Elucidating zinc-ion battery mechanisms in freestanding carbon electrode architectures decorated with nanocrystalline ZnMn₂O₄

Abstract

Rechargeable zinc-ion batteries represent an emerging energy-storage technology that offers the advantages of low cost, use of abundant and nontoxic materials, and competitive energy content in lightly packaged forms. Nanoscale manganese oxides are among the most promising positive-electrode materials for zinc-ion cells, and their performance is further enhanced when these oxides are expressed as conformal deposits on porous carbon architectures, such as carbon nanfoam paper (CNF). We describe an “in-place” conversion of nanometric birnessite Na⁺-MnOx@CNF to crystalline spinel ZnMn₂O₄@CNF, a manganese oxide polymorph that nominally contains sites for Zn²⁺ insertion. The ZnMn₂O₄@CNF cathodes are electrochemically conditioned in two-terminal cells and ex situ characterized using X-ray diffraction, scanning electron microscopy, energy-dispersive spectroscopy, and X-ray photoelectron spectroscopy. Despite specific Zn²⁺ insertion sites in ZnMn₂O₄, we demonstrate that the predominant discharge mechanism involves coupled insertion of protons and precipitation of Zn₄(OH)₆SO₄·xH₂O; upon recharge, protons deinsert and Zn₄(OH)₆SO₄ dissolves.

---

Introduction

Manganese oxides (MnOx) have a long history as charge-storing materials in devices ranging from primary alkaline Zn/MnO₂ cells¹ to rechargeable Li-ion batteries² to aqueous-electrolyte electrochemical capacitors.^3–6 Interest in these oxides is on the rise because of their prospective use as positive electrodes in rechargeable Zn-ion cells versus a Zn metal negative electrode in Zn²⁺-based aqueous electrolytes.^7–15 This cell chemistry inherits the advantages of the ubiquitous alkaline Zn/MnO₂ battery—low-cost, abundant components and the ability to deliver moderately high specific energy—but uses an even safer mild-pH electrolyte and is extensively rechargeable (hundreds of cycles). Other metal oxides,^16–18 sulfides,^19–21 and phosphates^22–24 are also under investigation, but MnOx is the most likely to transition to commercial Zn-ion batteries because of its lower cost and favorable redox potential (discharge voltage on the order of 1.3 V vs. Zn/Zn²⁺).

A key advancement toward rechargeable Zn-ion batteries was the recognition that nanostructured forms of MnOx undergo reversible redox reactions when electrochemically cycled in mild-pH aqueous electrolytes that contain Zn²⁺ salts (e.g., ZnSO₄). Early reports suggested that insertion/intercalation of Zn²⁺ into MnOx, coupled with Mn^3+/4+ redox,^16–18 provides reversible cycling to relatively high MnOx-specific capacity (>200 mA h g⁻¹). Other studies, however, show evidence for a multistep reaction with comparable specific capacity that involves proton insertion at MnOx; the coupled increase in local pH drives the precipitation of a hydrated Zn₄(OH)₆SO₄ at the electrode surface.^25–33 This complex discharge reaction can often be reversed by re-oxidizing MnOx, resulting in the release of protons and at least partial dissolution of the Zn₄(OH)₆SO₄·xH₂O precipitate. In reality, both mechanisms may be operative for a given MnOx material,³⁴ particularly those that are disordered, nanoscale, and/or porous. Optimizing the performance of MnOx-based positive electrodes for Zn-ion batteries requires understanding the influences of MnOx polymorph and electrode structure on the charge-storage mechanism, which ultimately impacts rate capability, capacity, and cycle life.¹⁰

Recently, we explored the electrochemical Zn-ion behavior of birnessite-like Na⁺-compensated manganese oxide (Na⁺–MnOx) distributed as ultrathin (<20 nm-thick) coatings throughout porous carbon nanofoam papers (MnOx@CNF).^35,36 These binder-free electrodes exhibit theoretical one-electron capacity (308 mA h g_MnO2⁻¹) at moderate rates (1C) in 1 M ZnSO₄.³⁵ When Na₂SO₄ is added to the electrolyte, high rate (20C) operation is enabled by pseudocapacitance mechanisms. Ex situ characterization after conditioning at pertinent cell voltages confirms that H⁺ insertion/de-insertion and subsequent Zn₄(OH)₆SO₄·xH₂O precipitation/dissolution is the dominant charge-storage mechanism for birnessite-like Na⁺-MnOx@CNF. The reversibility of these complex multiphase reactions depend on electrolyte composition and the pore structure of the CNF-based architecture.³⁶

Herein, we investigate Zn-ion charge-storage mechanisms for another MnOx polymorph, spinel-type ZnMn₂O₄, which contains tetrahedral sites that nominally accommodate Zn²⁺ insertion for divalent charge storage.^37–42 The disordered birnessite-like Na⁺-MnOx coatings on CNFs used in our previous study are readily converted to spinel ZnMn₂O₄via topotactic ion-exchange (Zn²⁺ for Na⁺), followed by mild thermal treatment. This transformation is achieved while maintaining the nanoscale, conformal nature of the as-deposited MnOx at the carbon surfaces and the through-connected pore structure of the CNF (Fig. S1, ESI†). We now have the opportunity to directly compare two distinct MnOx polymorphs (birnessite vs. spinel ZnMn₂O₄), but expressed in identical multifunctional electrode architectures.

We first examine key electrochemical properties of ZnMn₂O₄@CNFs in two-terminal cells with an aqueous Zn-ion electrolyte using cyclic voltammetry, AC electrochemical impedance, and galvanostatic charge–discharge for long-term cycling. Ex situ characterization via diffraction, microscopy, and spectroscopy of electrochemically conditioned ZnMn₂O₄@CNFs reveals that the dominant charge-storage mechanism is similar to that of Na⁺-MnOx@CNF, despite the presence of specific Zn²⁺ insertion sites in nanocrystalline ZnMn₂O₄ spinel. The charge-storage mechanism involves H⁺-insertion/de-insertion and subsequent precipitation/dissolution of Zn₄(OH)₆SO₄·xH₂O at the electrified interfaces.

