High-capacity, fast-charging and long-life magnesium/black phosphorous composite negative electrode for non-aqueous magnesium battery

Abstract

Secondary non-aqueous magnesium-based batteries are a promising candidate for post-lithium-ion battery technologies. However, the uneven Mg plating behavior at the negative electrode leads to high overpotential and short cycle life. Here, to circumvent these issues, we report the preparation of a magnesium/black phosphorus (Mg@BP) composite and its use as a negative electrode for non-aqueous magnesium-based batteries. Via in situ and ex situ physicochemical measurements, we demonstrate that Mg ions are initially intercalated in black phosphorus two-dimensional structures, forming chemically stable Mg_xP intermediates. After the formation of the intermediates, Mg electrodeposition reaction became the predominant. When tested in the asymmetric coin cell configuration, the Mg@BP composite electrode allowed stable stripping/plating performances for 1600 h (800 cycles), a cumulative capacity of 3200 mAh cm⁻², and a Coulombic efficiency of 99.98%. Assembly and testing of the Mg@BP | |nano-CuS coin cell enabled a discharge capacity of 398 mAh g⁻¹ and an average cell discharge potential of about 1.15 V at a specific current of 560 mA g⁻¹ with a low decay rate of 0.016% per cycle for 225 cycles at 25 °C.

Uneven Mg plating behaviour at the negative electrode leads to high plating overpotential and short cycle life. Here, to circumvent these issues, authors report the preparation of a magnesium/black phosphorus composite and its use as a negative electrode for non-aqueous magnesium-based batteries.

Introduction

The recent growth in electric transportation and grid energy storage systems has increased the demand for new battery systems beyond the conventional non-aqueous Li-ion batteries (LIBs)^1,2. Non-aqueous magnesium batteries have emerged as an attractive alternative among “post-lithium-ion batteries” largely due to the intrinsic properties of the magnesium (Mg) negative electrode. Supplementary Table 1 summarizes the physical and electrochemical properties of the Mg negative electrode and other metal negative electrodes. The element Mg is abundant in nature, with a concentration of ~2.0 wt% in the earth’s crust, which is >1000 times that of lithium, making Mg a cost-effective alternative negative electrode. Meanwhile, metal Mg negative electrode exhibit a relatively low reduction potential (–2.37 V vs. SHE) and an appealing volumetric and specific capacity of 3832 mAh cm⁻³ and 2205 mAh g⁻¹, respectively. Additionally, the potentially dendrite-free nature of Mg negative electrode means fewer safety concerns in practical applications^3,4.

Although metal Mg negative electrode has many advantages, its practical application in batteries with liquid non-aqueous electrolyte solutions remains rather limited to date. The bivalent nature of Mg²⁺ results in more than double the charge density of Li⁺ (120 C mm⁻³ for Mg²⁺ vs. 52 C mm⁻³ for Li⁺), which significantly raises the energy barrier required to break the solvation sheath, leading to sluggish migration and diffusion kinetics, as well as severe polarization in Mg batteries^5–7. To address these issues, various non-aqueous liquid electrolyte solutions have been developed and studied^8–14. However, current Mg negative electrode materials, including the metal Mg negative electrode and Mg_xM alloys (where M represents Pb, Ga, Bi, and Sn)^15–18, have generally shown poor compatibility with different kinds of liquid electrolyte solutions. Additionally, some issues such as severe side reactions and large volume changes of the negative electrode still exist, leading to electrolyte solution depletion and structure collapse of the negative electrodes. The above issues result in low Mg utilization efficiency and rapid performance degradation upon prolonged cell cycling¹⁹.

From the perspective of high energy density and cost-effectiveness, direct use of metal magnesium as a negative electrode is regarded as the best choice for rechargeable magnesium batteries (RMBs), but significant technical obstacles remain to be overcome or circumvented. In particular, thin Mg foil (<25 μm) is required to ensure a suitable negative (N)/positive (P) ratio for high-energy RMBs. A metal Mg negative electrode with a thickness of approximately 9.1 μm is demonstrated to be sufficient to meet the area capacity of ~3.5 mAh cm⁻² in practical application²⁰. Unfortunately, the process of rolling ultrathin metal Mg foil is extremely challenging because of the densely packed hexagonal lattice structure of Mg²¹. To circumvent this technical obstacle, one effective strategy is to construct a Mg composite negative electrode by electrodepositing Mg on a substrate with a precisely controllable area capacity. However, this preparation process requires a considerable amount of time, and the deposition capacity on the substrate is limited²².

Another challenge of applying metal Mg negative electrode is that Mg deposits tend to form loose and inhomogeneous morphologies on the metal Mg negative electrode and the commonly used substrates including copper (Cu), particularly at high current densities and high area capacities²³. This unsatisfactory plating behavior results in large volume expansion, progressive side reactions, and the formation of electrochemically inactive “dead Mg” during cycling. As it stands, there is an urgent need to develop a novel Mg composite negative electrode that simultaneously meets the requirements of facile synthesis, uniform deposition behavior, long lifespan, fast-charging properties, and good electrolyte compatibility.

For these reasons, here we report the development of a magnesium@black phosphorous (Mg@BP) composite negative electrode using black phosphorus (BP) as the substrate. The mechanism involved in the construction of the composite negative electrode was thoroughly studied by combining in situ spectroscopies, ex situ TEM, and theoretical simulations and calculations. We demonstrated that Mg²⁺ was first partially intercalated into the crystal structure of BP, resulting in the formation of chemically stable MgxP intermediates with metallic properties (Stage I, limited intercalation). Then metallic Mg was electrochemically deposited on the early-formed MgxP/BP to fabricate the Mg@BP composite negative electrode (Stage II, stable plating). Such magnesium utilization mechanism of BP-based materials deviates from the state-of-the-art theoretical prediction of the magnesium storage in black phosphate-based materials^24–29. Taking advantage of the magnesiophilic property and fast interface transfer properties of the Mg@BP composite negative electrode, it displayed uniform Mg nucleation and deposition behavior rather than inhomogeneous and loose features on the metal Mg negative electrode and other composite negative electrodes^17,22.

When tested in symmetrical cell configuration, the Mg@BP composite negative electrode enabled a cycling life of 1600 h with a cumulative capacity as high as 3200 mAh cm⁻². Even at 16.0 mA cm⁻² with plating capacity of 16.0 mAh cm⁻², the composite negative electrode still maintained stable cyclability for 800 h with nearly 100% Coulombic efficiency (CE). In addition to rapid preparation and fast charging potential, along with precisely adjustable plating/stripping capacity, the Mg@BP composite negative electrode exhibited good electrolyte compatibility, and was successfully paired with a copper sulfide and sulfur positive electrode for the assembly of non-aqueous Mg-based batteries. The assembled Mg@BP | |nano-CuS battery delivered a high specific capacity of 398 mAh g⁻¹ at 560 mA g⁻¹ with a low decay rate of 0.016% per cycle, as well as an initial specific energy of 339 Wh kg⁻¹ calculated based on total mass of cathode and anode materials (including electrochemically inactive components inside).

