A self-assembled nanoflower-like Ni5P4@NiSe2 heterostructure with hierarchical pores triggering high-efficiency electrocatalysis for Li–O2 batteries

The remarkably high theoretical energy densities of Li–O2 batteries have triggered tremendous efforts for next-generation conversion devices. Discovering efficient oxygen reduction reaction and oxygen evolution reaction (ORR/OER) bifunctional catalysts and revealing their internal structure-property relationships are crucial in developing high-performance Li–O2 batteries. Herein, we have prepared a nanoflower-like Ni5P4@NiSe2 heterostructure and employed it as a cathode catalyst for Li–O2 batteries. As expected, the three-dimensional biphasic Ni5P4@NiSe2 nanoflowers facilitated the exposure of adequate active moieties and provide sufficient space to store more discharge products. Moreover, the strong electron redistribution between Ni5P4 and NiSe2 heterojunctions could result in the built-in electric fields, thus greatly facilitating the ORR/OER kinetics. Based on the above merits, the Ni5P4@NiSe2 heterostructure catalyst improved the catalytic performance of Li–O2 batteries and holds great promise in realizing their practical applications as well as inspiration for the design of other catalytic materials.


Introduction
Nowadays, the excessive consumption of fossil energy and serious environmental pollution has drawn our attention to developing energy storage devices with high energy densities [1][2][3]. It is evident that Li-O 2 batteries hold great potential for nextgeneration battery systems, mainly due to their extra-high theoretical energy density (3500 Wh kg −1 ), which is related to the reversible redox reaction of 2Li + + 2e − + O 2 ↔ Li 2 O 2 [4][5][6]. However, some problems still need to be solved, including low specific capacities, inferior rate capacity, high discharge/charge overpotentials and limited cycle life, which can be generated from the slow reaction kinetics towards oxygen evolution reaction (OER) and oxygen reduction reaction (ORR) processes [7][8][9]. To overcome these obstacles, numerous researches have been carried out in recent years, in which the construction of efficient electrocatalysts can not only significantly improve the sluggish kinetics towards ORR/OER, but also limit the adverse parasitic reactions [10][11][12]. In other words, exploiting appropriate catalysts is crucial for improving the performance of Li-O 2 batteries.
Among them, noble metals (Pd, Pt, Au) [13][14][15] are considered as ideal cathode catalysts to mitigate polarization and improve battery efficiency cycling performance, but the high price and scarcity restrain their large-scale application. Various alternatives have thus been extensively studied in  batteries, such as carbon composites [16][17][18], alloys [19,20], transition metal oxides [21][22][23], nitrides [24][25][26], sulfides [27][28][29], carbides [30][31][32] and phosphide [33][34][35], etc. Actually, it is well known that carbon materials are too sensitive to generate unwanted by-products, which could largely deteriorate the battery performance. According to recent literature reports, it is universally acknowledged that transition metal compounds are promising catalyst materials for electrical storage and electrocatalytic systems due to their excellent physicochemical properties, including tunable active centers and high catalytic activity. Among the transition metals, Ni element are moderately reserved and more affordable than Co element, and Ni-based compounds exhibit high catalytic activity when used as redox reaction sites [36,37]. Most importantly, the presence of Ni 3+ and Ni 2+ redox couples could be easily obtained in the catalyst materials, realizing impressive electrocatalytic activities through promoting the formation/ decomposition of Li 2 O 2 [38]. Besides, Ni-based compounds have been intensively investigated and evaluated as cathode catalysts of Li-O 2 batteries due to the environmental benignity, high chemical and thermal stability, as well as facile fabrication protocols [39][40][41]. As reported, transition metal selenides typically exhibit superior electrical conductivity due to their exceptional D-electron configuration and suitable energy position, which in turn leads to excellent electrocatalytic performance [42,43]. Yoo's [44] groups demonstrated that FeSe hollow spheroids delivered excellent stable cycle performance without significant changes in the overpotentials during cycling in Li-O 2 batteries, and the SeO x on the surfaces of FeSe hollow spheroids contributed to facilitating ORR/OER bifunctional catalytic activities. Notably, transition metal phosphide surface polarization at the phosphorus terminus normally leads to negatively charged phosphorus centers, and the P sites with high electronegativity usually act as proton receptors, which facilitates the adsorption and desorption of intermediate species in oxygen electrocatalysis [35,45]. For example, Du et al [46] successfully synthesized concave polyhedrons CoP with a high-index facet (211), which presented favorable electrocatalytic ability in Li-O 2 batteries.
