1 Introduction

Manganese dioxide (MnO2) is the most studied metal oxide for redox-based electrochemical capacitors [1]. The main attractivities of MnO2 are related to its low cost, environmental compatibility and tunable physical and chemical properties [2], among other advantages. The forehand MnO2 energy storage application date from the late 1970s in lithium intercalation batteries [3,4,5]. Years later, in 1999, Goodenough and coworkers [6] observed the good capacitive response of amorphous MnO2 as electrochemical capacitor electrode while looking for cheaper replacers to the expensive ruthenium oxides. Since then, several workers [7, 8] have dedicated attention to the understanding, synthesis and application of such extraordinary material.

MnO2 structures are based in interlinked MnO6 octahedrons ([MnO6]), formed by 6 oxygen atoms on the vertices and a manganese atom on the center. Depending on the arrangement of the octahedrons, different layers and tunnels configurations are formed [9, 10]. Among several possible arrangements and stoichiometries, birnessite, also known as δ-MnO2, is a layered structure with monoclinic arrangement of [MnO6] sheets, with C2/m space group [11], containing mixtures of Mn(IV)/Mn(III) or Mn(IV)/Mn(II) and structural water in non-stoichiometric compositions [12, 13].

Electrochemical capacitors are considered a key solution for the energy storage systems required in the sustainable and renewable energy scenario that society needs to move into [1]. Two different mechanisms of capacitance are possible to occur in an electrode/electrolyte interface: the accumulation of ions in a electrochemical double layer (ECDL) or electrons transference via redox reactions, known as pseudocapacitance [14]. Due to its layered structure, δ-MnO2 is claimed to be one of the few materials that shows both capacitive mechanisms concomitantly [15, 16]. However, specific capacitance of birnessites, and other MnO2 crystallographic phases as well, are highly dependent of several features [10]. The morphology [8, 17], the amount of defects [8], the Mn(IV)/Mn(III) relation [18], among other parameters are decisive in the material’s capacity of storing tens or hundreds of farads per gram.

Thus, the search for phase selective MnO2 synthesis methods is still a challenging and relevant topic [7, 19]. Moreover, new properties may arise with new synthesis routes [8, 10]. Besides that, the often used birnessite synthesis routes present some drawbacks, particularly related to long-time duration. Hydrothermal routes may be relatively fast, lasting about 30 min, but require elaborated high temperature and pressure apparatuses [20,21,22,23]. Sol–gel processes may require even higher temperatures, several hours of reaction and steps [24]. Even newer methods, that avoid the use of high-temperature or pressure treatment, still have duration as disadvantage. For instance, a polyol-reflux method requires a 1 h reaction [25] and another demands a statically aging of 1 day [15]. Electrochemical deposition [18, 26], in its turn, is limited to the production of low dimensions films [27].

Therefore, in this work we suggest a novel and simple chemical route for selective δ-MnO2 synthesis with high specific capacitance. Our method is based on the reduction of potassium permanganate (KMnO4) [28] with nascent hydrogen. Nascent hydrogen (H*) reduction is a well spread method for the obtainment of several compounds [29, 30]. The metastable hydrogen form, produced on the reduction of metals in acid medium, tends to combine quickly to form H2, but during its short lifetime it is a highly reactive reduction agent [30, 31]. The production of δ-MnO2 by such method has the advantages of being a very quick, one-pot reaction, sprinted at low temperature. Reaction duration is lower than ten minutes. In addition, the final product may be used as suspension (as synthesized) or as powder (after drying) and procedure required only simple laboratory glassware. Such innovative method was firstly developed by our group for the obtainment of transition metal oxide and graphene nanocomposites, showing singular results [32]. To the best of our knowledge, this is the first time that reduction by H* is employed to the fabrication of any kind of manganese oxide.