Results and discussion

We previously demonstrated crystal engineering of disordered birnessite Na⁺-MnOx@CNF to crystalline spinel LiMn₂O₄@CNF;^43–45 here we show that this approach can be generalized to produce the Zn²⁺-containing spinel analogue, ZnMn₂O₄@CNF (Fig. 1a). The first step of the process involves electroless redox deposition from aqueous permanganate to generate nanoscale Na⁺-MnOx coatings on the carbon surfaces throughout the CNF paper.^46,47 The resulting Na⁺-MnOx@CNFs are soaked in 1 M ZnSO₄ (aq) to exchange Na⁺ in the lamellar MnOx domains with Zn²⁺, then copiously rinsed and dried to obtain birnessite-like Zn²⁺-MnOx@CNFs. The nanoscale nature of the oxide coating facilitates crystallization at a relatively mild temperature (300 °C), which minimizes particle ripening of the MnOx coating (Fig. 1b and c), as previously observed with LiMn₂O₄@CNFs.^43,44 The thermal processing step under flowing argon (low p_O2) reduces Mn from its initial mixed-valent Mn oxidation state of +3.7 in birnessite Na⁺-MnOx@CNF⁴³ to the expected +3 Mn oxidation state as verified by XPS (Fig. S1, ESI†).

Fig. 1 (a) Schematic of in-place conversion from birnessite Na⁺-MnOx to spinel ZnMn₂O₄; (b and c) Scanning electron micrographs at low and high magnification (inset) of (b) Na⁺-MnOx@CNF and (c) ZnMn₂O₄@CNF; (d) X-ray diffraction patterns of Na⁺-MnOx@CNF, Zn²⁺-MnOx@CNF, and ZnMn₂O₄@CNF. The diffraction peaks for ZnMn₂O₄@CNF index to the tetragonal spinel ZnMn₂O₄ (ICDD# 01-071-2499).

Tracking the progress of phase conversion from birnessite Na⁺-MnOx to spinel ZnMn₂O₄ with powder X-ray diffraction reveals that exchanging Na⁺ for Zn²⁺ does not significantly alter the XRD pattern; both Na⁺- and Zn²⁺-MnOx@CNF display two broad peaks at 37 and 66° 2θ, associated with the disordered birnessite MnOx phase (Fig. 1d). Following thermal treatment, the disordered lamellar MnOx phase transforms to crystalline spinel that indexes to tetragonal ZnMn₂O₄ (Fig. 1d). The average crystallite size is 8 nm, as calculated from whole-pattern fitting, confirming that the coating remains nanoscale during transformation from 2D lamellar to 3D spinel.

While no other crystalline phases are observed in the ZnMn₂O₄@CNF XRD pattern, the retention of some minor fraction of disordered MnOx cannot be precluded. Elemental analysis via inductively coupled plasma–atomic emission spectroscopy of ZnMn₂O₄@CNF yields a Mn:Zn ratio of 2.4, higher than the expected 2.0 for complete conversion. If we assume that all Zn in the sample exists as ZnMn₂O₄, 7.7% of the Mn remains unassociated with the ZnMn₂O₄ phase (Table S1 and eqn (S1), ESI†). Quantitative analysis of the XPS peaks for Mn 2p_3/2, Zn 2p_3/2, and oxide O 1s indicates a composition of ZnMn_2.1O_3.8, in relative agreement with the expected ZnMn₂O₄ stoichiometry (Table 1 and Fig. S1, ESI†). However, we note that pair-distribution function analysis of the in situ crystal engineering of our disordered Na⁺-MnOx@CNF to nanocrystalline LiMn₂O₄@CNF found that the first plane of MnOx, which forms when MnO₄⁻ oxidizes the carbon surface, retains a lamellar morphology that serves as the base of the 3D spinel phase. This foundational plane of MnOx accounts for the presence of a minor fraction of Mn remaining in a non-spinel form.⁴⁴

Table 1 Composition and structural properties of ZnMn₂O₄@CNF as a function of electrochemical conditioning in 1 M ZnSO₄

Echem cond.	Mn:Zn^a	S:Zn^a	Lattice parameter^ba|b|c	Unit cell vol.^b	Zn4(OH)6SO4 precipitate observed? (Method)
a Determined via XPS. b Extracted from XRD. c Cell held at specified voltage for 30 min.
Uncycled	2.1	—	5.74|5.74|9.24	305	No
OCV → 1.75 V^c	1.8	0.04	5.74|5.74|9.23	304	No
1.75 V^c	2.0	0.08	5.74|5.74|9.19	303	No
1.3 V^c	0.9	0.1	5.75|5.75|9.21	305	Nanoscale (?) (XPS, EDS)
0.9 V^c	0.07	0.3	5.77|.77|9.23	308	Macroscale (XRD, SEM)

---

With confirmation of successful phase transformation to spinel, we evaluated the electrochemical performance of ZnMn₂O₄@CNF in two-electrode cells versus a Zn foil anode and using 1 M ZnSO₄ (aq) electrolyte. Because as-synthesized ZnMn₂O₄@CNF is fully discharged with Mn in the +3 state, the electrochemical cells were first scanned to voltages positive of open circuit (∼1.5 V). Somewhat surprisingly, the first positive-going voltammetric scan shows no well-defined anodic peak (Fig. 2a), as would be nominally expected for oxidation of Mn³⁺ sites to Mn⁴⁺, accompanied by de-insertion of Zn²⁺ for charge balance.

Fig. 2 (a) First three cyclic voltammograms of ZnMn₂O₄@CNF in 1 M ZnSO₄ at 0.5 mV s⁻¹. Atomic percent of each element derived from ex situ X-ray photoelectron spectroscopy of (b) uncycled ZnMn₂O₄@CNF and (c) ZnMn₂O₄@CNF held at 1.75 V after linearly scanning directly from OCV in 1 M ZnSO₄.

To gain insight into this unexpected first-scan behaviour, we performed electrochemical impedance spectroscopy (EIS) of ZnMn₂O₄@CNF at 1.75 V (scanned directly from open circuit) followed by ex situ XPS, SEM/EDS, and XRD characterization of the conditioned electrode. The Nyquist plot reveals a high charge-transfer resistance (R_CT) of 26 Ω cm², indicative of significant impediment to multivalent ion extraction from the ZnMn₂O₄ domains (Fig. S2, ESI†).⁴² X-ray photoelectron spectroscopy and EDS corroborate this finding, as the Zn content after 1.75 V conditioning is qualitatively similar to the uncycled ZnMn₂O₄@CNF (Fig. 2b, c and Fig. S3, ESI†), revealing that minimal Zn²⁺ is removed from the spinel lattice during the initial charge. This finding is in agreement with that of Manthiram and co-workers, in which they revealed Zn²⁺ is not removed from ZnMn₂O₄ by NO₂BF₄, a chemical mimic for electrochemical Mn oxidation.⁴⁸ Furthermore, no significant changes in lattice parameters or structure are detected by XRD between the initially charged 1.75 V sample and an uncycled ZnMn₂O₄@CNF (Table 1 and Fig. S4, ESI†).

