Fast Microwave-Assisted Hydrothermal Synthesis of Pure Layered δ-MnO₂ for Multivalent Ion Intercalation

Abstract

This work reports on the synthesis of layered manganese oxides (δ-MnO₂) and their possible application as cathode intercalation materials in Al-ion and Zn-ion batteries. By using a one-pot microwave-assisted synthesis route in 1.6 M KOH (Mn_VII:Mn_II = 0.33), a pure layered δ-MnO₂ birnessite phase without any hausmannite traces was obtained after only a 14 h reaction time period at 110 °C. Attempts to enhance crystallinity level of as-prepared birnessite through increasing of reaction time up to 96 h in 1.6 M KOH failed and led to decreases in crystallinity and the emergence of an additional hausmannite phase. The influence of Mn_II:OH⁻ ratio (1:2 to 1:10) on phase crystallinity and hausmannite phase formation for 96 h reaction time was investigated as well. By increasing alkalinity of the reaction mixture up to 2.5 M KOH, a slight increase in crystallinity of birnessite phase was achieved, but hausmannite formation couldn’t be inhibited as hoped. The as-prepared layered δ-MnO₂ powder material was spray-coated on a carbon paper and tested in laboratory cells with Al or Zn as active materials. The Al-ion tests were carried out in EMIMCl/AlCl₃ while the Zn-Ion experiments were performed in water containing choline acetate (ChAcO) or a ZnSO₄ solution. Best performance in terms of capacity was yielded in the Zn-ion cell (200 mWh g⁻¹ for 20 cycles) compared to about 3 mAh g⁻¹ for the Al-ion cell. The poor activity of the latter system was attributed to low dissociation rate of tetrachloroaluminate ions (AlCl₄⁻) in the EMIMCl/AlCl₃ mixture into positive Al complexes which are needed for charge compensation of the oxide-based cathode during the discharge step.

1. Introduction

The permanent increase in energy storage devices is a very ambitious challenge that obviously cannot be overcome with established lead-acid, NiMH, and Li-ion technologies alone, mostly due to limited supplies of raw material resources and a lack of efficient recycling processes, especially in the case of Li-ion batteries. Therefore alternative electrochemically-active and abundant materials such as Na, Mg, Mn, Al, and Zn are appealing candidates. However, intercalation of multivalent ions into common host matrixes such as manganese oxide spinel is strongly affected by higher diffusion barriers than monovalent ones [1]. Nonetheless, the search for more adapted insertion structures than e.g., Mn₂O₄ for zinc and aluminum ion appears to be a very interesting strategy.

The aluminum-ion battery (AIB) is a very promising post lithium-ion battery (LIB) due to the high volumetric energy density of aluminum (~8 Ah L⁻¹ vs. 2 Ah L⁻¹ for LIB), its well established manufacturing and recycling processes, as well as its global availability and abundance. Thus, the AIB covers all prerequisites to develop a sustainable and ecologically-benign battery. Historically, this concept was limited due to the lack of electrolytes for reversible aluminum deposition and dissolution [2]. In the presence of oxygen and/or water, a thin passivation layer of Al₂O₃ is instantaneously formed on the surface of aluminum that can be dissolved by the usage of highly acidic solutions such as AlCl₃ in EMIMCl or urea. Thirteen years ago, reversible aluminum deposition was successfully carried out by Endres et al. [3,4,5] in EMIMCl/AlCl₃ ionic liquid (IL). Recently, Abbott [6] & Angell et al. [7] demonstrated aluminum stripping/deposition in urea/AlCl₃ deep eutectic solvents (DES); kinetics in DES were clearly slower than that in IL electrolyte, so the majority of published investigations on AIB were carried out in an EMIMCl/AlCl₃ electrolyte. The best result so far was presented by Lin et al. [8] who developed an AIB using an aluminum foil anode, EMIMCl/AlCl₃ (1:1.3) electrolyte, and a pyrolytic graphite (PG) or graphitic foam cathode. The latter generally implicates anion de-intercalation during discharging step, and more especially in this case, that of large tetrahedral AlCl₄⁻ (0.53 nm) ions, with an oxidation state of carbon being n < 1, according to

$${\mathrm{C}}_{\mathrm{x}}^{+\mathrm{n}}{\left[{\mathrm{AlCl}}_4^{-}\right]}_{\left(\mathrm{s}\right)}{+\mathrm{e}}^{-}\to {\mathrm{C}}_{\mathrm{x}}{}_{\left(\mathrm{s}\right)}{+\mathrm{AlCl}}_4^{-}{}_{\left(\mathrm{l}\right)}$$

(1)

which further react with Al anode to form aluminum heptachloride according to reaction below:

$${\mathrm{Al}}_{\left(\mathrm{s}\right)}{+7\mathrm{AlCl}}_4^{-}{}_{\left(\mathrm{l}\right)}\to {4\mathrm{Al}}_2{\mathrm{Cl}}_7^{-}{}_{\left(\mathrm{l}\right)}{+3\mathrm{e}}^{-}$$

(2)

Aluminum deposition requires the presence of Al₂Cl₇⁻ active species that only exist under acidic conditions. To keep the electrolyte acidic, at least a 1:1.1 EMIMCl/AlCl₃ molar ratio is required for reversible aluminum deposition/stripping, so that the theoretically-available reversible capacity is defined firstly by an excess concentration of Al₂Cl₇⁻ ions in the electrolyte, and secondly, by theoretical maximal insertable anions into the cathode matrix. This AIB battery offers high stability (7000 cycles), even at very high power densities up to 3000 Wh kg⁻¹ [8], which is comparable to that of lead acid batteries or super capacitors. However, maximum capacity and energy density is limited to 60–66 mAh g⁻¹ and 147–162 mWh g⁻¹ normalized to graphite mass [8] due to the large size of solvated AlCl₄⁻ ion (530 pm) and resulting lower coordination number compared to that of Li⁺ (LiC₆) in graphite matrix. In this work, we opted for the 1:1.5 EMIMCl/AlCl₃ mixture that provides equimolar Al₂Cl₇⁻ and AlCl₄⁻ concentration, as can be seen in Figure A1 (Appendix B), and consequently, might deliver higher capacity than the 1:1.3 one. If overall cell energy density is considered, cell performance with the 1:1.5 electrolyte exceeds that of cells with 1:1.3 in all published works.

