Sodium-Ion Battery Anode Construction with SnPx Crystal Domain in Amorphous Phosphorus Matrix

Institute of Advanced Battery Materials and Devices, Faculty of Materials and Manufacturing, Beijing University of Technology, Beijing 100124, China Key Laboratory of Advanced Functional Materials, Ministry of Education, Beijing University of Technology, Beijing 100124, China College of Chemical and Pharmaceutical Engineering, Hebei University of Science and Technology, Shijiazhuang 050018, China College of Materials Science & Technology, Jiangsu Key Laboratory of Materials and Technology for Energy Conversion, Nanjing University of Aeronautics & Astronautics, Nanjing 210016, China Center for High Pressure Science & Technology Advanced Research, Beijing 100094, China


Introduction
Sodium-ion batteries (NIBs) are expected to be a vital alternative to the extensively used lithium-ion batteries (LIBs) by considering the high abundance and low cost of sodium resources [1][2][3][4]. On the survey of electrode materials for NIBs, phosphorus (P) and tin (Sn) have gained wide attentions as anode materials because of their high theoretical capacities, low discharge voltages, and low costs. However, both of them suffer from large volume expansions when alloying with Na, leading to the structural collapse, particle pulverization, electrochemical separation of active materials, and thus rapid performance decay [5,6]. Moreover, the nonmetallic P and phosphides have poor conductivities, while the metallic Sn and its sodiation products can easily aggre-gate; both effects significantly reduce the electrochemical efficiency of NIBs [7][8][9][10]. It has been demonstrated that the cooperation of P and Sn, through either physical incorporation or chemical alloying, can attenuate the intrinsic drawbacks of individual P or Sn. Particularly, the compound Sn 4 P 3 has been shown an appealing anode material for NIBs because of the synergetic electrochemistry of P and Sn elements [7,[11][12][13][14]. Still, the elegant design of P/Sn composite anode materials, in terms of the chemical component, ratio, crystallinity, sizes, distribution, uniformity, interface, etc., is essential to further optimize the overall performance in NIBs [11,[15][16][17][18].
To synthesize high-performance P-/Sn-based anode materials, many approaches have been employed, such as high-energy ball milling (HEBM) [19,20], construction of yolk-shell spheres [21][22][23][24], and hydrothermal synthesis of heterostructures [25]. Among them, HEBM is of excellent practicability owing to its low cost, easy scale-up, and little reaction media involved [26,27]. However, the precise control of P-/Sn-based composite structures by HEBM, often combining carbon materials to improve the conductivity, is challenging because of the complicate and inhomogeneous mechanochemical reactions between P and Sn, which generate diverse components including P, Sn, and SnP x with complex structural ordering states [28]. Mulder's group [26] has fabricated a high-rate and stable Sn 4 P 3 -P@graphene nanocomposite from red P and Sn with a molar ratio of 3 : 1 by a three-stepped HEBM method for NIBs. However, the roles of the P/Sn ratio on the composition and microstructures of the P-/Sn-based composites and subsequent electrochemical performance in NIBs have not been studied. Therefore, it is essential to rationalize the components and microstructures of the P-/Sn-based composites, especially by the P/Sn ratio, to further optimize the performance.
The P-/Sn-based composite materials possess poor longrange ordering, and thus, their elegant structural identification, in terms of components, ratios, localized structures, etc., can hardly be well elucidated by characterizations such as X-ray diffraction (XRD) and transmission electron microscope (TEM). Therefore, it is required to develop new technologies for the structural investigation and quantification. The pair distribution function (PDF) is a powerful characterization technique, which can not only provide local structure information for complicated systems but also quantify the components in ordered or even disordered materials [29][30][31][32][33]. However, owing to the light weight of P, PDF has rarely been used in P-/Sn-based anode materials, making the further structural transformation and reaction mechanism indistinct [13,34]. Therefore, further attempts on the identification and quantification of microstructures, by means of new characterizations such as PDF, as well as the subsequent mechanism studies, are strongly required for P-/Sn-based anode materials.