The ill-defined first-scan voltammetry is consistent with other reports on the initial cycling behaviour of ZnMn₂O₄.^28,38 The absence of Mn^3+/4+ redox in the first positive scan is in contrast to our previous report with analogous LiMn₂O₄@CNF, where lattice-sited Li⁺ was easily removed upon initial electrochemical oxidation of the mixed-valent Mn³⁺/Mn⁴⁺ oxide.⁴³ Unlike LiMn₂O₄@CNF, the voltammetric peaks of ZnMn₂O₄@CNF are not well-defined in the first cycle, with only a single reduction peak at ∼0.9 V observed. We attribute this reduction peak to H⁺ insertion, with the supply of protons arising from the mild acidity of Zn(H₂O)₆²⁺ in the aqueous Zn-based electrolyte.⁶⁰ The expected redox peaks become well-defined on the second and subsequent cycles, with a single anodic peak paired with two cathodic peaks (Fig. 2a). Such redox peaks are commonly attributed to extraction/insertion of Zn²⁺,^38,49 but more recently co-insertion/extraction of H⁺ and Zn²⁺ into ZnMn₂O₄ has been proposed.²⁸

To elucidate the specific ZnMn₂O₄@CNF charge-storage mechanism in aqueous Zn²⁺-containing electrolytes, we use a multi-pronged approach that includes EIS and ex situ characterization of cells conditioned at pertinent voltages. For data reported in the following sections, all ZnMn₂O₄@CNF-based cells are subjected to: (i) a 10-cycle voltammetric break-in; (ii) a linear scan to the voltage of interest (depicted in Fig. 3a); and (iii) potentiostating at that voltage either for 10 min prior to EIS data acquisition or for 30 min for ex situ characterization. Cells are quickly disassembled after voltage conditioning and the ZnMn₂O₄@CNF electrode is rinsed copiously with ultrapure water and dried at 50 °C under flowing N₂(g).

Fig. 3 (a) Cyclic voltammogram showing voltages applied for EIS and subsequent ex situ characterization and (b) Nyquist plot at voltages specified in (a) in 1 M ZnSO₄.

Electrochemical impedance spectroscopy provides insights into the charge-storage mechanism when performed as a function of applied DC voltage. Nyquist plots from this series of cells reveal significant changes in R_CT during charge (e.g., 1.75 V, “a” in Fig. 3a) and discharge (e.g., 1.3 V, “b”, and 0.9 V, “c”, in Fig. 3a). Upon charging at 1.75 V, the R_CT is relatively low at 5 Ω cm², compared to the 26 Ω cm² measured after initial charging from open circuit (Fig. S2, ESI†), revealing that voltammetric break-in enhances performance. Discharging from 1.75 V to 1.3 V, increases the R_CT 4× to 20 Ω cm² with a further increase to 38 Ω cm² after discharging at 0.9 V (Fig. 3b). We previously observed qualitatively similar results for Na⁺-MnOx@CNF conditioned in 1 M ZnSO₄,³⁶ where R_CT increases significantly when fully discharged, arising from the precipitation of electronically insulating Zn₄(OH)₆SO₄·xH₂O at the electrode surface. For ZnMn₂O₄@CNF, the origin of the increase in R_CT upon discharge is attributed to either Zn²⁺ transport hindrances and/or precipitation of passivating Zn₄(OH)₆SO₄·xH₂O.

We use ex situ X-ray diffraction to monitor the expansion/contraction of the ZnMn₂O₄ lattice that would arise from Zn²⁺-insertion/extraction and the appearance of Zn₄(OH)₆(SO₄)·xH₂O.^27,36,39,50 All ZnMn₂O₄@CNFs harvested from conditioned cells show the main XRD peaks for spinel ZnMn₂O₄, indicating that the core crystal structure remains intact through the charge–discharge process (Fig. 4). After charging at 1.75 V, the main ZnMn₂O₄ peaks at 29.3, 33.1, and 36.4° 2θ shift to slightly higher 2θ compared to the uncycled ZnMn₂O₄@CNF, concomitant with an increase in the a and b lattice parameters and a decrease in the c lattice parameter and unit cell volume (Table 1). Discharging at 1.3 V does not alter either the main peak positions or corresponding cell parameters (Table 1), revealing that it is unlikely that Zn²⁺ inserts into the lattice; higher resolution synchrotron experiments are planned in the future to verify this finding.

Fig. 4 X-ray powder diffraction patterns of ZnMn₂O₄@CNF electrodes after conditioning at specified voltages in 1 M ZnSO₄. The blue diamond denotes peak positions indexed to tetragonal ZnMn₂O₄ and the pink asterisk denotes peak positions indexed to Zn₄(OH)₆SO₄·5H₂O.

Upon complete discharge at 0.9 V, additional diffraction peaks appear that index to Zn₄(OH)₆(SO₄)·xH₂O. Because of the overlap of the XRD reflections for ZnMn₂O₄ and Zn₄(OH)₆SO₄·xH₂O, we are unable to confidently fit XRD data with respect to determining changes in lattice parameters of the ZnMn₂O₄ phase, precluding the determination of Zn²⁺ insertion into the lattice at 0.9 V (Table 1). The presence of Zn₄(OH)₆SO₄·xH₂O for cells discharged at 0.9 V is the likely origin of the significant increase in R_CT (Fig. 3b).

Ex situ SEM visualizes morphological changes in ZnMn₂O₄@CNF electrodes as a function of cell voltage. Energy-dispersive X-ray spectroscopy provides a means to elucidate the reaction mechanism by monitoring for the appearance of sulfur, as mapped onto the micrographs, in which sulfur serves as an elemental marker for Zn₄(OH)₆SO₄·xH₂O. Uncycled ZnMn₂O₄@CNF provides the baseline for both morphology and sulfur content.

