Cooperative Jahn-Teller effect and engineered long-range strain in manganese oxide/graphene superlattice for aqueous zinc-ion batteries

Abstract

The Jahn-Teller and cooperative Jahn-Teller effects are phenomena that induce asymmetry in individual ions and solid-state lattices and are commonly observed in structures containing specific transition metals, such as copper and manganese. Although the Jahn-Teller effect causes lattice distortions that stress electrode materials in rechargeable batteries, strategically utilising the strain generated by cooperative Jahn-Teller distortions can enhance structural stability. Here we introduce the cooperative Jahn-Teller effect on MnO₂ by constructing a two-dimensional superlattice structure with graphene crated in the bulk MnO₂/graphene composite material. The strong interaction between MnO₂ and graphene increases the concentration of high-spin Mn³⁺ ions, creating orderly long-range biaxial strains that are compressive in the out-of-plane direction and tensile in the in-plane direction. These strains mitigate Zn²⁺ intercalation stress and proton corrosion, enabling over 5000 cycles with 165 mAh g⁻¹ capacity retention at 5 C (1 C = 308 mA g⁻¹) in aqueous zinc-ion batteries. Our approach offers an effective strategy to significantly enhance the lifetime of rechargeable batteries by introducing the cooperative Jahn-Teller effect that overcomes the stress of ion insertion in electrode materials.

Introduction

The structural integrity of electrode materials plays a critical role in determining the cycle life and overall performance of rechargeable batteries, such as lithium-ion batteries, sodium-ion batteries, and aqueous zinc-ion batteries^1,2,3,4. A significant challenge lies in the mechanical degradation of electrode materials caused by volumetric changes and stress resulting from the insertion and extraction of ions during the charge/discharge cycling process^5,6,7. Strain engineering has been employed to solve this conundrum⁸: this includes inducing lattice mismatch using heterostructures^9,10,11, introducing low-strain/zero-strain pillar structures¹², or applying specific substrates with different properties, notably thermal expansion coefficients^13,14. While these methods focus on surface or interface area, there is an opportunity to improve effectiveness by mitigating diffusion-induced lattice stress within localised regions. Introducing long-range strain offers a more effective approach to stabilize crystal structures by uniformly distributing strain to prevent the concentration of stress that typically leads to cracking and deformation. Current strategies for achieving long-range strain engineering primarily focus on mechanical transfer techniques or epitaxial growth utilizing substrates⁸. While these methods have shown promise in enhancing the performance of various functional materials, such as superconductivity, electrical, magnetic, and optical properties¹⁵, engineering of the long-range strain within the bulk materials, that are critical for electrode materials, remains a significant challenge.

The cooperative Jahn-Teller effect (CJTE), commonly observed in Mn-based oxides like typical perovskite-type LaMnO₃, could be utilized to generate a long-range strain field within crystals^16,17,18,19. The Birnessite δ-MnO₂ phase (illustrated in Fig. 1a) is considered a promising cathode material for aqueous zinc-ion batteries but suffers from structural degradation during cycling²⁰. The ideal crystal structure of layered MnO₂ comprises edge-sharing [MnO₆] octahedra with exclusively Mn^IV centres, which should theoretically prevent the Jahn-Teller effect (JTE) and maintain strain-free flat MnO₂ planes. However, the intercalation of ions, such as Li⁺, Na⁺, or Zn²⁺ can result in an increase in high-spin d⁴(t_2g³e_g¹) Mn³⁺ ions, as observed in a typical Na-exchanged Birnessite with the formula Na_0.31(Mn^IV_0.69Mn^III_0.31)O₂·0.4H₂O²¹. The high-spin Mn³⁺ ions prompt the JTE due to the occupation of one electron in the doubly degenerated e_g orbital, resulting in the elongation of two longitudinal Mn−O bonds in the [Mn^IIIO₆] octahedra (as depicted in Fig. 1b)^22,23. The occurrence of CJTE not only distorts individual [Mn^IIIO₆] octahedra but also affects the orbital ordering of the neighboring [Mn^IVO₆] octahedra through the shared edges, inducing strain in the surrounding matrix and lowering the overall energy of the system¹⁹, thus CJTE has been applied to overcome the charge disproportionation of MnO₂ for aqueous zinc-ion batteries²⁴. However, in partially CJTE, the distribution of JTE-active centres is localized, limiting its ability to effectively propagate long-range structural coherence. If the Mn^III:Mn^IV ratio reaches 1:1, the MnO₂ lattice could adopt an alternating arrangement of [Mn^IIIO₆] (JTE-active) and [Mn^IVO₆] (CJTE-affected) octahedra, defined as fully CJTE (f-CJTE). This structural alignment ensures a more continuous and extended interaction between JTE-active and non-active centres, reinforcing long-range orbital and lattice distortions across the MnO₂ plane, as illustrated in Fig. 1a. The introduction of f-CJTE is crucial for maximizing the effectiveness of CJTE, as it eliminates isolated, short-range distortions and instead creates a globally interconnected strain network. This long-range ordered distortion acts as a framework to link separate local strain areas, culminating in a continuous biaxial strain field.

Fig. 1: Theoretical prediction of CJTE-induced long-range biaxial strains.

a Schematic illustration depicting the distinct structural scenarios in layered MnO₂, including no Jahn-Teller effect (JTE), partially cooperative Jahn-Teller effect (CJTE), and fully CJTE. The corresponding strain distributions are also highlighted: an unstressed lattice for MnO₂ without JTE, localized strains regions for partially CJTE, and a continuous long-range biaxial strain field for fully CJTE. No JTE-affected Mn^IV atoms, JTE centre Mn^III atoms, CJTE-affected Mn^IV atoms, and O atoms are marked in purple, gold, green, and red, respectively. b Schematic of formed JTE in [MnO₆] octahedra (where Mn and O atoms are marked in purple and red, respectively), where a, b and c are crystal axes and the distortion vector ω^r (r represents x, y, or z axis in [MnO₆] octahedra) can be projected to the c axis and ab plane as ω^r_c and ω^r_ab. c Statistics of log(|ω^r_c|/|ω^r_ab|) values for each Mn sites in simulated bilayer MnO₂ and MnO₂/graphene superlattice.