In this work, a series of P-/Sn-based composite anode materials have been prepared by a one-step HEBM to rationalize the P/Sn ratios and the resultant microstructures for the optimization of electrochemical performance in NIBs. PDF has been employed as a hardcore quantitative technique to elucidate the structures of these materials, suggesting the formation of Sn 4 P 3 and Sn crystalline domains embedded inside amorphous P/carbon matrices with adjustable compositions. It was found that the P-excess composite with the largest amount of crystalline Sn 4 P 3 in the P/C matrix can deliver the best overall electrochemical performance. Based on the derived structural information, the reaction mechanism of the optimal anode material has been elucidated by 23 Na and 31 P solid-state nuclear magnetic resonance (NMR) investigations.

Material Preparation.
The composite anode materials, including C1, C2, C3, C4, C5, and C6, were prepared by the one-step HEBM of red phosphorous (P), tin (Sn), and super P (C) in different proportions, as listed in Table 1. In detail, raw materials in a certain ratio were transferred into a Wolfram carbide container with 10 mm and 1 mm Wolfram carbide milling balls inside an argon glovebox (H 2 O and O 2 contents < 0:1 ppm). The ball-to-material ratio was kept constant to be 40 : 1. The milling container was then transferred from the glovebox into the ball miller. For all composites, the HEBM was carried out at 600 rpm for 12 h valid milling time. After the milling, the container was disassembled inside the glovebox, and the obtained composite anode was collected and sealed before use.

Electrode Preparation and Electrochemical Measurements.
Half-cells were assembled to evaluate the electrochemical performance of all composites. For a typical electrode, a mixture of 80 wt.% active composite and 10 wt.% super P carbon as the conductive additive was prepared by hand milling in an agate crucible for 0.5 h, which was then added with 10 wt.% polyacrylic acid (PAA) solution and stirred for 8 h to get a homogenous slurry. A copper foil was uniformly covered with the obtained slurry by a scraper and dried inside a vacuum oven at 80°C overnight. The mass loading of the active material was~1.0 mg cm −2 . Finally, the electrode was then cut into small circular sheets and compressed at 10 MPa.
R2032-type coin cells (MTI Co.) were assembled in argon glove box by using the prepared sheets as the working electrode, sodium foil as the counter electrode, and Whatman paper as the separator. A solution of 1.0 mol L −1 NaPF 6 in propylene carbonate (PC) with 5 vol.% fluoroethylene carbonate (FEC) was used as the electrolyte. The electrochemical performances of all synthesized materials were tested at 25°C by Neware Test System in the voltage range of 0.01~3.0 V (vs. Na + /Na). The specific capacity was calculated based on the active composited material. Cyclic voltammetry (CV) was performed in a coin cell using a CHI600c electrochemical work station (Chen Hua Instrument Co., China). Galvanostatic intermittent titration technique (GITT) was performed by Neware Test System with constant current pulse for 1 h and followed relaxation for 2 h to allow the system to reach the electrochemical equilibrium. Full cells were assembled by coupling the representative C3 anode with Na 3 V 2 (PO 4 ) 3 cathode in the R2032-type coin cell configuration and tested at 25°C by Neware Test System. The preparation of the cathode electrode is the same as that of the anode electrode, except that the binder uses polyvinylidene fluoride, collector uses aluminum foil, and the conducting carbon uses acetylene black.

Material Characterization.
The crystal structures of all composites were characterized by XRD (Bruker, D8 Discover) with Cu Kα radiation (λ = 0:154 nm). The morphologies were studied using scanning electron microscopy (SEM) (FEI, Quanta 650) and TEM (JOEL, JEM-2010). Energy-dispersive spectroscopy (EDS) was carried out to evaluate the distribution of elements in the composite anodes. Ex situ X-ray photoelectron spectroscopy (XPS, Thermo Scientific ESCALAB 250Xi) was used to investigate the chemical structures of C3 at different electrochemical states.

2
Energy Material Advances 23 Na and 31 P NMR was employed to study the mechanism of the sodiation/desodiation process, which were acquired on a Bruker 400 MHz spectrometer (AVANCE HD III, Germany) with 4.0 mm HX MAS probe. The Larmor frequencies of 23 Na and 31 P were 105.85 and 162.0 MHz, respectively. All the studied electrodes were filled into 4.0 mm rotors inside an argon glovebox. The spinning rate was set to be 8 kHz. Single pulses of 1.4 and 0.95 μs were set for 90 degree to polarized 23 Na and 31 P magnetization, respectively. Total 1024 scans were recorded for each measurement to get reasonable signal-to-noise (S/N) ratio. 23 Na was externally referenced to a 1 mol L −1 NaCl solution at 0 ppm, and the peak of 85 wt.% H 3 PO 4(aq) at 0 ppm was set as the 31 P spectral reference.
