Highly improved supercapacitance properties of MnFe2O4 nanoparticles by MoS2 nanosheets

Manganese ferrite (MnFe2O4) nanoparticles were synthesized via a hydrothermal method and combined with exfoliated MoS2 nanosheets, and the nanocomposite was studied as a supercapacitor. X-ray diffractometry and Raman spectroscopy confirmed the crystalline structures and structural characteristics of the nanocomposite. Transmission electron microscopy images showed the uniform size distribution of MnFe2O4 nanoparticles (~ 13 nm) on few-layer MoS2 nanosheets. UV–visible absorption photospectrometry indicated a decrease in the bandgap of MnFe2O4 by MoS2, resulting in a higher conductivity that is suitable for capacitance. Electrochemical tests showed that the incorporation of MoS2 nanosheets largely increased the specific capacitance of MnFe2O4 from 600 to 2093 F/g (with the corresponding energy density and power density of 46.51 Wh/kg and 213.64 W/kg, respectively) at 1 A/g, and led to better charge–discharge cycling stability. We also demonstrated a real-world application of the MnFe2O4/MoS2 nanocomposite in a two-cell asymmetric supercapacitor setup. A density functional theory study was also performed on the MnFe2O4/MoS2 interface to analyze how a MoS2 monolayer can enhance the electronic properties of MnFe2O4 towards a higher specific capacitance.

www.nature.com/scientificreports/ MnFe 2 O 4 nanoparticles would prevent the MoS 2 nanosheets from restacking, and in a similar manner, the MoS 2 nanosheets would act as a substrate on which the MnFe 2 O 4 nanoparticles can be uniformly distributed so that they would not be agglomerated. The mutual effect would result ultimately in a larger active surface area that can promote electrolyte access and provide more channels for migration of ions and electrons 32 .
To the best of our knowledge, there is not yet any report on supercapacitance properties of the composite of MoS 2 nanosheets (as a TMD) decorated with MnFe 2 O 4 (as a metal ferrite) nanoparticles, and it is thus interesting to see how the probable synergistic effect of them can be useful to achieve improved electrochemical energy storage performance. Here, we reported the successful fabrication of the MnFe 2 O 4 /MoS 2 nanocomposite on nickel foam via a facile hydrothermal method and tested the nanocomposite as a supercapacitor electrode.
Synthesis. Few-layer MoS 2 nanosheets were exfoliated from bulk MoS 2 powder in mixed solvents based on the work of Lu et al. 33 . First, 20 mg MoS 2 powder was mixed in 0.5 ml acetonitrile and it was ground for 1 h. The obtained powder was mixed in a solution of 30 wt% H 2 O 2 and NMP (H 2 O 2 :NMP volume ratio = 1: 19), and it was stirred for 10 h at 35 °C to be exfoliated. The mixture was dried in a furnace for 5 h at 300 °C to evaporate NMP. To in-situ synthesize the MnFe 2 O 4 /MoS 2 nanocomposite on a Ni foam substrate using a hydrothermal method, 0.1 g of the obtained MoS 2 nanosheets was first dispersed into 40 ml deionized water, and 0.4 g Fe(NO 3 ) 3 .9H 2 O, 0.125 g Mn(NO 3 ) 2 .4H 2 O, and 0.125 g CTAB were dissolved into the dispersion by stirring for 2 h. Next, 1 ml of 25% ammonia solution was added into the prepared mixture under vigorous stirring until its pH reached ~ 9. The obtained mixture was transferred into a Teflon-lined autoclave and a nickel foam substrate, cut in the size of 1 × 2 cm 2 , and cleaned with deionized water, acetone, and ethanol, was put into it. The autoclave was subsequently heated in an oven at 180 °C for 15 h, and then it was allowed to cool to room temperature in ambient air. Finally, the MnFe 2 O 4 /MoS 2 -coated nickel foam was washed several times with deionized water and ethanol and dried in an oven at 80 °C for 2 h. For comparative purposes, MnFe 2 O 4 -coated nickel foam was also synthesized via a similar method without adding MoS 2 . The prepared substrates were used for further characterizations. Figure 1 shows a schematic of various steps followed in our synthesis procedure.