The exterior surface of the uncycled ZnMn₂O₄@CNF is featureless at low magnification and as expected, only adventitious sulfur is detected (Fig. 5a and b). Higher magnification of the exterior surface reveals the through-connected pore structure (Fig. S5, ESI†), which is also visible in the cross-section (Fig. 5c); minimal sulfur is detected in the interior of the uncycled sample (Fig. 5d).

Fig. 5 Scanning electron micrographs and sulfur elemental maps of the exterior surface (top two rows) and interior surface (bottom two rows) of uncycled ZnMn₂O₄@CNF (a–d) and after conditioning for 30 min at 1.75 V (e–h), 1.3 V (i–l), and 0.9 V (m–p) in 1 M ZnSO₄.

Charging at 1.75 V does not yield any significant changes in the morphology of the exterior or interior surfaces (Fig. 5e, g and Fig. S5, ESI†), but a slight increase in sulfur content is detected (Fig. 5f and h). Upon discharging at 1.3 V, minimal-to-no-change in morphology is observed on either the exterior or the interior surfaces (Fig. 5i, k and Fig. S5, ESI†); however, an increase in sulfur content is visible in the EDS maps (Fig. 5j and l).

A significant change in morphology is observed after discharging at 0.9 V, with large plate-like precipitates visible that extensively cover the exterior surface (Fig. 5m and Fig. S5, ESI†); some of these precipitates protrude into the underlying pore structure (Fig. S5, ESI†). The cross-sectional micrograph reveals that this layer lies on top of the electrode surface and is ∼3 μm thick (Fig. 5o). A significant increase in sulfur is detected on both exterior and interior surfaces (Fig. 5n and p), but with significantly more sulfur concentrated in the 3 μm-thick exterior layer (Fig. 5p).

The SEM, EDS, and XRD results for 0.9 V-conditioned ZnMn₂O₄@CNF substantiate that the micrographically observed exterior layer comprises Zn₄(OH)₆SO₄·xH₂O. In addition to confirming the presence of sulfur, the EDS spectra show an increase in Zn and decrease in Mn (Fig. S6, ESI†) and the XRD data (Fig. 4) corroborates the presence of crystalline Zn₄(OH)₆SO₄·xH₂O. A similar plate-like morphology was also visible on the exterior surface for fully discharged birnessite-type Na⁺-MnOx@CNF electrodes cycled in 1 M ZnSO₄.³⁶

The presence of sulfur throughout the interior of the 0.9 V-conditioned electrode (Fig. 5p) coupled with the observation that Zn₄(OH)₆SO₄ crystallites are oriented orthogonally to the electrode surface, and do not completely occlude the underlying pore structure (Fig. S5, ESI†), reveals that a large fraction of the electrode volume remains accessible to the electrolyte and available to participate in the charge-storage reaction. The absence of visible Zn₄(OH)₆SO₄·xH₂O precipitates in the interior voids of the ZnMn₂O₄@CNF stems from the fact that there is only 10% of the required Zn²⁺ for the reaction inside these pores (eqn (S2), ESI†).³⁶

Tracking the atomic ratios (Mn:Zn and S:Zn) by ex situ XPS provides further insight into the charge-storage mechanism. The Mn:Zn ratio of uncycled ZnMn₂O₄@CNF is 2.1 and after conditioning at 1.75 V decreases slightly to 2, revealing a general return to the starting state after voltammetric break-in and charging (Table 1 and Fig. 6a, b). A small amount of sulfur persists after charging at 1.75 V, detectable by both XPS (S:Zn ratio of 0.08) and EDS (Fig. 5f, h and Fig. S6, ESI†), which we attribute to a patchy <7 nm-thick Zn₄(OH)₆SO₄·xH₂O layer present at a level below the detection limit of XRD; this insulating coating could also be the source of the contrast differences observed in the corresponding micrograph (Fig. 5e). Discharging at 1.3 V decreases the Mn:Zn ratio to 0.9, concomitant with an increase in the S:Zn ratio to 0.10 (Table 1 and Fig. 6c). The binding energy of the S 2p_3/2 peak is 168.8 eV, consistent with sulfate, indicating either that SO₄²⁻ associates at edge sites or that nanoscale Zn₄(OH)₆SO₄·xH₂O is present below the detection limits of XRD. The Mn:Zn ratio decreases by over an order of magnitude to 0.07 upon discharge at 0.9 V (Table 1 and Fig. 6d), attributed to screening of the underlying ZnMn₂O₄ by the 3 μm-thick Zn₄(OH)₆(SO₄)·xH₂O overlayer (Fig. 5o). By measuring a Zn- and S-rich surface (S:Zn ratio = 0.3; Fig. 6d), the XPS data are consistent with the presence of Zn₄(OH)₆SO₄·xH₂O.

Fig. 6 Atomic percent of each element derived from ex situ XPS of ZnMn₂O₄@CNF electrodes as a function of voltage conditioning: (a) uncycled, (b) 1.75 V, (c) 1.3 V, and (d) 0.9 V in 1 M ZnSO₄.

Although ZnMn₂O₄ contains specific insertion sites for Zn²⁺, our data confirm that the dominant charge-storage mechanism for ZnMn₂O₄@CNF is H⁺ insertion/de-insertion with subsequent precipitation/dissolution of Zn₄(OH)₆(SO₄)·xH₂O, similar to that observed on our birnessite-like Na⁺-MnOx@CNF.³⁶ This same reaction mechanism has been proposed for VO₂,^51,52 V₃O₇·H₂O,^53,54 V₂O₅,⁵⁵ NaV₃O₈,⁵⁶ V₁₀O₂₄·12H₂O,⁵⁷ and Co₃O₄⁵⁸ cathode materials, revealing that pH changes upon H⁺ insertion/de-insertion is a general charge-storage mechanism for oxide-based materials in aqueous ZnSO₄ electrolytes, as recently suggested by Kundu and co-workers.⁵³

Circumventing this general precipitation/dissolution process would be advantageous from a performance standpoint (e.g., long-term cycling and rate), but swapping NO₃⁻ for SO₄²⁻ is not feasible, as the former is too oxidizing for the Zn anode. Buffering the SO₄²⁻ electrolyte, however, may be an effective strategy to suppress the precipitation of the Zn₄(OH)₆SO₄·xH₂O salt and is the focus of future experiments.