Herein, we successfully introduce long-range biaxial strain via f-CJTE by constructing two-dimensional MnO₂/graphene superlattices using monolayer-stacked δ-MnO₂ and graphene nanosheets. Both theoretical calculations and experimental investigations confirm the existence of f-CJTE of MnO₂ in the superlattices. The derived long-range biaxial strains, consisting of compressive out-of-plane and tensile in-plane strains on the MnO₂ layers, were further studied by transmission electron microscopy (TEM) and geometric phase analysis (GPA) technique in nanoscale and Raman spectroscopy in macroscale. This superlattice structure exhibits good cycling stability over 5000 cycles (lasting more than 1200 h) in aqueous zinc-ion batteries. This outstanding performance is attributed to the biaxial strains counteracting the stresses induced by ion intercalation, as evidenced by in situ static ultraviolet-visible (UV-vis) spectroscopy and in situ synchrotron radiation X-ray powder diffraction (SR-XRPD) characterizations.

Results

Existence of CJTE in MnO₂/graphene superlattice

Density functional theory (DFT) calculations were first employed to predict the existence of the CTJE in the MnO₂/graphene superlattice. The bilayer MnO₂ derived from Na-exchanged Birnessite, and a MnO₂/graphene superlattice unit, consisting of MnO₂ and graphene monolayers, were adopted for the calculations (Supplementary Fig. 1 and Supplementary Data 1). Compared with bare MnO₂, the Mn 3d states in the MnO₂/graphene superlattice exhibit a significant downward shift in energy, as shown in the projected density of states (PDOS) (Supplementary Fig. 2). This shift could be caused by charge redistribution and electronic hybridization between MnO₂ and graphene, leading to a partial reduction of Mn species. The introduction of Mn³⁺ induces additional electronic states near the conduction band, leading to a reduced band gap in the MnO₂/graphene superlattice. This suggests that introducing graphene causes a stronger JTE due to the lower energy from completely reduced degeneracy²⁵. This JTE distortion is further verified by the calculated shape variations within [MnO₆] octahedra (Supplementary Fig. 3). As shown in Fig. 1b, [MnO₆] octahedra in Birnessite δ-MnO₂ normally experiences Q₃ Jahn-Teller distortion with a predominant elongation of d_z²-orbital ordering¹⁷. Every orbital axis (aligned with Mn−O bonds) has the chance for stretching, accompanied by the shrinkage of a corresponding perpendicular plane, where the increment is indicated as the distortion vector ω^r. To elucidate the relationship with strain within the crystalline configuration, the distortion vector should be projected on the crystal axis as ω_c^r and ω_ab^r, considering the 2D structure of δ-MnO₂, as shown in an example of [Mn(34)O₆] octahedra (Supplementary Fig. 3). The type of strain can be identified from the comparison between magnitudes |ω_c^r| and |ω_ab^r|. When |ω_c^r|>|ω_ab^r|, it suggests a local tensile strain on [MnO₆] octahedra along the c axis, while conversely, the ab plane experiences tensile strain when |ω_c^r|<|ω_ab^r|. The calculation results are shown in Fig. 1c and Supplementary Tables 1 & 2. A long-range biaxial strain may exist in MnO₂/graphene superlattice for the consistency of [MnO₆] orbital ordering, while only localized strains occur in bilayer MnO₂ since nearly one of third [MnO₆] octahedra possess strains with the opposite direction.

The MnO₂/graphene superlattice materials were prepared by a flocculation approach via a typical electrostatic self-assembly^26,27,28 between negative-charged monolayer-stacked δ-MnO₂ (Supplementary Fig. 4 and Supplementary Note 1) and positive-charged graphene (p-Gr) nanosheets, obtained by modifying reduced graphene oxide with cationic polymer poly(diallyldimethylammonium chloride (PDDA) (Supplementary Fig. 5 and Supplementary Note 2). The thicknesses of monolayer-stacked δ-MnO₂ and p-Gr nanosheets were determined to be 0.95 nm and 1.61 nm, respectively, by atomic force microscopy (AFM in Supplementary Fig. 4 & 5). The colloidal solutions of these two species were mixed with a mass ratio of 2.8:1, determined by the theoretical crystal lattice matching²⁶. The actual weight percentage of MnO₂ was calculated to be 74.7% from thermogravimetric analysis (TGA) results (Supplementary Fig. 6), close to the theoretical value of 73.7%. The flocculation-like sediment (Supplementary Fig. 7 and Supplementary Note 3) was separated from the clear supernatant via centrifugation. X-ray diffraction (XRD) and TEM characterizations indicated the successful synthesis of MnO₂/graphene superlattices. The powder XRD in Supplementary Fig. 8 shows that the obtained MnO₂/graphene superlattice has a (002) peak located at 6.88° with a d spacing of 1.28 nm, corresponding to a secondary order of reflection for infinite (00 l) planes. Accordingly, the thickness for one repeat unit in the MnO₂/graphene superlattice structure, for example the d spacing for (001) facet, is estimated to be 2.57 nm, which is consistent with the combined thicknesses of monolayer-stacked δ-MnO₂ and p-Gr nanosheets. The cross-section TEM images and the associated line profiles of bare MnO₂ nanosheets (Fig. 2a) and MnO₂/graphene superlattice (Fig. 2b) are consistent with the XRD result. The strong and weak peaks in the intensity profile of MnO₂/graphene superlattice correspond to δ-MnO₂ and p-Gr nanosheets, respectively.