2 Materials and methods

Firstly, 100 mL of an aqueous solution of 10 mmol L−1 KMnO4 (≥ 99.0%, Sigma-Aldrich) was prepared and transferred to a round-bottomed flask. The same volume of a previously cooled 2 mol L−1 hydrochloric acid (Synth) solution was added to the flask. The resulting solution was kept in ice bath until temperature stabilization in approximately 10 °C. Then, 1.65 g of magnesium metallic foils (Vetec) was added to the solution, starting the reaction. By the end of the exothermic reaction, that lasted about 10 min, a dark brown precipitate was formed. The product was washed with deionized water by centrifugation. After that, the precipitate was dried at 100 °C by 24 h. The whole procedure is illustrated in Fig. 1.

Fig. 1
figure 1

Schematic representation of the δ-MnO2 synthesis procedure

X-ray diffractogram was acquired in a diffractometer (Shimadzu XRD-6000) with CuKα radiation (λ = 1.5418 Å), 40 kV voltage and 40 mA current. For all other characterization techniques, the powder was dispersed in ethanol, drop casted in silicon (Si/SiO2) wafers and dried for 24 h at 65 °C. Raman spectrum was collected in a Raman spectrometer (Witec UHTS 300 VIS) with 488 nm nominal wavelength laser excitation line. Attenuated Total Reflectance Fourier Transformed Infrared (ATR-FTIR) spectrum was collected in a spectrophotometer (Shimadzu IRAffinity-1S FTIR) with 4 cm−1 resolution and 32 accumulations. X-ray photoelectron spectroscopy (XPS) analysis was performed on a photoelectron spectrometer (Thermo Scientific™ XPS K-Alpha™) using a 72 W Al Kα X-ray source and a monochromator with 400 μm spot size with 5 μm step. Curve fitting and background subtraction were accomplished using Avantage 5.977 (Thermo Scientific™) software. Scanning electron microscopy (SEM) images were obtained in a scanning electron microscope (TESCAN MIRA3) operated at 10 kV with an In-Beam detector.

Electrochemical measurements were performed in a potentiostat/galvanostat (Metrohm Autolab PGSTAT302N) in three electrodes cell. For working electrodes preparation, the obtained δ-MnO2 powder was dispersed in deionized water, drop casted in a 1 cm2 area of fluorine doped tin oxide (FTO) substrate and dried for 24 h at 65 °C. The electrolyte was an aqueous solution of 0.5 mol L−1 sodium sulfate (Na2SO4). Reference and counter electrodes were silver/silver chloride (Ag/AgCl) and platinum wire, respectively.

3 Results and discussion

Nascent hydrogen could be produced from the oxidation of metals as magnesium, aluminum, iron or zinc [30]. For convenience we choose Mg, which standard reduction potential is \({E}^{0}\left({\mathrm{Mg}}^{2+}/\mathrm{Mg}\right)= -2.372 \mathrm{V}\) [33], compatible with the H+/H* couple estimated \({E}^{0}\left({\mathrm{H}}^{+}/{\mathrm{H}}^{*}\right)= -2.106 \mathrm{V}\) [31]. Thus, the formation of the nascent hydrogen is spontaneous upon Mg oxidation in acid media [31], following the Eq. 1.

$${\text{Mg }} + { }2{\text{H}}^{ + } { } \to {\text{ Mg}}^{2 + } { } + { }2{\text{H}}^{*}$$
(1)

This was confirmed through vigorous effervescence of hydrogen bubbles and a small increase in the temperature was also observed when the metal was added to the solution. The reaction was considered ended few minutes later, when the effervescence was finished, and all the Mg was consumed. By that moment, it was noticeable that the previous vivid-purple KMnO4 solution changed to a transparent solution with a large amount of a brownish precipitate, as described in Fig. 1.

In order to confirm the formation of the desired birnessite and to acquire detailed information about the product, we performed structural and morphological characterizations.