Results

Composite material preparation and Mg plating/stripping performances

Orthorhombic black phosphorus (BP), one of the most stable allotropes of the element phosphorus, possesses a relatively high electrical conductivity of ~300 S m⁻¹ at 25 °C³⁰ and has been applied and theoretically predicted to be a promising electrode material for advanced energy storage systems^31–33. In view of its appealing physical and chemical properties, BP was employed as a substrate to construct a Mg@BP composite negative electrode using electrochemical deposition method. As shown in Fig. 1a, the Mg@BP composite negative electrode was fabricated by plating Mg inside the BP electrode using the asymmetric BP | |Mg cell (for more detail, see Methods).

Fig. 1. Composite negative electrode preparation method and plating/stripping performances in asymmetric Mg | |BP coin cell configuration.

a Schematic illustration of the fabrication method to prepare Mg@BP composite negative electrode. b SEM images and corresponding EDS mappings of the BP electrode before cycling. c Coulombic efficiencies (CE) of the asymmetric BP | |Mg cells at I_D of 2, 4, 8, and 16 mA cm⁻², respectively. d Voltage profiles at different I_D and Mg deposition overpotential during preparing the Mg-plated (composite) negative electrodes with BP, Cu, and metal Mg substrates, e, Nucleation overpotential. f Overpotential-1/2. The numbers in the legends in (d) and the numbers above the columns in (e, f) showed the applied I_D. The error bar in (e, f) represents standard deviation calculated from three tests conducted under the same condition. All the tests were conducted at 25 °C using an APC electrolyte solution.

Specifically, BP nanosheets were exfoliated from the bulk BP via solvent-assisted ultrasonication³⁴. Compared with bulk BP (Supplementary Fig. 1a, b), the exfoliated BP has a 2D layer-stacked structure (Supplementary Fig. 1c−h) and consists of < 25 phosphorene layers, as determined by atomic force microscopy (AFM) image and height profiles (Supplementary Fig. 1i, for detail discussion, see Supplementary Note 1). High resolution transmission electron microscopy (HRTEM) images revealed a lattice fringe with a d-spacing of 0.26 nm, assignable to the (040) crystal planes (Supplementary Fig. 2). The BP electrode was fabricated by pasting collected BP inks on copper foil. Scanning electron microscopy (SEM) and corresponding energy-dispersive X-ray spectroscopy (EDS) mappings in Fig. 1b and Supplementary Fig. 3 revealed that the fabricated BP electrode had a flatter surface than a polished metal Mg negative electrode. Additionally, freestanding BP electrode (F-BP) without applying a Cu foil could also be prepared in large sizes with good flexibility (Supplementary Fig. 4).

Mg plating/stripping performances of the Mg@BP composite negative electrode were then evaluated in asymmetric BP | |Mg cells and symmetric Mg@BP | |Mg@BP cells. The asymmetric BP | |Mg cells were firstly studied at various current densities (I_D) and area capacities (C_a) with 1 h-plating/1 h-stripping per cycle (t_c-1 h). A stable long cyclic performance was achieved for 1600 h at I_D of 2.0 mA cm⁻² and C_a of 2.0 mAh cm⁻² (Fig. 1c and Supplementary Fig. 5), corresponding to a cumulative capacity as high as 3200 mAh cm⁻². At higher C_a of 4.0, 8.0 and 16.0 mAh cm⁻², the asymmetric cells still delivered cycling lives of nearly 1400, 1200 and 800 h, respectively (Fig. 1c and Supplementary Figs. 6−8). The coulombic efficiencies (CEs) of all these cycles approached to nearly 100%, indicating good Mg plating/stripping reversibility of the BP electrode and the prepared Mg@BP composite negative electrode. The good reversibility of the Mg@BP composite negative electrode was also confirmed by an optical observation. The deposition area in the Mg@BP composite negative electrode turned white after Mg depositing, reflecting the color of metallic Mg deposits (Supplementary Fig. 9a), while the color of the electrode recovered to black without any white residues after Mg stripping (Supplementary Fig. 9b). To determine whether Cu foil has an effect on CE, we also recorded the cycle performance and CE of the F-BP electrode, which showed a high CE of > 95% after 5 cycles stabilization (Supplementary Fig. 10a), indicating no apparent influence of Cu foil on the electrochemical performance. In a comparison, the Mg plating/stripping performances of commonly used substrates such as Cu, Al, Mo or carbon fabrics displayed significantly lower CEs, less than 80% during Mg plating and stripping cycling (Supplementary Fig. 10b-f, for detail discussion, see Supplementary Note 2).

The BP electrode in the asymmetric Mg cell configuration can deposit up to 80 mAh cm⁻² of Mg at I_D of 8.0 mA cm⁻² with a cycle life of 1100 h (Supplementary Fig. 11a), fully demonstrating its Mg storage capability. Ex situ cross-section SEM image and its corresponding EDS mappings revealed that at this deposition capacity, the Mg@BP composite negative electrode still exhibited dense Mg plated structure after long-time cycling, which stood contrast to the porous stacking structure of the pristine BP electrode (Supplementary Fig. 11b, c).

The plating voltage profiles and corresponding deposition overpotentials (nucleation overpotential and half deposition process-overpotential) of the Mg@BP composite negative electrode were smaller than that of pristine Mg and Cu foils, as summarized in Fig. 1d−f. Furthermore, the BP | |Mg cell showed higher average CE of 98.5% than that of only 81.6% of the Cu | |Mg cell (calculated based the results of Fig. 2a, b, respectively), as well as the highest exchange current density (0.95 mA cm⁻² vs. 0.71 mA cm⁻² of Cu | |Mg cell and 0.56 mA cm⁻² of Mg | |Mg cell, Fig. 2c). These results preliminarily revealed the stable nucleation/plating process as well as the a fast charge transfer kinetics³⁵ of the BP substrate than other substrates for preparing Mg plated (composite) negative electrodes.

Fig. 2. Evaluation of the Mg plating and stripping capability of the Mg@BP composite negative electrode.

Voltage profile to calculate average CE (a) in the asymmetric BP | |Mg cell and (b) in the asymmetric Cu | |Mg cell. c Polarization profiles of asymmetric BP | |Mg, Cu | |Mg, and symmetric Mg | |Mg batteries to prepare the composite negative electrode. Rate performances of the asymmetric BP | |Mg cell to prepare Mg@BP composite negative electrode (d) at I_D from 2.0 to 20.0 mA cm⁻² with fixed t_c of 1 h and (e) at fixed I_D of 4.0 mA cm⁻² with gradually increased t_c. f Cycling performances of the symmetric Mg@BP | |Mg@BP compared with symmetric Mg | |Mg cell. g Radar plot to summarize Mg plated capability of Mg@BP composite negative electrode from the perspective of Mg plating capacity, I_D, cycling life, initial planting overpotential (Over-V), and CE. Same colored lines represent the same independent set of experiment results. h Comparison of Mg plated capability of the Mg@BP composite negative electrode with current Mg composite negative electrode^20,38–42 and Li composite negative electrode^{11,39,43–49}. The number above and below each column represents the accurate cycling life (h) and the applied I_D (mA cm⁻²), respectively. All the tests were conducted at 25 °C using an APC electrolyte solution.

