Revealing the Various Electrochemical Behaviors of Sn4P3 Binary Alloy Anodes in Alkali Metal Ion Batteries

Sn4P3 binary alloy anode has attracted much attention, not only because of the synergistic effect of P and Sn, but also its universal popularity in alkali metal ion batteries (AIBs), including lithium‐ion batteries (LIBs), sodium‐ion batteries (SIBs), and potassium‐ion batteries (PIBs). However, the alkali metal ion (A+) storage and capacity attenuation mechanism of Sn4P3 anodes in AIBs are not well understood. Herein, a combination of ex situ X‐ray diffraction, transmission electron microscopy, and density functional theory calculations reveals that the Sn4P3 anode undergoes segregation of Sn and P, followed by the intercalation of A+ in P and then in Sn. In addition, differential electrochemical curves and ex situ XPS results demonstrate that the deep insertion of A+ in P and Sn, especially in P, contributes to the reduction in capacity of AIBs. Serious sodium metal dendrite growth causes further reduction in the capacity of SIBs, while in PIBs it is the unstable solid electrolyte interphase and sluggish dynamics that lead to capacity decay. Not only the failure mechanism, including structural deterioration, unstable SEI, dendrite growth, and sluggish kinetics, but also the modification strategy and systematic analysis method provide theoretical guidance for the development of other alloy‐based anode materials.


Introduction
Alkali metal ion batteries (AIBs), such as lithium-ion batteries (LIBs), sodium-ion batteries (SIBs), and potassium-ion batteries (PIBs), exhibit high working voltages, owing to the high electronegativity of alkali metals (A) and rapid kinetics on account of the monovalent alkali metal ions (A + ). [1][2][3] (namely, graphite for LIBs and PIBs, and hard carbon for SIBs), their low specific capacity limits their energy density. [13,14] Phosphorus (P), with a multi-electron alloying mechanism, shows a high theoretical specific capacity in AIBs (2590 mAh g -1 for A 3 P as the final discharge products). [3] However, the huge volume change (≈300% in LIBs and ≈400% in SIBs) and low conductivity (10 −14 S cm -1 for red P and 0.2-3.3 S cm -1 for black P) restrict the cycling and rate performance of P in AIBs. [15] Contemporary modification strategies mainly include nanocrystallization, carbon coating, alloying, and functional electrolyte and binder exploration. [16][17][18][19][20][21] Alloying can not only enhance the bulk conductivity and buffer volume expansion of P, but also sometimes provide additional capacity depending on whether the second phase metal is participating in the formation of the alloy. Although the theoretical gravimetric capacities of tin (Sn) are lower than those for P, the volumetric capacities of Sn can compensate for this disadvantage, and the conductivity of Sn is better than that of P. [22][23][24] Based on this, many studies have been reported on P-Sn binary alloys that show excellent performances owing to the synergistic effect of P and Sn in AIBs. [22,[25][26][27] Although P-Sn alloys can store Li + , Na + , and K + separately, there is a notable difference in their electrochemical behaviors, such as the specific capacity, voltage platform, polarization voltage, Coulombic efficiency, cycling, and rate performance in various AIBs. [25,26] Comparing these disparities can be conducive to determine exactly what kind of alkali metal ion it best suits for storage. [3] However, very few studies have been systematically conducted to explore and explain these diversities. Here, we use Sn 4 P 3 as an example to detect the various electrochemical behaviors of Sn-P binary alloys in AIBs. The origin of the discrepancies in their electrochemical behaviors is then analyzed from the perspective of structural transformation, dynamics mechanism, solid electrolyte interface (SEI) evolution, alkali metal deposit, and density functional theory (DFT) calculations. Finally, we present our attempt to improve the capacity retention of Sn 4 P 3 through carbon hybridization.