Fig. 2: Experimental observations of CJTE in superlattice.

Cross-section TEM images (scale bar 5 nm) and line profiles (scanning at the arrow areas) with a corresponding schematic illustration of a restacked MnO₂ nanosheets and b MnO₂/graphene superlattice (where p-Gr is positively charged graphene). The purple, red, and brown spheres represent Mn, O, and C atoms, respectively. c Magnetization (M) versus temperature (T) curves with dM/dT versus T curves (inset) and d 1/χ (where χ is susceptibility) versus T curves with Curie-Weiss fitting of bare MnO₂ and MnO₂/graphene superlattice in an applied magnetic field of 10 Oe.

X-ray photoelectron spectroscopy (XPS) characterizations (Supplementary Fig. 9) were conducted to quantify the valence states of Mn in bare MnO₂ and MnO₂/graphene superlattice. In the high-resolution Mn 2p spectra, the Mn 2p_1/2 and 2p_3/2 peaks of bare MnO₂ are located at 654.1 and 642.4 eV, respectively. The corresponding peaks in the MnO₂/graphene superlattice are blue-shifted by 0.2 eV, suggesting a slight decrease in overall Mn valence. This phenomenon is also found in the Mn 3s spectra, with a higher magnitude blue-shift of 0.4 eV. Analysis of simulated peaks shows an increased content of Mn^III species in MnO₂/graphene superlattice, resulting in a lower Mn valence state of +3.52 compared to +3.84 for bare MnO₂.

Zero field cooling (ZFC) and field cooling (FC) magnetization curves in Fig. 2c were employed to measure the magnetic properties of MnO₂/graphene superlattice and bare MnO₂. There is an anomaly broad peak at 33.4 K before T_N temperature (the transition temperature of “spin glass status” to antiferromagnetic/ferromagnetic interaction) in the ZFC/FC curve, which is more apparent in dM/dT curves. According to previous reports, a specific spin ordering in superlattice structure may exist, related to the magnetoelastic coupling^29,30,31. The magnetoelastic effect, namely the inverse magnetostrictive effect, is the change in the magnetic susceptibility of a material when subjected to mechanical stress. The anomaly in ZFC and FC magnetization curves confirms the existence of CJTE in the MnO₂/graphene superlattice³⁰.

The Curie-Weiss fitting in Fig. 2d was applied to measure the average oxidation state of mixed-valent Mn in both samples. The 1/χ versus T curve was fitted via Curie-Weiss law as shown in the following Eq. 1:

$$\chi=\frac{C}{T-\theta }+{\chi }_{0}$$

(1)

where θ is the Curie-Weiss temperature, χ₀ corresponds to the temperature-independent factors to the susceptibility, and C is the Curie-Weiss constant that is associated with the effective magnetic moment in Eq. 2:

$$C=\frac{n{{\mu }_{{{{\rm{eff}}}}}}^{2}{{\mu }_{{{{\rm{B}}}}}}^{2}}{3{k}_{{{{\rm{B}}}}}}$$

(2)

where n is the number of magnetic atoms per cell unit, k_B is the Boltzmann constant, and μ_eff and μ_B are the effective magnetic moment and Bohr magneton, respectively. The Curie constant C can be calculated from fitted slope 1/C in Fig. 2d with values of 1.765 and 1.400 emu·K·Oe⁻¹·mol⁻¹ for bare MnO₂ and MnO₂/graphene, respectively. Thus, the effective magnetic moments μ_eff of bare MnO₂ and MnO₂/graphene are determined as 3.76μ_B and 3.35μ_B, respectively. Both samples have mixed-valent Mn, including Mn⁴⁺ (3.87μ_B) and low-spin Mn³⁺ (2.83μ_B), compared with standard effective moments. The presence of low-spin Mn³⁺ rather than high-spin Mn³⁺ (4.90μ_B) suggests the existence of JTE under octahedral fields^32,33,34. The proportions of Mn⁴⁺ and Mn³⁺ in two samples can be estimated based on their contributions. The ratio of Mn³⁺/Mn⁴⁺ in MnO₂/graphene is calculated to be 1:1, which matches the theoretical threshold of f-CJTE.

DFT-based charge density difference analysis (Supplementary Fig. 10) was conducted to investigate the electronic interaction between MnO₂ and graphene monolayers. In the MnO₂/graphene superlattice, yellow regions represent areas of electron accumulation, whereas blue regions denote electron depletion. Notably, pronounced electron accumulation appears near the MnO₂ surface at the interface, while the graphene surface exhibits a depletion zone, indicative of significant hole localization. In comparison, no significant charge redistribution is observed in a typical MnO₂ bilayer structure. This distinct interfacial behaviour is attributed to substantial charge transfer via a Faradaic process (Supplementary Fig. 11 and Supplementary Note 4), which alters the occupancy of Mn 3d orbitals.

CJTE-induced long-range biaxial strains in MnO₂/graphene superlattice

The CJTE-induced lattice distortion of MnO₂ in the superlattice structure has been directly observed from the aberration-corrected high-angle annular dark field scanning TEM (HAADF-STEM) and inverse fast Fourier transformed (iFFT) image (marked in yellow square) in Fig. 3a, where the bright spots indicate the positions of Mn atoms in the lattices. The arrangement of Mn atoms exhibits deviation with periodic lattice twisting, which is indicative of a standard CJTE¹⁸. Figure 3b shows the corresponding FFT pattern, revealing asymmetric hexagonal diffraction spots that match a single crystal MnO₂. The angle between (100) and (−110) facets is measured to be 113°, which is much lower than 120° in a typical hexagonal Birnessite phase³⁵. The deformations of Mn-centred [MnO₆] octahedra confirm the existence of severe lattice distortion for MnO₂ in the superlattice structure.