To circumvent some inconvenient aspects during aluminum stripping/deposition such as passivation, H₂ evolution, and dendrite formation, the use of an insertion anode material such as TiO₂ might be an attractive strategy. It should be noted that several published contributions on this material focused mostly on half-cell measurements, and therefore, should be carefully interpreted. Recent investigations reported on reversible Al intercalation into TiO₂ nanoleaves [9], in which 270 mAh g⁻¹ at 50 mA g⁻¹ with 8.4% capacity fading over 300 cycles were yielded in a 1 M Al(NO₃)₃ aqueous solution. If TiO₂ is used as the anodic intercalation material, however, cell voltage drop is about 1.10 V vs. an aluminum foil, since Ti³⁺/Ti⁴⁺ amounts to −0.56 V vs. RHE. Interestingly, Holland et al. recently [10] demonstrated the feasibility of an aqueous Al ion cell under full-cell condition by using 1 M KCl/1 M AlCl₃ as an electrolyte, anatase TiO₂ as the insertion anode, and copper hexacyanoferrate (CuHCF) as the cathode. Charge/discharge tests revealed an average discharge capacity of ≈10 mAh g⁻¹ with only 7% capacity fade at an average cell voltage of 1.5 V by 333 mA g⁻¹ for over 1750 cycles. The authors pointed out that they were not able to elucidate whether the main mechanism charge storage occurs via Al³⁺ intercalation or the surface adsorption reaction. Half-cell measurements of CuHCF as the cathode material and aqueous 0.5 M Al₂(SO₄)₃ as the electrolyte revealed high cyclability but poor capacity (41–22 mAh g⁻¹ after 1000 cycles) [11].

In order to increase cell capacity, cobalt-free oxide-based electrodes are preferentially needed. Reversible Al³⁺ intercalation into cathodic VO₂ nanorods (116 mAh g⁻¹, 50 mA g⁻¹, 100 cycles) was reported by Wang et al. [12]. A very promising AIB with very high theoretical capacities and energy densities (400 mAh g⁻¹, 2.65 V cell voltage, 1060 Wh kg⁻¹) was published by Brown et al. [13], who patented a battery system in which good cyclability of cathodic Mn₂O₄ spinel in ionic liquid EMIMCl/AlCl₃ under half–cell conditions is claimed.

Substitution of cobalt in an established LiCoO₂ intercalation electrode by less expensive and more abundant Mn is an aspired strategy. Meanwhile, manganese oxide-based insertion materials such as layered LiNi_xMn_yCo_zO₂ (NMC) are commercially-available cathode materials in LIB for electric vehicles because of their high capacity ≈ 160 Ah kg⁻¹, 3.7 V cell voltage, and specific energy ≈ 592 Wh kg⁻¹ [14]. Electrochemically active lithium-MOs are mainly divided into three different crystal morphologies: 2D layered (LiMnO₂, LiCoO₂ and NMC), 3D spinel (LiMn₂O₄), and linear 1D tunnel (LiMnPO₄). Doped (Cr; Ni; Al; Fe) lithium-MOs are more or less prone to fading, and undergo unwanted growth of solid-electrolyte interface (SEI) or irreversible delithiation, resulting in electrochemically inactive species [15]. However, because of larger and extendable pore sizes compared to spinel and tunnel structures, layered MOs appear to be the most promising candidate for the AIB and ZIB [16].

The feasibility of using MO-based cathodes as insertion materials in zinc-ion battery with a zinc metal anode and an aqueous electrolyte was already demonstrated. Depending on the electrolyte used, (ZnSO₄ or ZnTfO₂/H₂O; ChOAc/ZnOAc/H₂O), a discharge voltage of 1.6–1.0 V (vs. Zn/Zn²⁺) through both proton and zinc ion intercalation was recorded.

Most of the recently published studies have shown a highly-reversible Zn²⁺ intercalation into tunnel α-MnO₂ nanorods [17] (233 Ah kg⁻¹, 83 mA g⁻¹, CE 100%), layered δ-MnO₂ [18] (252 Ah kg⁻¹, 100 cycles) and tunnel α-MnO₂ (140 Ah kg⁻¹, 2000 cycles) [19]. In contrast to water-free δ-MnO₂ cathodes in non-aqueous organic electrolytes (LIB, etc.), the well-studied birnessite crystal structure incorporates water in between the layers. As a consequence, the inter-layer distance increases Å by 0.8, which might be beneficial for zinc ion diffusion.

Synthesis of nanoscale oxides is most favorable at temperatures below 200 °C, which can be achieved by hydrothermal and/or microwave-assisted syntheses. Several different methods have already been published. A good overview of current synthesis routes is given by Wei et al. [20]. δ-MnO₂ was formed under acidic conditions at 160 °C for at least 24 h [21]. Hydrothermal reaction (24 h, room temperature) of Mn(II) species at different pH (6–11) led to polymorphous crystalline MOs [22]. Amorphous MO nanopowders were synthesized by the reduction of potassium permanganate in organic solvents at room temperature within 2 h [23]. Amorphous nanoscale powders were synthesized by microwave-assisted hydrothermal comproportionation of MnO₄⁻ and Mn(II) species in neutral aqueous solutions at 75 °C within 30 min [24]. Microwave-assisted hydrothermal reduction of MnO₄⁻ with NaNO₂ under acidic conditions led to amorphous MO at 90–170 °C within 8–20 min [25,26] according to following equation

$$\begin{array}{c}{2\mathrm{KMnO}}_4{}_{\left(\mathrm{aq}\right)}{+3\mathrm{NaNO}}_2{}_{\left(\mathrm{aq}\right)}{+\mathrm{H}}_2{\mathrm{SO}}_4{}_{\left(\mathrm{aq}\right)}{+2\mathrm{H}}_2{\mathrm{O}}_{\left(\mathrm{aq}\right)}\\ \to {2\mathrm{MnO}}_2{}_{\left(\mathrm{s}\right)}{+3\mathrm{NaNO}}_3{}_{\left(\mathrm{aq}\right)}{+\mathrm{K}}_2{\mathrm{SO}}_4{}_{\left(\mathrm{aq}\right)}{+3\mathrm{H}}_2{\mathrm{O}}_{\left(\mathrm{aq}\right)}\end{array}$$

