In recent years, the demand for electrochemical energy storage devices is more and more extensive, and the requirements for production efficiency are higher and higher, which promotes the development of new materials, and expands the exploration of their novel and simple modification strategies [1-3]. Owing to the short cycle life and inferior capacity (i.e., 372 mAh/g), graphite, as the anode material for lithium-ion batteries (LIBs), could not meet the increasing demand [4, 5]. In contrast, transition metal oxides (TMOs) have become a research hotspot due to their high theoretical capacity and good safety [6-8]. However, due to the structural limitations of TMOs, they are faced with larger bottlenecks. For example, most TMOs belong to wide gap semiconductors or even insulators, which usually exhibit extremely poor electrical conductivity. The weak electronic conductivity not only affects the dynamic performance of the battery, but also leads to the accumulation of joule heat in the electrode, thus causing safety problems in the charging/discharging process. In addition, at the micron scale and block scale, the specific surface area of these materials is relatively small, which is prone to volume expansion, and the movement speed of Li⁺ and electrons will be relatively slow, leading to poor cycling stability, poor initial coulombic efficiency and poor rate performance [9-11]. In order to solve the above-mentioned problems, many methods have been adopted, among which the formulation of composite materials is an attractive strategy to improve electrochemical properties [12]. In general, due to the synergistic advantages, multicomponent hybrids show better performance than each component through mutual reinforcement or modification [13]. Therefore, exploring the rational design of new hybrid metal/metal oxides for multicomponent combinations has become a research direction [14, 15]. For example, CoO@BNG yolk-shell nanotubes [16], hierarchical α-MnS@GSC/GF [17], MOF-derived Co₃O₄/Co-Fe oxide double-shelled nanoboxes [18], MXene/Si@SiO_x@C layerby-layer assembling superstructure [19], etched PBAs-derived porous FeMnO₃@NC nanocages [20], yolk-shell Bi@void@C nanospheres [21] and flowerlike Ti-MoO₃ [22] were previously designed to improve the electrochemical performance. Inspired by the above, the multiple components of the hybrid compounds play a synergistic role, which makes up for the deficiency of the single component compounds, and provides an effective research strategy for the rational design of new TMOs [23, 24].

Metal-organic frameworks (MOFs) are porous crystalline materials with intramolecular pores formed by self-assembly of metal ions and organic ligands through coordination bonds [25]. They usually have large specific surface area, evenly distributed channels and strong adsorption capacity [26-28]. As a kind of MOFs, Prussian blue analogous (PBAs) have obtained wide application prospects in the fields of gas storage/separation, catalysis, and electrochemical storage [29-31]. According to the composition, PBAs can provide carbon and metal sources. Furthermore, their unique composition, hybrid structure of atomic mixing, and suitable chemical stability provide special advantages for in situ immobilization of the TMO nanostructures into porous carbon skeleton with adjustable structural topologies through direct pyrolysis [32-35]. Therefore, the synthesis of ideal multifunctional nanocomposites through the thermolysis of the PBAs is still a research hotspot, which offers a unique opportunity to develop a class of highly adjustable functional materials [36-38].

Herein, we used the MnFe–PBA as a sacrificial template and carbon source to obtain a series of Mn–Fe oxides-based hybrids, and further explored the effects of microstructures and electrochemical performance for the products obtained at different calcination temperatures. Notably, the Fe–Fe_0.33Mn_0.67O/C (i.e., M600) sample exhibited excellent rate performance and long cycle life (~890 mAh/g at 0.1 A/g, 626.8 mAh/g after 1000 cycles at 1.0 A/g with 99% capacity retention). The excellent lithium storage performance of M600 sample is attributed to the reasonable design of the porous structures and the synergetic effect of different functional components.