The rate performance of the Mg@BP composite negative electrode in asymmetric BP | |Mg cells was further studied by two test methods. The first one was performed at a fixed t_c (1 h) with gradually increased I_D. The other test was conducted at a fixed I_D (4.0 mA cm⁻²) with gradually increasing t_c. The results showed that the Mg@BP composite negative electrode maintained stable plating/stripping cycles until I_D increasing to 20.0 mA cm⁻² (Fig. 2d) and Mg plating C_a increasing to 36.0 mAh cm⁻², respectively (Fig. 2e). These values are higher than the reported substrates to prepare composite metal negative electrodes via the electrodeposition method^22,36,37. Careful examination of the voltage profiles in these rate tests revealed that short circuits occurred when Mg²⁺ was stripped from the Mg@BP composite negative electrode and plated onto the metal Mg counter electrode (Fig. 2d, e), demonstrating improved stability of the Mg@BP composite negative electrode than Mg counter electrode.

Unlike alkali metal ion batteries, very few Mg-rich positive electrode materials of RMBs were developed so far, so the negative electrode materials must be in Mg-rich states. The electrochemical performances of symmetric Mg@BP | |Mg@BP cell were further investigated with C_a of 1 mAh cm⁻² and 4 mAh cm⁻², respectively. The depth of discharge (DOD) was set as 90%, meanings 90% of pre-plated Mg was stripped and plated during cycling. As shown in Fig. 2f and Supplementary Fig. 12, the symmetric Mg@BP | |Mg@BP cell with C_a of 1 mAh cm⁻² can stably cycled for 1000 h with a small polarization voltage of 101 mV. In contrast, the symmetric Mg | |Mg cell exhibited noisy voltage profiles after 210 h cycling. Even with high C_a of 4 mAh cm⁻² and I_D of 3.6 mA cm⁻², the symmetric Mg@BP | |Mg@BP cell still exhibited stable life-span of 200 h, which surpassed that of the symmetric Mg | |Mg cell suffering a sudden short circuit after 81 h (Supplementary Fig. 13). Additionally, the initial and cyclic plating overpotentials of the Mg@BP negative electrode in this test were smaller than that of the pristine Mg negative electrode: 0.44 V for the Mg@BP negative electrode while 1.87 V for the pristine Mg negative electrode.

To sum up, the as-prepared Mg@BP composite negative electrode enable efficient Mg plating performances with low Mg nucleation and plating overpotentials, nearly 100% CEs, as well as long cycling life under high I_D and C_a, as summarized in the radar plot in Fig. 2g. These performances are above the state-of-the-art for most of the Mg composite negative electrodes^20,38–42 and Li composite negative electrodes^{11,39,43–49} reported in the literature (Fig. 2h and Supplementary Table 2).

Mg plating/stripping behaviors and charge transfer kinetics

To identify the critical factors of the electrochemical performances, Mg plating/stripping behaviors of Mg@BP composite negative electrode were systematically studied via structural analysis and theoretical simulations. As schemed in Supplementary Fig. 14a, Mg was initially deposited on the BP nanosheets and then filled the space between the nanosheets as the deposition capacity increased. To be more specific, the nanosheets became obviously thicker with a C_a of 3 mAh cm^‒2 (Fig. 3a and Supplementary Fig. 14b−d) than that of the pristine BP nanosheets (Supplementary Fig. 14e, f). When Mg deposition capacity increased to 6 mAh cm^‒2, the interlayer space between BP nanosheets was reduced by filling with the Mg (Supplementary Fig. 14g−i). Further increasing the deposition capacity to 9 mAh cm^‒2, the cross-section structure of the obtained Mg@BP composite negative electrode became dense, but the layered structure of BP could still be recognized as compared with the pristine BP structure (Fig. 3b and Supplementary Fig. 14g−I). This Mg deposition behavior could be identified by the conformal element P and Mg distribution on the BP nanosheets after Mg depositing (Supplementary Fig. 15). The dense-packed Mg deposition behavior could also be found in the Mg@F-BP composite negative electrode (Supplementary Fig. 16).

Fig. 3. Mg plating behavior and the stability of Mg@BP composite negative electrode.

Cross-section morphologies of Mg@BP composite negative electrode with plated capacity of (a) 3 mAh cm⁻² and (b) 9 mAh cm⁻². c Comparison of the surface morphology of (c) Mg@BP composite negative electrode and d, metal Mg negative electrode after plating Mg with C_a of 3 mAh cm⁻². e Schematic illustration of Mg plating behavior of Mg@BP composite negative electrode. Theoretical simulations of Mg plating behaviors of (f) Mg@BP composite negative electrode and (g) Mg@Cu composite negative electrode. Nyquist profiles of the symmetric Mg@BP | |Mg@BP cell and Mg@F-BP | |Mg@F-BP cell (h) before and (i) after 5 cycles. Comparison of R_s and R_ct values of Mg@substrate composite negative electrodes (j) before and (k) after cycling. The number above the columns represents the fitting value of R_s and R_ct. All the tests were conducted at 25 °C using an APC electrolyte solution.

Top-view SEM results can allow to distinguish the difference of Mg deposits with different substrates. Uniform Mg deposits with a flat surface can be observed on the surface of Mg@BP composite negative electrode with plated capacity of 3.0 mAh cm⁻², revealing homogenous mass transport and good magnesiophilic of BP (Fig. 3c, Supplementary Fig. 17a, b). However, for the commonly used Cu and pristine Mg substrates, spherical Mg deposits were observed on their surfaces (Fig. 3d, Supplementary Fig. 17c−f, Supplementary Fig. 18); further increased the plating capacity to 5.0 mAh cm⁻², random Mg spherical electrodepositions coalesced with each other into a loose structure and detached from Cu and metal Mg substrates (Supplementary Fig. 19a−j) while Mg@BP composite negative electrode still kept dense and uniform deposition surface (Supplementary Fig. 19k−o). To compare the Mg deposition affinity of the BP nanosheets and carbon-based materials in one electrode, control Mo, Al, and BP-carbon electrode was prepared to confirm the Mg deposition affinity of these substrates. The deposition results showed that uneven Mg deposition and agglomeration were found on the Mo, Al or carbon-based substrates (Supplementary Figs. 20−23, for detail discussion, see Supplementary Note 3−6). Electrochemically inactive electrode regions and detrimental side reactions can be triggered in these situations, leading to low CE and short cycling life. Therefore, BP nanosheets are considered as a substrate for Mg deposition, while the interlayer space between BP sheets is functioned as a plating space until it is completely filled. As schemed in Fig. 3e, this Mg plating behaviors can avoid the dead Mg formation in the Mg@BP composite negative electrode and effectively relieve the volume expansion during cycles, which is the critical reason for its stable long-time cycling performances.