(3)

Comproportionation reactions of Mn(VII) and Mn(II) species in strongly alkaline media such as NaOH or KOH lead to the formation of sodium birnessite Na_0.55Mn₂O₄ * 1.5 H₂O (PDF 00-043-1456) or potassium birnessite K_0.46Mn_1.54Mn_0.46O₄ * 1.4 H₂O (PDF 00-080-1098), as studied in detail by Boumaiza et al. [27]. After 16 h reaction time at 60 °C, crystalline K-birnessite with some hausmannite impurities was obtained. Luo et al. [28,29] found that at a Mn(VII):Mn(II) molar ratio below 0.24, preferentially β-MnOOH (feitknechtite PDF 00-018-0804) and Mn₃O₄ (hausmannite PDF 00-024-0734) are formed, while above 0.40, mostly amorphous material was yielded. They found an optimum at 0.32 molar ratio. Luo also studied the influence of the alkalinity at 0.40 Mn_VII:Mn_II molar ratio on the nature of the MnO₂ crystal phase, and concluded that at a molar ratio of 0.40 and OH⁻ concentration of ≥1.6 mol L⁻¹, the formation of K & Na birnessite is favored. Reported reaction times ranged from least 16 h [27] to several days [28,29].

This work aims at both further increasing crystallinity and reducing the reaction time of layered MnO₂ synthesis by using microwave-assisted hydrothermal comproportionation of MnO₄⁻ and Mn(II) species in an alkaline solution, according to following reaction

$$\begin{array}{c}3\mathrm{Mn}{\left({\mathrm{NO}}_3\right)}_2{}_{\left(\mathrm{aq}\right)}{+2\mathrm{KMnO}}_4{}_{\left(\mathrm{aq}\right)}{+4\mathrm{KOH}}_{\left(\mathrm{aq}\right)}\\ \to {5\mathrm{MnO}}_2{}_{\left(\mathrm{s}\right)}{+6\mathrm{KNO}}_3{}_{\left(\mathrm{aq}\right)}{+2\mathrm{H}}_2{\mathrm{O}}_{\left(\mathrm{aq}\right)}\end{array}$$

(4)

2. Materials and Methods

2.1. Experimental

First, 5.50 g potassium permanganate (Merck, Darmstadt, Germany) was weighed in a glass beaker that was filled with 100 mL ultrapure water (ELGA Purelab©, Veolia Water, Paris, France) to get a stock solution. A 1 g mL⁻¹ manganese nitrate (Sigma-Aldrich, Schnelldorf, Germany) solution and a 15.2 mol L⁻¹ potassium hydroxide (Merck) solution were prepared in ultrapure water. A typical procedure consisted of pipetting and mixing 10 mL KMnO₄, 1.8 mL Mn(NO₃)₂, 3.9 mL KOH and 21.2 mL ultrapure water.

Microwave-assisted hydrothermal comproportionation was carried out at 110 °C for 14 and 96 h after 1 h aging under vigorous stirring in a microwave (MLS Ethos.lab, Leutkirch, Germany). The reaction mixture was cooled, and then washed with ultrapure water and centrifuged two times. The coating ink was prepared by adding 30 wt% carbon (C65, Emerys, Bironico, Switzerland) and 10 wt% PTFE (Sigma Aldrich, Schnelldorf, Germany), and hereafter air-sprayed on a 36 cm² gas diffusion layer (GDL H2315I2, Freudenberg, Weinheim, Germany). The coated electrodes were punched into disks with a diameter of 18 mm. The electrodes were assembled together with a glass fiber separator (thickness: 1.5 mm), 300 mg of respective AIB or ZIB electrolyte, and an 18 mm metal foil anode disk (Al or Zn). For half cell measurements, a pseudo-reference metal wire (Al or Zn) was applied. All components were assembled into EL-ECC battery cells (El-Cell, Hamburg, Germany). AIB cells were assembled inside a glovebox in an Argon atmosphere (H₂O and O₂ < 0.1 ppm)

2.2. Characterization Techniques

Thermo gravimetric analysis (TGA) was performed with a NETSCH Jupiter F3 apparatus (Selb, Germany). The samples were heated from 30 to 600 °C with a temperature ramp of 10 °C min⁻¹. The nitrogen flow rate was set to 40 mL min⁻¹.

XRD analysis was carried out with a BRUKER D8 Advance diffractometer (Billerica, MA, USA) with a goniometer radius of 300 mm and Cu K_α radiation (λ = 1.5418740 Å). Standard diffraction patterns were taken at room temperature between 2θ = 10–80° at an increment of 0.02° and 1.5 s exposure time for standard resolution and at an increment of 0.01° and 12 s exposure time for high resolution (HR-XRD) experiments. For XRD analysis at elevated temperatures, a special vacuum oven chamber (HTK1200, BRUKER, Billerica, MA, USA) was used. The temperature profile was increased by 5 °C min⁻¹ with a conditioning time of 2 h at three preselected temperatures (250, 400, 600 °C) before high-resolution XRD measurement. XRD patterns were taken between 2θ = 10–80° at an interval of 0.02° with 1.5 s exposure time. All XRD patterns were processed with Match! software (3.5.2.104 version; Crystal Impact, Bonn, Germany).

Scanning electron microscopy (SEM) images were taken with a Hitachi 4100 (Tokyo, Japan) at 20 mm WD and 20 kV to observe morphology of the powder samples.

Specific surface analysis of the powder samples was performed on a Quantachrome Autosorb-iQ-MP-XR (Boynton Beach, FL, USA) with N₂ as adsorbent.