A charge-storage mechanism that involves precipitation/dissolution of Zn₄(OH)₆(SO₄)·xH₂O on the surface of the ZnMn₂O₄@CNF electrode influences both the capacity and rate of Zn-ion cells; and as such, 3D electrode–architecture designs play a role in energy-storage performance. Cyclic voltammometric examination of ZnMn₂O₄@CNF as a function of scan rate reveals that ZnMn₂O₄-based redox peaks are discernible at scan rates as high as 10 mV s⁻¹ (Fig. 7a). This impressive rate performance, for a nominal battery material, is due to sufficient counter-ion compensation from both an adequate volume of electrolyte within the pores and to rapid transport of ions to the ZnMn₂O₄ domains via the through-connected pore structure of the underlying CNF. A 3D design-enabled performance we have demonstrated with CNFs modified with other MnOx polymorphs.⁵⁹

Fig. 7 (a) Cyclic voltammograms and (b) specific capacity of ZnMn₂O₄@CNF in 1 M ZnSO₄ as a function of scan rate.

A specific capacity of 119 mA h g_ZnMn2O4⁻¹ is delivered at 1 mV s⁻¹ (Fig. 7b), on par with the 1-electron theoretical capacity of ZnMn₂O₄ (116 mA h g⁻¹), further confirming that the charge-storage mechanism is H⁺ insertion/Zn₄(OH)₆SO₄ precipitation. As the scan rate increases to 10 mV s⁻¹, the capacity decreases to 62 mA h g⁻¹ (Fig. 7b). The realization of theoretical specific capacity is a consequence of the 3D multifunctional electrode architecture (Fig. S7, ESI†). The 20 nm-thick ZnMn₂O₄ domains are well-wired to the underlying carbon current collector, as it is generated from the precursor MnOx phase that is deposited via MnO₄⁻¹ redox deposition. In this deposition, the carbon in the nanofoam serves as a sacrificial reductant, and thus the first few layers of the MnOx are embedded into the carbon current collector.^43,46,47 The through-connected pore volume/structure of the 3D multifunctional electrode ensures an adequate supply and rapid transport of ions to the ZnMn₂O₄ domains,⁵⁹ supporting rapid charge–discharge at nominally high rates (1 mV s⁻¹ = 28 min charge/discharge) for a battery material. This 1-electron high-capacity at high rate is in agreement with our previous results for both crystalline spinel LiMn₂O₄@CNF and Na⁺-birnessite-type MnOx@CNF, where these electrodes deliver full theoretical capacity (148 mA h g⁻¹ at 2 mV s⁻¹ and 308 mA h g⁻¹ at 1C, respectively) when cycled in Li⁺- and Zn²⁺-containing aqueous electrolytes, respectively.^35,43

Long-term electrochemical stability is a key requirement for MnOx-based active materials used in aqueous Zn-ion cells. We cycle ZnMn₂O₄@CNF in two-terminal cells with a Zn-foil anode and 1 M ZnSO₄ electrolyte at 1C (136 mA h g_ZnMn2O4⁻¹) for 400 cycles (7 weeks). Capacity decays significantly over the first 200 cycles and plateaus from cycle 200 to 400, leading to a 50% decrease in total capacity (Fig. 8).

Fig. 8 Capacity versus cycle number for ZnMn₂O₄@CNF in 1 M ZnSO₄ at a 1C rate.

The mildly acidic nature of 1 M ZnSO₄ (aq) is known to promote reductive dissolution of MnOx as Mn³⁺ disproportionates to generate soluble Mn²⁺, which could be the source of the capacity fade. To assess this hypothesis, we soaked ZnMn₂O₄@CNF in 1 M ZnSO₄ for 13 days and upon addition of potassium periodate to a portion of the solution, the electrolyte changed from colorless to magenta, indicating the presence of soluble Mn²⁺ species (Fig. S8, ESI†), confirming the disproportionation reaction resulting from the mild acidity of 1 M ZnSO₄. Gravimetric analysis of the ZnMn₂O₄@CNF after soaking in 1 M ZnSO₄, revealed a 6 wt% loss, leaving 35 wt% ZnMn₂O₄ for charge-storage.

This long-term stability problem has been previously addressed by adding Mn²⁺ (e.g., 0.010–0.05 M MnSO₄) to the electrolyte to drive the equilibrium back toward Mn^3+/4+ oxide, resulting in extended cycle life.³⁸ We do not obtain such improvements when using 0.05 M MnSO₄ + 1 M ZnSO₄ in our cycling studies. For our particular electrode structure, we calculate that if 10% of the nanofoam-supported ZnMn₂O₄ were to dissolve, the Mn²⁺ concentration would reach 0.7 M inside the pores (eqn (S3), ESI†). This degree of dissolution likely represents an extreme condition, but reveals that 0.05 M Mn²⁺ is insufficient at suppressing the disproportionation reaction in this porous CNF architecture. In on-going experiments, we are exploring other strategies, including buffering the electrolyte,^30,60 as well as methods to form nanoscale protective coatings at the oxide surface, as previously achieved using bicarbonate electrolyte additives at LiMn₂O₄@CNF.⁴³

Conclusions

Our ability to crystal engineer 2D lamellar birnessite-like Na⁺-MnOx inside high surface-area CNF is extended to generate nanocrystalline ZnMn₂O₄@CNF. We show that despite specific lattice sites for Zn²⁺ insertion into the spinel, the dominant charge-storage mechanism of ZnMn₂O₄@CNF in 1 M ZnSO₄ remains H⁺ insertion/de-insertion coupled with precipitation/dissolution of Zn₄(OH)₆SO₄·xH₂O. A 50% decrease in capacity is observed over 400 cycles when cycled at 1C, which is attributed to dissolution of ZnMn₂O₄, the bulk of which resides in the CNF interior, via disproportionation of electrogenerated Mn³⁺ to Mn⁴⁺ and soluble Mn²⁺.

Experimental

Chemicals and materials

Resorcinol (Sigma Aldrich, 99%), formaldehyde (Sigma Aldrich, 37 wt% in H₂O, 10–15% methanol stabilizer), sodium carbonate (Aldrich Chemical Company, Inc., 99.5 + %), Na₂SO₄ (Sigma Aldrich, ≥99.0%), NaMnO₄·H₂O (Sigma Aldrich, ≥97%) and ZnSO₄ (Sigma Aldrich, ≥99.0%) were used as received. Carbon fiber papers (Lydall Technimat), cellulose acetate filters (1.2 μm pores, SterliTech Corporation), and 0.25 mm Zn foil (Alfa Aesar, 99.98% metal basis) were used as described.