High-energy synchrotron total scattering measurements were carried out at the 11-ID-C beamline of the Advanced Photon Source (APS), Argonne National Laboratory. The wavelength of the X-ray is 0.1173 nm. Experimental geometries were calibrated using the CeO 2 powder sample and Fit2D [35] freeware. Fit2D was also used to integrate collected 2D data. The PDF was calculated by a Fourier transform using PDFgetX3 [36] from the integrated data with Qmax up to 27 Å −1 . PDFgui [37] was used to simulate PDFs and to fit data of the structures.

Results and Discussion
A series of P-/Sn-based composite materials with different compositions and crystalline phases have been prepared by HEBM of red P and Sn (total 70 wt.%) in different ratios together with super P carbon (30 wt.%) ( Table 1), which are termed as C1, C2, C3, C4, C5, and C6 with the increased content of red P. The XRD patterns in Figure 1(a) show that C1, C2, and C3 contain crystalline phases, while C4, C5, and C6 only contain amorphous structures. Although the P/Sn molar ratio in C1 (0.66) is close to the stoichiometric P/Sn ratio in Sn 4 P 3 (0.75), only the Sn pattern could be identified (Figure 1(b)), which implies that the mechanically generated SnP x alloys, if any, are dominantly amorphous. With the increase of P/Sn molar ratio to 1.53 in C2 (Figure 1(c)), the Sn pattern becomes noticeably weaker, while a new Sn 4 P 3 structure appears, indicating more Sn undergoes mechanochemical alloying with red P. When the P/Sn molar ratio is further increased to 2.87 in C3 (Figure 1(d)), the Sn pattern disappears, while the Sn 4 P 3 pattern strengthens, suggesting nearly all Sn has been consumed in the alloying reaction with red P. The absence of observable XRD peaks in C4-C6 suggests the contents of crystalline Sn 4 P 3 were ultralow owing to the large excess of amorphous P.
Although the crystalline phases in C1-C3 have been shown by XRD, much information about nanocrystallites as well as amorphous structures can hardly be revealed. In addition, the noncrystalline structures reduce the signal-tonoise (S/N) ratios of XRD patterns, making them less reliable. PDF analysis based on synchrotronic total scattering experiments is powerful to provide atomic structure information for crystalline and disordered materials [32,33,38]. We further performed PDF analysis on the prepared materials. As shown in Figure 2(a), regular peaks can be seen for C1-C3, indicating the existence of ordered structures. In contrast, no noticeable peak above 5 Å is observable for C4-C6, suggesting their dominant disordered features.
More detailed structural information can be extracted by fitting the PDF curves based on the crystal structures of Sn and Sn 4 P 3 . The structural model of Sn in Figure 2(b) displays a tetragonal structure (space group: I4 1 ), with representative Sn-Sn bonds having distances of 3.03 Å ((Sn-Sn) 1 ) and 4.43 Å ((Sn-Sn) 2 ) [34]. Similarly, the structural model of Sn 4 P 3 in Figure 2(c) displays a trigonal structure (space group: R 3m), with representative bonds having distances of 3.95 Å ((Sn-Sn) 3 ), 3.95 Å (P-P), and 2.61 Å (Sn-P) [13]. Note the (Sn-Sn) 3 bond has the same distance as the P-P bond in Sn 4 P 3 . All these peaks can be identified in C1, C2, and C3 in Figure 2(d), suggesting the existence of Sn and Sn 4 P 3 alloys. With the decrease of Sn elemental contents from C1, C2, to C3, the intensities of (Sn-Sn) 1 and (Sn-Sn) 2 bonds from the metallic Sn become weaker, in line with the further alloying of Sn crystallites with P when more P was added. In contrast, the intensities of (Sn-Sn) 3 /P-P and Sn-P become stronger from C1, C2, to C3, which demonstrates the formation tendency of Sn 4 P 3 with the increase of P. Based on the metallic Sn and Sn 4 P 3 models, the PDF curves of C1, C2, and C3 can be well fitted, providing quantitative information about the components in these materials. The crystalline Sn/Sn 4 P 3 mass ratios in C1, C2, and C3 are calculated to be 94 : 6, 37 : 63, and 2 : 98, respectively.