Theoretical simulation was carried out to investigate the deposition process. The simulated Mg deposition contours were consistent with the SEM observations (Fig. 3f, g and Supplementary Fig. 24). It showed that BP triggered a uniform Mg deposition with moderate interface current density distribution (Fig. 3f and Supplementary Fig. 24a−d). However, at the interfaces of the Cu and Mg electrodes, the current densities were unevenly distributed, resulting in loose semi-spherical and dead Mg deposition (Fig. 3g and Supplementary Fig. 24e−l).

The charge transfer kinetics of different Mg negative electrodes were further studied by electrochemical impedance spectroscopy (EIS). By fitting the EIS raw data (reported in Nyquist plots) with equivalent circuit models (Supplementary Fig. 25), Mg@BP and Mg@F-BP composite negative electrodes displayed similar solid ohmic resistance (R_s) and interfacial charge transfer resistance (R_ct) values before or after cycling (shown in Fig. 3h, i). This result support the claims again that the conductivity and charge transfer properties of BP electrode are not affected by the Cu foil. R_s and R_ct values of the studied composite negative electrode in this work were summarized in Fig. 3j, k and Supplementary Table 3. The R_ct values of BP based composite negative electrodes were reduced by one to two orders of magnitude than those of Mg@Cu, Mg@Mo, Mg@Al and Mg@C composite negative electrodes, which support the claims of improved charge transfer kinetics at the interface of the BP-based electrode.

Fundamental investigations on the composite negative electrode formation and utilization mechanism

The initial Mg plating mechanism of BP to form the Mg@BP composite negative electrode was investigated by in situ/ex situ measurements and theoretical calculations. In the in situ XRD and in situ Raman experiments, Mg was plated onto the BP electrode at a low I_D of 3.0 mA g⁻¹ to examine the reaction process. The set I_D and the achieved specific capacity were all based on the mass of BP in the electrode during the in situ investigations. The analysis of the voltage profile in Fig. 4a allows to distinguish the formation process into two stages: the first stage is the intercalation process, which occurs at the voltage profile above 0 V (labeled Stage I). This is followed by the Mg nucleation and deposition process (labeled Stage II), which takes place at the voltage profile below 0 V. During the in situ XRD experiment, the capacity contribution of Stage I and Stage II were calculated to be 2.4 and 39.9 mAh g⁻¹ based on the mass of BP in the electrode, respectively.

Fig. 4. In/ex situ experiments to discover the constructing process of the Mg@BP composite negative electrode.

a In situ XRD detection. b In situ confocal Raman spectra. Ex situ TEM images of BP nanosheets at (c) original state, d Stage (I), and (e) Stage II. f Schematic illustration of the construction process of the Mg@BP composite negative electrode. The electrochemical tests were conducted at 25 °C using an APC electrolyte solution.

In situ X-ray diffraction (XRD) detected the presence of magnesium phosphide (MgxP) species at around 4.2° during the entire Mg plating process, while the signal between 16.0° and 16.4° representing the metallic Mg deposition became much stronger in Stage II (Fig. 4a)³². It is inferred that Mg could be intercalated into BP to form MgxP phosphide intermediates in Stage I, which remained stable during the entire Mg plating process. The constructing process of the composite negative electrode was also analyzed by in situ Raman spectra, where the electrodeposited Mg phase was observed at 100−125 cm⁻¹ in Stage II (Fig. 4b and Supplementary Fig. 26)^50,51. This was in good agreement with the in situ XRD result and suggests that the Mg nucleation and deposition mainly occur in Stage II. The characteristic peaks of BP (300−500 cm⁻¹), including $A_1^{\mathrm{g}}$, B_2g, and $A_2^{\mathrm{g}}$ peaks^52,53, remained consistent throughout Stage I and Stage II. At the same time, Raman images also captured the Mg phase accumulation process in the Stage II, as evidenced by the increasing image signals (Supplementary Fig. 27).

Ex situ XPS and TEM were further employed to investigate the structural evolution of the BP nanosheets during the Mg@BP composite negative electrode formation process. XPS investigations analyzed the changes of chemical environment during formation of the Mg@BP composite negative electrode. According to the photoelectron emission spectrum of P 2p in Supplementary Fig. 28, phosphide intermediates could be detected at Stage I and remained stable at the Stage II, even after Mg stripping. These results further confirmed that the MgxP intermediates formed via intercalation reaction in Stage I are electrochemically stable during the stripping process, which can further promote the reversible Mg deposition/stripping behaviors during cycling (For detail discussion, see Supplementary Note 7). TEM images revealed that compared with the exposed lattice fringe of the original BP (0.26 nm, as shown in Fig. 4c), nanodots are formed on the BP matrix in Stage I, exhibiting expanded lattice fringes of 0.28 and 0.29 nm (Fig. 4d). This can be assigned to the formation of metal phosphides (MgxP) via Mg²⁺ intercalation reaction. The nanodot intermediates persisted throughout the Mg deposition process (Supplementary Fig. 29), which confirms the stability of the MgxP intermediates, consistent with the above results. Subsequently, the in situ formed MgxP/BP can ensure a stable Mg deposition, as evidenced by the exposed (100) lattice plane of Mg in the HRTEM image of the Mg@BP composite negative electrode (Fig. 4e)⁵⁴.

Based on the above studies, a two-stage process of constructing the Mg@BP composite negative electrode is proposed, as schematically illustrated in Fig. 4f. In the Stage I-limited intercalation process, Mg²⁺ intercalation occurs and forms stable phosphide intermediates on the BP nanosheet matrix to produce a MgxP/BP co-matrix. This process provides a limit capacity contribution of 2.4 mAh g⁻¹. Stage II-metallic Mg plating process, Mg is electrodeposited on the above MgxP/BP matrix with a controllable depositing capacity. Considering the negligible capacity contribution of Stage I and the effective Mg electrodeposition performance, it is more viable for BP to construct a substantial Mg@BP composite negative electrode rather than work as a Mg²⁺ storage material in RMBs. The chemical process of forming the Mg@BP composite negative electrode is different from the theoretical predictions regarding the potential of BP materials to function as an intercalation negative electrode in RMBs^24–29.

The properties of the magnesium phosphide intermediates formed during the initial Mg plating process were further studied by density functional theory (DFT) calculations. Previous studies have emphasized that no passivation layers are produced on the surface of Mg with the all-phenyl complex (APC) electrolyte solutions^6,8,55. Hence, the chemical compositions of phosphide intermediates can be precisely identified by inductively coupled plasma-optical emission spectrometry (ICP-OES). According to the determined molar ratio of Mg and P, the intermediates at the end of Stage I are likely in the form of Mg_0.06P (Supplementary Table 4). Representative intermediates of Mg_0.03P, Mg_0.06P and Mg_0.12P were theoretically studied to further elucidate the Mg utilization mechanism. The negative formation energies of these species indicate the Mg²⁺ insertion process is energetically favorable (Fig. 5a). Corresponding optimized structures are showed in the insets of Fig. 4a. Density of states (DOS) of the pristine BP show typical semiconducting properties with a band gap of 0.068 eV (Fig. 5b). After Mg²⁺ insertion, the Fermi level of Mg_0.06P (Fig. 5c) shifted to the conduction band, indicating a transition from semiconductor BP to metallic compound, which is beneficial for electron conduction in the electrodeposition process. Combined with the high Mg²⁺ adsorption energy and mild Mg²⁺ diffusion energy barrier (Fig. 5d, e), the Mg²⁺ insertion reaction tends to cutoff when nominated Mg_0.06P intermediate forms, followed by steady Mg²⁺ electrodeposition.