The facile preparation process is schematically demonstrated in Fig. 1a. First, the MnFe–PBA precursor was synthesized via a typical coprecipitation method. Subsequently, the prepared MnFe–PBA precursor was calcined in N₂ atmosphere at different temperatures (from 300 to 800 ℃) to obtain a series of composite materials, named as M300, M400, M500, M600, M700 and M800, respectively. The scanning electron microscopy (SEM) and transmission electron microscopy (TEM) were employed to characterize the MnFe–PBA precursor. As shown in Fig. S1a (Supporting information), it can be clearly seen that the MnFe–PBA precursor shows the very smooth surface and has good uniformity with an average size of 1 µm. TEM image indicates that it is a solid structure (Fig. S1b). The X-ray diffraction (XRD) pattern of MnFe–PBA precursor is shown in Fig. S2 (Supporting information), exhibiting a typical face-centered cubic structure. The strong diffraction peaks show good crystallinity and no other impurities are detected.

The electrochemical performances of the MnFe–PBA precursor were further tested. Fig. S3a (Supporting information) exhibits the cycling performances of the MnFe–PBA sample at 100 mA/g for 100 cycles. Although the MnFe–PBA sample has a high initial discharge specific capacity (1072.5 mAh/g), the capacity decays rapidly, only 164 mAh/g after 100 cycles, indicating that the sample has poor cyclic stability. Fig. S3b (Supporting information) shows the Nyquist plots of the sample before and after 20 cycles. The curve is composed of a semicircle and a straight line. The semicircle in the high frequency region is related to the charge transfer process of the electron and Li⁺. The smaller the radius is, the smaller the charge transfer resistance (R_ct) is. The lines in the low frequency region are related to the solid diffusion process of Li⁺ in the active material [39]. The larger the slope of the line, the smaller the resistance (R_s) of the electrolyte solution [40, 41]. Obviously, the semicircle diameter of MnFe–PBA sample increases significantly after cycling, indicating that the R_ct value increases, which is not conducive to electron transmission and may be the reason for the poor cyclic stability.

As shown in Fig. 1b1, the SEM image of the M300 sample suggests that it well inherits the morphology of MnFe–PBA precursor while the magnified SEM image (Fig. 1c1) reveals that its surfaces become shrunk slightly and rough with a large number of small nanoparticles attached. For the M400 and M500 samples (Figs. 1b2–b3 and c2–c3), these rough surfaces are actually composed of a large number of small particles, and numerous pores are distributed on the compact surfaces, which are due to the gas release during the annealing treatment. Furthermore, part of the MnFe–PBA sample shrinks into smaller amorphous carbon particles after calcination, and the porous structure is also formed in this process due to the change of volume [42]. As shown in Figs. 1b4 and c4, the M600 sample becomes rougher and has obvious cracks on its surface. When the temperature rises to 700 ℃, the cubic morphology still maintains well (Figs. 1b5 and c5) while the nanoparticles welded on the surface become larger, resulting in a rougher surface. Notably, the initial cubic skeleton has completely collapsed and some irregular agglomerates appear in the product when the annealing temperature is increased to 800 ℃ (Figs. 1b6 and c6). In addition, the TEM images indicate that with the increase of the calcination temperature, these samples tend to become hollow structure, especially for the M600 sample, where obvious mesoporous can be observed inside the cubes (Figs. 1d1–d6).

To further investigate the structural details, the structure characteristics of the M300–M800 samples are studied by the analysis of interplanar distances and selected area electron diffraction (SAED) patterns analysis. As shown in Figs. 2a1–a6, the edges of these samples gradually change from a large number of small particles to a stack of thin sheets. Furthermore, as shown in Figs. 2b1 and 2c1, MnO₂ and Mn₂O₃ phases are identified by interplanar distances in the M300 sample. The (200) and (111) planes of Fe_0.33Mn_0.67O, with interplanar spacing of 0.22 nm and 0.25 nm, are all detected in the M400–M700 samples (Figs. 2b2–b5 and 2c2–c5). In addition, (101) planes of MnO₂ with interplanar spacing of 0.403 nm, (330) planes of Mn with 0.210 nm, (110) planes of Fe with 0.203 nm, and (311) planes of Fe₃Mn₃O₈ with 0.256 nm, are also detected in M400, M500, M600 and M700, respectively. As shown in Figs. 2b6 and c6, the lattice fringe distance of 0.221 nm matches well with the (200) plane of MnO. The SAED patterns of the M300–M700 samples demonstrate the polycrystalline structures, while the M800 sample shows the characteristics of single crystal, illustrating that samples calcined at high temperature have better crystallinity (Figs. 2c1–c6). Fig. 2d displays the element mapping image of M600 sample. The result reveals the homogeneous distribution of Mn, Fe, O, N, and C elements in the material structure. The uniform distribution of carbon is beneficial to avoid the aggregation of in situ generated nanoscale metal oxide particles and endows the battery with good conductivity, which is conducive to the rate capability and durability of batteries [43].