Characterization. Crystalline structures of the samples were identified using a PANalytical X'pert MPD (Philips) diffractometer with a Cu-Kα radiation source (λ = 0.15406 nm). Structural fingerprints of the ferrites and the MoS 2 nanosheets were investigated by a Takram P50C0R10 Raman spectrometer (Teksan Co., Iran) employing an Nd:YAG laser (λ ex = 532 nm) at room temperature. To observe structural shapes ph of the prepared nanostructure, field-emission scanning electron microscopy (FESEM) and transmission electron microscopy (TEM) images were taken by MIRA3TESCAN-XMU and PHILIPS CM30 NETHERLANDS instruments, respectively. The elemental compositions of the samples were analyzed by energy-dispersive X-ray spectroscopy (EDS) mapping using a BRUKER XFlash 6 I10 instrument. The topographical information of MoS 2 nanosheets  www.nature.com/scientificreports/ was acquired by atomic force microscopy (AFM, Veeco Autoprobe CP-research). The optical properties of the nanocomposites were examined using a Unico 4802 UV-Vis photospectrometer.
Electrochemical tests. The supercapacitive performance of the samples was investigated using a threeelectrode setup containing the coated Ni foam substrate (1 cm 2 ) as the working electrode, a square-shaped platinum sheet (1 cm 2 , 99.99%) as the counter electrode, and Ag/AgCl as the reference electrode in a 3 M KOH solution at room temperature. Although Ni foam shows a battery-like behavior 34 , we chose it because of its large specific surface area that can accommodate more parts of active materials. In this regard, the calculated specific capacitance is better not to be compared with other literature. Nevertheless, we aim at comparing the specific capacitances of our samples with each other to find how the incorporation of MoS 2 nanosheets can enhance the specific capacitance of MnFe 2 O 4 nanoparticles. The electrochemical measurements involved cyclic voltammetry (CV), galvanostatic charging/discharging (GCD), and electrochemical impedance spectroscopy (EIS) techniques using a VSP-300 Multichannel Potentiostat/Galvanostat/EIS instrument (Bio-Logic Science Instruments). The CV measurements were recorded at different scan rates (5-100 mV/s) within the potential window of 0-0.55 V. The GCD measurements were recorded at different current densities with the potential window of 0-0.4 V.
Asymmetric two-electrode supercapacitor setup. An asymmetric two-electrode supercapacitor device was assembled by using activated carbon (AC) as the negative electrode and the MnFe 2 O 4 /MoS 2 nanocomposite as the positive electrode. The electrodes were separated by a filter paper wetted with 3 M KOH solution as the electrolyte. The AC electrode was prepared from the activated carbon and PVDF, as a binder, with the weight ratio of 95:5 dispersed in NMP. The prepared dispersion was coated on a nickel foam substrate by a brush and the obtained electrode was dried in an oven at 60 °C for 10 h. The masses of the positive and negative electrodes were balanced according to the following equation 14 : where m is the mass, C s is the specific capacitance, ∆V is the potential window, and ( +) and ( −) denote the positive and the negative electrodes, respectively. The coated mass on the negative electrode was ~ 3 mg. The CV measurements were recorded at different scan rates (5-100 mV/s) within the potential window of 0-1.5 V. The GCD measurements were recorded at different current densities with the potential window of 0-1.5 V.
Computational methods. First-principles calculations were performed in the framework of density functional theory (DFT), as implemented in the Quantum Espresso package (version 6.3) 35 , using the plane-wave basis set and ultrasoft pseudopotentials 36 . The spin polarization was included in both geometry optimizations and electronic structure calculations. The generalized gradient approximation (GGA) developed by Perdew, Burke, and Ernzerhof (PBE) 37 was applied for electron exchange-correlation functionals with the on-site Coulomb repulsion U terms 38 of U(Mn) = 3.9 eV and U(Fe) = 5.3 eV to reproduce experimental data 14 . The kinetic energy cutoffs for wavefunctions and charge densities were set to 50 and 450 Ry, respectively. To sample the first Brillouin zone for electronic structure calculations, we adopted the k-point grid of 9 × 9 × 1 for the unit cell of the MoS 2 monolayer, 6 × 6 × 5 for the bulk MnFe 2 O 4 , and the k-point grid of 6 × 6 × 1 for the MnFe 2 O 4 /MoS 2 interface. All structures were fully relaxed until the convergence criteria of energy and force became less than 10 -6 Ry and 10 -3 Ry/Bohr, respectively. All crystal images were produced by VESTA (version 3.4.5) 39 .