Fig. 5. Theoretical study of the nominated intermediates of Mg_0.03P, Mg_0.06P, and Mg_0.12P.

a Formation energy (Form. Energy) of the intermediates. The red spheres represent the P atom and the blue spheres represent the P atoms. DOS of (b) pristine BP and (c) Mg_0.03P. d Mg²⁺ adsorption energy (Adsor. Energy) of the intermediates. e Mg²⁺ diffusion energy barrier of the intermediates.

Then, series of ex situ XRD and SEM postmortem analyses further revealed the stability of BP-based electrodes. As ex situ XRD patterns shown in Supplementary Fig. 30, there was a presence of unreacted BP residue during preparing the composite negative electrode and cycling process. The metallic Mg phase was detected in the Mg@BP composite negative electrode and undetected after the stripping process, implying adequate Mg plating and stripping reversibility of the Mg@BP composite negative electrode. The phases of unreacted BP and Cu current collector in the Mg@BP composite negative electrode remained unchanged after first stripping and long-time cycling and no other phase was detected. Because of the negligible content of phosphide intermediates in the composite negative electrode and the open testing condition, no corresponding phases were detected in the ex situ XRD results.

The cycling and chemical stability of the Mg@BP composite negative electrode were identified by the postmortem analysis within two asymmetric BP | |Mg cells after long-time cycling. Firstly, the electrochemical cycling performances of these two cells exhibited well consistency as shown in the voltage profiles in Supplementary Fig. 31a, b. The cycled cells were disassembled and observed at Mg plated and stripped state, respectively. Even after long-time cycling, Mg still could be uniformly plated on the BP nanosheets and completely stripped as evidenced by the dense deposition morphology at plated state (Supplementary Fig. 31c) and the clean BP nanosheets at stripped state (Supplementary Fig. 31d).

Mg@BP | |CuS lab-scale cells assembly and testing

Non-aqueous Mg batteries were assembled and tested to estimate the practical application potential of the constructed Mg@BP composite negative electrode after pairing with a suitable positive electrode, as shown in Fig. 6a. Although Cheverel Mo₆S₈ remains by far the best known positive electrode material for RMBs, its relatively low theoretical capacity of 110 mAh g⁻¹ make it hard to narrow the energy density gap between RMBs and LIBs⁵⁶. Copper sulfide (CuS) was considered as a promising conversion positive electrode due to its high theoretical capacity of 560 mAh g⁻¹, together with the advantages of high electronic conductivity (10³ S cm⁻¹ at 25 °C) and abundant resources⁵⁷. However, the polarization and Coulombic resistance of Mg²⁺ ions resulted in sluggish reaction kinetics and low redox reversibility in bulk CuS^58,59. It was suggested that a rationally controlling the size of the electrode material, could not only enhance the ingress/egress of Mg²⁺ ions but also shorten ion diffusion lengths during cycling^54,60,61.

Fig. 6. Electrochemical testing of the Mg@BP | |nano-CuS battery.

a Schematic model of the Mg@BP | |nano-CuS cell. b CV profiles. c Rate performances. d Voltage profiles of the 1st, 2nd, 3rd, and 5th cycles during cycling in (e). e Cycling performances at specific current of 560 mA g⁻¹. The insets in (c, e) are the Coulombic efficiencies (CE) during cycling. All the tests were conducted at 25 °C using an APC electrolyte solution.

Accordingly, a nano-size CuS (nano-CuS) was synthesized here using a modified reflux and hydrothermal sulfurization method (see Methods)^62,63 to assemble the Mg@BP | |nano-CuS battery, as shown in Fig. 6a. The synthesized nano-CuS exhibited a uniform secondary particle diameter of about 300 nm, which is up to ten times smaller than that of commercial micro-size CuS (micro-CuS), as shown in Supplementary Fig. 32a−c. XRD patterns of both our nano-CuS and commercial micro-CuS revealed the typical covellite CuS diffraction peaks, matching well with the JCPDF#79-2321 pattern (Supplementary Fig. 32d). TEM and HR-TEM images in Supplementary Fig. 32e, f clearly reveal that the nano-CuS clusters are composed of numerous ultrafine nanocrystals which can effectively shorten the ion diffusion pathway and improve the Mg²⁺ transport kinetics.

Cyclic voltammetry measurements of the Mg@BP | |micro-CuS and Mg@BP | |nano-CuS cells were carried out. With the micro-CuS positive electrode, it presented only one reduction peak and one oxidation peak, with decreasing peak currents during cycling, suggesting an insufficient conversion reaction and weak reaction reversibility (Supplementary Fig. 33a). In sharp contrast, the CV curve of the nano-CuS (Fig. 6b) displayed three distinct reduction peaks located at 1.45, 1.08, and 0.63 V and one oxidation peak located at 1.91 V⁶⁴. The overlapping voltage profiles indicated the good reversibility of the nano-CuS positive electrode during cycling. Meanwhile, the batteries with the nano-CuS positive electrode also delivered improved specific capacity and CEs, as well as better cycling performances than those with the micro-CuS positive electrode (Supplementary Fig. 33b, c), revealing the faster reaction kinetics of the nano-CuS material. Moreover, the electrochemical performances of the magnesium batteries with the Mg@BP-based negative electrodes are better than those equipped with Mg metal negative electrode.

The ratio of negative to positive electrodes (N/P ratio) is a crucial parameter of the battery design, and is related to the discharge/charge capability, energy density, and cycling lifespan. According to the practical design principles of Li metal batteries, the N/P ratio should be set within the range of 1 < N/P ratio < 2 with conversion positive electrodes^65,66. Therefore, in this study, we set the N/P ratio of the Mg@BP | |nano-CuS batteries at 1.5 to ensure an adequate amount of Mg in the negative electrode side and further evaluate the practical application feasibility of the Mg@BP composite negative electrode. As shown in Fig. 6c, the Mg@BP | |nano-CuS battery delivered high discharge capacities of 439, 328, 258, 226 and 173 mAh g⁻¹ at specific currents of 100, 200, 500, 1000 and 2000 mA g⁻¹, respectively. More importantly, all capacity could be recovered when specific current was reverted to 100 mA g⁻¹, indicating the good working tolerance of different kinds of working conditions.