Since electrochemical reactions in Li-O 2 batteries occur essentially at the three-phase interfaces, surface modifications, including defect engineering, heterogeneous atom doping and heterostructure construction, are proposed to effectively improve catalytic performance. In recent years, heterojunction engineering has received considerable research interests due to its unique physicochemical properties and practicality in designing unique electrocatalysts [47][48][49][50]. First, the built-in electric fields at the heterointerfaces could modulate the interfacial electronic structure and promote the reaction kinetics in the ORR/OER processes [51][52][53]. Besides, due to the high difference of electronegativity between Se (2.55) and P (2.19), the heterojunction interfaces could present two electrical regions of opposite charges [54,55]. The strongly charged regions of Ni 5 P 4 show the potential to optimize the chemisorption of reaction intermediates, and the electrondeficient regions of NiSe 2 may act as active sites for continuity and accessibility via electron transfer, thus positively affecting the performance of Li-O 2 batteries [56]. The conventional strategies for forming heterogeneous interfaces, however, are generally to employ the epitaxial growth methods in solution, and they cannot be widely applied in practical production due to the complicated processes [57,58]. Therefore, it remains a challenge to effectively fabricate heterostructured catalysts with rich heterogeneous interfaces.
Herein, nanoflower-like Ni 5 P 4 @NiSe 2 heterostructure was prepared and acted as a cathode catalyst for Li-O 2 batteries, which delivered superior specific capacities and extended cycling life, compared with the Ni 5 P 4 and NiSe 2 counterparts. The improved catalytic activity of the nanoflowerlike Ni 5 P 4 @NiSe 2 heterostructure mainly stemmed from the built-in electric fields at the heterojunction interfaces, which effectively enhanced the electrical conductivity and thus improved the slow reaction kinetics in the charge and discharge processes. Moreover, the disordered atomic arrangement and the slight lattice distortion triggered by the Jahn-Teller effect at the heterogeneous interfaces could enable additional active sites to facilitate the regulation of the adsorption of oxygen-containing intermediates, which significantly improving the ORR/OER bifunctional catalytic activity. In addition, the constructed flower-like structure facilitated the construction of three-dimensional (3D) diffusion paths of Li + /O 2 and provided sufficient room for the storage of discharge products. These results inspire promising strategies to develop new sufficiently stable electrocatalysts for Li-O 2 batteries.

Fabrication of Ni(OH) 2 precursor
The precursor solution was obtained by dissolving 1 mmol Ni(NO 3 ) 2 ·6H 2 O, 8 mmol CH 4 N 2 O and 3 mmol NH 4 F into 30 ml of deionized water at room temperature with stirring for 20 min. Then, it was transferred into a 50 ml Teflon-lined stainless-steel autoclave and heated at 120 • C for 24 h. After cooling down to room temperature, the as-prepared Ni(OH) 2 was washed with deionized water and ethanol three times and dried at 50 • C for 12 h in a vacuum oven.

Synthesis of Ni 5 P 4 @NiSe 2 nanoflowers
Ni 5 P 4 @NiSe 2 nanoflowers were prepared by simultaneous phosphorylation and selenization treatment. In brief, its fabrication was carried out in a tube furnace with as-prepared Ni(OH) 2 precursor at the downstream and a mixture of Se powder and NaH 2 PO 2 at the upstream at 350 • C for 2.5 h with a heating rate of 2 • C min −1 under an Ar atmosphere.