Results and discussions
The XRD patterns of the samples are shown in Fig where θ hkl is the diffraction peak angle, β hkl is the full-width at half maximum of the (hkl) diffraction peak, D is the crystallite size, K is the shape factor approximated to 0.9, λ is the incident X-ray wavelength (1.5406 Å), and ε is the lattice strain 41 . Accordingly, D and ε are calculated from the Y-intercept and the slope of the line fitted on the plot of 4sinθ versus βCosθ, respectively 32,41 . Therefore, the crystallite size of the MnFe 2 O 4 nanoparticles was calculated ~ 9.5 nm with a compressive strain of − 0.0144. In the XRD pattern of the MnFe 2 O 4 /MoS 2 -coated Ni foam, all of the peaks related to MnFe 2 O 4 are seen with a peak at 2θ = 14° related to few-layer MoS 2 nanosheets and two other peaks at 2θ = 43° and 53° corresponding to the Ni foam substrate. This indicates that the nanocomposite has been successfully prepared, with no other impurities. Figure 3 shows the Raman spectra of the samples. The Raman spectrum of few-layer MoS 2 nanosheets (Fig. 3A) show two peaks at 385 and 407 cm -1 attributed to the in-plane (E 2g ) and the out-of-plane (A 1g ) vibration modes, respectively, of few-layer MoS 2 32 . In the Raman spectrum of the bulk MoS 2 powder, the www.nature.com/scientificreports/  www.nature.com/scientificreports/ lower-wavenumber mode shifts slightly towards a lower wavenumber (387 cm −1 ), and the higher-wavenumber mode shifts slightly towards a higher wavenumber (411 cm −1 (1), E g , and F 2g (2) modes, respectively), which can be due to the applied strain when they are composited with MoS 2 nanosheets. On the other hand, no Raman mode is expected from the nickel foam, as all metals with one atom per unit cell, like Ni, are Raman inactive 44 .
UV-vis absorption spectroscopy is a powerful tool to investigate the optical properties of semiconductor materials 45 . UV-Vis absorption spectra of the samples with their corresponding Tauc plots are shown in Fig. 4. Optical bandgaps of the samples were estimated using the classical Tauc relation.
where α, ν, n, B, h, and E g is the absorption coefficient, the photon frequency, a constant that depends on the bandgap type (1/2 and 2 for direct and indirect band gaps), a constant, the Planck's constant, and the optical bandgap, respectively. The optical band gap is estimated from an extrapolation of the linear part of (αhν) 2 versus the photon energy (hν) for direct bandgaps. The UV-Vis spectrum of MoS 2 nanosheets shows four characteristic peaks at 684, 625, 481, and 399 nm, corresponding to four different electronic transitions denoted with A, B, C, and D, consistent with previously reported values 33 . The few-layer MoS 2 nanosheets exhibit a bandgap of 1.7 eV, consistent with the literature 33 Fig. 5A,B are sufficiently wide with a lateral size of about 2 to 3 µm, which is appropriate as a substrate to hold other nanoparticles. Some obvious foldings can be seen in these MoS 2 nanosheets, and one can conclude that they are few-layer. Figure 5C shows an AFM micrograph of an exfoliated MoS 2 nanosheets on a mica substrate. According to the height profile of the shown dashed line drawn, the thickness of the sheet is ~ 5 nm, indicating that the exfoliated MoS 2 nanosheets are few-layer. Figure 5D shows that the MnFe 2 O 4 nanoparticles are highly uniform in size with an average diameter of ~ 10 nm, which is very close to the size calculated from the XRD spectrum (~ 9.5 nm).