A coin-type Mg@BP | |nano-CuS battery with nano-CuS loading of 5.8 mg cm⁻² delivered a high initial capacity of 398 mAh g⁻¹ at specific current of 560 mA g⁻¹ (refers to 1 C rate). As shown in Fig. 6d, the voltage profiles of the 1^st, 2^nd, 3^rd, and 5^th cycle overlapped well, which demonstrated good reversibility of the magnesium battery. After 200 cycles, the battery was still able to deliver a capacity of 385 mAh g⁻¹, corresponding to a capacity decay rate of 0.016% per cycle (Fig. 6e). To more accurately assess the specific energy of the battery, we applied a nano-CuS positive electrode with a stainless-steel current collector in the Mg@BP | |nano-CuS battery and tested its electrochemical performance. As shown in Supplementary Fig. 34, the Mg@BP | |nano-CuS battery delivered a specific capacity of 552 mAh g⁻¹ in the first activation cycle, which corresponds to a calculated specific energy of 339 Wh kg⁻¹ based on the total mass of the positive electrode and negative electrode materials and an average cell discharge potential of about 1.15 V of the battery. The calculated specific energy of 339 Wh kg⁻¹ of the lab-scale Mg@BP | |nano-CuS cell is well-positioned when compared to other aqueous and non-aqueous lab-scale battery systems reported in the literature (Supplementary Fig. 35a)^67–74.

According to the above discussion, the Mg@BP | |nano-CuS battery exploited the advantages of both the Mg@BP composite negative electrode and the nano-CuS positive electrode. As depicted in Supplementary Fig. 35b, the Mg@BP composite negative electrode exhibited fast charge transfer kinetics and magnesiophilic electro-plating/stripping behaviors. On the positive electrode side, the nano-CuS crystal provided a short ion diffusion path and avoided the conventional activation/pulverization process during cycling. All these advantages combined to ensure Mg@BP | |nano-CuS batteries with high capacity and stable cycling life.

The Mg@BP | |nano-CuS battery was also assembled in single-layer pouch cell configuration. A soft packaged BP | |Mg cell was first applied to prepare a Mg@BP negative electrode (Supplementary Fig. 36a). As shown in Supplementary Fig. 36b, large area BP negative electrode still kept low nucleation potential and stable plating process, sufficiently to construct a large area Mg@BP composite negative electrode. Single layer Mg@BP | |nano-CuS pouch cells were then assembled with 3 cm × 2.5 cm electrodes. The Mg@BP | |nano-CuS pouch cell was able to deliver a total capacity of 372 mAh before failure (Supplementary Fig. 36c, d). Two pouch cells in series connection with an open circuit voltage of 3.6 V were able to stably power a string of colored light emitting diodes for 2 h (Supplementary Fig. 36e and Supplementary Movie 1).

We further assessed the compatibility of BP substrate and the Mg@BP composite negative electrode using a borate-based non-aqueous electrolyte solution⁷⁵. As demonstrated in Supplementary Fig. 37, Mg can be reversibly plated and stripped on the BP electrode with CEs of > 95% during cycling. Subsequently, the achieved Mg@BP composite negative electrode was paired with a sulfur (S) positive electrode to assemble the Mg@BP | | S coin cell. The S content in the positive electrode was about 63.8% according to the thermogravimetric measurements result of the S positive electrode (Supplementary Fig. 38). Electrochemical performances (Supplementary Fig. 39) showed that the Mg@BP | | S battery was able to deliver an initial specific capacity of 627 mAh g⁻¹ at specific current of 840 mA g^–1.

Discussion

Black phosphorus (BP) nanosheets were employed to prepare Mg@BP composite negative electrode. The preparation mechanism of the composite negative electrode was investigated with state-of-the-art techniques, experimentally and theoretically. It was demonstrated that Mg²⁺ was partially inserted into the BP to form MgxP intermediates in the early stage (denoted as limited intercalation). Subsequently, Mg was densely and uniformly plated on the MgxP/BP co-substrate, completely different from the prediction about the Mg²⁺ storage chemistry of BP. The limited intercalation process triggered a transition from a semiconductor BP to a metallic compound, endowing the Mg@BP negative electrode with magnesiophilic and fast charge transfer properties, thus facilitating uniform Mg nucleation and growth behavior that are difficult to achieve on the pristine metal Mg negative electrode and other composite negative electrodes.

The designed Mg@BP composite negative electrode was able to deliver stable Mg plating and stripping performance for 1600 h with a cumulative capacity as high as 3200 mAh cm⁻², and about 800 h stable cycling even at 16.0 mA cm⁻² and 16.0 mAh cm⁻² with nearly 100% CE. In addition, the Mg@BP composite negative electrode exhibited good electrolyte compatibility, and non-aqueous magnesium battery in combination with a nano-CuS positive electrode at a low N/P ratio of 1.5 enabled a calculated specific energy of 339 Wh kg⁻¹ (for the first cycle at 25 °C) and a low capacity decay rate of 0.014% during cycling.

Methods

Materials and electrodes preparation

Preparation of the black phosphorus (BP) slurry

To synthesize the BP nanosheets, bulk BP (50 mg, 99.9%, Zhongke Scientific and Technical Co., Ltd.) was exfoliated in a N-methyl-2-pyrrolidone (NMP, Aladdin, 50 mL) solution under ultra-sonication (Elmasonic, 300 W) for 12 h. The entire exfoliation process was performed under argon atmosphere and the temperature was maintained below 15 °C. After that, the as-formed dispersion was centrifuged at 7000 rpm and washed with acetone (Aladdin) and ethanol (Aladdin) several times in air. The BP nanosheets were collected after drying in a vacuum oven for 12 h at 45 °C. 90 mg BP nanosheets were fully milled using planetary ball mill (Kejing, SFM-1) for 6 h with 10 mg polyvinylidene difluoride (PVDF) in 0.5 mL NMP to form homogenous BP slurry. The milling process took place in an agate mortar with 2 mm diameter agate balls, without any protective atmosphere.

Preparation of the BP-based electrode

BP electrode

The BP slurry was pasted onto Cu current collectors (Canrud, 99.99%, thickness of 12 μm) and then dried in a vacuum oven at 80 °C for 24 h to get the BP electrode mainly discussed in this study. The mass loading of BP on the current collectors was estimated to ~0.5 mg cm⁻². The thickness of the BP layer is controlled by the casting amount of the slurry which were controlled as about 11.5 μm in this work expect for the large capacity deposition test (80 mAh cm⁻², Supplementary Fig. 11).

Freestanding BP electrode (denoted as F-BP electrode)

The BP slurry was pasted on to a PTFE film (Canrud, thickness of 500 μm) and then peeled after being fully dried to form the freestanding electrode.

BP/carbon electrode

A control BP-carbon slurry (mass ratio of BP: PVDF: carbon black = 8:1:1) was pasted on the Cu current collector and fully dried to get the BP/carbon electrode. Conductive carbon black (Canrud, particle size of 100‒200 nm) were chosen as the carbon material for preparing this electrode.

Preparation of the Mg composite negative electrodes

A series of asymmetric substrates | |Mg cells (the substrates refer to the BP electrode, F-BP electrode, Cu, Mo, Al, and carbon fabric) were assembled to prepare the Mg composite negative electrodes by electrochemical deposition method, as scheme illustrated in Fig. 1a. Mg foil (purity: 99.99%, thickness: 100 μm), Mo foil (purity: 99.95%, thickness: 100 μm), Al foil (purity: 99.98%, thickness: 100 μm), and carbon fabric (purity: 99%, thickness: 100 μm) were all purchased from Canrd New Energy Technology Co., Ltd., China.