The thermal stability of MnFe–PBA precursor is detected via thermogravimetric (TG) analysist (Fig. 3a). The weight loss of 22.72%, from 27.3 ℃ to 200 ℃, is attributed to the removal of adsorbed and coordinated water molecules. The second weight loss of 7.81% from 200 ℃ to 300 ℃, indicates that the precursor has begun to undergo thermal decomposition, which is a mixture of PBA and metal oxides. After heating above 300 ℃, the organic linker is oxidized to CO₂ and NO_x, resulting in complete decomposition and weight loss of up to 22.06%. The powder XRD patterns of the as-prepared M300–M800 samples are shown in Fig. 3b. The compositions of the M300–M800 products are MnO₂₋Mn₂O₃/C, MnO₂–Fe_0.33Mn_0.67O/C, Mn–Fe_0.33Mn_0.67O/C, Fe–Fe_0.33Mn_0.67O/C, Fe₃Mn₃O₈–Fe_0.33Mn_0.67O/C and MnO–Fe₃C–Fe/C, respectively. This is consistent with the above structural analysis. The associated Raman spectra are shown in Fig. 3c, where two peaks are observed at 1353 cm⁻¹ (D-band) and 1579 cm⁻¹ (G-band), which are characteristic of carbon [44, 45]. The M300 sample did not exhibit carbon peaks because of the presence of a very small amount of carbon. Elemental analysis (EA) is used to test the C content of M300–M800 samples, and the results are shown in Table S1 (Supporting information). We found that with the increase of calcination temperature, the C content of the samples also increased, which was consistent with the Raman result.

The surface composition and chemical states of the M300–M800 samples are investigated by X-ray photoelectron spectroscopy (XPS). As shown in Fig. S4a (Supporting information), the survey spectrums manifest the presence of Fe, Mn, O, N and C in the samples. The XPS spectrums of Mn 2p (Fig. 3d) show Mn 2p_3/2, Mn 2p_1/2 peaks and two satellite peaks (labeled Sat.). The Mn 2p_3/2 and Mn 2p_1/2 peaks can further be divided into two spin-orbit doublets, indicating the coexistence of Mn²⁺ and Mn³⁺ in the composite [46]. It is worth noting that for the M500 sample, two small peaks at 639.5 eV and 650.4 eV are attributed to metallic Mn⁰ peaks. The low intensity can be attributed to metallic Mn⁰ which is covered by metal oxides. As exhibited in Fig. 3e, two major peaks appearing at 721.3 eV and 708.6 eV are ascribed to Fe 2p_1/2 and Fe 2p_3/2 of Fe²⁺, respectively. The other two peaks at 710.6 eV and 723.7 eV belong to Fe 2p_3/2 and Fe 2p_1/2 of Fe³⁺, respectively [47, 48]. Besides, a small satellite peak is observed at 713.8 eV. Obviously, for the M600 and M800 samples, metallic Fe0 peaks are observed at 704.2 eV and 719.2 eV, which is consistent with the XRD analysis results. The O 1s peak for the M300–M800 samples (Fig. S4b in Supporting information) can be fitted to three peaks at approximately 529.9, 531.6, and 532.6 eV, which are related to the metal oxide (Fe–O–Fe, Mn–O–Mn, Mn–O–Fe), carbonyl oxygen, and chemisorbed oxygen species (H–O–H), respectively [49-51]. The C 1s spectrums exhibit two peaks at 284.0 eV and 284.8 eV (Fig. 3f), which can be ascribed to C–C and C–N, respectively [52]. And the pyridinic N and pyrrolic N are detected at 397.5 eV and 399.3 eV, respectively, in the N 1s spectrums (Fig. S4c in Supporting information) [53].