Physical and Electrochemical Characterization of Sn 4 P 3
Sn 4 P 3 was prepared by a facile mechanical ball milling method with stoichiometric P and Sn power, and its crystal structure, morphology, and electrochemical properties were obtained using X-ray diffraction (XRD), scanning electron microscopy (SEM), transmission electron microscopy (TEM), and 2032 coin cell tests. The XRD results (Figure 1d and Figure S1: Supporting Information) show that the synthesized material is pure Sn 4 P 3 (hexagonal system, R-3m space group, PDF 73-1820) with a potato-like morphology and particle size of ≈3 µm (Figure 1a,b and Figure S2: Supporting Information). The lattice fringe in the high-resolution TEM (HRTEM) image ( Figure 1c) can be indexed in the (107) plane, which is consistent with the strongest diffraction peak in the XRD pattern. Physical and electrochemical characterization of Sn 4 P 3 . a,b) SEM (a) and TEM (b) images of Sn 4 P 3 . c) HRTEM image of selected area of (b). Inset: the corresponding FFT pattern. d) XRD spectrum with the standard PDF card of Sn 4 P 3 . e) Galvanostatic discharge-charge curves at 0.1 C (1 C = 1100 mA g -1 for LIBs, 1 C = 600 mA g -1 for SIBs, 1 C = 400 mA g -1 for PIBs) between 0.005 and 2 V for the Sn 4 P 3 anode at the first cycle in AIBs. f) Statistical data with error bars of reversible specific capacity, Coulomb efficiency, and polarization voltage drawn from (e). All data in this paper was repeated at least three times and is shown with error bars if possible. g-i) Simulation discharge-charge (g), V versus dQ/dV (h), and Q versus dV/dQ (i) curves for better understanding of their respective roles.
The inclined discharge-charge curves (Figure 1e) indicate the multistep reaction mechanism of Sn 4 P 3 in LIBs, delivering a large reversible specific capacity (1150 mAh g -1 ), high Coulomb efficiency (87.2%), and low polarization voltage (0.23 V vs Li/Li + ) (Figure 1f and Figure S3: Supporting Information). In contrast, although Sn 4 P 3 displays a low average sodiation voltage (≈0.4 V vs Na/Na + ), which is related to the high formation energy of the discharge species in SIBs (Figure 3a), its flat reversible capacity (654 mAh g -1 ), gentle Coulomb efficiency (80.1%), and large polarization (0.4 V vs Na/Na + ) will cancel out its low-potential advantage. As for PIBs, while their insertion plots are almost the same as those of SIBs, they demonstrate the worst reversible capacity (386 mAh g -1 ), Coulomb efficiency (56.7%), and polarization voltage (0.62 V vs K/K + ) among all AIBs considered herein.
Differential discharge-charge patterns, mainly V versus dQ/dV and Q versus dV/dQ (where V and Q represent the voltage and capacity of a special electrode material), will be frequently used for exploring the reaction mechanism of Sn 4 P 3 in this paper, and simulation curves (Figure 1g-i) are depicted here to better understand their respective roles in advance. It can be clearly seen that the platform in the discharge-charge curves, i.e., the phase transition process, is reflected in the form of a "peak" in the V versus dQ/dV patterns, and in the form of a "Valley" in the Q versus dV/dQ curves. In addition, the V versus dQ/dV and Q versus dV/dQ patterns are mainly utilized to analyze the potential and capacity when the phase transition occurs, respectively (more details about the V versus dQ/dV and Q versus dV/dQ curves can be found in Figure S4, Supporting Information).

Structural Transformation of Sn 4 P 3 Anodes during the Initial Cycle in AIBs
In order to reveal the origin of the various electrochemical behaviors for Sn 4 P 3 anodes in AIBs, a combination of the V versus dQ/dV curves, ex situ XRD, and TEM techniques was explored to detect its structural evolution during the insertion-extraction process. (We failed to conduct in situ XRD tests owing to the low crystallinity of the P-and Sn-based species during electrochemical processes.) The V versus dQ/dV curves for the first cycle in LIBs exhibit four typical pairs of peaks, and each one corresponds to a redox process, labeled as Li1, Li2, Li3, and Li4 (Figure 2a). During the stage of the discharge from 2-0.83 V (vs Li/Li + ), the flat shape of the V versus dQ/dV curve and the absence of an updated Bragg peak at 0.83 V compared with the initial XRD Figure 2. Structural transformation of Sn 4 P 3 anodes in AIBs. a-c) V versus dQ/dV curves for detecting the position at which phase transformation occurs and the corresponding ex situ XRD plots for determining the specific species formed in these positions of Sn 4 P 3 cycled at various states in LIBs (a), SIBs (b), and PIBs (c). d) Schematic diagram of the discharge mechanism based on (a-c) for the Sn 4 P 3 anode in AIBs. The Sn 4 P 3 anode exhibits various ion storage mechanisms in AIBs, delivering four (labeled as Li1, Li2, Li3, and Li4), three (labeled as Na1, Na2, and Na3) and two steps (labeled as K1 and K2) in LIBs, SIBs and PIBs, respectively. Pink, grey, cyan, dark blue, and dark yellow balls represent the P, Sn, Li, Na, and K atoms, respectively. With continued discharge to 0.63 V, all diffraction peaks belong to Sn (PDF 86-2265), except that of the Cu current collector, indicating the decomposition of Sn 4 P 3 into metallic Sn and amorphous low lithiation phosphides (such as LiP 7 , LiP 5 , Li 3 P 7 , and LiP) during the Li1 process (0.83-0.63 V). In fact, the existence of Sn is detectable in the entire intercalation-deintercalation process, resulting in better conductivity, unless damage occurs from losing partial capacity. Fresh Bragg peaks with low crystallinity at 0.41 V could be indexed to Li 3 P (PDF 74-1160), demonstrating the conversion from an amorphous low lithiation phosphide to Li 3 P during the Li2 process (0.63-0.41 V). Thereafter, at 0.27 V new XRD peaks cannot be determined exclusively, owing to the similar standard diffraction peaks of the low lithiation stannide species, including Li 5 Sn 2 (PDF 29-0839), Li 13 Sn 5 (PDF 29-0838), and Li 7 Sn 2 (PDF 29-0837) (Table S1, Supporting Information). Thus the Li3 process (0.41-0.27 V) represents the transformation from Sn to low lithiation stannide compounds. The complete lithiation species could be attributed to Li 3 P and Li 22 Sn 5 (PDF 18-0753), based on the diffraction peaks at 0.005 V, revealing the conversion from low lithiation stannide to Li 22 Sn 5 (Li4 process, 0.27-0.005 V). The subsequent delithiation process is an almost complete reverse of the lithiation process, since the resulting compounds are made up of Sn nanocrystalline surrounded by an amorphous P matrix rather than Sn 4 P 3 , as evidenced by using combined characterization of ex situ XRD and HRTEM at 2 V ( Figure S5, Supporting Information).
Analogical analysis methods have also been applied to the Sn 4 P 3 anodes in SIBs and PIBs (Figure 2b,c), and their reaction mechanism has been determined using the following equation together with Figure 2d.