Fig. 3: Identification of long-range biaxial strain in MnO₂/graphene superlattice via TEM.

a HAADF-STEM image (scale bar, 1 nm) recorded along the [002] direction with iFFT image (in the yellow square) showing the existence of CJTE in MnO₂/graphene superlattice along with b its corresponding FFT pattern (scale bar, 5 nm⁻¹). The calculated c ε_xx, d ε_yy, and e ε_xy strain components via GPA. f Distribution and g statistics analysis of ratios of orthometric strain ε_c versus ε_ab.

The local strain field relaxation distribution in the MnO₂/graphene superlattice was characterized by the GPA technique³⁶ using the HAADF-STEM image, with the x- and y-axes aligned parallel to the [100] and [−110] zone axes, respectively. The simulated strain field maps, presented in Fig. 3c-e, illustrate the ε_xx, ε_yy, and ε_xy strain components, where ε_xy reflects shear strain within the xy-plane. The overall strain values range from −1.5 to 1.5%, with negative values indicating compressive strains and positive values indicating tensile strains. The strain distribution appears relatively uniform, with dominant tensile strain along the [−110] directions, while both compressive and tensile strains coexist along the [100] direction. The combination of ε_xx and ε_yy represents in-plane strain distribution within the hexagonal lattice. According to Poisson’s ratio (υ) for layered structures³⁷, defined as:

$$\upsilon=-\frac{{{{{\boldsymbol{\varepsilon }}}}}_{{\mbox{in}}-{\mbox{plane}}}}{{{{{\boldsymbol{\varepsilon }}}}}_{{\mbox{out}}-{\mbox{of}}-{\mbox{plane}}}}=-\frac{{{{{\boldsymbol{\varepsilon }}}}}_{{ab}}}{{{{{\boldsymbol{\varepsilon }}}}}_{c}}=-\frac{{{{{\boldsymbol{\varepsilon }}}}}_{{xx}}+{{{{\boldsymbol{\varepsilon }}}}}_{{yy}}}{{{{{\boldsymbol{\varepsilon }}}}}_{{zz}}}$$

(3)

where ε_in-plane and ε_out-of-plane represent strains within the layered [MnO₆] plane and perpendicular to the plane, which can be defined in terms of crystallographic axes as ε_ab and ε_c, respectively. The ε_ab component is given by the sum of ε_xx and ε_yy, while ε_c corresponds to ε_zz. Therefore, ε_zz was calculated based on a Poisson’s ratio of 0.14, as reported for δ-MnO₂³⁸. The mapping results of ε_zz, shown as a contour plot in Supplementary Fig. 12, indicate a strain field significantly dominated by compressive strains.

Similar to the orthometric vector components of ω^r_ab and ω^r_c within [MnO₆] octahedra in the aforementioned theoretical calculations, the comparison of |ε_c| and |ε_ab| values was conducted and presented as mapping results in Fig. 3f and statistics distribution in Fig. 3g. In the bar chart, red and blue indicate locations where |ε_c|>|ε_ab| and |ε_c|<|ε_ab|, respectively. The statistical data demonstrate a Gaussian distribution of |ε_c|/|ε_ab| ratios, suggesting a random and even arrangement within MnO₂ nanosheets. The map is dominated (97.0%) by blue regions, confirming that |ε_c|<|ε_ab|. As CJTE is widely accepted to be the origin of the strain, especially in materials containing Mn^16,39,40, the overall TEM and GPA results confirmed the hypothesis that the f-CJTE generates long-range tensile strains within the ab plane and the perpendicular compressive strains along the c direction in the crystal lattice.

Raman spectroscopy was employed to analyze the Mn−O bonding in bare MnO₂ and MnO₂/graphene superlattice. Two dominant Mn−O stretching modes are identified in [MnO₆] octahedra. These include: the in-plane stretching vibration (ν_par), associated with the A_g symmetric stretching vibration of [MnO₆] octahedra, which occurs parallel to the linked [MnO₆] sheets and is originated from the shear motions of O atoms within the layers, and the out-of-plane mode (ν_perp) perpendicular to the sheets and corresponds to expansion and compression motions of [MnO₆] sheets along the layer stacking direction (Fig. 4a, b)^41,42. Previous investigations revealed that both ν_par and ν_perp stretching vibrations are very sensitive to the intercalated species among interlayers^42,43. In Birnessite-type MnO₂, the larger cation insertion or substitution can weaken interlayer interaction, resulting in a redshift of ν_perp stretching vibrations⁴⁴. In the case of MnO₂/graphene superlattice, as shown in Fig. 4c, the remarkable blue-shift of the ν_perp band from 631 to 653 nm⁻¹ is attributed to the intercalation of p-Gr nanosheets. Meanwhile, the ν_par counterpart is redshifted.

Fig. 4: Evidence of long-range biaxial strain via Raman spectroscopy.

The schematic illustrating ν_par and ν_perp stretching modes (corresponding to the in-plane stretching vibration parallel to [MnO₆] sheets and out-of-plane stretching vibration perpendicular to the sheets, respectively) of Mn−O bonds in Raman spectra for a bilayer MnO₂ and b MnO₂/graphene superlattice structures. The purple, red, and brown spheres represent Mn, O, and C atoms, respectively. c Raman spectra of bare MnO₂ and MnO₂/graphene superlattice with statistics distribution of ν_par/ν_perp ratios of bare MnO₂ and MnO₂/graphene superlattice in d contour mapping and e bar chart.