Since in AIB cells, EMIMCl (1-ethyl-3-methyl-imidazonlinumchloride)/AlCl₃ is usually used as the electrolyte, the cathode material has to be water-free. Aluminum tetrachloride reacts heavily with residual water under the formation of aluminum hydroxide and hydrochloric acid gas:

$${\mathrm{AlCl}}_4^{-}{}_{\left(\mathrm{l}\right)}{+3\mathrm{H}}_2{\mathrm{O}}_{\left(\mathrm{aq}\right)}\to \mathrm{Al}{\left(\mathrm{OH}\right)}_{3\left(\mathrm{s}\right)}{+3\mathrm{HCl}}_{\left(\mathrm{g}\right)}{+\mathrm{Cl}}^{-}{}_{\left(\mathrm{aq}\right)}$$

(5)

To avoid this reaction, and as consequence heavy corrosion induced by HCl gas, all powder samples were dried at 120 °C under a vacuum (as determined in Section 3.5) at the schlenk line (<0.3 mbar) for at least 24 h prior to assembly and tests in laboratory cells.

Electrochemical analysis was carried out in EL-ECC battery cells (El-Cell) with a BioLogic BCS-810 battery testing system (Seyssinet-Pariset, France). Cyclic voltammograms were run at 10 mV s⁻¹ and charge/discharge profiles from 2.4 to 0.8 V vs. Al/Al³⁺ in AIB and from 1.8 to 1.0 V vs. Zn/Zn²⁺ in ZIB. Charge/discharge experiments were run at 2.5–30 mA g⁻¹ current densities normalized to active cathode mass.

3. Results & Discussion

3.1. Influence of KOH Concentration on MnO₂ Powder Synthesis

Formation of qualitative, highly-pure birnessite was achieved by Luo et al. [28] after 75 days at room temperature. In order to reduce the reaction time significantly, the influence of temperature and alkalinity on δ-MnO₂ formation was investigated in this work. In preliminary experiments, the concentration of KOH was varied between 0.6 and 2.5 mol L⁻¹, while temperature was fixed to 110 °C and reaction time set to 96 h as shown in Figure 1. The XRD spectra depict as the main phase birnessite, the hydrated layered MnO_2. Due to the low-temperature synthesis which results in low crystalline order and very small particle sizes, the XRD patterns show broad peaks of birnessite. In order to determine the optimal KOH concentration, the main diffraction peak at 2θ = 25° in XRD spectra, which is typical for birnessite-type manganese, was first considered and evaluated in terms of intensity and broadening (crystallinity). For the 2.5 M KOH sample, a sharper peak was observed in comparison to those obtained at more diluted concentrations, which indicates more pronounced crystallinity. The higher the molarity, the narrower the (001) peak.

Additionally, the formation of hausmannite-type manganese phase increased with decrease of alkalinity, so that a compromise between crystallinity and purity level was adopted. Since in 2.5 M KOH, soluble manganate species were observed due to incomplete comproportionation reaction, and in 0.6 M KOH highest proportion of hausmannite was detected, optimal concentration was set to 1.6 M KOH for further experiments. A qualitative evaluation of the as-synthesized MnO₂ and 14 h reaction time composition (arbitrary unit = au) based on the peak area of each single phase A_i divided by total peak area A_total according to following equation (au = A_i/A_total) is shown in Figure 2d. Because hausmannite has a denser 3D crystal structure (4.59 g cm⁻³) [30] compared to 2D-birnessite (3.40 g cm⁻³) with large interplanar space, the hausmannite diffraction pattern intensity is significantly more pronounced than that of birnessite. Thus, a quantitative determination of phase composition of the as-prepared materials is not straightforward.

3.2. XRD Spectra of MnO₂ Powder Synthesized in 1.6 M KOH for 14 h

HR-XRD measurements of the 14 h/1.6 M KOH sample of as-synthesized MnO₂ powder revealed characteristics of a highly pure δ-MnO₂ phase with significant peaks for birnessite-type MnO₂ at 2θ = 12.5° and 25° corresponding to (001) and (002) reflection pairs, respectively, and surprisingly, without any additional phase impurities (Figure 2a,b). The three additional peaks marked with * at higher 2θ values also belonging to birnessite structure cannot be clearly indexed. The d-spacing of the δ-MnO₂ sheets was determined to 7.1 Å. By plotting d-spacing from HR-XRD measurements, typical reflexes for monoclinic (A) d₂₀₀ ≈ 2.52, d₁₁₀ ≈ 2.48 Å and d₃₁₀ ≈ 1.43 Å, turbostratic (B) d₁₀₀ ≈ 2.42 Å and d₁₁₀ ≈ 1.42 Å and hexagonal (C) d₁₀₂ ≈ 2.0 Å and d₁₀₃ ≈ 1.7 Å crystal systems were identified (Figure 2b) according to the study on the occurrence of different crystal phases in synthetic Na-rich birnessite and hexagonal birnessite published by Drits et al. [30].

Additionally, the significantly narrower shape of a (001) peak at 2θ = 12.5° is a clear indication for higher crystallinity level after only 14 h reaction time compared to that after 96 h (Figure 2c). This obvious result confirms that the microwave technique is well-suited for the fast synthesis of highly pure birnessite manganese oxides.

The appearance of a predominant monoclinic crystal phase besides the hexagonal one (shown in Figure 2b) might be explained by post-synthesis treatment thereof. Once the residual interlayer water in-between hexagonal layers has been removed during drying step, the potassium ions are anchored in MnO₂ matrix, as shown in Figure 3, and as a consequence, the basal planes of birnessite shift along the b-axis due to the negative and repellent charge of emergent oxygen atoms from the MnO₂ plane (Figure 3).

This assumption is in good accordance with the obtained HR-XRD diffraction pattern (Figure 2a,b). In the presence of water, hexagonal crystal phase prevails.

3.3. Influence of Hydroxide Concentration on Reaction Kinetics

In order to get a better insight into thermodynamic driving forces of each reaction step, free Gibbs enthalpies were calculated for all considered reactions based on equal concentrations of 1 M at 298 K and given in

Table 1. Overview of standard free Gibbs enthalpies of reaction compounds involved in MnO₂ formation [32,33,34]. Formulas and oxidation states of all involved compounds are listed in Table A1.

Compound	Mn2+	Mn(OH)2	MnO4−	MnO42−	MnO43−	MnOOH	MnO2	Mn3O4	H2O	OH−
∆rG/kJ mol−1	−228	−615	−447	−501	−722	−548	−465	−1283	−237	−157

.