To investigate the localized structures of the P-/Sn-based anode materials, the representative composite C3 was examined by high-resolution transmission electron microscopy (HRTEM) (Figures 3(a)-3(c)) as well as SEM. The SEM image ( Figure S1, supporting information (SI)) shows the agglomerated C3 with irregular shapes comprised of micrometer-sized particles. As shown in the inset of   Figure 3(a) shows an amorphous master matrix with intimately imbedded crystallites of 10 nm. The two embedded crystallites are identified to be Sn 4 P 3 and Sn according to their lattice fringes with interplanar spacings, as illustrated in the enlarged images in Figures 3(b) and 3(c), respectively. The amorphous matrix is likely composed of amorphous P and carbons. The EDS elemental maps in Figures 3(d)-3(g) reveal that P, Sn, and C are relatively homogeneously distributed in C3, suggesting that Sn 4 P 3 and metallic Sn crystalline nanoparticles are nearly evenly distributed in the amorphous matrix.
Such a configuration with nanosized crystal domains diffusely imbedded in a conductive amorphous matrix can combine the effective ion/electron transfer in metallic/semimetallic crystallites, and the isotropic volume variations of amorphous structures, which will synergistically improve the electrochemical efficiency, mitigate the volume change, and improve the stability. Also, the amorphous P can separate the metallic/semimetallic crystal domains from aggregation, while the amorphous carbon improves the electron conductivity of the composite [28]. The continuous amorphous matrix can also promote the uniform formation of charge/discharge products by tuning the nucleation process [39]. Moreover, one component in the composite may attenuate the volume change, inherent stress, and mechanical instability of another component because of their asynchronous and distinct reactions [40]. With the elegant synergy of crystalline and amorphous structures, composite C3 is expected to deliver better electrochemical performance than that of the all-crystalline or all-amorphous P-/Sn-based anode materials.
The electrochemical performance of these composite anode materials has been evaluated in half-cell NIBs ( Figure 4). As shown in the CV curves of the representative 10    Energy Material Advances C3 in Figure 4(a), an irreversible broad cathodic peak at 1.0 V appears in the first sodiation process, which could be attributed to the electrochemical decomposition of the electrolyte and the formation of a solid electrolyte interphase (SEI) film [41]. During the sodiation process, Na + ions insert into the C3 composite through the multistep alloying process to form Na-P and Na-Sn species, with the final reaction products to be Na 3 P and Na 15 Sn 4 , respectively [19,42]. In the anodic scan, two peaks at 0.4 V and 0.7 V in the first cycle correspond to the two-step desodiation processes that Na + ions extract from the Na-P and Na-Sn alloys [14]. The CV curves become almost constant from the second cycle, suggesting the good reversibility of the electrochemical conversions of C3 after its initial stabilization.
The effect of the P/Sn ratio on the electrochemical performance was studied, showing that a balanced P/Sn ratio, i.e., a rational structure comprised of crystal domains embedded in the amorphous matrix, is critical to obtain the optimal performance. As shown in Figure S2, C1 and C2 exhibit stable cycle performance, but the capacities are low because of the high ratios of Sn. On the contrary, owing to the large content of the high-capacity P, C6 has the highest specific capacity but decays rapidly. Similarly, C4 and C5 exhibit poor capacity retention in NIBs. Therefore, the most balanced electrochemical performance with both high capacity and cycling stability could be realized in the C3 composite.