Materials characterizations
Field-emission scanning electron microscope (FESEM, Hitachi, S-4800, Japan) coupled with energy-dispersive x-ray spectroscope (EDX, Oxford Materials Analysis, UK) and high-resolution transmission electron microscope (HRTEM, JEOL-JEM 2100F, 200 kV, Japan) were used to investigate the morphologies and structures of the samples. The crystalline structures were recorded via x-ray diffraction (XRD, D/Max-IIIC, 36 kV and 20 mA, Japan). The Brunauer-Emmett-Teller (BET) specific surface areas and pore size distribution were examined by nitrogen adsorption/desorption isotherm (BET, Micromeritics ASAP2020). X-ray photoelectron spectroscopy (XPS, ESCALAB Xi+) was collected to characterize the surface chemical states, and all binding energies of the XPS spectra were adjusted by the carbon peak (C 1s) at around 284.8 eV. Exact two-phase ratio of Ni 5 P 4 @NiSe 2 was obtained by inductively coupled plasma-atomic emission spectrometer (ICP-AES, Agilent-5110, USA).

Electrochemical measurements
To evaluate the electrochemical performance of the cathode catalysts for Li-O 2 batteries, modified 2032 coin-type cells with holes on the cathode lid were assembled. To prepare the Ni 5 P 4 @NiSe 2 cathodes, 40 wt % Ni 5 P 4 @NiSe 2 powder, 40 wt % Ketjen black (KB) and 20 wt % poly-1,1,2,2-tetrafluoroethylene were mixed in 3 ml isopropanol under ultrasonic condition. The slurry was then uniformly dispersed on carbon papers and dried under vacuum at 120 • C for 12 h. The Ni 5 P 4 and NiSe 2 cathodes were also prepared by the same method as above for comparison. The Li-O 2 cells were assembled in a glovebox (Mbraun) with the prepared cathodes, Li sheet anodes and glass fiber separators with 1 M lithium bis(trifluoromethanesulfonyl)imide in triethylene glycol dimethyl ether (LTFSI/TEGDME) electrolyte. The electrochemical performance was measured by using a multi-channel cell test system (LAND CT 2001A). Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) were performed on an electrochemical workstation (CHI 660E, frequency region: 10 5 -0.01 Hz, amplitude voltage: 10 mV).

Result and discussion
Here, heterostructures can be described as the unique structures that consist of heterointerfaces formed by different materials through chemical or physical combinations. It is evident that the different work function (Φ) of Ni 5 P 4 and NiSe 2 are 4.97 and 6.88 eV, respectively [59,60]. Meanwhile, NiSe 2 presents semiconductor characteristics with a wide band gap energy (E g ) of about 1.96 eV, while Ni 5 P 4 shows metallic properties due to the Femi level of Ni 5 P 4 passed through the conduction band [61,62]. Therefore, when they contacted with each other closely, a thermodynamic equilibrium was gained, and a space charge region would be formed, in which electrons were injected from NiSe 2 to Ni 5 P 4 [63]. And the builtin fields work mechanism of interfacial NiSe 2 and Ni 5 P 4 in equilibrium as shown in scheme 1. As a result, the surface of Ni 5 P 4 @NiSe 2 heterostructure exhibits enriched electron density, and this implies a significant increase in electrical conductivity, which can further improve the electrocatalytic activity of the cell. The synthetic process of the Ni 5 P 4 @NiSe 2 heterostructure was schematically illustrated in scheme 1, and its fabrication photograph was included in figure S1. First, Ni(NO 3 ) 2 ·6H 2 O, CH 4 N 2 O and NH 4 F were ultrasonically dispersed in deionized water under stirring, and the precursor of Ni(OH) 2 was achieved after the hydrothermal treatment, evidenced by the data in figure S2. It was then further converted to Ni 5 P 4 @NiSe 2 heterostructure by a simultaneous phosphorylation and selenization process at 350 • C under the Ar atmosphere for 2.5 h. For comparison, Ni 5 P 4 and NiSe 2 nanoflowers were also obtained via the same fabrication route with phosphorylation or selenization, respectively.