There are significant intensity changes of ν_par and ν_perp stretching vibrations when comparing Raman spectra between bare MnO₂ and MnO₂/graphene superlattice. To validate the intensity variation of these two stretching modes, the intensity ratios, ν_par/ν_perp, were statistically calculated based on areal-scanned Raman spectra for both bare MnO₂ and MnO₂/graphene superlattice, presented in contour graphs (Fig. 4d) and a statistical bar chart (Fig. 4e). Bare MnO₂ has significantly higher overall ν_par/ν_perp ratios than MnO₂/graphene superlattice, which is attributed to the greater Mn^IV occupying rate⁴² shown by the XPS (Supplementary Fig. 9) and the Curie-Weiss fitting (Fig. 2d) results. Another contributing factor could be related to the superlattice architecture, in which the sizeable lateral-sized graphene nanosheets cap every monolayer-stacked MnO₂ nanosheet via a strong interaction, including electrostatic force and possible hydrogen bonding to surface hydroxyl groups of MnO₂⁴³. Furthermore, the relatively lower ν_par/ν_perp ratios are irregularly distributed on aggregated MnO₂ nanosheets, while the distribution of ν_par/ν_perp ratios on the MnO₂/graphene superlattice is in a narrow range, as observed from the contour graph and statistical bar chart that statistically obeys a Gaussian distribution. Moreover, based on Badger’s rule^45,46,47 (Supplementary Note 5), which describes the relationship between force constant and bond length, the Raman spectra analyses elucidate that the overwhelmingly increased out-of-plane and decreased in-plane stretching vibrations in MnO₂ after coupling with graphene are ascribed to the overall elongated and compressed Mn−O bonds along the same orientations. These experimental results also align with the category of |ω^r_c|<|ω^r_ab| observed in the theoretically calculated results, providing strong evidence for the existence of long-range biaxial strain in the superlattice.

Electrochemical performance of MnO₂/graphene superlattice as cathodes in aqueous zinc-ion batteries

The structural stability of bare MnO₂ and MnO₂/graphene superlattices in aqueous electrolyte (2 M ZnSO₄ with 0.1 M MnSO₄ additive) was evaluated using in situ UV-vis characterizations, with schematic illustrations of the configuration and setup provided in Supplementary Fig. 13. The measurements were conducted both in static conditions (Supplementary Fig. 14) and under electrochemical control (Supplementary Fig. 15). To assess static structural stability (Supplementary Fig. 14 and Supplementary Note 6), both electrodes were immersed in the aqueous electrolyte following the specific ratio described in Methods section. The Mn dissolution status was reflected in the UV-vis spectra by the variation of peak position, corresponding to the emergence of soluble Mn²⁺ ions, δ-MnO₂ or p-Gr fragments. Notably, the MnO₂/graphene electrode remained stable over 20 h with slight dissolution of Mn into Mn²⁺ ions, while bare MnO₂ experienced a severe erosion in aqueous electrolyte, resulting in structural collapse evidenced by the appearance of δ-MnO₂ fragments. Coupled with Zn metal anodes, home-made two-electrode cells were employed for in situ UV-vis tests under electrochemical control at a specific current of 1 C (308 mA g⁻¹) for both electrodes, as shown in Supplementary Fig. 15. While Mn dissolution is inherently limited by the oxidation of Mn²⁺ ions back to MnO₂ at certain voltages during the charge process⁴⁸, the overall spectral shifts in the UV-vis spectra for the MnO₂/graphene superlattice remain significantly smaller than those of bare MnO₂ over 20 cycles, indicating its enhanced ability to suppress Mn dissolution.

The electrochemical performance of bare MnO₂ and MnO₂/graphene superlattice cathodes was evaluated in full aqueous zinc-ion batteries using zinc foil as the anode. Supplementary Fig. 16 presents cycling behaviour at a current rate of 1 C. The MnO₂/graphene superlattice electrode delivers an initial specific capacity of 270 mAh g⁻¹ with an initial Coulombic efficiency of 93.4%. After a short activation stage, the capacity stabilizes with Coulombic efficiencies approaching 100% and retains 91.2% of its capacity after 700 cycles. In contrast, the bare MnO₂ electrode exhibits lower initial capacity, a prolonged activation phase, and a rapid capacity decay, retaining only ~50 mAh g⁻¹ after 500 cycles. These differences are further reflected in the galvanostatic discharge curves. The MnO₂/graphene superlattice electrode shows highly consistent discharge profiles throughout cycling (Fig. 5a), with a notable inflection point at ~1.25 V, separating two distinct discharge plateaus. This transition marks a shift in the dominant charge storage mechanism: proton insertion above 1.25 V and Zn²⁺ insertion below it⁴⁹. For the bare MnO₂ electrode (Supplementary Fig. 17), the separation between these processes becomes increasingly blurred with cycling, as the faster proton insertion reaction begins to dominate⁵⁰. The decline in Zn²⁺-insertion contribution and overall capacity is attributed to progressive MnO₂ inactivation. In situ UV-vis spectroscopy further confirms that Mn²⁺ dissolution and fragmentation of the MnO₂ structure are the main causes of its limited cycling stability.

Fig. 5: Suppressed Mn dissolution of MnO₂/graphene superlattice in aqueous zinc-ion batteries.

a Galvanostatic discharge profile of MnO₂/graphene superlattice electrodes within various cycles. b Long-term cycling life after a short activation (Supplementary Fig. 18) at 1 C (308 mA g⁻¹) for 10 cycles of bare MnO₂ and MnO₂/graphene superlattice electrodes tested at a specific current of 5 C. c In situ SR-XRPD patterns of MnO₂/graphene superlattice within a full cycle at a specific current of 1 C under 25°C±5°C in the ambient condition. Schematic illustration of d the structural degradation of bare MnO₂ electrodes within two steps and e good structural stability of MnO₂/graphene superlattice electrodes.