The electrochemical performance of C3 was thus carefully evaluated. GITT was performed to investigate the kinetic behavior of C3, as shown in Figure S3. The diffusion coefficient of Na + (D Na+ ) can be estimated based on Fick's second law according to the equation in SI. The D Na+ of C3 maintains a relatively stable level with an order of about 10 −16 -10 −15 cm 2 s −1 during cycling. Figure 4(b) shows that C3 can deliver a high capacity of 996.8 mA-h g −1 at a current density of 200 mA g −1 . Note that super P carbon can deliver a capacity of only~75 mA-h g −1 in NIBs at 1 A g −1 (Figure S4), suggesting the minor capacity contribution from the carbon component in C3. The rate performance and the corresponding charge/discharge curves of C3 have been tested between 0.01 V and 3.0 V at current densities ranging from 50 to 2000 mA g −1 (Figures 4(c) and 4(d)). At current densities of 50, 100, 200, 500, 1000, and 2000 mA g −1 , C3 exhibits specific discharge capacity of 1009.4, 923.2, 847.3, 682.6, 607.7, and 493.7 mA-h g −1 , respectively. Moreover, when the current density was returned to 50 mA g −1 , the discharge capacity could largely recover to 947.2 mAh g −1 . The good rate performance demonstrates that C3 favors the reversible sodiation/desodiation transformation at high current densities.   Aside from the 1 st cycle, C3 can work stably in the NIBs, delivering discharge capacities of 673.9 and 422.3 mAh g −1 in the 2 nd and 300 th cycles, respectively. Moreover, the Coulombic efficiency of C3 reaches to above 99.5% from~26 th cycle, further suggesting the high reversibility of C3 as the NIB anode material. Figure S5 shows the surface morphologies of C3 before and after 300 cycles analyzed by SEM. In the fresh electrode, C3 was blended with polyacrylic acid binder and conductive super P. However, the cycled electrode was obviously coated with SEI layer and there was no obvious crack, suggesting the good electrochemical reversibility of C3. The promising cycle reversibility of C3 in NIBs indicates that the Sn 4 P 3 as well as Sn crystal domains embedded inside an amorphous P/carbon matrix significantly improve the electrochemical stability. Table S1 shows the performance comparison of the phosphorus-based anode for NIBs published in recent years, among which C3 has advantages in initial Coulombic efficiency and capacity.
We further explored the electrochemical performance of full cells, which were assembled using C3 as the anode and Na 3 V 2 (PO 4 ) 3 as the cathode. The electrochemical behavior of Na 3 V 2 (PO 4 ) 3 was first investigated in half-cells as shown in Figure S6. In order to balance the capacity between the two electrodes in the full cell, the N/P (anode/cathode) ratio was set as 1.2. The full cell was measured in the voltage range of 2.0-4.0 V at a current density of 50 mA g −1 and delivered an initial charge and discharge capacity of 111.6 and 110.3 mAh g −1 (based on the mass of Na 3 V 2 (PO 4 ) 3 ), respectively, with an initial Coulombic efficiency of 98.8%. A discharge capacity of 107.8 mAh g −1 could be obtained after 50 cycles, corresponding to a capacity retention of 97.7%, suggesting the effectiveness of C3 as anode materials in full cells.
It is challenging to precisely probe the structural evolution of P-/Sn-based anode materials with poor long-range ordering upon electrochemical cycling by conventional diffraction and microscopy techniques [43]. For example, the representative C3 anode material showed no correlated signal except for the Cu foil in the ex situ XRD as shown in Figure S7. Ex situ XPS test result of the C3 anode is shown in Figure S8. It is observed that the P 2p 1/2 (132.7 eV) and PO x (136.5 eV) peak weaken during the discharge process to 0.01 V and increase during the charge to 3.0 V. Since XPS is surface sensitive and cannot provide bulk information about the structural evolution of the anode material, we further applied ex situ NMR to investigate C3  Energy Material Advances at different electrochemical states. NMR is powerful to probe the local atomic environments and provide detailed structural information for noncrystalline materials [44,45]. Well-resolved 23 Na and 31 P NMR spectra obtained under magic angle spinning (MAS) were taken to monitor the structural evolution of C3 during the sodiation/desodiation (Na x P phases) processes at different charge/discharge stages ( Figure 5). Figure 5(a) shows various charge/discharge states of C3 for the NMR investigation. As shown in Figure 5(b), the pristine C3 anode shows a single 31 P resonance centering at δ = 0 ppm and spreading from −500 to 500 ppm. The broad linewidth and asymmetric peak reflect different chemical environments due to the noncrystalline nature [43]. For the electrode discharged to 1.0 V, a broad tail is observed at the right shoulder of the 31 P spectrum ( Figure 5(b)), suggesting the formation of amorphous Na x P [45][46][47]. The minor sharp peak at −245 ppm is probably originated from impurities during the sample treatment. Further discharge to 0.01 V, a variety of Na x P alloys could be formed. Since the electrodes studied here are highly amorphous and have complicated environments, it is impractical to deconvolute these peaks, similar to the cycled NiP electrode [48]. However, it is assignable when referenced to the overlapping signals of crystalline NaP and Na 3 P [47,49]. The purple and green shadows represent NaP and Na 3 P positions, respectively. Since the test powder is from the whole electrode plate, the voltage will rise back to~0.2 V from 0.01 V when set rest after discharged to the final state. The phase of the final discharge state is composed of NaP and Na 3 P. The sharp peak at 14 ppm likely originated from unreacted phosphorus which will decrease during the alloying reaction to form Na x P.