Mg@BP-based composite negative electrode

Mg was plated onto the prepared BP-based electrodes in the BP-based electrode | |Mg cell at precise current density and area capacity. After deposition process, the Mg@BP-based composite negative electrode could be easily got by disassembling the asymmetric cell. The obtained Mg composite negative electrodes were named the Mg@BP and Mg@F-BP composite negative electrode.

Other control composite negative electrode

With the same method, Mg was also plated onto the commonly used current collectors of Cu, Mo, Al, and carbon fabric (C) to construct the composite Mg negative electrodes in corresponding symmetric cells. The achieved composite negative electrodes were named as Mg@Cu, Mg@Mo, Mg@Al and Mg@C composite negative electrode, respectively.

Synthesis of nano-CuS

Nano-CuS was prepared according to the literature^63,64. Ammonium hydroxide (Aladdin, 5.0 mL, 25 wt% in water) solution was added to copper acetate (Aladdin, 99.99%, 0.8 g) in ethanol (Aladdin, 99.5%, 50 mL) solution. After fully stirring, the solution was refluxed at 100 °C for 3 h. The as-prepared precursor was washed with ethanol three times, and dried at 60 °C for 12 h. The powder was dispersed in a mixed solvent of deionized (DI) water (18.2 MΩ∙cm) and ethylene glycol (Aladdin, 40 mL, 3:1, v:v) solution with cetyltrimethylammonium bromide (Sigma-Aldrich, 95%, 0.2 g) and thiourea (0.5 g) while stirring for 1 h. Then, the homogenous solution was transferred into a Teflon-lined stainless-steel autoclave (50 mL) and heated to 160 °C for 3 h. To collect nano-CuS, the precipitate was washed with water and ethanol for at least 5 times and then dried at 60 °C under vacuum overnight.

Preparation of nano-CuS and micro-CuS positive electrode

Nano-CuS positive electrode slurry was prepared by mixing the as-prepared nano-CuS, carbon black, and PVDF with a mass ratio of 8:1:1 in a NMP solvent. After the slurry was fully milled, the slurry was pasted onto a copper foil and dried at 80 °C overnight to get the nano-CuS positive electrode film. The micro-CuS positive electrode was prepared using same procedure, expect the nano-CuS active material was replaced with micro-CuS active material. The micro-CuS positive electrode material was a commercial micro-CuS (99.0%) which was purchased from Macklin with average diameter of about 11 µm and directly used for comparison. The positive electrode films were cut into discs with a diameter of 12 mm for further coin magnesium battery assembling.

Preparation of the S positive electrode

The sulfur and carbon black (mass ratio, 8:2) were mixed in a CS₂ solution to ensure the uniform loading of sulfur inside the carbon black. After CS₂ fully evaporating in the chemical hood, the above mixture were heated at 155 °C for 12 h with Ar atmosphere protection. The obtained S/C active material was then hand-milled with carbon black and PVDF at a mass ratio of 8:1:1 in the NMP solution within an agate mortar to get the positive electrode slurry. The positive electrode slurry was pasted onto the carbon fabric and dried at 45 °C in a vacuum oven for 24 h to get the S positive electrode film. The S positive electrode film was cut into discs with a diameter of 12 mm for further coin magnesium battery assembling. The sulfur mass loading was calculated according to the TGA test, to be about 1.6 mg cm⁻².

Electrochemical measurements

Cells assembly

All coin cells (CR2032) were assembled in an argon-filled glove box with oxygen and H₂O content of <0.1 ppm. A glass fiber separator (Whatman GF/A, thickness of 50 μm, average pore size of 1.6 μm) and all-phenyl complex (APC electrolyte, Suzhou DoDoChem. Co., Ltd., China, 0.4 M PhMgCl in tetrahydrofuran with 0.25 M AlCl₃) were mainly used for the cell assembly in this work. For the borate-based electrolyte solution was prepared by dissolving tris (hexafluoroisopropyl) borate [B(HFP)₃, TCI, 95%] and MgCl₂ (Macklin, 99.99%) with different molar ratios in pure tetraglyme (Aladdin, 99%). Mg foil was cut into ribbons (1 mm × 10 mm) and excessively added into the above solution and stirred for 48 h at 25 °C. After that, the unreacted Mg ribbons were filtered and clear borate-based electrolyte solution was obtained⁷⁵. The water content in the electrolytes was <10 ppm. The electrolyte volume was uniformly controlled at 200 microliters for the cell assembly in this work.

Asymmetric Mg cells assembly

Asymmetric cells were assembled by sandwiching the glass fiber separator (GF) between the studied substrates and metal Mg counter electrode with 100 μL electrolyte to examine the Mg deposition and stripping performances, CEs, rate performances, and cycling stability. The studied substrates include BP-based electrodes (BP, BP/C, and F-BP) and some compared substrates (Cu, Mo, Al, or carbon fabric) were included in the asymmetric coin cells to construct the Mg composite negative electrodes. The APC electrolyte solution was used in the cells unless the specified. The soft-packaged BP | |Mg cell was assembled by stacking BP electrode, GF separator, and Mg electrode layer by layer with 500 μL APC electrolyte, and then were sealed by PE film.

Symmetric Mg coin cells assembly

Symmetric Mg coin cells were assembled by sandwiching the GF separator between two Mg electrodes with 100 μL electrolyte. Similarly, the symmetric coin cells with the composite negative electrode were assembled by sandwiching the GF separator between two Mg composite negative electrodes with 100 μL electrolyte. The symmetric batteries were assembled and tested to study the Mg deposition and stripping performances and Electrochemical impedance spectroscopy (EIS) measurements. The APC electrolyte solution was used in the cells unless the specified.

Mg@BP | |CuS battery assembly

Mg@BP | |nano-CuS coin cell was assembled by sandwiching a GF separator between the as-prepared Mg@BP negative electrode and nano-CuS positive electrode with 100 μL APC electrolyte. The Mg@BP | |micro-CuS, Mg | |nano-CuS coin cell, and Mg | |micro-CuS coin cell were assembled using same procedure by replacing corresponding positive electrode and negative electrode in the battery setup. The soft-packaged nano-CuS | |Mg@BP battery was assembled by stacking nano-CuS positive electrode, GF separator, and Mg@BP negative electrode layer by layer with 500 μL APC electrolyte, and then were sealed by PE film.

Mg@BP | | S battery assembly

Mg@BP | | S coin cell was assembled by sandwiching a GF separator between the as-prepared Mg@BP negative electrode and S positive electrode with 100 μL borated-based electrolyte.