The concentration of OH⁻ was set to 1.6 mol L⁻¹ because at 2.5 mol L⁻¹, a stable green colored manganate(VI) solution was obtained as mentioned, according to following reaction steps:

$${\mathrm{Mn}}^{2+}{}_{\left(\mathrm{aq}\right)}{+2\mathrm{OH}}^{-}{}_{\left(\mathrm{aq}\right)}\stackrel{{\mathrm{k}}_1}{\to }\mathrm{Mn}{\left(\mathrm{OH}\right)}_{2\left(\mathrm{s}\right)}$$

(6)

$${\Delta }_{\mathrm{r}}\mathrm{G}=-615\frac{\mathrm{kJ}}{\mathrm{mol}}-\left(-228\frac{\mathrm{kJ}}{\mathrm{mol}}+2\times \left(-157\frac{\mathrm{kJ}}{\mathrm{mol}}\right)\right)=-73\frac{\mathrm{kJ}}{\mathrm{mol}}$$

(6.1)

$$\mathrm{Mn}{\left(\mathrm{OH}\right)}_{2\left(\mathrm{s}\right)}{+\mathrm{MnO}}_4^{-}{}_{\left(\mathrm{aq}\right)}{+\mathrm{OH}}^{-}{}_{\left(\mathrm{aq}\right)}\stackrel{{\mathrm{k}}_2}{\to }{\mathsf{\beta }-\mathrm{MnOOH}}_{\left(\mathrm{s}\right)}{+\mathrm{MnO}}_4^{2-}{}_{\left(\mathrm{aq}\right)}{\left(\mathrm{green}\right)+\mathrm{H}}_2{\mathrm{O}}_{\left(\mathrm{aq}\right)}$$

(7)

$$\begin{array}{c}{\Delta }_{\mathrm{r}}\mathrm{G}=\left(-548\frac{\mathrm{kJ}}{\mathrm{mol}}+\left(-501\frac{\mathrm{kJ}}{\mathrm{mol}}\right)+\left(-237\frac{\mathrm{kJ}}{\mathrm{mol}}\right)\right)-\\ \left(-615\frac{\mathrm{kJ}}{\mathrm{mol}}+\left(-447\frac{\mathrm{kJ}}{\mathrm{mol}}\right)+\left(-157\frac{\mathrm{kJ}}{\mathrm{mol}}\right)\right)=-67\frac{\mathrm{kJ}}{\mathrm{mol}}\end{array}$$

(7.1)

Reactions (6) and (7) are favored due to negative free enthalpies, and therefore, an essential indicator for assumed first reaction steps during MnO₂ formation. Furthermore, Mn(OH)₂ forms with an excess of OH⁻, a very stable [Mn(OH)₃]⁻ complex with a stability constant value of 3.98 × 10¹⁶. Hence, once the solution reaches the OH⁻ saturation level, the complex is in equilibrium with its precipitate Mn(OH)₂.

$$\mathrm{Mn}{\left(\mathrm{OH}\right)}_{2\left(\mathrm{s}\right)}{+\mathrm{OH}}^{-}{}_{\left(\mathrm{aq}\right)}\stackrel{{\mathrm{k}}_3}{\leftrightarrow }{\left[\mathrm{Mn}{\left(\mathrm{OH}\right)}_3\right]}^{-}{}_{\left(\mathrm{aq}\right)}$$

(8)

Reformed Mn(OH)₂ can further react with MnO₄⁻ under formation of MnO₄²⁻ as described before in Reaction (7). Manganate(VI) MnO₄²⁻ subsequently reacts with Mn(OH)₂ under the formation of β-MnOOH and blue-colored hypomangante(V) MnO₄³⁻ as follows

$$\mathrm{Mn}{\left(\mathrm{OH}\right)}_{2\left(\mathrm{s}\right)}{+\mathrm{MnO}}_4^{2-}{}_{\left(\mathrm{aq}\right)}{+\mathrm{OH}}^{-}{}_{\left(\mathrm{aq}\right)}\stackrel{{\mathrm{k}}_4}{\leftrightarrow }{\mathsf{\beta }-\mathrm{MnOOH}}_{\left(\mathrm{s}\right)}{+\mathrm{MnO}}_4^{3-}{}_{\left(\mathrm{aq}\right)}{\left(\mathrm{blue}\right)+\mathrm{H}}_2{\mathrm{O}}_{\left(\mathrm{aq}\right)}$$

(9)

$$\begin{array}{c}{\Delta }_{\mathrm{r}}\mathrm{G}=\left(-548\frac{\mathrm{kJ}}{\mathrm{mol}}+\left(-722\frac{\mathrm{kJ}}{\mathrm{mol}}\right)+\left(-237\frac{\mathrm{kJ}}{\mathrm{mol}}\right)\right)-\\ \left(-615\frac{\mathrm{kJ}}{\mathrm{mol}}+\left(-501\frac{\mathrm{kJ}}{\mathrm{mol}}\right)+\left(-157\frac{\mathrm{kJ}}{\mathrm{mol}}\right)\right)=-234\frac{\mathrm{kJ}}{\mathrm{mol}}\end{array}$$

(9.1)

Hypomanganate has a high oxidation potential of E⁰ = 0.96 V (vs. NHE) and reacts with β-MnOOH (E⁰ = 0.95 V) to birnessite type δ-MnO₂ according to Equation (10):

$${\mathsf{\beta }-\mathrm{MnOOH}}_{\left(\mathrm{s}\right)}{+\mathrm{MnO}}_4^{3-}{}_{\left(\mathrm{aq}\right)}{+\mathrm{H}}_2{\mathrm{O}}_{\left(\mathrm{aq}\right)}\stackrel{{\mathrm{k}}_5}{\leftrightarrow }{2\mathrm{MnO}}_{2\left(\mathrm{s}\right)}{+3\mathrm{OH}}^{-}{}_{\left(\mathrm{aq}\right)}$$