It can be observed in figure S3 that the precursors were assembled in a nanoflower-like structure with an average diameter of 5 µm, consisting of many smooth nanosheets with the thickness of 3-5 nm. Compared with the smooth and granular  nanosheets of Ni 5 P 4 in figure S4 and NiSe 2 in figure S5, those of the Ni 5 P 4 @NiSe 2 heterostructure in figures 1(a) and (b) shows smooth surfaces with many nanopores and large available space, which could be conducive to effective electrolyte penetration, boosted mass transfer and effective discharge products accommodation [64][65][66]. The N 2 adsorption/ desorption isotherms ( figure 2(b)) of Ni 5 P 4 @NiSe 2 heterostructure show the IV-type H3 hysteresis loop. In the observed isotherms, the hysteresis curves exhibit a saturated adsorption plateau, indicating homogeneous pore formation [4,67]. Its pore size distribution result demonstrates that the pores are mainly mesoporous, which could provide abundant mass diffusion tunnels and expose more active sites for LOBs. The overall molar ratio of Ni 5 P 4 /NiSe 2 is approximately 57/43 according to EDX result in figure 1(c), which is close to those of XPS and ICP-AES results in figure S6. The EDX-elemental mapping images of Ni 5 P 4 @NiSe 2 heterostructure suggest that the Ni, P, and Se elements are uniformly dispersed on the whole architecture.
To further investigate the more detail microstructure of Ni 5 P 4 @NiSe 2 heterostructure, the TEM image in figure 1(d) shows 3D hierarchical porous nanoflower-like morphology, which is well consistent with the SEM results. In figure 1(e), the lattice fringes with well-defined interfacial distances of 0.223 nm can be clearly described to the spacing of the (210) crystal planes of Ni 5 P 4 , meanwhile the lattice fringes of approximately 0.299 nm can be consistent with the (200) planes of NiSe 2 . Interestingly, figure 1(e) also depicts clear interfacial regions owing to the mismatch of the different phases, and the resulting strong electronic interaction between Ni 5 P 4 and NiSe 2 may lead to an increase in the active sites [43,68,69]. Figure 1(f) exhibits the corresponding selected area electron diffraction pattern of Ni 5 P 4 @NiSe 2 heterostructure, which can be unambiguously indexed into (002), (201), (211), (302), (204) planes of the Ni 5 P 4 and (200), (211) planes of the NiSe 2 , respectively, further demonstrating that Ni 5 P 4 @NiSe 2 heterostructure was successfully synthesized.
The crystalline structure and phase of Ni 5 P 4 @NiSe 2 , Ni 5 P 4 and NiSe 2 were tested by XRD measurement. As depicted in figure 2(a), all diffraction peaks of Ni 5 P 4 @NiSe 2 heterostructure correspond perfectly to hexagonal Ni 5 P 4 (JCPDS. no 18-0883) and cubic NiSe 2 (JCPDS. no 89-3058), which is identical to the HRTEM data. As we all know, electrocatalytic reactions in Li-O 2 batteries generally mainly occur at the three-phase interfaces, and it is thus critical to analyze the surface elemental states of different samples [70]. XPS testing was used to study the bonding configuration and elemental composition of Ni 5 P 4 @NiSe 2 and NiSe 2 . Figure 2(c) reveals the presence of C, O, Ni, P and Se elements on the asprepared Ni 5 P 4 @NiSe 2 heterostructure. Its Ni 2p spectrum in figure 2(d) shows two spin-orbit peaks, which are assigned to 2p 3/2 and 2p 1/2 signals. Moreover, the peaks can be respectively fitted to Ni 3+ (855.9 and 873.9 eV), Ni 2+ (852.7 and 869.8 eV) and the associated satellite (861.1 and 879.6 eV) peaks. Compared with those of pure NiSe 2 , the two-orbit doublets in the XPS spectrum of Ni 5 P 4 @NiSe 2 heterostructure are slightly shifted to negative binding energies [55], which can be attributed to electronic structure changes caused by the interfacial charge redistribution of Ni 5 P 4 and NiSe 2 . The high-resolution XPS spectrum of Se 3d (figure 2(e)) splits into two-component peaks at about 54.5 and 53.5 eV, related to Se 3d 3/2 and Se 3d 1/2 of Se 2− . The Se 3d peak of Ni 5 P 4 @NiSe 2 is normally accompanied by a 58.5 eV characteristic peak assigned to the Se-O bond, confirming that the surfaces of some Se species were oxidized to SeO x during the synthesis route [44,66]. As can be seen in the high-resolution XPS spectrum of P 2p in figure 2(f), the peak of Ni 5 P 4 @NiSe 2 heterostructure at 129.35 eV is ascribed to the Ni-P bond, and the peak at 133.7 eV demonstrates the presence of oxidation on the material surfaces [55,71].