As shown in Fig. 4a, the nitrogen adsorption/desorption isotherms of the M300−M800 samples exhibit obvious hysteresis loops, illustrating the presence of mesopores [54]. Meanwhile, it can be inferred from Fig. 4b that the average pore size of the M600 sample is slightly higher (17.00 nm) than that of other samples (10.52, 19.73, 16.43, 16.14 and 12.28 nm for the M300, M400, M500, M700 and M800 samples, respectively). Further measure the contact angle of electrolyte on the surface of M600 sample. Fig. S5a (Supporting information) shows the electrolyte just dripping onto the electrode sheet, and Fig. S5b (Supporting information) shows the process of wetting M600 sample with electrolyte. After 350 ms, the electrolyte completely penetrated into M600 sample (Fig. S5c in Supporting information), indicating that the electrolyte has good wettability to M600 sample, which is due to its porous structure. In addition, the specific surface area of M600 obtained from Brunauer-Emmett-Teller (BET) method is approximately 35.01 m²/g, which is higher than that of M300 (10.06 m²/g), M400 (24.93 m²/g), M500 (20.42 m²/g), M700 (16.18 m²/g), and M800 (7.36 m²/g). In general, mesoporous can alleviate the structural strain caused by repeated Li⁺-insertion/extraction processes, so as to expand cycle life of the sample and improve cycle capacity even at high current rate. The high specific surface area can provide more active sites in electrochemical reactions, which is conducive to the storage of electrostatic charges. Obviously, the M600 sample has some advantages over other samples in structural characteristics.

Fig. 5a displays the first three cyclic voltammetry (CV) curves of the M600 electrode at 0.1 mV/s between 0.01 V and 3 V. In the first cycle, the broad peak located in 1.65 V belongs to the reduction of Mn³⁺ and Fe³⁺ to Mn²⁺ and Fe²⁺, and the intensive peak at 0.21 V can be assigned to the further reduction of Mn²⁺ and Fe²⁺ to metallic Mn and Fe, respectively. Moreover, the small peak at 0.52 V can be attributed to the formation of solid electrolyte interface (SEI), produced by the irreversible decomposition of the electrolyte. The broad anode peak at about 1.31 V can be assigned to the oxidation of the metallic Mn and Fe to Mn²⁺ and Fe²⁺, respectively. Meanwhile, the minor peak at 1.72 V is due to the oxidation of Mn²⁺ and Fe²⁺ to Mn³⁺ and Fe³⁺, respectively. The CV curve of the third scan almost overlaps with that of the second scan, indicating the good reversibility. In addition, Figs. S6−S10 (Supporting information) display the CV curves of the M300, M400, M500, M700 and M800 samples, respectively.