The elemental composition and distribution of different atoms in the prepared MnFe 2 O 4 /MoS 2 nanocomposite (MnFe 2 O 4 nanoparticles attached on MoS 2 nanosheets) were also analyzed by energy-dispersive X-ray spectroscopy (EDS) mapping, shown in Fig. 6. It is seen that Mo and S atoms are uniformly distributed on the entire mapped window, indicating the presence of MoS 2 nanosheets. Furthermore, the distribution of Mn, Fe, and O atoms are almost uniform and similar, with some aggregates of them in the lower-left part of the mapped window. This uniform and dense distribution of the ferrite nanoparticles enabled by the presence of MoS 2 nanosheets can provide a larger surface to volume ratio, required for an enhanced charge transfer and, in turn, a higher capacitance. Figure 7 shows TEM images of the few-layer MoS 2 nanosheets, the MnFe 2 O 4 nanoparticles, and the MnFe 2 O 4 / MoS 2 nanocomposite. Figure 7A shows a few-layer MoS 2 nanosheet with some wrinkles, indicating its low thickness. The MnFe 2 O 4 nanoparticles in Fig. 7B are nearly uniform in diameter, with a mean diameter of ~ 13 nm, which is very close to the sizes found from XRD calculations and FESEM measurements. Figure 7C shows the MnFe 2 O 4 /MoS 2 nanocomposite, where both MoS 2 nanosheets and MnFe 2 O 4 nanoparticles can be seen. Figure 8 shows the CV curves of the samples at various scan rates, where two peaks are seen with positive and negative currents corresponding to oxidation and reduction processes at the electrode's surface, respectively 46 . As it is seen, by increasing the scan rate the oxidation and reduction peaks shift to higher and lower potentials, respectively, because a shorter time would be available for the electrolyte ions to access the electrode's surface. However, there is a trade-off between potential and time. Besides, at higher scan rates, both the area under the CV curve and the current increase. Nevertheless, it is seen that the area enclosed in a CV curve (or equivalently, the specific capacitance) decreases as the scan rate increases. This is because at higher scan rates, due to the fast migration of ions, some parts of the active surface areas become inaccessible for the charge storage process 14,18,47 . Figure 8 also represents the galvanostatic charge/discharge (GCD) curves of the samples at different current densities in a potential window of 0 to 0.4 V. It is seen that the discharge time of the samples decreases as the current density increases, explained above. The specific capacitance of the electrodes was calculated from their discharge curves according to the equation 14,18 : www.nature.com/scientificreports/ at different current densities, where C sp , I/m, Vdt , and ΔV are the specific capacitance (F/g), the current density (A/g), the area under the discharge curve, and the active potential window, respectively. It should be noted that for battery-type materials, where they have a plateau during their charging and discharging, the capacity should not be calculated using Eq. (4) 34 . However, in our GCD curves, there is no plateau, and instead, an oblique part is seen, consistent with a mostly-pseudocapacitive behavior. Such behavior is more observable in our two-electrode cell measurements (in Fig. 10). Therefore, it is safe to use the formula to calculate the specific capacitance values. The specific capacitances of the samples are reported in Table 1. www.nature.com/scientificreports/ As it is seen in Fig. 9A, which compares the GCD curves of the samples at the current density of 1 A/g, the discharge time, or equivalently the specific capacitance, of the MnFe 2 O 4 /MoS 2 composite is more than those of both MoS 2 and MnFe 2 O 4 . This can be for several reasons. Firstly, MoS 2 is itself an active material for electrochemical reactions due to its electronic structure as well as layered structure, because it can accommodate electrolyte ions in its layered structure. Secondly, due to its versatile redox states, MnFe 2 O 4 has a pseudocapacitive behavior. Therefore, both MoS 2 and MnFe 2 O 4 contribute to the supercapacitance of the MnFe 2 O 4 /MoS 2 nanocomposite. On the other hand, MoS 2 nanosheets prevent the MnFe 2 O 4 from agglomeration and aggregation. Similarly, MnFe 2 O 4 nanoparticles prevent the MoS 2 nanosheets from restacking. This will increase the active surface area for charge storage processes. In the next section on DFT calculations, we will see how MoS 2 can redistribute the stored charges on MnFe 2 O 4 to achieve an improved capacitance. As can be seen in Fig. 9B, the specific capacitance decreases as the current density increases, which is due to the ion diffusion mechanism. In other words, at a