The 23 Na NMR spectra of C3 at different cycling stages were further performed. Since the pristine C3 anode does not contain Na, the 23 Na NMR spectrum was not collected.
After being discharged to 1.0 V, a peak at 6 ppm presenting in 23 Na NMR spectrum ( Figure 5(c)) consolidates the formation of NaP. Upon further discharge to 0.01 V, the NaP peak almost disappeared and a new peak at 50 ppm noticeably appeared due to the generation of Na 3 P [43,46]. This further validates the analysis of the curve displayed in Figure 5(b) when discharged to 0.01 V. The sharp peak at −10 ppm in 23 Na NMR ( Figure 5(c)) is attributed to the residual Na ions in the unwashed electrolyte.
A reversible evolution of C3 is observed from both 31 P and 23 Na NMR spectra upon subsequent charging. Refer to the spectra of 0.5 V and 1.5 V, resonances of 31 P show a downfield shift, whereas 23 Na move towards the upfield. The sodium is mostly extracted after one cycle to 1.5 V, which results in a dominant signal at −10 ppm for Na in the residual electrolyte. In sum, the amorphous sodiation/desodiation of P-/Sn-based anode material is clearly characterized by the combination of solid-state 23 Na and 31 P NMR.

Discussion
In summary, a series of P-/Sn-based composite anode materials from different P/Sn ratios were prepared by the onestep high-energy ball milling method. With a rationalized P/Sn ratio, a configuration of Sn 4 P 3 and Sn crystalline domains embedded inside an amorphous P/carbon matrix could be realized, combining the synergistic functions of crystalline structures and amorphous matrix. The composite structures were elucidated by a series of characterizations. Especially, PDF studies provide clear structure information not only about the formation of Sn 4 P 3 and Sn crystal domains but also the quantitative information. The P-excess composites containing the highest amount of Sn 4 P 3 delivered the best overall performance, with balanced capacities and cycle stabilities of above 422.3 mA h-g −1 for 300 cycles at 1000 mA g −1 . Combined with a Na 3 V 2 (PO 4 ) 3 8 Energy Material Advances cathode, a full cell delivered a discharge capacity of 110.3 mAh g −1 (based on the Na 3 V 2 (PO 4 ) 3 ) at 50 mAg −1 with a capacity retention of 97.7% after 50 cycles, suggesting the high effectiveness of the composite anode in full cells. The reaction mechanism of the optimal composite has been unraveled by 23 Na and 31 P NMR, showing a stepped sodiation process from NaP to Na 3 P during the discharge of the composite. The study well adjusts and identifies the components and microstructures of P-/Sn-based anode materials, providing insights into the design of high-performance anode materials for NIBs.

Data Availability
The data and materials used to support the findings of this study are available from the corresponding author upon request. Figure S1: SEM images of the composite C3 at magnification of (a) ×10 K and (b) ×100 K. Figure S2: cycle performance of Sn/P/C composites C1-C6 at 500 mA g −1 . Figure S3: (a) the GITT curves and (b) corresponding Na + diffusion coefficients of C3. Figure S4: cycle performance and Coulomb efficiency of super P carbon at 1000 mA g −1 . Figure S5: SEM observation of the surface of electrode: (a, b) before cycling and (c, d) after 300 cycles. Figure S6. (a) charge and discharge curves of Na 3 V 2 (PO 4 ) 3 and C3 in half cells are shown in blue and orange lines, respectively. (b) Cycle performance of Na 3 V 2 (PO 4 ) 3 at 50 mA g −1 . Figure