The electrocatalytic activity of Ni 5 P 4 @NiSe 2 heterostructure was measured in Li-O 2 batteries placed in a testing box purchased from NJZH (Shenzhen) Scientific Ltd, as displayed in figure S7. CV profiles of Ni 5 P 4 @NiSe 2 , Ni 5 P 4 , NiSe 2 and pure KB cathodes within 2.35-4.5 V at 0.15 mV s −1 are depicted in figure 3(a). The Ni 5 P 4 @NiSe 2 cathode exhibits the highest ORR/OER current densities, proving that it can significantly promote the electrocatalytic reaction kinetics. Specifically, the Ni 5 P 4 @NiSe 2 cathode distinctly exhibits two negative peaks for OER, which are attributed to the different decomposition stages of discharge products [72][73][74]. It is proposed that the peak at ∼4.2 V is attributed to the delithiation process of Li 2 O 2 (Li 2 O 2 → Li 2−x O 2 + x Li + + x e − ), while the lower peak at 3.78 V is ascribed to a further delithiation process (Li 2 [7,8]. The discharge/charge plots of Li-O 2 batteries based on various cathodes were tested at the current density of 100 mA g −1 with the voltage range of 2.35-4.5 V versus Li + /Li. It is apparent in figure S8 that the initial discharge/charge specific capacities of carbon paper cathodes are negligible, which proves their contribution mainly comes from the active materials. As presented in figure 3(b), the Ni 5 P 4 @NiSe 2 cathode exhibits the largest discharge/charge capacities of 19 090/19 031 mAh g −1 at 100 mA g −1 , while those of Ni 5 Figure 3(c) shows the rate performance of Ni 5 P 4 @NiSe 2 , Ni 5 P 4 , NiSe 2 and pure KB cathodes under different current densities with the cutoff capacity of 1000 mAh g −1 . It can be seen that at current densities of 100, 200, 400, 800, 1000 and 100 mA g −1 , the Ni 5 P 4 @NiSe 2 cathode exhibits the largest/lowest terminal discharge/charge voltages. When the current density returns to 100 mA g −1 , the terminal voltages remained almost unchanged compared to the initial values. Those differences in electrochemical properties demonstrate that the built-in electric fields with charge redistribution on heterostructure could improve the electrical conductivity of the cathode catalyst materials, thus effectively facilitating the formation and decomposition of discharge products [69]. In addition, the disordered atomic arrangement and the slight lattice distortion triggered by the Jahn-Teller effect at the heterogeneous interfaces could increase the reaction activity centers to boost electrocatalytic reactions [75][76][77]. It is thus concluded that the synergy of these two factors endowed the heterostructure cathodes excellent electrocatalytic performance. Figure 3(d) further shows the rate capability of Ni 5 P 4 @NiSe 2 cathodes at different current densities under the voltage window of 2.35-4.5 V, and the pure KB cathode (figure S9) was also tested under the same conditions. As the current densities increased from 100 to 800 mA g −1 , the ORR/OER overpotentials of Ni 5 P 4 @NiSe 2 cathode increased to 0.20/1.45 V, and impressive discharge/charge specific capacities of 19 090/19 031, 18 026/17 802, 16 620/15 788 and 14 379/12 844 mAh g −1 at the current densities of 100, 200, 400 and 800 mA g −1 were also yielded, respectively. Moreover, the voltage platforms in the galvanostatic ORR/OER profiles of the Ni 5 P 4 @NiSe 2 cathodes match well with the peaks of CV curves.