lower current density, the electrolyte ions have enough time to penetrate into the active sites on the electrode material, leading to a higher specific capacitance 14 . Figure 9C compares the CV curves of the prepared electrodes at the scan rate of 5 mV/s. The reduction and oxidation peaks are seen around 0.14 V and 0.4 V, respectively. As it is seen, the MnFe 2 O 4 /MoS 2 nanocomposite shows a larger CV-enclosed area than that of MnFe 2 O 4 , which is itself larger than that of MoS 2 . This could indicate the higher capacitance of the MnFe 2 O 4 /MoS 2 nanocomposite than MnFe 2 O 4 . The incorporation of MoS 2 nanosheets largely increased the specific capacitance of MnFe 2 O 4 from 600 to 2093 F/g at 1 A/g. The power densities of the MoS 2 , MnFe 2 O 4 , and MnFe 2 O 4 /MoS 2 samples were obtained at the current density of 1 A/g, listed in Table 2, and their Ragone plots (energy density versus power density) are shown in Fig. 9D. The energy and power densities of the samples were calculated based on www.nature.com/scientificreports/ respectively, where ΔV and t are the potential windows and the discharge time (h) 48 . Cyclic stability tests were performed at the current density of 20 A/g for 2000 GCD cycles, as shown in Fig. 9E. It is seen that the incorporation of MoS 2 nanosheets considerably improved the cycling stability of the pure MnFe 2 O 4 nanoparticles. This can be due to the fact that MoS 2 prevents the MnFe 2 O 4 nanoparticles from detaching from the electrode into the electrolyte, which can improve the capacitance stability of the composite. Electrochemical impedance spectroscopy (EIS) was used to analyze the resistance information of samples. The frequency range of the impedance measurements is 10 MHz-100 kHz. An EIS curve typically consists of two parts: (1) the high-frequency region is a semicircle and (2) the low-frequency region is a straight line, indicating the charge-transfer resistance, and an inclined line, indicating the diffusion of ions into the electrolyte 49,50 . According to the EIS plots in Fig. 9F, the charge transfer resistance of the samples is negligible. The internal resistance can be obtained from the slope of the curves intersecting the x-axis. According to the EIS plots, the internal resistance of the MnFe 2 O 4 / (6) P = E t   Next, we assembled the (MnFe 2 O 4 /MoS 2 )//AC asymmetric supercapacitor in a two-electrode setup, as discussed in the "Experimental" section. Figure 10A shows the CV curves of the device for incremental voltages to confirm its operating potential. It is seen that the capacity increases as the potential window increases, indicating the ability of the device to perform faradaic processes at higher voltages. Figure 10B shows the CV curves of the device for various scan rates at the potential window of 0-1.5 V, showing quasi-rectangular shapes that illustrate the good electrochemical reversibility of the device. Figure 10C shows the GCD curves of the device for various current densities. The charge and discharge parts of the GCD curves are almost symmetric, which demonstrates a small internal resistance drop, indicating the contributions from both the faradaic processes and the double layer capacitance 51 . Figure 10D illustrates the lighting up of a green light-emitting diode (LED) using the (MnFe 2 O 4 /MoS 2 )//AC asymmetric supercapacitor.

DFT study
MoS 2 is a layered transition metal dichalcogenide where its layers are bound together by weak van der Waals (vdW) forces. Each MoS 2 monolayer consists of three atomic layers in the sequence of S-Mo-S. Bulk MoS 2 is a p-type indirect-bandgap semiconductor (1.23 eV), and its bandgap slightly increases to 1.8 eV as the number of layers decreases to one 52,53 . Figure 11A shows top and side views of the optimized structure of the MoS 2 monolayer and its atom-projected density of states. We found the optimized lattice constant of 3.18 Å and the Mo-S bond length of 2.41 Å, and the MoS 2 monolayer shows a direct bandgap of 1.76 eV at the K point, consistent with 53 . On the other hand, MnFe 2 O 4 is an insulating, soft ferrimagnetic spinel ferrite, which crystallizes in a mixed-phase spinel structure 54 with an almost low inversion degree of 0.2, where 80% and 20% of Mn 2+ ions  www.nature.com/scientificreports/ occupy the tetrahedral sites and octahedral sites, respectively, and Mn 2+ and Fe 3+ ions are distributed in the remaining tetrahedral and octahedral sites. Recently, we compared the experimental and theoretical results of MnFe 2 O 4 and showed that the true XRD pattern of MnFe 2 O 4 is a combination of normal and inverse spinel XRD patterns 14 . However, for the sake of simplicity, we considered here the normal