Figure 2a shows the diffractogram for the MnO2 powder. The diffractogram shows four distinct peaks centered in 12.08°, 23.11°, 37.96° and 66.22°. Several authors [2, 10, 11, 22, 34] attributed similar patterns to K-birnessite-type manganese dioxide, following JCPDS 80–1098 card. According to the card, those peaks may be ascribed to (001), (002), (−111) and (020) planes, respectively. We highlight that powdered specimens contain very small crystallites arranged in completely random orientations. Because of that, powder lines have a significant width and tend to merge or overlap [35], becoming broad peaks. That may have happened with the several lines that occur in the (−111) and (020) peaks region, resulting in the intense and broad 37.96° and 66.22° peaks in our diffractogram. Besides that, the diffractogram presents a significant high signal–noise ratio. That is also a consequence of powder diffraction, where many reflections with low resolution are available [35]. Regarding to the structure, according to previous studies, the alkaline cation K+ in the medium acts as a nanosheet stabilizer, thus favoring the formation of the layered structure [22]. The position of the (001) peak reveals an interlayer distance of 7.32 Å, in accordance with other birnessites reported in the literature [2, 10, 13, 22].

Fig. 2
figure 2

a X-Ray diffractogram of δ-MnO2 (red line) and pattern of related JCPDS card No. 80–1098 (blue line); b Raman spectrum of δ-MnO2, and c ATR-FTIR spectrum of δ-MnO2

Although birnessite has low Raman activity and accuracy, the technique was helpful to further understanding the oxide structure. The Raman spectrum of the δ-MnO2 is presented in Fig. 2b. Three of the expected bands aroused in 645, 568 and 514 cm−1. The 645 cm−1 band was assigned to the A1g symmetric mode, ascribed to the symmetric stretching vibration ν2(Mn–O) of [MnO6] groups. The 568 cm−1 band was assigned to the ν3(Mn–O) symmetric stretching vibration in the basal plane of [MnO6] sheets [13]. On the other hand, the red shift of the 514 cm−1 band (ν4), comparing to other reports, and the appearance of a fourth mode (ν7) at around 356 cm−1 may be related to ν(K–O) vibrations [13, 23].

Combined with others, FTIR is a powerful technique to evaluate materials structure. The ATR-FTIR spectrum, illustrated in Fig. 2c, exhibits six assignable bands: 635, 1001, 1419, 1539, 1635 and 3360 cm−1. All these modes were compatible with birnessite samples [36, 37]. The band at 635 cm−1 was attributed to ν(Mn–O) lattice vibrations of poorly ordered birnessites. The band at 1001 cm−1 was related to ν(Mn(III)–OH) in the layer basal plane, thus evidencing the presence of Mn(III) simultaneously with the predominant Mn(IV). The bands within the 4000–1400 cm−1 range were assigned to.

–OH stretching (ν) and bending (δ) vibrational modes [36]. The one in 1419 cm−1 is related to hydroxide groups located at specific crystallographic sites, while the others are related to less ordered water groups [37]. The presence and position of those four Raman and these six FTIR modes ratified the DRX findings about the δ-MnO2.

The surface chemical composition of the δ-MnO2 was further analyzed by XPS to clarify the oxidation states of the oxide components. As shown in Fig. 3a, Mn2p, O1s and Mg1s peaks were present on the survey. From that, composition of the sample surface was determined as 31.7% of Mn, 66.1% of O and 2.2% of Mg. Magnesium in the structure, evidenced by the peak at 1303.15 eV, derived from the metal/acid reaction. From Fig. 3b, the position of the Mn2p main peaks – Mn2p1/2 at 654.1 eV and Mn2p3/2 at 642.3 eV and the spin-energy separation of 11.8 eV between them are in accordance with the main MnO2 XPS studies [38,39,40,41]. Similarly, works dedicated to the synthesis of MnO2 [42,43,44,45] found similar values. Mn2p3/2 was further analyzed for the determination of the different Mn oxidation states composition. According to Nesbitt and Banerjee [38], the Mn2p3/2 peak of birnessites can be deconvoluted in five multiplets for each manganese valence. Following these authors, we deconvoluted the Mn2p3/2 peak fixing the FWHM of the multiplets as done by Sun and coworkers [46], as shown in Fig. 3c. The resulting Mn valences composition was calculated as 58.3% Mn(IV), 35.0% Mn(III) and 6.7% Mn(II). Comparing with Sun and coworkers [46], we observed a higher Mn(III) content to the detriment of Mn(IV). However, this composition is similar to some of the verified by Ilton and coworkers when obtained from the Mn2p3/2 peak deconvolution [40]. Moreover, Athouël and coworkers [16] proved that higher Mg(III)/Mg(IV) ratio is usual when the synthesis of birnessite occurs in the presence of Mg. Following the same methodology, we proceeded the analysis of the O1s peak. As shown in Fig. 3d, the O1s peak showed a maximum at 529.9 eV and a broad shoulder at higher energies, being this another evidence that the synthesized oxide is a birnessite [38]. The main contribution of O1s peak is given by the oxide structural oxygen (Mn–O–Mn), representing 46.4% of the spectrum. The two other peaks with maximums in 530.1 and 532.4 eV were attributed to the presence of structurally bounded hydroxide (–OH) (39.6%) and adsorbed water (H–O–H) (14.0%), respectively. The structural oxygen is lower than the reported by other authors [46, 47]. However, it is a consequence of the lower Mn(IV) availability. Besides all the XPS being compatible with birnessite-type MnO2, these results agreed with the previously attested on FTIR.

