitance of 6.67 F cm −2 (166.7 F g −1 ) was achieved at 2 mV s −1 for TP 3 for the electrode prepared by Method 2. Turning again to the data presented in Figure 7A, it is seen that very small variations in the capacitance were observed for TP1 and TP3 for scan rates of 20-100 mV s −1 . The capacitance measurements at different scan rates for TPs 3-5 did not show significant variations in capacitances.   Figure 7A shows integral capacitances, calculated from the CV data for the electrode prepared by Method 2. The capacitance for TP 1 (Figure 7A(a)) showed a maximum for a scan rate of 10 mV s −1 . A similar maximum was observed for TP 1 for the electrode prepared by method 1 (Figure 5A(a)). As pointed out above, such a maximum resulted from the activation of the electrode by cycling at low scan rates. Therefore, some activation occurred for the electrodes prepared by Method 2. However, it should be noted that the capacitance obtained for the first cycle at 2 mV s −1 for TP 1 for electrode prepared by Method 2 was 5.46 F cm −2 (136.4 F g −1 ), which is higher than the capacitance of 4.87 F cm −2 (121.8 F g −1 ) at 2 mV s −1 for TP 5 for the electrode prepared by Method 1. The highest capacitance of 6.67 F cm −2 (166.7 F g −1 ) was achieved at 2 mV s −1 for TP 3 for the electrode prepared by Method 2. Turning again to the data presented in Figure 7A, it is seen that very small variations in the capacitance were observed for TP1 and TP3 for scan rates of 20-100 mV s −1 . The capacitance measurements at different scan rates for TPs 3-5 did not show significant variations in capacitances.
The results of capacitance measurements from the EIS data ( Figure 7B,C) correlated with the results obtained by CV. The real part of capacitance for TP 1 at 10 mHz for the electrode prepared by Method 2 was 4.75 F cm −2 (118.8 F g −1 ), which is higher than the capacitance of 3.77 F cm −2 (94.25 F g −1 ) for the electrode prepared by Method 1 at the same frequency and TP 5. The analysis of EIS data for TP 1 and TP 3 revealed changes in both real and imaginary capacitance, which indicates that some activation process occurs for the electrodes prepared by Method 2. The highest real part of the capacitance obtained at 10 mHz for TP 3 for the electrode prepared by Method 2 was found to be 5.70 F cm −2 (142.5 F g −1 ). EIS capacitance data did not show significant variation for TP 4 and TP 5 compared to TP 3. The results of capacitance measurements from the EIS data ( Figure 7B,C) correlated with the results obtained by CV. The real part of capacitance for TP 1 at 10 mHz for the electrode prepared by Method 2 was 4.75 F cm −2 (118.8 F g −1 ), which is higher than the capacitance of 3.77 F cm −2 (94.25 F g −1 ) for the electrode prepared by Method 1 at the same frequency and TP 5. The analysis of EIS data for TP 1 and TP 3 revealed changes in both real and imaginary capacitance, which indicates that some activation process occurs for the electrodes prepared by Method 2. The highest real part of the capacitance obtained at 10 mHz for TP 3 for the electrode prepared by Method 2 was found to be 5.70 F cm −2 (142.5 F g −1 ). EIS capacitance data did not show significant variation for TP 4 and TP 5 compared to TP 3.
The results of CV and EIS data indicate that Method 2 resulted in a significant acceleration of the activation process. This can potentially eliminate the need in the time-consuming activation process for Mn3O4-based electrodes. Indeed, relatively small variations in capacitance were obtained for the electrodes prepared by Method 2. The first cycle of capacitance measurements for the electrodes prepared by this method showed higher capacitance than that for TP 5 for the electrode prepared by Method 1. Turning again to the comparison of MnO2 and Mn3O4 electrodes, it should be noted that some activation process was also reported for the MnO2 electrodes, which also exhibited a small increase in the capacitance during initial cycling [47]. Such a capacitance increase in the MnO2 electrodes was attributed to the microstructure changes during initial cycling [47]. GCD testing of the electrodes prepared by Method 2 showed linear charge-discharge curves (Figure 7D), indicating good capacitive behavior. The capacitance of 7.55 F cm −2 (188.8 F cm −2 ) was achieved at a current density of 3 mA cm −2 . The electrodes prepared by Method 2 showed energy density of 18.8 Wh kg −1 at power density of 0.11 kW kg −1 . The results of CV and EIS data indicate that Method 2 resulted in a significant acceleration of the activation process. This can potentially eliminate the need in the time-consuming activation process for Mn 3 O 4 -based electrodes. Indeed, relatively small variations in capacitance were obtained for the electrodes prepared by Method 2. The first cycle of capacitance measurements for the electrodes prepared by this method showed higher capacitance than that for TP 5 for the electrode prepared by Method 1. Turning again to the comparison of MnO 2 and Mn 3 O 4 electrodes, it should be noted that some activation process was also reported for the MnO 2 electrodes, which also exhibited a small increase in the capacitance during initial cycling [47]. Such a capacitance increase in the MnO 2 electrodes was attributed to the microstructure changes during initial cycling [47]. GCD testing of the electrodes prepared by Method 2 showed linear charge-discharge curves ( Figure 7D), indicating good capacitive behavior. The capacitance of 7.55 F cm −2 (188.8 F cm −2 ) was achieved at a current density of 3 mA cm −2 . The electrodes prepared by Method 2 showed energy density of 18.8 Wh kg −1 at power density of 0.11 kW kg −1 .