spinel configuration for MnFe 2 O 4 (see Fig. 11B). We considered a 28-atom unit cell for the bulk MnFe 2 O 4 as half of a simple cubic structure. Figure 11B shows the unit cell and the atom-projected density of states of the bulk MnFe 2 O 4 . It is seen that the structure is an insulator with a direct bandgap of 1.41 eV. The a and c lattice constants were found 6.14 and 8.68 Å, respectively. Next, we created the MnFe 2 O 4 /MoS 2 interface (see Fig. 11C). The lattice mismatch between the MnFe 2 O 4 surface and the MoS 2 monolayer was ~ 5%. We applied the strain to the MoS 2 monolayer because it only affects its bandgap and cannot change its semiconducting nature 55 . In the optimized structure, the smallest distance between Mn and S atoms is 2.79 Å, which is larger than the sum of the covalent radii of both atoms (1.   www.nature.com/scientificreports/ Figure 12A shows the differential charge density of the MnFe 2 O 4 /MoS 2 interface (i.e., the charge density of the MnFe 2 O 4 /MoS 2 interface minus those of the isolated MnFe 2 O 4 slab and the isolated MoS 2 monolayer). It is seen that the electrons just below the MnFe 2 O 4 surface have been depleted, while they have been accumulated on the MoS 2 surface, more on the nearest sulfur layer, which is due to the higher electron affinity of S as compared to Mo. The integrated charge density difference was calculated using the following equation: where ρ interface , ρ MnFe 2 O 4 , and ρ MoS 2 denote the charge densities corresponding to the MnFe 2 O 4 /MoS 2 interface, the MnFe 2 O 4 slab, and the MoS 2 monolayer. The result is shown in Fig. 12B. The net charge transfer from the MnFe 2 O 4 slab to the MoS 2 monolayer was calculated as 2 electrons. Figure 12C shows the in-plane averaged electrostatic potential of the MnFe 2 O 4 /MoS 2 interface along the z-direction. It is seen that the interface develops a very large difference (16.53 eV) in the in-plane averaged electrostatic potential across the interface, leading to a large intrinsic built-in electric field (E in ) from MoS 2 to MnFe 2 O 4 . This large built-in electric field drives electrons from MnFe 2 O 4 towards MoS 2 , enhancing the interlayer coupling. The attraction of charge density from MnFe 2 O 4 to MoS 2 would help to enhance the charge storage of the composite. When charging, due to the constant transfer of electrons from the electrolyte to MnFe 2 O 4 and in turn to MoS 2 , a longer time would be needed to reach the charge saturation state (as seen in GCD curves in Fig. 9). When discharging, the process is reversed and a long time would be needed for a fully discharged state (again consistent with GCD curves in Fig. 9). This can enhance the specific capacitance of the pure MnFe 2 O 4 .
To sum up, according to the hybrid experimental and computational work, one can conclude that MoS 2 can enhance the charge storage capability and the specific capacitance of MnFe 2 O 4 for several reasons: (1) MoS 2 itself can exhibit a supercapacitance behavior, (2) MoS 2 nanosheets will act as a substrate to hold MnFe 2 O 4 nanoparticles uniformly so that they will not be agglomerated, (3) in a similar manner, the MnFe 2 O 4 nanoparticles can prevent the MoS 2 nanosheets from restacking, (4) MoS 2 nanosheets provide a significantly higher active surface area, (5) MoS 2 would create several energy levels near the Fermi energy of MnFe 2 O 4 that are suitable for charge (8)   www.nature.com/scientificreports/ storage, and (6) MoS 2 will attract the electron charge density from MnFe 2 O 4 and constantly redistribute the stored charges. These all can synergistically enhance the specific capacitance of pure MnFe 2 O 4 nanoparticles.

Conclusions
MnFe 2 O 4 nanoparticles were in-situ synthesized on pre-exfoliated few-layer MoS 2 nanosheets via a simple hydrothermal method, and the synthesized MnFe 2 O 4 /MoS 2 nanocomposite was studied for supercapacitor applications. We found that owing to the effect of MoS 2 , the MnFe 2 O 4 /MoS 2 nanocomposite demonstrates a considerably higher (~ 3.5 times) specific capacitance and better charge-discharge cycling stability as compared to pure MnFe 2 O 4 . Using DFT calculations, we attributed the improvement to the energy levels of MoS 2 near the Fermi level of the composite, making it a conductor, and the attraction of electron charge density from MnFe 2 O 4 to MoS 2 , which will help the redistribution of electrons between MoS 2 and MnFe 2 O 4 when charging and discharging. Figure S1 (in the supplementary material) provides a summary schematic of the research findings.

Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable requests. www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.