ZnMn₂O₄@CNF synthesis

One-ply 40/500 carbon nanofoam papers (CNF) were fabricated using a previously reported protocol.⁶¹ Briefly, the 40/500 resorcinol–formaldehyde (RF) sol was prepared by mixing 10 g resorcinol + 14.74 g formaldehyde + 0.0177 g sodium carbonate + 13.9 g water and stirring on a magnetic stir plate set at 250 rpm for 30 min, followed by a 2.5 h resting period. Carbon fiber papers (2.5 × 4.5 cm²) were exposed to an air–ice RF plasma (Harrick PlasmaFlo PDC-FMG) for 45 min to introduce oxygen functionalities on the carbon fiber surfaces. The CFPs were then vacuum-infiltrated with the RF sol to generate RF-CFPs and placed between two glass slides with each glass slide edge secured with a mini binder clip. The glass slide assembly was then sealed in duct tape. The RF-CFPs were placed in an aluminum foil pouch with ∼2 mL of water and allowed to cure under ambient conditions for 20 h, then placed in a pressure cooker (Nesco 3-in-1, Target) for 9.5 h at “slow cook” (∼88–94 °C) and then at “warm” until removed. The RF-CFPs were removed from the glass slides, soaked in nanopure water and acetone, each for 1 h, and dried under ambient conditions. Pyrolysis of the RF-CFPs was performed in a tube furnace (Thermo Scientific Lindberg Blue M) under flowing argon by ramping to 1000 °C at a 1 °C min⁻¹ and held at 1000 °C for 2 h to generate carbon nanofoam papers (CNFs).

Manganese oxide (birnessite-like Na⁺-MnOx) was electrolessly deposited onto the “one-ply 40/500” CNF by soaking under vacuum in 0.1 M Na₂SO₄ for 20 h and then in 0.1 M NaMnO₄·H₂O + 0.1 M Na₂SO₄ for 20 h, generating Na⁺-MnOx@CNF.⁴⁶ The Na⁺-MnOx@CNF were removed from the NaMnO₄ solution, thoroughly rinsed with nanopure water, vacuum infiltrated with nanopure water and soaked under vacuum for 1 h; the rinse/soak process was repeated a total of three times. The Na⁺-MnOx@CNF were dried at 50 °C under flowing N₂(g) for 20 h.

To generate ZnMn₂O₄@CNF, the Na⁺-MnOx@CNFs were vacuum infiltrated with 1.0 M ZnSO₄ solution and soaked under vacuum for 24 h, removed from the 1 M ZnSO₄ solution, rinsed copiously with nanopure water, and soaked in nanopure water under vacuum for 1 h, with the nanopure water rinse/soak step repeated two more times. The Zn²⁺-MnOx@CNFs were dried at 50 °C under flowing N₂(g) for 12 h. Next, the Zn²⁺-MnOx@CNF papers were placed in a tube furnace under flowing argon, ramped to 300 °C at a rate of 2 °C min⁻¹, held at 300 °C for 4 h, and then cooled to ambient temperature before removing from the furnace.

Elemental analysis

A ZnMn₂O₄@CNF sample was analyzed to quantitatively determine Mn, Na, and Zn content (sent to Galbraith Laboratories, Inc.). Prior to analysis by inductively coupled plasma–atomic emission spectroscopy, the samples were dried under vacuum.

Electrochemical characterization

Prior to electrochemical tests, the ZnMn₂O₄@CNF electrode was vacuum-infiltrated with 1 M ZnSO₄ for 4 h. Two-electrode Zn-ion Swagelok cells were fabricated with a ZnMn₂O₄@CNF cathode (1/2′′ diameter circle), a cellulose acetate filter wetted with 1 M ZnSO₄ as the separator, and a 0.25 mm-thick Zn foil as the anode. A Gamry REF 600 potentiostat was used to collect cyclic voltammetry, linear sweep voltammetry, chronoamperometry, and AC electrochemical impedance spectroscopy data. Cyclic voltammetry was carried out from 0.9 V to 1.75 V at scan rates of 0.5, 1, 2, 5, and 10 mV s⁻¹. Ex situ and EIS data on ZnMn₂O₄@CNF samples were generated by first doing a 10-cycle voltammetric break-in from 1.75 to 0.9 V to 1.75 V at 2 mV s⁻¹, followed by linear-scan voltammetry at 0.5 mV s⁻¹ to a specified voltage (1.75 V, 1.3 V, or 0.9 V) and holding at that voltage for either 30 min for ex situ characterization samples or 10 min for EIS. After electrochemical conditioning, the cell voltage was terminated, the ZnMn₂O₄@CNF was immediately removed, rinsed well with nanopure water, and dried under flowing N₂(g) for 12 h prior to analysis by XPS, XRD, and SEM/EDS.

Long-term cycling

Two-terminal Zn-ion cells with ZnMn₂O₄@CNF cathodes assembled as described above were galvanostatially cycled at 1C (136 mAh g_ZnMn2O4⁻¹) on an Arbin battery cycler.

Scanning electron microscopy

Exterior surface samples were cut with clean scissors and secured to aluminum stubs with conductive carbon tape (Ted Pella). Cross-sectional samples were prepared by immersing uncycled and conditioned ZnMn₂O₄@CNF samples in liquid nitrogen for 1 min, fractured with a new razor blade, and secured to a 45/90° aluminum stub with conductive carbon tape. Carbon paint was used to make an electrical connection between the exposed surface and the SEM stub, especially critical for imaging samples with electrically insulating Zn₄(OH)₆(SO₄)·xH₂O precipitates. All samples were imaged with a Leo Supra 55 SEM at 20 keV equipped with an Oxford Instruments Aztec energy-dispersive X-ray detector.

X-ray photoelectron spectroscopy

Elemental analyses of the surface of the electrodes were performed using XPS (Thermo Scientific K-Alpha X-ray) equipped with a monochromatic Al Kα source (1486.68 eV) and a 400 μm elliptical spot size. High-resolution spectra over the C 1s, O 1s, Mn 2p, S 2p, and Zn 2p regions were obtained. The instrument was operated using a low-energy electron flood gun; the resulting spectra were not peak-shifted prior to quantitative analysis. Ratios of Mn:Zn and S:Zn were tracked to monitor the degree of precipitated film formation of Zn₄(OH)₆(SO₄)·xH₂O. The spectra were analyzed with Avantage (version 5.35).