The cycle stability of different cathodes were evaluated with the fixed capacity of 1000 mAh g −1 at 200 mA g −1 , as shown in figure 3(e). The terminal discharge/charge plots of Ni 5 P 4 @NiSe 2 cathode was the most stable during cycling and can be effectively cycled up to 128 cycles, while those of Ni 5 P 4 , NiSe 2 and pure KB cathodes dropped down quickly after 64, 28 and 20 cycles, respectively. Moreover, the excellent cycling performance of 202 cycles at a lower limiting capacity of 600 mAh g −1 with 100 mA g −1 was  Table S1 shows a comparison of the battery performance in this work with previously reported similar counterparts. It can be well noticed that the Ni 5 P 4 @NiSe 2 cathode exhibits long cycle stability and ultra-high discharge specific capacities, compared to those of transition metal phosphide and selenide cathodes. It is believed that the nanoflower-like Ni 5 P 4 @NiSe 2 heterostructure with more free space and nanopores could construct a large amount of 3D channels for the fast Li + /O 2 transport. Additionally, the built-in electric fields between Ni 5 P 4 and NiSe 2 can largely improve the slow redox kinetics of the cathode reactions, resulting in excellent electrochemical performance of Li-O 2 batteries.
To explain the formation/decomposition mechanism and reason for the excellent electrocatalytic performance of Ni 5 P 4 @NiSe 2 cathodes, ex-situ XRD, XPS and EIS at different stages during cycling were characterized. Figure 4(a) shows the XRD patterns of the Ni 5 P 4 @NiSe 2 cathode at different stages, where two new diffraction peaks located at 32.9 and 35.0 • after first discharging, indexed to (100) and (101) planes of Li 2 O 2 (JCPDS. no 09-0355), respectively, and those of the fresh carbon paper cathode was also given in figure S10 for comparison. After 1st and 60th recharging, the diffraction peaks of Li 2 O 2 fully disappeared, and almost no peaks of high crystalline by-products were traced. Meanwhile, EIS testing was carried out to measure the intrinsic kinetics characteristic at different discharge/charge stages. Typically, the EIS diagram contains two parts, the semicircle at high-frequency is associated with charge-transfer resistance (R ct ), and the diagonal line at low-frequency is related to the ion diffusion properties. The date was fitted using the equivalent series circuit (inset of figure 4(b)). As shown in figure 4(b), the semicircles in the Ni 5 P 4 @NiSe 2 -based cell were greatly enlarged from 14.5 to 82.5 Ω after first discharging, mainly due to the coverage and accumulation of insulating Li 2 O 2 [18,23]. Besides, it also shows that after first recharging and even after 100 cycles of prolonged cycling, the R ct of Ni 5 P 4 @NiSe 2 cathode is close to its initial stage, demonstrating that the discharge products were almost decomposed. To further shed light on the composition of discharge products on Ni 5 P 4 @NiSe 2 cathode, ex situ XPS experiments were studied, and the discharge/charge profiles with corresponding selected states are listed in figures 4(c)-(f). For discharged figure 4(e) relative to the initial stage, indicates that the main discharge products were Li 2 O 2 and almost no by-products generation at this stage. After recharging (figure 4(f)), this peak no longer appeared, whereas the Se 3d 3/2 and Se 3d 1/2 peaks of the pristine cathode (figure 4(d)) can be detected at 54.5 and 53.5 eV. Those results further explain the ultra-long cycle life and stable cycling performance of Ni 5 P 4 @NiSe 2 cathodes.