When tested at a high rate of 5 C (Fig. 5b) after an activation at 1 C for ten cycles (Supplementary Fig. 18), the specific capacity of bare MnO₂ electrodes quickly dropped to 100 mAh g⁻¹ within 400 cycles, and only 60 mAh g⁻¹ was maintained after 5000 cycles. Conversely, the MnO₂/graphene superlattice electrodes presented a similar initial specific capacity as bare MnO₂ electrode with a value of 175 mAh g⁻¹, but much superior long-term cycling stability over 5000 cycles with a high reversible capacity of 165 mAh g⁻¹. It should be noted that the operating time of full cells using MnO₂/graphene superlattice electrodes is over 1200 h here. Comparative data for previously reported manganese oxide cathodes are provided in Supplementary Table 3. The long lifespan of the superlattice electrode firmly proves the good suppression effect on Mn dissolution and stabilization of MnO₂ in aqueous zinc-ion batteries.

In situ SR-XRPD was employed to investigate the Zn storage mechanism of the MnO₂/graphene superlattice (Fig. 5c and Supplementary Fig. 19). The SR-XRPD patterns can be divided into three regions in terms of H⁺ insertion, Zn²⁺ insertion, and H⁺/Zn²⁺ extraction. The conventional XRD pattern (Supplementary Fig. 8) indicates the (002) peak of MnO₂/graphene superlattice is located at 6.88°, while the synchrotron powder diffraction (PD) patterns reflect the (003) peak of superlattices at 10.34°, associated with the initial d spacing of 2.57 nm as well. During the galvanostatic discharge/charge processes, the interspacing of the superlattice almost remained within the H⁺ insertion stage due to the ultrasmall proton radius, but the intensities of H⁺-intercalated (003) peaks increased. The ionic radius of Zn²⁺ ions (~0.75 Å)⁵¹ is much larger and thus causes a dramatic peak shift to 10.13°, accompanied by continuous Zn²⁺ ion insertion. Interestingly, the (003) peak vanishes and is then replaced by the (006) peak at 20.82° when the concentration of Zn²⁺ in the superlattice is high enough. Another finding is the emergence and disappearance of the Groutite-phase MnOOH phase (JCPDS no. 12-0733) at ~43.24°, which is convincing evidence of the intercalation and deintercalation of H⁺ ions and closely dependent on the contents of H⁺ ions in the MnO₂ lattice⁴⁹. The disappearance of the MnOOH phase is accompanied by the emergence of a small peak at 10.21°, which may result from the migration of Mn³⁺ ions from the MnO₂ plane into the interlayer space, creating metal vacancies^52,53. The presence of this peak is likely influenced by the relatively small ionic radius of high-spin Mn³⁺ (0.64 Å)⁵⁴. The overall peak alterations are reversible in the extraction procedure. However, no apparent peak variations appear in the SR-XRPD patterns of the bare MnO₂ electrode (Supplementary Fig. 20) during the H⁺/Zn²⁺ insertion/extraction. There is an inconspicuous reversible peak shift from 43.16° to 43.19° in the bare MnO₂ patterns, suggesting the slight shrinkage of (220) facets and the layered [MnO₆] plane in Zn_xMn₂O₄ phase (for example, ZnMn₂O₄ JCPDS no. 12-0733). The strain distribution of Zn²⁺-inserted MnO₂/graphene superlattice was analyzed in Supplementary Fig. 21 and Supplementary Note 7. After full discharge, although MnO₂ converts into Zn_xMn₂O₄, the superlattice retains its 2D structure and biaxial strains, with tensile strain in-plane and compressive strain out-of-plane, without localized strain accumulation. The overall strain is reduced to the range of −0.2% to 0.2%, indicating that the pre-introduced biaxial strain effectively mitigates Zn²⁺ insertion-induced stress.

It is evident that the long-range biaxial lattice strains induced by the f-CJTE play a crucial role in mitigating the dissolution of Mn and maintaining the structural integrity of the electrode material during the H⁺/Zn²⁺ insertion/extraction. The deterioration of the pristine MnO₂ structure is illustrated in Fig. 5d. In this system, protons infiltrate the MnO₂ layers through their edges, concurrently subjecting the [MnO₆] cells on the surface to direct attack (Step I), as reported in previous studies⁵⁵. Consequently, the resultant MnOOH species are rapidly corroded when exposed to the corrosive proton-rich environment. Simultaneously, the migration of Mn³⁺ from the MnO₂ plane to the interlayer space, and eventually to the surface and edges, further destabilizes the local structure. This corrosion process ultimately leads to the irreversible transformation of MnO₂ layers into soluble Mn²⁺ ions and MnO₂ fragments (Step II). In stark contrast, the MnO₂ layers within the MnO₂/graphene superlattice are effectively shielded, thanks to the pre-existing lattice strain, as illustrated in Fig. 5e. When confronted with attempts at H⁺ insertion, the out-of-plane compressive strain primarily regulates the intercalation of H⁺ ions at the edge sites. Consequently, the MnOOH phases gradually develop toward the central regions, inducing a lattice mismatch at the interface between MnO₂ and MnOOH. The presence of graphene and the associated compressive strain help confine Mn³⁺ migration locally, preventing its movement toward the surface and edges, where further corrosion occurs. Simultaneously, the formation of layered Zn_xMn₂O₄ may also introduce lattice distortion due to the increased concentration of Jahn-Teller Mn³⁺ ions within the Zn²⁺ insertion⁵⁶. In both scenarios, stress propagates from the edges inward, but the pre-existing in-plane tensile strain effectively counteracts these stresses, thereby stabilizing the superlattice structure throughout reversible cycles.

We further investigated the kinetics of bare MnO₂ and MnO₂/graphene superlattice electrodes, as shown in Supplementary Fig. 22 and Supplementary Table 4, and discussed in Supplementary Note 8. The MnO₂/graphene superlattice electrodes deliver a better rate performance than the bare MnO₂ electrodes at various specific currents. The MnO₂/graphene superlattice electrodes deliver specific capacities of about 180 and 130 mAh g⁻¹ at high rates of 5 and 10 C, respectively. In contrast, the bare MnO₂ electrodes only achieve 100 and 50 mAh g⁻¹ at the same specific currents. The superior rate performance of MnO₂/graphene superlattice electrodes is attributed to the high electronic conductivity of the graphene matrix and, more importantly, their enhanced ion transport kinetics, supported by a series of electrochemical characterization techniques, including galvanostatic intermittent titration technique, cyclic voltammetry (CV), and electrochemical impedance spectra. The superior kinetics stem from the much lower Zn²⁺ ion diffusion barrier in the MnO₂/graphene superlattice, as shown by DFT calculations (Supplementary Fig. 23 and Supplementary Note 9).