Fig. 5b shows the discharge/charge profiles of M600 for the 1^st, 2^nd, 5^th, 10^th, 20^th and 40^th cycles at 100 mA/g. The first discharge and charge capacities are 1195.2 mAh/g and 824.1 mAh/g, leading to an initial coulombic efficiency of 68.95%. The low initial coulombic efficiency may be related to the formation of SEI film and the decomposition of electrolyte [55]. The discharge/charge curves of M600 in the 2^nd, 5^th, 10^th, 20^th and 40^th almost overlap and maintain stable plateaus, which is consistent with the CV curves. The discharge/charge curves of other samples are displayed in Figs. S11−S15 (Supporting information). It is not difficult to find that M600 sample has the best performance. Furthermore, the cycling performances of the M300−M800 samples at 100 mA/g are investigated, as shown in Fig. 5c. The M600 nanocubes achieve a capacity of 890.5 mAh/g at the 2^nd cycle and slightly decrease to 730.4 mAh/g after 70 cycles. However, only 430.8, 609.8, 580.4, 582.0 and 268.7 mAh/g are left for the M300, M400, M500, M700 and M800 samples, respectively, after 70 cycles. In addition, Fig. S16 (Supporting information) shows the Nyquist plots of the M300−M800 samples, where the semicircle region represents the charge transfer resistance R_ct and the slope represents the Warburg impedance Z_W [56, 57]. The calculation indicates that the R_ct value of the M600 anode is 6.04 Ω, which is much smaller than that of the M300 (28.94 Ω), M400 (10.27 Ω), M500 (16.57 Ω), M700 (21.74 Ω), and M800 (44.05 Ω) anodes. Meanwhile, the slope of the M600 sample is the largest, indicating that its impedance is the smallest. This result further confirms that metallic Fe increases the electronic conductivity of the M600 composite electrode and provides faster Li⁺ diffusion and electron transport property. In addition, although there is metallic Mn in M500 sample, the conductivity of Mn is not as good as that of Fe. Meanwhile, the M600 sample has more structural advantages, so the M600 sample has more remarkable electrochemical properties.

The rate capability of M600 anode is evaluated from 100 mA/g to 1600 mA/g. As shown in Fig. 5d, the specific capacities are 785.9, 738.3, 689.2, 643.8 and 589.2 mAh/g at 100, 200, 400, 800 and 1600 mA/g, respectively. When the current density is restored to 100 mA/g, the discharge capacity resumes to 776.3 mAh/g, suggesting a good rate capability. The long cycling performance of M600 anode at the larger current density of 1 A/g up to 1000 cycles is also shown in Fig. 5e. In the first 200 cycles, the specific capacities gradually decrease, which can be attributed to the formation of SEI layer in the activation process. Then, the capacities slowly increase and stabilize, reaching 626.8 mAh/g after 1000 cycles, indicating the good cycle life. In addition, Table S2 (Supporting information) compares the electrochemical performance of different PBAs derived metal oxides in LIBs, from which it can be seen that M600 sample has excellent cycling performance.

In order to explore the energy storage mechanism, we tested the XPS patterns of M600 sample during cycle, where D1 = 1.6 V, D2 = 0.2 V, and C1 = 1.3 V. As shown in Figs. 6a and b, when the voltage is discharged from 1.6 V to 0.2 V, the formation of metal Fe and metal Mn can be observed in Fe 2p and Mn 2p patterns, respectively. Subsequently, when the voltage is charged from 0.2 V to 1.3 V, the metal Fe and metal Mn are oxidized, which is consistent with the CV result. Based on the above analysis, the reaction equations for the M600 sample during Li⁺-(de)insertion process may be as follows:

First discharge:

Afterwards:

The Fe formed in the reduction process not only enhances the conductivity, but also effectively prevents the aggregation of the generated Mn [58]. At the same time, the MnO_x and FeO_x products can be evenly distributed after the oxidation. Thus, the M600 sample is an ideal anode material for LIBs.

The morphological changes of M300−M800 electrodes are shown in Figs. 6c and d and Fig. S17 (Supporting information). Before cycling, the surfaces of all samples are uniform and compact without cracks. Except for the collapse of M800 sample, the other samples are original cube structure. Besides, the nanoparticles distributed around these cubes are the binder used in the fabrication of the electrode materials. After 50 cycles, cracks of different degrees appeared on the surfaces of M300, M400, M500, M700 and M800 electrodes, and the nanocubes are damaged and seriously agglomerated [59]. Surprisingly, the M600 sample can well maintain the original texture characteristics in terms of shape, size, and structural integrity after 50 cycles (Figs. 6c and d), not only because of its unique structural characteristics, but also because the conductive carbon layer formed by the carbonization of organic ligands in M600 also acts as a physical buffer matrix which prevents the interparticle agglomeration of Fe and Fe_0.33Mn_0.67O nanoparticles during cycling.