According to previous reports in the literature, the morphology of Li 2 O 2 plays an essential role in affecting the subsequent charging process. Whether discharge products Li 2 O 2 formed film-or disc-shaped was controlled by the adsorption energy of LiO 2 . The disc-shaped Li 2 O 2 grew through the solution growth model due to the weak adsorption energy of the LiO 2 intermediate, while the film-shaped Li 2 O 2 was formed via a surface mechanism with strong adsorption energy of LiO 2 . As can be seen in figure 4(h), the surfaces of the Ni 5 P 4 @NiSe 2 cathode were covered with film-like discharge products with great contact after being fully discharged distinct different from that at the fresh stage in figure 4(g). It is noteworthy that the evenly deposited film-like discharge products could facilitate the establishment of a good contact interface with the Ni 5 P 4 @NiSe 2 cathode, which can make fully utilized of active centers and maximize the synergistic effect between Ni 5 P 4 , NiSe 2 and the heterogeneous interfaces [7]. At the following recharging in figure 4(i), the film-like discharge products completely disappeared, showing almost the same nanoflower-like morphology as initial appearance, indicting the excellent reversibility of Li-O 2 batteries with Ni 5 P 4 @NiSe 2 cathodes.
Based on the above results, the superior electrocatalytic performance of Ni 5 P 4 @NiSe 2 cathodes would be attributed to their unique architecture. The 3D nanoflower-like structure with self-assembled nanosheets not only promoted the diffusion of O 2 /Li + throughout the cathode, but also provided sufficient active sites for storing the discharge products. Besides, the excellent electrical conductivity of the heterostructure can accelerate the charge transfer during the charge/discharge processes and enhance the electrochemical reaction kinetics [78,79]. More importantly, the unique heterostructure shows a significant effect on the electron redistribution and disordered atomic arrangement, which can provide additional active sites to improve the ORR/OER bifunctional catalytic activity [66]. Additionally, the Ni 5 P 4 and NiSe 2 heterostructure can modulate the growth pathway of Li 2 O 2 and induce their tight coating with low crystallinity structure along the 3D selfassembled nanosheets, building homogeneous low-impedance cathode/Li 2 O 2 interfaces and promoting the efficient formation/decomposition of Li 2 O 2 [80]. The possible formation/ decomposition mechanisms of the Li 2 O 2 on the Ni 5 P 4 @NiSe 2 cathode are shown in figure S11. First, O 2(sol) was adsorbed to the active sites to form adsorbed oxygen (O 2 * , * represents surface adsorbed species) based on equation (1): Second, O 2 * . captured one electron and reacted with Li + to generate LiO 2 * based on equation (2): Third, Li 2 O 2 * was formed by electrochemical reduction of LiO 2 * based on equation (3):

Conclusion
In summary, the nanoflower-like Ni 5 P 4 @NiSe 2 heterostructure was successfully synthesized via hydrothermal method combining simultaneous phosphating/selenization treatment. The 3D hierarchical porous structure of Ni 5 P 4 @NiSe 2 heterostructure can facilitate barrier-free Li + /O 2 transport and provide sufficient specific surface area for the storage of discharge products. Moreover, the unique heterostructure shows a significant effect on the electron redistribution and disordered atomic arrangement, which can provide additional active sites to perfect the ORR/OER bifunctional catalytic activity. The Ni 5 P 4 @NiSe 2 cathode delivered superior electrochemical performance, including an ultra-high discharge/charge specific capacity of 19 090/19 031 mAh g −1 and extended cycling life of 202 cycles at 100 mA g −1 . The above results demonstrate that interfacial electron structure modulation by the construction of the heterogeneous structure with rational architecture design is a promising way of developing highly-efficient bifunctional electrode materials, which can also be expected to be employed in other energy catalytic applications.