Discussion

In this study, the introduction of long-range biaxial strain, consisting of in-plane tensile and out-of-plane compressive stains, in MnO₂ layers was achieved by constructing a superlattice structure via electrostatic self-assembly of monolayer-stacked δ-MnO₂ and p-Gr nanosheets. Within the superlattice structure, the strong charge transfer between MnO₂ and graphene induces a significant valence decrease of Mn ions and, consequently, a fully cooperative Jahn-Teller effect on MnO₂ layers. Both nano- and macro-scale characterizations, including TEM and Raman spectra techniques, confirmed the existence of long-range biaxial strains along the entire superlattice. When applying MnO₂/graphene superlattice electrodes in aqueous zinc-ion batteries, this introduced long-range biaxial strain efficiently protects the Mn sites in [MnO₆] octahedra from H⁺ corrosion, offsets the Zn²⁺ insertion-induced stress and maintains the integrity of δ-MnO₂. Hence, the MnO₂/graphene superlattice exhibited good cycling stability over 5000 cycles at 5 C. Ionic diffusion kinetics are also enhanced. This work provides an innovative strategy to construct long-range strains via the cooperative Jahn-Teller effect and opens a promising prospect for regulating ion insertion-induced stresses in electrode materials for battery applications.

Methods

Preparation of δ-MnO₂ nanosheets

Two precursor solutions were prepared prior to synthesis. Solution A comprised 0.6 M tetramethylammonium hydroxide (TMAOH, 25 wt.% in H₂O, Sigma-Aldrich) and 3 wt.% hydrogen peroxide (H₂O₂, 30 wt.%, Sigma-Aldrich), while solution B consisted of freshly prepared 0.3 M MnCl₂ (99%, Sigma-Aldrich) in deionized water. Under constant stirring at ambient temperature, 20 mL of solution A was gradually introduced into 10 mL of solution B. The resulting dark brown suspension was stirred in air for 12 h. The precipitate was isolated by centrifugation at 10,000 rpm (10,621 × g) for 30 min and washed three times with deionized water under the same conditions to eliminate excess TMA⁺ ions. The final product was redispersed in water and centrifuged at 5000 rpm (2655 × g) for 10 min to collect a reddish-brown supernatant, representing a colloidal suspension δ-MnO₂ nanosheets.

Preparation of positively charged graphene nanosheets

Graphene oxide was converted to positively charged graphene via surface functionalization with PDDA. A 500 mL suspension of graphene oxide (0.2 mg mL⁻¹), synthesized using a modified Hummer’s method²⁶, was refluxed with 3.75 mL of PDDA solution (20 wt.% in H₂O, Sigma-Aldrich) and 100 μL hydrazine hydrate (N₂H₄·H₂O, 50–60 wt.%, Sigma-Aldrich) at 90 °C for 3 h. The product was collected by centrifugation at 10,000 rpm for 30 min to remove unbound PDDA. The residue was redispersed in water and subjected to a second centrifugation at 5000 rpm for 10 min to yield a stable suspension of positively charged graphene nanosheets.

Preparation of MnO₂/graphene superlattices

The δ-MnO₂ nanosheets and positively charged graphene suspensions were mixed at a mass ratio of ~2.8:1 (δ-MnO₂:graphene), based on a previously reported area-matching model²⁶, and stirred at room temperature. Electrostatic self-assembly occurred rapidly, forming a flocculent precipitate within seconds. The composite was recovered by centrifugation at 5000 rpm for 5 min, followed by washing with deionized water to remove residual ions and unbound components.

Materials characterizations

XRD measurements were carried out using a Bruker D8 Discover diffractometer with Cu Kα radiation (λ = 1.5406 Å). Data were collected over a 2θ range of 2–80° with a scan rate of 0.5° min⁻¹ and a step size of 0.04°. AFM via a Park XE7 was applied to characterize thicknesses and lateral sizes. SEM via a Zeiss Supra 55VP and TEM via a FEI Tecnai G2 F30 S-TWIN were carried out to elucidate morphology at a lower resolution, while FEI Titan Cubed Themis G2 300 was used to observe atomic-level STEM images. SEM-EDX data were collected with the beam voltage of 30 keV. XPS measurements were performed using a Themo Scientific ESCALAB 250Xi spectrometer equipped with a monochromatic Al Kα X-ray source (hυ = 1486.6 eV). Powder samples were mounted on conductive carbon tape and transferred to the XPS chamber through ambient air. The binding energies were calibrated using the C 1 s peak at 284.8 eV to correct for surface charging. A Renishaw inVia Raman microscope obtained Raman spectra. The Raman mapping was conducted in areas with the size of 52 × 52 μm and the collecting increment of 4 μm. An Agilent Cary 60 UV-Vis spectroscopy tested ultraviolet-visual spectra. An SDT-2960 system performed TGA in air from room temperature to 800 °C with a temperature rising rate of 5 °C min⁻¹.

Magnetic measurements

Magnetic characterization was conducted using a Quantum Design MPMS-3 system to evaluate both temperature- and field-dependent magnetization behaviours. Powder samples were loaded into gelatin capsules, which were secured in the magnetometer using a non-magnetic straw holder. For ZFC measurements, samples were first cooled to 1.8 K in the absence of an external magnetic field. Upon reaching this temperature, a magnetic field (H) was applied, and the magnetization (M) was recorded during subsequent warming to 300 K. FC measurements were obtained by cooling the sample under the same applied field from room temperature to 1.8 K, followed by magnetization recording over the same temperature range.