Electrochemical measurements and calculations

The cycling and rate performances of the asymmetric cells were tested on a Neware battery testing system by discharging at I_D and C_c, and then charging with a cutoff voltage of 1.0 V. The polarization profiles of the asymmetric BP | |Mg, Cu | |Mg and Mg | |Mg cells were obtained by linear sweep voltammetry (LSV) tests at a sweep rate of 1.0 mV s⁻¹ from −100 − 100 mV. The average Coulombic efficiency (CE) of the Mg plated on BP nanosheets and Cu was tested using asymmetric BP | |Mg and Cu | |Mg cells as shown in Fig. 2e. Specifically, 4.0 mAh cm⁻² Mg was first deposited on the BP and Cu electrodes and cycled 10 times. Then, 0.4 mAh cm⁻² (I_D: 0.4 mA cm⁻² for discharging and charging 1 h each) Mg was stripped and deposited repeatedly for another 10 times. Finally, the Mg residue was stripped with cut-off voltage of 1 V. The cycling performances of symmetric Mg | |Mg cell and symmetric Mg@BP | |Mg@BP cell (90% DOD) were tested by galvanostatic discharging and charging.

Electrochemical impedance spectroscopy measurements (EIS) of the symmetric cells were obtained and fitted on an Auto Lab NOVA 2 station under an AC voltage of 5 mV from 1 MHz to 0.01 Hz and to analyze the interfacial properties of the prepared Mg composite negative electrodes (25 °C). The equivalent circuit model, which includes a solid bulk resistance (R_s), charge-transfer resistance (R_ct), a constant phase element (CPE) and Warburg impedance (Z_w), was employed to fit the obtained Nyquist plots.

CV results of the full CuS | |Mg@BP batteries were collected from 0.2−2.5 V (vs. Mg/Mg²⁺) at a scanning rate of 0.2 mV s⁻¹ (25 °C). The cycling and rate performances of the full batteries were tested in the voltage range of 0.4−1.95 V at different charge and discharge current density (room temperature). The Coulombic efficiency of Mg@BP | |nano-CuS battery was calculated as the ratio of the discharge capacity to the charge capacity in each cycle. The specific energy of the Mg@BP | |nano-CuS battery was calculated using the following equation⁷².

$$E=\frac{{VCm}_{\mathrm{CuS}}}{\sum m_i}$$

where, V was chosen from the second platform long voltage value (1.25 V), C is the specific mass capacity (mAh g⁻¹), m_CuS is the mass loading of CuS (g cm⁻²), ∑m_i is the total mass per unit square (mg cm⁻²). The total mass to calculate the specific energy were based on the contents of the electrode materials, including the active CuS, carbon black, and PVDF binder in the positive electrode, as well as the deposited Mg, BP nanosheets, and PVDF binder in the negative electrode. Specifically, the mass loading of active CuS was 5.8 mg cm⁻², and the carbon black and PVDF could be calculated based on the mass ratio (CuS: carbon black: PVDF, 8:1:1). For the Mg@BP composite negative electrode side, the mass loading of the BP substrate was 2.4 mg cm⁻² and the active Mg loading was 2.2 mg cm⁻² which was loaded according to the N/P ratio (1.5) of the battery.

Physicochemical characterizations

SEM was carried out on a Hitachi SU7000 field emission SEM instrument, while TEM was performed on a Tecnai G2 F20 X-Twin operating at 200 keV. Atomic force microscopy images were collected using a Bruker Dimension Icon microscope under ambient conditions. In situ measurements were consecutively conducted upon cycling. An in situ XRD experiment was carried out on a Bruker D8-Advanced X-ray diffractometer with Mo Kα radiation (λ = 0.07093 nm) in the 2θ range of 4−18 degrees. In situ confocal Raman tests were conducted on a confocal Raman microscope WITec Alpha 300 R system (WITec GmbH, Germany) at room temperature. The collected data were processed with Project FIVE+ software (Version 5.2, WITec GmbH, Germany). X-ray photoelectron spectroscopy (XPS) was employed to reveal the chemical environment of BP and Mg@BP composite at different plating state, using a KRATOS AXIS DLD spectrometer. Ex situ SEM, TEM, and XPS tests were used to study the Mg deposition behavior on the studied substrates. Asymmetric substrate | |Mg cells (substrate including BP, Mg, Cu, Al, Mo, BP/C, and carbon fabric) were assembled. The above cells were disassembled after Mg plating process and the resulting Mg@substrate (composite) negative electrodes were sealed with the lamination film in the glove box.

For the ex situ tests, the sealed Mg@substrate (composite) negative electrodes were removed from the seal films and transferred quickly (<1 min) into the equipment chamber to minimize air exposure. Thermal-gravimetric analysis (TGA) was performed using a TG 209 F3 Tarsus (NETZSCHC, Selb, Germany) under an argon protection atmosphere to determine the sulfur loading in the sulfur positive electrode. The temperature range was set from room temperature to 550 °C with a heat rate of 10 °C min^‒1.

Theoretical calculations

Mg deposition behavior and the surface current density distribution of the BP nanosheets, Cu, and Mg substrates were simulated using the electrodeposition module in COMSOL. The simulation module was composed of BP nanosheets (Cu, or Mg substrates) and liquid electrolyte impregnated inside. In the finite element simulation, simplified three-dimensional models of the BP nanosheets were used to obtain the mechanics information using the Solid Mechanics Interface Module. A loading of 2% was applied to the models to check the mechanical property change after Mg storage. Specifically, the whole simulation area was set to linear elastic material with interlacing strain. The solver used the fully coupled method to automatically determine the damping factor of the Newton method in each iteration to get the von Mises normal form stress, and the volume strain of the test samples.

Density functional theory (DFT) calculations were performed in the commercial Cambridge sequential total energy package program (CASTEP). A generalized gradient approximation with the Perdew–Burke–Ermzerh of functional was adopted for the total energy calculations. The ultrasoft pseudopotential was used to treat the core electrons. The energy cut-off was set to 550 eV. The vacuum region between slabs was 15 Å. The Brillouin zone of the surface unit cell was sampled using Monkhorst–Pack grids. The MP grids were set as 4 × 4 × 1 for all the surfaces and slabs, respectively.

Author contributions

Qiannan Zhao and Prof. Chaohe Xu conceived the idea. Prof. Ronghua Wang, Prof. Chaohe Xu and Prof. Jong-Beom Baek supervised this work and provided financial support. Qiannan Zhao, Kaiqi Zhao, Gao-Feng Han, Ming Huang, and Zhongting Wang helped the electrochemical tests and theoretical calculations. Qiannan Zhao, Zhiqiao Wang, Wang Zhou, Yue Ma, and Jilei Liu, Ronghua Wang, and Prof. Jong-Beom Baek helped in situ tests and corresponding data analysis. Qiannan Zhao, Guangsheng Huang, Jingfeng Wang, and Fusheng Pan helped SEM, TEM and data analysis. Qiannan Zhao, Ronghua Wang, Chaohe Xu, and Prof. Jong-Beom Baek prepared the manuscripts. All authors participated in discussing and approving the final manuscript.

Peer review

Peer review information

Nature Communications thanks Ki-Joon Jeon, Shanghai Wei, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

All data analyzed and generated during this study are included in the article and its Supplementary Information. The data that support the plots within this paper and other findings of this study are available from the corresponding author upon request. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-024-52949-4.