Fig. 3
figure 3

a Survey XPS spectrum of the δ-MnO2; b Mn2p; c deconvoluted Mn2p3/2 and d deconvoluted O1s XPS spectra of the δ-MnO2

The morphology of the δ-MnO2 synthesized through H* reduction of KMnO4 was investigated by SEM. The lower magnification image, shown in Fig. 4a, revealed aggregates of randomly shaped structures with some microns of dimension. Higher magnification image, shown in Fig. 4b, indicated that the oxide structure is porous and built up of many nanosheets, similar to nanoflower structures. These features are typical of δ-MnO2 obtained by hydrothermal routes [20, 21, 23].

Fig. 4
figure 4

SEM images of δ-MnO2 with a lower and b higher magnification

As the structural and morphological characterization are in accordance with the δ-MnO2 features, we evaluated the electrochemical performance of this structure as electrode for electrochemical capacitors.

According to the literature, sodium sulfate is one of the most suitable electrolyte for manganese oxides [48]. Na+ cations are appropriate to be intercalated in the bulk of MnO2 upon reduction and deintercalated upon oxidation [48, 49] following the Eq. 2 or to be adsorbed on the surface of the oxide by following the Eq. 3.

$${\text{MnO}}_{2} + {\text{Na}}^{ + } + e^{ - } \leftrightarrow {\text{ MnOONa}}$$
(2)
$$\left( {{\text{MnO}}_{2} } \right)_{{{\text{surface}}}} { } + {\text{ Na}}^{ + } + e^{ - } { } \leftrightarrow { }\left( {{\text{MnO}}_{2}^{ - } {\text{Na}}^{ + } } \right)_{{{\text{surface}}}} .$$
(3)

In general, MnO2 electrodes present pseudocapacitive behavior, showing nearly rectangular-shaped voltammograms [10, 18, 20, 25]. In birnessite arrangement, however, the electron transfer that takes place in the Mn(IV)/Mn(III) redox system, described by the Eq. 2, is usually visible through cathodic and anodic peaks [10, 15, 50]. Depending on the involved constituents, sometimes even more than one peak of each type may arise [16, 49, 51, 52]. Figure 5a shows the cyclic voltammogram of the δ-MnO2 between − 0.1 and 0.9 V in different scan rates (5–100 mV s−1). Both expected electrochemical processes were perceptible. The anodic peak occurred between 0.55 and 0.71 V, and the cathodic peak between 0.35 and 0.51 V, depending on the scan rate. This redox process may reflect the redox transitions of the manganese, including Mn3+/Mn2+, Mn4+/Mn3+ and Mn6+/Mn4+ or can be related to the cation deintercalation upon oxidation and cation insertion upon reduction [10]. These values are highly comparable with the ones of the cited reports [10, 15, 50].