To sum up, compared with MnFe−PBA precursor, the electrochemical performance of calcined samples is significantly improved, especially M600 samples, which can be mainly attributed to the following reasons: (1) Compared with other samples, M600 sample has the largest pore size (17.00 nm) and the largest specific surface area (35.01 m²/g). The porous structure and large specific surface area can provide shorter Li⁺ diffusion paths and abundant active sites. (2) The presence of Fe nanoparticles gives the M600 sample a minimal R_ct value (6.04 Ω), which further enhances the conductivity and the electrochemical activity, which accelerates the electrochemical reactions. (3) The carbon layer produced by high temperature calcination can buffer the volume change and prevent aggregation of Fe_0.33Mn_0.67O nanoparticles during the discharge/charge cycles.

To investigate the comprehensive electrochemical performance of the M600 electrode, we recorded the CV curves of the M600 sample at scanning rates from 0.1 mV/s to 1.0 mV/s, as displayed in Fig. 7a. Eq. 3 reveals the relationship between the peak current (i) and sweep rate (v) [60].

According to Fig. 7b, the b values at peak I, peak II, and peak III are calculated to be 0.896, 0.803 and 0.792, respectively. Based on previous reports, the b value is between 0.5 and 1, and the closer the value is to 1.0, the more it indicates that it is a capacitive dominated process. On the contrary, the closer the b value is to 0.5, the more it manifests that it is the electrochemical reaction controlled by ion diffusion [61]. Therefore, a typical pseudocapacitive phenomenon occurs on the M600 electrode, which is beneficial to rapid charge storage and long-term cycling capability, especially at high current densities [62].

To further quantitatively distinguish the two processes of the total charge storage, we divided Eq. 4 into two parts, where k₁ν represents the capacitive-controlled process and k₂ν^1/2 is the diffusion-controlled behavior [63]. As exhibited in Fig. 7c, with the increase of scanning rate, the capacitive contribution to total charge storage increases significantly, even reaching 89.4% at 0.5 mV/s (Fig. 7d). The great contribution of pseudocapacity to the overall capacity can be derived from the porous structure and the participation of N–doped C layer, which make the Fe–Fe_0.33Mn_0.67O/C nanocubes exhibit satisfactory lithium storage performance.

In summary, we synthesized a series of Mn–Fe oxides-based hybrids using Mn–Fe PBA as a template and an organic carbon source by calcination. Owing to the structural and compositional benefits, the as-derived porous Fe–Fe_0.33Mn_0.67O/C nanocubes (i.e., M600) exhibited more excellent rate capability and longer cycle life than other samples (~890 mAh/g at 0.1 A/g, 626.8 mAh/g after 1000 cycles at 1.0 A/g with 99% capacity retention). The impressive electrochemical performance can be attributed to the following points: (1) The porous structure can not only provide shorter Li⁺ diffusion path and abundant void space, but also promote the penetration of electrolyte. (2) The presence of Fe nanoparticles enhances the conductivity and electrochemical activity, which accelerates the electrochemical reactions. (3) The N−doped C layer can buffer the volume change and prevent aggregation of Fe_0.33Mn_0.67O nanoparticles during the discharge/charge cycles. (4) The strong synergistic effect among the components of the M600 sample can make up for the poor stability of metal oxides as anodes of LIBs. Therefore, the key to obtain ideal electrochemical properties is reasonable design of microstructure and compositions of functional nanocomposites.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

This work was supported by the National Natural Science Foundation of China (NSFC, Nos. 21901222, U1904215 and 21671170), Lvyangjinfeng Talent Program of Yangzhou, the Top-notch Academic Programs Project of Jiangsu Higher Education Institutions (TAPP), Program for Young Changjiang Scholars of the Ministry of Education, China (No. Q2018270), and Natural Science Foundation of Jiangsu Province (No. BK20200044).

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2022.04.045.