(10)

$$\begin{array}{c}{\Delta }_{\mathrm{r}}\mathrm{G}=\left(2\times \left(-465\frac{\mathrm{kJ}}{\mathrm{mol}}\right)+3\times \left(-157\frac{\mathrm{kJ}}{\mathrm{mol}}\right)\right)-\\ \left(-548\frac{\mathrm{kJ}}{\mathrm{mol}}+\left(-722\frac{\mathrm{kJ}}{\mathrm{mol}}\right)+\left(-237\frac{\mathrm{kJ}}{\mathrm{mol}}\right)\right)=+106\frac{\mathrm{kJ}}{\mathrm{mol}}\end{array}$$

(10.1)

Since standard potentials of MnO₄⁻ (E⁰ = 0.558 V) and MnO₄²⁻ (E⁰ = 0.60 V) in alkaline media are not high enough for MnOOH oxidation, the limiting reaction step is the formation of hypomanganate(V), despite relative high exothermic reaction enthalpy value (−234 kJ mol⁻¹). In our experiments at OH⁻ = 0.6 and 0.8 mol L⁻¹, other manganese oxides like hausmannite (Mn₃O₄) were yielded (Figure 3). This may result from too low hypomanganate(V) concentrations. We assume that the reaction constant k₄ of Equation (9) increases with higher [OH⁻] concentration. This assumption is supported by the increase in crystallinity of formed birnessite at 1.6, and especially, 2.5 mol L⁻¹ (Figure 3).

The law of mass action for Equations (9) and (10) and the calculated endothermic reaction enthalpy (+106 kJ mol⁻¹) of MnO₂ formation explain the extremely long reaction time of several days at room temperature reported by Luo et al., and at least 14 h at 110 °C in microwave synthesis in this work.

$${\mathrm{K}}_4=\frac{\left[\mathsf{\beta }-\mathrm{MnOOH}\right]\times \left[{\mathrm{MnO}}_4^{3-}\right]{\times [\mathrm{H}}_2\mathrm{O}]}{\left[\mathrm{Mn}{\left(\mathrm{OH}\right)}_2\right]\times \left[{\mathrm{MnO}}_4^{2-}\right]{\times [\mathrm{OH}}^{-}]}$$

(11)

$${\mathrm{K}}_5=\frac{{\left[{\mathrm{MnO}}_2\right]}^2\times {\left[{\mathrm{OH}}^{-}\right]}^3}{\left[\mathsf{\beta }-\mathrm{MnOOH}\right]\times \left[{\mathrm{MnO}}_4^{3-}\right]{\times [\mathrm{H}}_2\mathrm{O}]}$$

(12)

As assumed, the formation of hypomanganate is directly correlated to the concentration of [OH⁻] and [MnO₄²⁻] (Equation (11)). Due to low hypomanganate, and therefore, low MnO₂ concentration at 0.6 & 0.8 mol L⁻¹ hydroxide, the equilibrium is massively shifted to the educt site (Equation (12)). This might explain the formation of crystalline birnessite phase MnO₂ at [OH⁻] concentrations ≥ 1.6 mol L⁻¹. The formation of hausmannite at [OH⁻] concentrations ≤ 1.6 mol L⁻¹ can be explained by the thermodynamically-favored reaction of residual Mn(OH)₂ with MnOOH (−46 kJ mol⁻¹) according to Reaction (13). The presence of residual Mn(OH)₂ can be explained by significantly-decreased hypomanganate formation, according to Equation (11). After 96 h of heating, the thermodynamically more stable hausmannite structure is formed due to the reduction and collapse of δ-MnO₂.

$$\mathrm{Mn}{\left(\mathrm{OH}\right)}_{2\left(\mathrm{s}\right)}{+2\mathrm{MnOOH}}_{\left(\mathrm{s}\right)}\stackrel{{\mathrm{k}}_6}{\to }{\mathrm{Mn}}_3{\mathrm{O}}_{4\left(\mathrm{s}\right)}{+2\mathrm{H}}_2{\mathrm{O}}_{\left(\mathrm{aq}\right)}$$

(13)

$$\begin{array}{c}{\Delta }_{\mathrm{r}}\mathrm{G}=\left(-1283\frac{\mathrm{kJ}}{\mathrm{mol}}+2\times \left(-237\frac{\mathrm{kJ}}{\mathrm{mol}}\right)\right)-\left(-615\frac{\mathrm{kJ}}{\mathrm{mol}}+2\times \left(-548\frac{\mathrm{kJ}}{\mathrm{mol}}\right)\right)=\hfill \\ -46\frac{\mathrm{kJ}}{\mathrm{mol}}\hfill \end{array}$$

(13.1)

The limiting reaction step is the hypomanganate MnO₄³⁻ formation that is directly dependent on the MnO₄²⁻ and [OH⁻] concentration. Our experiments confirm that strongly-alkaline environments between 2.5 and 0.8 with an optimum at 1.6 mol L⁻¹ and elevated temperatures are required for the fast formation of layered δ-MnO₂. However, higher temperatures than 110 °C lead to the formation of hausmannite due the collapse of the birnessite crystal structure.

3.4. TGA Measurements

TGA measurements were carried out in order to determine the water content of the as-prepared powder material. The initial mass loss of about 10% in a temperature range between 60 and 200 °C is accompanied by a low endothermic change than can be associated with interlayer water evaporation from manganese matrix (Figure 4). At temperatures above 200 °C, a pronounced endothermic process with less intense mass loss is assigned to a crystal phase change from monoclinic to indefinable layered MnO₂ and hausmannite (see XRD spectra in Figure A2 in Appendix B).

3.5. In-Situ XRD Spectra

In-situ XRD measurements were carried out to evaluate thermal material stability, and to determine optimal drying parameters for removing all interlayer water, before it can be assembled to an aluminum-ion battery together with the highly water-sensitive electrolyte EMIMCl/AlCl₃. In-situ XRD measurements took place under a vacuum and small heating steps at a low heating rate (5 °C min⁻¹), combined with 2 h equilibration time at respective temperatures of up to 600 °C.

The application of a vacuum at 30 °C led to a shrinkage of the d-spacing from 7.1 to 6.45 Å (Figure 5a) that can be correlated to the removal of interlayer water. Further heating to 250, 400, and 600 °C influenced neither the interlayer distance nor the crystal phase. In Figure 3, the d-spacing for the monoclinic structure without any interlayer water amounts theoretically to 6.35 Å. The slightly higher measured d-spacing value of 6.45 Å within 30–600 °C temperature range might be explained by the formation of a birnessite polymorph K_xMnO₂ with x ≤ 0.5. Thus, the as synthesized δ-MnO₂ does not fit perfectly with the reference spectra of hydrated or dried single phase monoclinic δ-MnO₂ (Figure 5b).