Fig. 5
figure 5

δ-MnO2 a cyclic voltammograms at 5, 10, 20, 50 and 100 V s−1 scan rate; b cyclic voltammetry anodic and cathodic peak currents (Ip) as function of the scan rate; c charge and discharge curves at 0.25, 0.50, 1.00 and 2.00 A g−1; d specific capacitances and standard deviation in each current density and e capacitance retention through 500 cycles. All performed in 0.5 mol L−1 Na2SO4 electrolyte

Figure 5b shows the plot of peak currents (Ip) versus potential scan rate. The linearity of the plots confirms the redox behavior of the storage reaction [53]. However, for an exclusively redox reaction the ratio of the modulus of the anodic and cathodic peak currents (Ip,a/Ip,c) in all scan rates should equal to 1 [53, 54]. Noticeably the modulus of the curves slopes differs from each other, thus implying in a varying Ip,a/Ip,c. Differences between anodic and cathodic peaks is a recurrent feature of birnessites [49, 51]. Kanoh and coworkers [51] suggested that it could be an implication of differences of mechanisms of kinetics between intercalation and deintercalation processes, what was later experimentally confirmed by Athouël and colleagues [49].

Specific capacitance is usually valued through galvanostatic charge/discharge measurement. Performed galvanostatic charge/discharge curves at different current densities (0.25–2.00 A g−1) are plotted in Fig. 5c. The shape of the charge and discharge curves of potential as function of time are typical of faradaic electrochemical capacitors [14]. Specific capacitances (Cs) were calculated using the Eq. 4, where i is the discharge current, Δt is the discharge time, m is the mass of the active material in the electrode, and ΔV is the potential window of cycling.

$$C_{s} = \left( { i\Delta t } \right) / \left( { m\Delta V } \right).$$
(4)

The results were plotted in Fig. 5d, where one can see that the best specific capacitance of 190±12 F g−1 was obtained at 0.25 A g−1 current density. When the current density was increased to 0.50, 1.00 and 2.00 A g−1, measured specific capacitances were 173±12, 158±12 and 136±8 F g−1, respectively. These high capacitance values are typical of δ-MnO2 thin film electrodes reported [16, 18, 50] with the additional advantages of being produced very quickly, without requiring external energy provision, long furnace dwell times nor sophisticated laboratory apparatuses.

It is worth mentioning that all the electrochemical reported data was recorded after an electrode stabilization. As evidenced on the long-term cycling stability shown in Fig. 5e, the electrode rapidly gains capacitance in the first cycles and remains relatively stable after this. Such phenomenon is well discussed in the literature, once many authors report the capacitances and capacitance retention only after stabilization [10]. Thus, the capacitance gain may be related to several ongoing processes, as wettability improvement and self-activation [55, 56] over the cycles. Beyond wettability improvement, the electroactivation in our birnessite may be related to the presence of Mg (evidenced by XPS). Athouël and colleagues [16] showed that magnesium is exchanged by sodium ions (Na+) by simply soaking the electrode in the electrolyte solution. Upon the exit of the retained Mg, especially in the initial cycles, more electroactive sites are released, increasing the capacitance.

4 Conclusions

We successfully synthesized a high capacitive birnessite-type MnO2 by a novel simple route based on the reduction of KMnO4 by nascent hydrogen. The volatile metastable hydrogen form was generated by the reduction of magnesium foil in acid media. DRX characterization showed that the as-synthesized MnO2 has a layered structure, with interlayer space of 7.13 Å and nanoscale sized crystallite. Raman and FTIR techniques confirmed DRX findings and evidenced the presence of Mn(III) and Mn(II) apart to the predominant Mn(IV). By XPS, it was confirmed the presence of Mn in different valence states and a residual Mg. Morphology was clarified by SEM, that showed randomly shaped porous structures formed by nanosheets aggregates.

Finally, we evaluated the electrochemical activity of the as-synthesized δ-MnO2 in Na2SO4. The oxide showed a redox behavior and specific capacitance as high as 190 F g−1 at 0.25 A g−1 current density. Thus, the developed synthesis route for high capacitive δ-MnO2 was found to be advantageous by being fast, selective, driven at low temperature with only simple laboratory glassware.