X-ray diffraction

X-ray diffraction patterns were collected for MnOx@CNF and ZnMn₂O₄@CNF series using a 3 kW Rigaku Smartlab X-ray diffractometer operating with a Cu Kα (λ = 1.5406 Å) radiation source in continuous mode. The samples were aligned with the incident X-rays by sandwiching each sample between a glass slide and the Rigaku reference sample holder. The average crystallite sizes of selected samples were calculated using peak broadening determined from whole pattern fitting in the Rigaku PDXL analysis software. The reference structure for the pattern fitting was ZnMn₂O₄ (ICDD# 01-071-2499).

Conflicts of interest

The authors have no conflicts to declare.

Acknowledgements

This work was supported by the U.S. Office of Naval Research.

Notes and references

1. Y. Chabre and J. Pannetier, Prog. Solid State Chem., 1995, 23, 1–130.
2. M. M. Thackeray, Prog. Solid State Chem., 1997, 25, 1–71.
3. D. Bélanger, T. Brousse and J. W. Long, ECS Interface Spring, 2008, 17, 49–52.
4. C. J. Xu, F. Y. Kang, B. H. Li and H. D. Du, J. Mater. Res., 2010, 25, 1421–1432.
5. M. Huang, F. Li, F. Dong, Y. X. Zhang and L. L. Zhang, J. Mater. Chem. A, 2015, 3, 21380–21423.
6. Q. Z. Zhang, D. Zhang, Z. C. Miao, X. L. Zhang and S. L. Chou, Small, 2018, 14, 1702883–1702897.
7. B. Tang, L. Shan, S. Liang and J. Zhou, Energy Environ. Sci., 2019, 12, 3288–3304.
8. X. Zeng, J. Hao, Z. Wang, J. Mao and Z. Guo, Energy Storage Mater., 2019, 20, 410–437.
9. J. Ming, J. Guo, C. Xia, W. Wang and H. N. Alshareef, Mater. Sci. Eng., R, 2019, 135, 58–84.
10. M. Song, H. Tan, D. Chao and H. J. Fan, Adv. Funct. Mater., 2018, 41, 1802564.
11. L. E. Blanc, D. Kundu and L. F. Nazar, Joule, 2020, 4, 771–799.
12. Y.-P. Deng, R. Liang, G. Jiang, Y. Jiang, A. Yu and Z. Chen, ACS Energy Lett., 2020, 5, 1665–1675.
13. Y. Wu, J. Fee, Z. Tobin, A. Shirazi-Amin, P. Kerns, S. Dissanayake, A. Mirich and S. L. Suib, ACS Appl. Energy Mater., 2020, 3, 1627–1633.
14. A. Dhiman and D. G. Ivey, Batteries Supercaps, 2020, 3, 293–305.
15. B. Yong, D. Ma, Y. Wang, H. Mi, C. He and P. Zhang, Adv. Energy Mater., 2020, 2002354.
16. J. Lee, J. B. Ju, W. I. Cho, B. W. Cho and S. H. Oh, Electrochim. Acta, 2013, 112, 138–143.
17. M. H. Alfaruqi, V. Mathew, J. Gim, S. Kim, J. Song, J. P. Baboo, S. H. Choi and J. Kim, Chem. Mater., 2015, 27, 3609–3620.
18. M. H. Alfaruqi, J. Gim, S. Kim, J. Song, J. Jo, S. Kim, V. Mathew and J. Kim, J. Power Sources, 2015, 288, 320–327.
19. H. Qin, Z. Yang, L. Chen, X. Chen and L. Wang, J. Mater. Chem. A, 2018, 6, 23757–23765.
20. Y. Cheng, L. Luo, L. Zhong, J. Chen, B. Li, W. Wang, S. X. Mao, C. Wang, V. L. Sprenkle, G. Li and J. Liu, ACS Appl. Mater. Interfaces, 2016, 8, 13673–13677.
21. H. Liang, Z. Cao, F. Ming, W. Zhang, D. H. Anjum, Y. Cui, L. Cavallo and H. N. Alshareef, Nano Lett., 2019, 19, 3199–3206.
22. G. Li, Z. Yang, Y. Jiang, C. Jin, W. Huang, X. Ding and Y. Huang, Nano Energy, 2016, 25, 211–217.
23. P. Hu, T. Zhua, X. Wang, X. Zhou, X. Wei, X. Yao, W. Luo, C. Shi, K. A. Owusu, L. Zhou and L. Mai, Nano Energy, 2019, 58, 492–498.
24. J. S. Ko, P. P. Paul, G. Wan, N. Seitzman, R. H. DeBlock, B. S. Dunn, M. F. Toney and J. N. Weker, Chem. Mater., 2020, 32, 3028–3035.
25. H. Pan, Y. Shao, P. Yan, Y. Cheng, K. S. Han, Z. Nie, C. Wang, J. Yang, X. Li, P. Bhattacharya, K. T. Mueller and J. Liu, Nat. Energy, 2016, 1, 16039.
26. B. Lee, H. R. Seo, H. R. Lee, C. S. Yoon, J. H. Kim, K. Y. Chung, B. W. Cho and S. H. Oh, ChemSusChem, 2016, 9, 2948–2956.
27. Y. Wu, K. Zhang, S. Chen, Y. Liu, Y. Tao, X. Zhang, Y. Ding, T. Hayat, A. M. Abusorrah and S. Dai, ACS Appl. Energy Mater., 2020, 3, 319–327.
28. Z. Yao, D. Cai, Z. Cui, Q. Wang and H. Zhan, Ceram. Int., 2020, 46, 11237–11245.
29. D. L. Chao, W. H. Zhou, C. Ye, Q. H. Zhang, Y. G. Chen, L. Gu, K. Davey and S. Z. Qiao, Angew. Chem., Int. Ed., 2019, 58, 7823–7828.
30. C. F. Bischoff, O. S. Fitz, J. Burns, M. Bauer, H. Gentischer, K. P. Birke, H.-M. Henning and D. Biro, J. Electrochem. Soc., 2020, 167, 020545–020553.