Considering that nearly all water can be removed just by applying a vacuum to the sample, we set the drying temperature to 120 °C, slightly above the boiling point of water, to be sure that the whole sample was completely water-free. If a faster drying step is needed, conditioning under a vacuum at higher temperatures would be also feasible.

3.6. BET Analysis

Specific surface area of as-prepared δ-MnO₂ powder analysis by N₂ adsorption revealed a total pore volume of 0.088 cm³ g⁻¹ and a surface area of 27 m² g⁻¹. Hence, the pore size distributions computed for an equilibrium and adsorption model show nearly the same local maxima of mesopores between 5 and 40 nm. The mean pore size was determined to 5.4–6.0 nm (Figure 6a,b). The mesoporous character is confirmed by the presence of hysteresis loop of pore volume for P/P₀ = 0.5–0.9 in adsorption/desorption isotherm profile shown in Figure 6c. The porous flake-like morphology of δ-MnO₂ can be seen in Figure 6d. The flakes have a diameter of up to 0.2–2 µm and a thickness of 50–100 nm. These results are in very good accordance with information reported in other works about birnessite type structures [35,36,37,38,39].

3.7. Electrochemical Measurements

3.7.1. Aluminum-Ion Battery

The influence of a potential window on cyclic voltammograms (CV) profiles of synthesized δ-MnO₂ in EMIMCl/AlCl₃ is presented in Figure 7a. The potential window limits were set to 2.45 and 0.80 V vs. Al/Al³⁺ that correspond to upper electrolyte stability window and irreversible manganese reduction at lower potential values than 0.8 V, respectively. A small anodic shoulder at 1.9–2.0 V, and corresponding cathodic peaks shifted from 1.6–1.8 V with increasing the potential window, are induced by the Mn²⁺/Mn³⁺ redox couple. An irreversible reduction peak is observed at 1.2 V. This might result from reduction of the electrode or from electrolyte impurities. Since the first and last scan between 0.8 and 2.1 V are identical, potential extension up to 2.3 V obviously didn’t affect the manganese oxide structure.

Charge/discharge experiments under full-cell conditions revealed poor energy density of 4.2–3.3 mWh g⁻¹ which corresponds to ~0.8% of theoretically value (524 mWh g⁻¹) based on Mn^+3/+4O₂ redox pair (Figure 7b). Within the first 14 cycles, the energy efficiency was about 56%, as shown in Figure 7c.

As determined in Section 3.5, the interlayer distance in dried layered MnO₂ is ~6.45 Å, which corresponds to a gallery height of ~3 Å, which is the distance in between the layers. However, for successful intercalation, the large AlCl₄⁻ (ø = 5.5 Å) has to dissociate into Al³⁺ (ø = 1.08 Å) to enter the gallery. We conclude that the central Al³⁺ ion is too strongly shielded by four Cl⁻ anions in the tetrahedral configuration. Consequently, the affinity of AlCl₄⁻ anions for dissociation into a positive charge species, which is needed for charge compensation of the cathode material during reduction step, is very poor.

In contrast to our oxide cathode, W. Wang et al. reported successful reversible Al³⁺ ion intercalation into a vanadium oxide (VO₂) tunnel structure that showed significantly higher capacities in EMIMCl/AlCl₃ (116 mAh g⁻¹ @ 0.9–0.1 V vs. Al/Al³⁺ for 100 cycles) [12]. However, the reported discharge voltage is significantly lower than in this work (2.0–0.8 V). In a similar work, H. Wang et al. claimed that the excellent reversible capacity of 239 mA g⁻¹ with a V₂O₅ cathode can be attributed to the presence of fully dissociated Al³⁺ ions in the electrolyte [40]. In both works, no proof for the presence of positive Al³⁺ ions in the electrolyte and/or oxide matrix was provided by e.g., Raman, NMR or XPS analysis, respectively.

3.7.2. Zinc-Ion Battery

In contrast to the AlCl₄⁻ ions in the AIB electrolyte, fully dissociated Zn²⁺ ions (ø = 1.48 Å) are present in aqueous electrolytes. Additionally, as shown in Section 3.5, hydrated layered MnO₂ has an increased interlayer distance of 7.1 Å, resulting in an accessible gallery height of ~3.8 Å. Thus, the aqueous ZIB appeared to be a priori a more suitable system to demonstrate the MnO₂ ability for multivalent ion intercalation.

Influence of a potential window on the electrochemical behavior of δ-MnO₂ in choline/zinc acetate and zinc sulfate electrolytes is shown in Figure 8a,b, respectively. In choline acetate, first cathodic peaks were detected at 1.2 V with the corresponding anodic reduction peak at 0.95 V and attributed to Mn²⁺/Mn³⁺. Above 1.6 V, the current starts to increase rapidly, indicating the deintercalation of H⁺ and Zn²⁺, as well as possible water splitting. The corresponding cathodic peaks shifted between 1.37–1.5 V. An unidentified small reduction peak was observed at 1.6–1.7 V. The overall specific current is significantly higher compared to that of the AIB (0.8–1.8 A g⁻¹ vs. 50–110 mA g⁻¹). Interestingly, in contrast to AIB-CVs in Figure 7a in which 1st and 4th curves are almost identical, change in CV shape between 1st and 4th curve after successively extending and narrowing the potential window down to 1.5 V in zinc-containing electrolyte is obvious, especially at 1.2 V. This may be an indicator for irreversible structural change in MnO₂ matrix.

The redox couple of Mn^3+/2+ visible in choline acetate electrolyte (Figure 8a) was not detectable in CV with zinc sulfate electrolyte (Figure 8b) also down to 0.2 V (not shown here). Since Zn²⁺ concentration in choline acetate is lower by a factor of 100 compared to that in zinc sulfate solution, this redox pair was not observed obviously due to a large overpotential arising from the stronger solvation of Zn²⁺ with SO₄²⁻ compared to single-charged acetate anions.