Cell assembly and electrochemical measurements

To prepare electrode slurries, active materials, carbon black (Super P, Timcal) and polyvinylidene fluoride (PVDF, M_w ~1,000,000, Arkema) were manually ground at a mass ratio of 7:2:1 in N-methyl-2-pyrrolidone (NMP, 99.5%, Sigma-Alrich) solvent with an approximate solvent-to-solid ratio of 2:1 by weight. The mixture was then single-side coated on hydrophobic carbon cloth current collectors (thickness of 454 µm, ELAT LT1400, FuelCellsEtc) with a blade and dried under vacuum at 100 °C overnight. The active material-coated carbon clothes were punched into 12 mm by using a precision disc cutter (MTI Corporation), and the average active material mass loading is ~2.5 mg cm⁻². The Zn||MnO₂ coin cells were assembled by Zn foil (thickness of 0.1 mm, 99.99%, SCI Materials Hub) counter electrodes, glass fibre separators (GF/D, Whatman), and working electrodes with ~75 μL 2 M ZnSO₄ (99.0%, Sigma-Aldrich) + 0.1 M MnSO₄ (99%, Sigma-Alrich) electrolyte, in which Zn foil was washed with ethanol, dried under vacuum, and then punched into disc with the diameter of 12 mm. The assembled batteries were tested at 25 ± 5 °C in the ambient condition without climatic/environmental chamber. For the electrochemical measurements, CV at various scanning rates (0.1–1.0 mV s⁻¹) in a potential window of 0.8–1.8 V vs. Zn/Zn²⁺ was performed on a Bio-Logic VMP3 workstation. All galvanostatic tests were accomplished on LAND battery testers (1 C = 308 mA g⁻¹), including galvanostatic discharge/charge profiles at 1 C, cycling life tests at 1 C and 5 C (including a ten-cycle activation at 1 C), and rate performance at various specific currents (0.2–5 C) between 0.8 V and 1.8 V. 3–6 of coin cells were tested for each single electrochemical experiment, and the presented data in the manuscript represent a statistical analysis. Electrochemical impedance spectroscopy (EIS) measurements were conducted at open circuit potential using a Bio-Logic VMP3 electrochemical workstation for cells cycled ten and 100 cycles at a specific current of 1 C. A potentiostatic signal with an AC amplitude of 5 mV was applied over a frequency range of 10 mHz to 100 kHz, with six data points collected per decade.

In situ UV-vis Tests

The schematic illustration of the in situ UV-vis setup is shown in Supplementary Fig. 13. The bare MnO₂ and MnO₂/graphene superlattice electrodes were coated onto hydrophobic carbon clothes using the same methods as mentioned above. The prepared electrodes were cut into pieces of 4 × 30 mm with an active material mass loading of ~2.5 mg cm⁻² and positioned at the side of UV-vis cuvettes, secured with Kapton tape. For static in situ UV-vis characterisation, the electrodes were immersed in the electrolyte (2 M ZnSO₄ + 0.1 M MnSO₄) with a relatively higher ratio of m(active materials):V(electrolytes) = 1 mg:0.5 mL compared to the standard conditions used in aqueous ZIBs (1 mg:0.03 mL in this work). The cuvette was then placed inside the UV-vis spectrometer chamber for measurement at 25 ± 5 °C in the ambient condition. For in situ UV-vis tests under electrochemical control, zinc metal anodes were used and fixed in place. A Neware battery tester was employed to supply a constant current and record electrochemical data. The electrolyte ratio remained the same as in the static tests. Once the UV-vis test commenced, a constant current of 1 C (1 C = 308 mA g⁻¹) was applied between 0.8 and 1.8 V to the home-made two-electrode cell for 20 cycles. The UV-vis spectra were automatically recorded within the 200–800 nm range, with a scanning speed of 300 nm min⁻¹ and a measurement interval of 10 min.

In situ SR-XRPD Characterizations

The mass loading of the electrodes for in situ SR-XRPD was ~3.0 mg cm⁻² with carbon clothes as current collectors. The working electrodes were assembled with Zn foil anodes and glass fibre separators in designed coin cells, which had windows on both cathode and anode parts and sealed by Kapton tapes. The in situ SR-XRPD characterizations were conducted at the PD Beamline in the Australian Melbourne Synchrotron. The wavelength was 0.6883 Å using the NIST LaB6 600b standard reference materials. The coin cells were placed on a homemade stage for testing with galvanostatic cycling at a rate of 0.25 C using a Neware battery tester at 25 ± 5 °C in the ambient condition. The PD data collection was scanned within 1 min twice with an interval of 8 min.

Computational methods

All the spin-polarized DFT⁵⁷ computations involved in this work were performed using plane-wave Vienna Ab initio Simulation Package^58,59. The projector augmented wave pseudopotential approach⁶⁰ was used to reflect the core and valence electrons of each element. The climbing image nudged elastic band (CI-NEB) method⁶¹ was used to calculate the ion diffusion property. The exchange-correlation interactions were described by generalized gradient approximation⁶² of the Perdew-Burke Enzerhor functional⁶³, and the plane-wave cutoff energy was set to 450 eV. Considering the long-range dispersion effect, empirical DFT-D3 scheme⁶⁴ was adopted to treat the van der Waals correction, and only the Gamma point was considered in the first Brillouin zone because of the large enough crystal lattice of heterojunction. The energy and residue force were converged until 10⁻⁴ eV and 0.02 eV Å⁻¹, respectively. Notably, in terms of electronic calculations, DFT plus Hubbard method (DFT + U)⁶⁵ was employed with 4.0 eV U value⁶⁶ added to the 3d orbital of Mn. The charge distribution was analysed by the Bader population using code developed by Henkelman et al. ⁶⁷.