31. D. Wu, L. M. Housel, S. J. Kim, N. Sadique, C. D. Quilty, L. Wu, R. Tappero, S. L. Nicholas, S. Ehrlich, Y. Zhu, A. C. Marschilok, E. S. Takeuchi, D. C. Bock and K. J. Takeuchi, Energy Environ. Sci., 2020, 13, 4322–4333.
32. T. Zhang, Y. Tang, G. Fang, C. Zhang, H. Zhang, X. Guo, X. Cao, J. Zhou, A. Pan and S. Liang, Adv. Funct. Mater., 2020, 3, 2002711.
33. L. Li, T. K. A. Hoang, J. Zhi, M. Han, S. Li and P. Chen, ACS Appl. Mater. Interfaces, 2020, 12, 12834–12846.
34. W. Sun, F. Wang, S. Hou, C. Yang, X. Fan, Z. Ma, T. Gao, F. Han, R. Hu, M. Zhu and C. Wang, J. Am. Chem. Soc., 2017, 139, 9775–9778.
35. J. S. Ko, M. B. Sassin, J. F. Parker, D. R. Rolison and J. W. Long, Sustainable Energy Fuels, 2018, 2, 626–636.
36. J. S. Ko, M. D. Donakowski, M. B. Sassin, J. F. Parker, D. R. Rolison and J. W. Long, MRS Commun., 2019, 9, 99–106.
37. N. Zhang, F. Cheng, Y. Liu, Q. Zhao, K. Lei, C. Chen, X. Liu and J. Chen, J. Am. Chem. Soc., 2016, 138, 12894–12901.
38. X. Wu, Y. Xiang, Q. Peng, X. Wu, Y. Li, F. Tang, R. Song, Z. Liu, Z. He and X. Wu, J. Mater. Chem. A, 2017, 5, 17990–17997.
39. L. Chen, Z. Yang, H. Qin, X. Zeng and J. Meng, J. Power Sources, 2019, 425, 162–169.
40. J.-W. Lee, S.-D. Seo and D.-W. Kim, J. Alloys Compd., 2019, 800, 478–482.
41. L. Chen, Z. Yang, H. Qin, X. Zeng, J. Meng and H. Chen, Electrochim. Acta, 2019, 317, 155–163.
42. H. Zhang, J. Wan, Q. Liu, W. He, Z. Lai, X. Zhang, M. Yu, Y. Tong and X. Lu, Energy Storage Mater., 2019, 21, 154–161.
43. M. B. Sassin, S. G. Greenbaum, P. E. Stallworth, A. N. Mansour, B. P. Hahn, K. A. Pettigrew, D. R. Rolison and J. W. Long, J. Mater. Chem. A, 2013, 1, 2431–2440.
44. M. D. Donakowski, J. M. Wallace, M. B. Sassin, K. W. Chapman, J. F. Parker, J. W. Long and D. R. Rolison, CrystEngComm, 2016, 18, 6035–6048.
45. J. S. Ko, M. B. Sassin, D. R. Rolison and J. W. Long, Electrochim. Acta, 2018, 275, 225–235.
46. A. E. Fischer, K. A. Pettigrew, D. R. Rolison, R. M. Stroud and J. W. Long, Nano Lett., 2007, 7, 281–286.
47. M. B. Sassin, C. N. Chervin, D. R. Rolison and J. W. Long, Acc. Chem. Res., 2013, 46, 1062–1074.
48. J. C. Knight, S. Therese and A. Manthiram, J. Mater. Chem. A, 2015, 3, 21077–21082.
49. S. Yang, M. Zhang, X. Wu, X. Wu, F. Zeng, Y. Li, S. Duan, D. Fan, Y. Yang and X. Wu, J. Electroanal. Chem., 2019, 832, 69–74.
50. V. Soundharrajan, B. Sambandam, S. Kim, S. Islam, J. Jo, S. Kim, V. Mathew, Y.-K. Sun and J. Kim, Energy Storage Mater., 2020, 28, 407–417.
51. Z. Li, S. Ganapathy, Y. Xu, Z. Zhou, M. Sarilar and M. Wagemaker, Adv. Energy Mater., 2019, 9, 1900237–1900246.
52. Q. Pang, H. Zhao, R. Lian, Q. Fu, Y. Wei, A. Sarapulova, J. Sun, C. Wang, G. Chen and H. Ehrenberg, J. Mater. Chem. A, 2020, 8, 9567–9578.
53. D. Kundu, S. H. Vajargah, L. Wan, B. Adams, D. Prendergast and L. F. Nazar, Energy Environ. Sci., 2018, 11, 881–892.
54. P. Oberholzer, E. Tervoort, A. Bouzid, A. Pasquarello and D. Kundu, ACS Appl. Mater. Interfaces, 2019, 11, 674–682.
55. Y. Dong, M. Jia, Y. Wang, J. Xu, Y. Liu, L. Jiao and N. Zhang, ACS Appl. Energy Mater., 2020, 3, 11183–11192.
56. X. Shan, S. W. Kim, A. M. M. Abeykoon, G. Kwon, D. Olds and X. Teng, ACS Appl. Mater. Interfaces, 2020, 12, 54627–54636.
57. W. Liu, L. Dong, B. Jiang, Y. Huang, X. Wang, C. Xu, Z. Kang, J. Mou and F. Kang, Electrochim. Acta, 2019, 320, 134565.
58. L. Ma, S. Chen, H. Li, Z. Ruan, Z. Tang, Z. Liu, Z. Wang, Y. Huang, Z. Pei, J. A. Zapiena and C. Zhi, Energy Environ. Sci., 2018, 11, 2521–2530.
59. M. B. Sassin, C. P. Hoag, B. T. Willis, N. W. Kucko, D. R. Rolison and J. W. Long, Nanoscale, 2013, 5, 1649–1657.
60. M. Mateos, N. Makivic, Y.-S. Kim, B. Limoges and V. Balland, Adv. Energy Mater., 2020, 10, 2000332.
61. J. C. Lytle, J. M. Wallace, M. B. Sassin, A. J. Barrow, J. W. Long, J. L. Dysart, C. H. Renninger, M. P. Saunders, N. L. Brandell and D. R. Rolison, Energy Environ. Sci., 2011, 4, 1913–1925.

---

Footnotes

† Electronic supplementary information (ESI) available: Tables of composition and structural properties; data from additional XRD, EDS; XPS, impedance, microscopy, and voltammetric analyses. See DOI: 10.1039/d1ma00159k

‡ Present address: Applied Physics Laboratory, Baltimore, MD USA.

---