In order to analyze the difference in the activation of electrodes prepared by methods 1 and 2, CV studies were performed for fresh electrodes at a scan rate of 50 mV s −1 for 2000 cycles, and the obtained capacitances were normalized by the capacitance obtained for the 2000th cycle ( Figure 8).
The normalized capacitance C N for the first cycle for the electrode prepared by method 1 was only 23% and it was slowly increased with cycle numbers. In contrast, the capacitance for the first cycle for the electrode prepared by Method 2 was 70% and rapidly increased with an increasing cycle number. The comparison with the data presented in Figure 7 also indicates that a lower scan rate can result in a faster activation process for the electrodes prepared by Method 2. 2000th cycle (Figure 8).
The normalized capacitance CN for the first cycle for the electrode prepared by method 1 was only 23% and it was slowly increased with cycle numbers. In contrast, the capacitance for the first cycle for the electrode prepared by Method 2 was 70% and rapidly increased with an increasing cycle number. The comparison with the data presented in Figure 7 also indicates that a lower scan rate can result in a faster activation process for the electrodes prepared by Method 2. The capacitances obtained from CV, EIS, and GCD data for electrodes prepared by Method 2 are significantly higher than the capacitances obtained by the same testing techniques for electrodes prepared by Method 1. The results of this investigation indicated that the use of murexide as a capping agent allowed for an enhanced performance of the Mn3O4-MWCNT electrodes. It should be noted that MWCNTs have a low electrical double-layer-type specific capacitance [48] of about 20 F g −1 . The use of MWCNTs as conductive additives is critically important for the utilization of capacitive properties of Mn3O4, which has low conductivity. Due to small MWCNT content in the Mn3O4-MWCNT electrode material, the high capacitance of the composite electrodes resulted from pseudocapacitive properties of Mn3O4.
A recent comprehensive review [6] summarized capacitances for Mn3O4 and MnO2 electrodes with high active mass reported in the literature. A comparison with the literature data for Mn3O4 showed that the areal capacitance of Mn3O4-MWCNT electrodes achieved in this investigation is significantly higher than in the literature data (Table 1). Moreover, the capacitance of the Mn3O4-MWCNT electrodes was higher than the capacitances of MnO2-MWCNT electrodes of a similar mass reported in the literature (Table 1).  The capacitances obtained from CV, EIS, and GCD data for electrodes prepared by Method 2 are significantly higher than the capacitances obtained by the same testing techniques for electrodes prepared by Method 1. The results of this investigation indicated that the use of murexide as a capping agent allowed for an enhanced performance of the Mn 3 O 4 -MWCNT electrodes. It should be noted that MWCNTs have a low electrical doublelayer-type specific capacitance [48] of about 20 F g −1 . The use of MWCNTs as conductive additives is critically important for the utilization of capacitive properties of Mn 3 O 4 , which has low conductivity. Due to small MWCNT content in the Mn 3 O 4 -MWCNT electrode material, the high capacitance of the composite electrodes resulted from pseudocapacitive properties of Mn 3 O 4 .
A recent comprehensive review [6] summarized capacitances for Mn 3 O 4 and MnO 2 electrodes with high active mass reported in the literature. A comparison with the literature data for Mn 3 O 4 showed that the areal capacitance of Mn 3 O 4 -MWCNT electrodes achieved in this investigation is significantly higher than in the literature data (Table 1). Moreover, the capacitance of the Mn 3 O 4 -MWCNT electrodes was higher than the capacitances of MnO 2 -MWCNT electrodes of a similar mass reported in the literature (Table 1). Therefore, the results of this investigation indicate that Mn 3 O 4 is a promising alternative to MnO 2 as a cathode material for asymmetric supercapacitors. The strategy developed in this investigation opens up an avenue for a further improvement of capacitive properties of Mn 3 O 4 -based electrodes. Of particular importance for future investigations is the ability of Mn 3 O 4 to form solid solutions with other spinel compounds. The fabrication and testing of such solid solutions can result in electrodes with higher capacitive properties, which can be combined with improved magnetic and other functional properties.

Conclusions
For the first time, murexide was used as a capping agent for the synthesis of Mn 3 O 4 nanoparticles and as a co-dispersant for Mn 3 O 4 and MWCNTs. The adsorption of murexide on Mn 3 O 4 and MWCNTs involved two different mechanisms and facilitated electrostatic co-dispersion of Mn 3 O 4 with MWCNTs and their enhanced mixing. The use of murexide as a capping agent in Method 2 allowed for a reduced agglomeration. As a result, the capacitance of the Mn 3 O 4 -MWCNT electrodes prepared by Method 2 was significantly higher than the capacitance of the Mn 3 O 4 -MWCNT electrodes prepared by Method 1. The simple approach developed in this investigation resulted in record-high capacitances of 6.67 F cm −2 obtained from cyclic voltammetry data at a scan rate of 2 mV s −1 and 7.55 F cm −2 obtained from the galvanostatic charge-discharge data at a current density of 3 mA cm −2 . The good electrochemical performance was achieved at a high active mass loading of 40 mg cm −2 . It was found that the time-consuming electrode activation procedure for Mn 3 O 4 electrodes can be avoided. The approach developed in this investigation paved the way for the development of advanced cathodes for asymmetric supercapacitors for operation in a neutral electrolyte. It is expected that future progress in the fabrication of Mn 3 O 4 electrodes will result in a superior performance compared to MnO 2 electrodes for practical applications. Further development of chelating dispersants offers a promising strategy for the synthesis and the colloidal processing of advanced energy storage materials. The ability to achieve high capacitance for a spinel material in a neutral electrolyte opens the door for the fabrication of multifunctional devices, combining capacitive, magnetic, and other functional properties.