At 1.65 V and 1.25–1.3 V, the Mn^3+/4+ redox pair was observed. The measured reduction potential at ≈1.3 V is in very good accordance with published de-intercalation potential (≈1.3–1.4 V) values of Zn²⁺ from MnO₂ matrix [41]. The measured currents are even higher than those in ChoAc, and reached almost 2 A g⁻¹ at 1.6 V, which may be related to a higher zinc concentration in sulfate solution. The blurred potential above 1.68 V might be explained by oxygen evolution due to water decomposition. The change in CV shape between the 1st and 4th curves indicates a structural change in MnO₂ matrix (Figure 8b) as observed in ChoAc electrolyte (Figure 8a).

In order to allow easier comparison between AIB and ZIB cyclic voltammograms, redox peak potential regions were normalized to the standard hydrogen reference electrode (SHE), and listed in

Table 2. Summary of AIB & ZIB (ChOAc & ZnSO₄) redox potential region from half-cell measurements in Figure 8; * Extrapolated potential values based on measured redox potentials in ZIB.

Mnx+ Redox Pair		SHE	Al	Zn ChOAc	Zn ZnSO4
		V	V vs. Al/Al3+	V vs. Zn/Zn2+	V vs. Zn/Zn2+
Mn3+/Mn2+	Anodic	0.3–0.4	1.96–2.03	1.19	n.d.
	Cathodic	0.1–0.2	1.75–1.82	0.94	n.d.
Mn4+/Mn3+	Anodic	0.8–1.0	[2.47–2.67 *]	1.6–1.8	1.64
	Cathodic	0.6–0.7	[2.27–2.37 *]	1.4–1.5	1.28

* Bold marked potential values were calculated based on the measured peak potentials of Mn⁴⁺/Mn³⁺ in ZIB with CHOAc as electrolyte.

. We assume that anodic peaks between 1.96 and 2.03 V and cathodic peaks at 1.66–1.82 V from the AIB experiments correspond to Mn³⁺/Mn²⁺ and not to the oxidation/reduction of Mn⁴⁺/Mn³⁺ in δ-MnO₂. The theoretical Mn⁴⁺/Mn³⁺ redox potential in AIB derived from the measurements in ZIB should be in the range of 2.3–2.7 V (see Table 2), whereas the decomposition onset potential of EMIMCl/AlCl₃ is ≈2.5 V (vs. Al/Al³⁺). Because of electrolyte decomposition and poor dissociation of AlCl₄⁻ into positively-charged species, EMIMCl/AlCl₃ appeared to be an inappropriate electrolyte for high-energy AIB batteries with metal oxide cathodes.

Results from galvanostatic charge & discharge experiments of zinc-ion cell with δ-MnO₂ as the cathode material are shown in Figure 9a. After the initiation phase within the first twenty cycles, the specific discharge energy was about 200 mWh g⁻¹ with an energy efficiency ≥ 83%. This specific energy corresponds to about 38% of the theoretical value of pure MnO₂ equal to ~527 mWh g⁻¹ by assuming a one-electron transfer step based on the Mn^4+/3+ redox reaction. The mean discharge voltage was around 1.48 V. The discharge curve shows a plateau region till 1.38 V. As published by Sun et al. [40], this flat plateau region is induced by H⁺ insertion (120–140 mWh g⁻¹) from the slightly acidic electrolyte (pH~4.7). Afterwards, the discharge curve became significantly steeper due to insertion of bigger Zn²⁺ ions into MnO₂ matrix (70–80 mWh g⁻¹) at ≤1.35 V. This voltage value is in very good accordance with measured ≈1.3 V intercalation potential as shown in Figure 8b. The potential drop could be explained by a higher energetic diffusion barrier of solvated zinc ions within layered MnO₂ matrix compared to that of protons. Since the ionic radii of K⁺ (138 pm) and H₂O (2.78 Å) defines the interlayer distance in δ-MnO₂ matrix, hydrogen (53 pm) and zinc (74 pm) can be a-priori intercalated. The gravimetric energy and power density was determined to 210 Wh kg⁻¹ (748 Wh L⁻¹) and 44.4 W kg⁻¹. After the 20th cycle, the capacity started to decrease continuously down to 72 mWh g⁻¹ after the 80th cycle (Figure 9b). After the 50th cycle, the energy density amounted to 110 Wh kg⁻¹ (392 Wh L⁻¹) which is ~48% less than that yielded during the first cycle.

4. Conclusions

A modified microwave-assisted hydrothermal route based on comproportionation of potassium permanganate and manganese nitrate at 110 °C within 96 h led to layered δ-MnO₂ birnessite phase with hausmannite Mn₃O₄ as side phase. The optimum KOH concentration in terms of hausmannite impurities was found to be 1.6 mol L⁻¹. Further efforts to reduce side-phase were very successful by only reducing reaction time down to 14 h. Experiments with shorter reaction time failed. These should be eventually performed at slightly higher temperature than 110 °C.

The as-synthesized manganese oxide powders have been tested as cathode material for both Al-ion and Zn-ion battery. In AIB, a very poor capacity was yielded due to large size (ø = 5.5 Å) and negative charge of AlCl₄⁻ in EMIMCl/AlCl₃ electrolytes which does not fit into the gallery of dried MnO₂ (d = 3 Å). It must be concluded that dissociation of the AlCl₄⁻ into the much smaller Al³⁺ ions (d = 1.08 Å) didn’t take place due to the strong shielding by chloride ions. In ZIB, the fully-dissociated Zn²⁺ ions (d = 1.48 Å) are readily able to intercalate into the hydrated interlayer space (d = 3.8 Å). A specific discharge energy of about 200 mWh g⁻¹MnO₂ at 1.48 V mean voltage and 85% energy efficiency was obtained within first 20 cycles in 1 M ZnSO₄. At C5 discharge current, capacity is induced by first proton intercalation between 1.65 and 1.3 V, followed by Zn intercalation at lower cell polarization values. Thus, the as-synthesized δ-MnO₂ is active for divalent-ion intercalation, and is a very promising candidate as cathode material in Zn-ion batteries. However, cyclability and cathode material conductivity and stability have to be significantly improved.