DFT Calculations of Sn 4 P 3 Anodes in AIBs
To further demonstrate the alkali metal ion storage mechanism of Sn 4 P 3 anodes in AIBs, we performed DFT calculations for A x Sn 4 P 3 with amorphous structural models, considering the pulverization of Sn 4 P 3 during the charge-discharge reactions (see the Methods section). [26,28] The calculated formation energy curves exhibit minimum values at x = 28.29, 23.95, and 10.03 (Figure 3a) with corresponding theoretical specific capacities of 1335, 1131, and 473 mAh g -1 for A = Li, Na, and K, respectively, which agrees well with our experimental values. The calculated voltage profiles are also consistent with the experimental measurements ( Figure 3b). The wide Bader charge distribution of P and Sn atoms in Figure 3c,d further demonstrates the coexistence of various Li-Sn and Li-P alloys during the entire alkali metal ion intercalation process. Some neutrally charged Sn atoms at x = 4 are located within Sn clusters, implying the segregation of Sn and P, consistent with our ex situ TEM measurements ( Figure S5, Supporting Information). In addition, P atoms when x ≥ 12 and Sn atoms at x = 28 deliver the majority of Bader charges of ≈−2.5 e and −3.4 e, giving evidence for the initial formation of Li 3 P followed by Li 22 Sn 5 (close to Li 28 Sn 4 P 3 ), validating the ex situ XRD observation. [29] The distribution histograms for PIBs ( Figure S6, Supporting Information) also reveal that KSn and K 3 P are formed (totally, close to K 10 Sn 4 P 3 ). A previous paper published by our collaborator demonstrated that the final discharge products in SIBs are Na 15 Sn 4 and Na 3 P (close to Na 24 Sn 4 P 3 ). [30] In Figure 3e the decreasing average Bader populations of Sn and P indicate the gradual reduction in these species and that P is more quickly saturated than Sn. In addition, the volumes of all the A x Sn 4 P 3 increase almost linearly with increasing x, whereas the slope depends on the type of alkali metal (Figure 3f). Specifically, the calculated maximum volume expansions are ≈326%, 391%, and 382% for Li 28 Sn 4 P 3 , Na 24 Sn 4 P 3 , and K 10 Sn 4 P 3 , respectively, resulting in restricted insertion processes of A + , especially for Na + and K + .

Capacity Decay Analysis of Sn 4 P 3 Anodes in AIBs
In addition to the initial reversible specific capacity, the cycling performance is also a crucial index for estimating a battery's performance. The disparate cycling behaviors of Sn 4 P 3 in AIBs are explored using Q versus dV/dQ curves to clarify the capacity contribution of each phase transition mentioned www.afm-journal.de www.advancedsciencenews.com 2102047 (5 of 13) © 2021 The Authors. Advanced Functional Materials published by Wiley-VCH GmbH above. The Sn 4 P 3 anode in LIBs presents an inferior cycling performance, with a capacity retention of ≈65.4% (700 mAh g -1 ) compared with the initial capacity (1070 mAh g -1 ), and a Coulombic efficiency of ≈99.2% after 50 cycles (Figure 4a). The shape of the corresponding discharge-charge curves at the 1st, 20th, and 50th cycles did not obviously change except for a gradually increasing polarization (Figure 4a, inset figure). Four valleys in the Q versus dV/dQ curves derived from the relevant charge curves reveal a four-step lithiation-delithiation mechanism for Sn 4 P 3 , in agreement with the result in Figure 2a, i.e., Li4→Li3→Li2→Li1 (representing the delithiation process, as shown in Figure 4d. It was found that the specific capacities attributed to the Li1 and Li3 phase transformations first increase and then decrease, while those resulting from the Li2 and Li4 processes decline continuously, indicating that the capacity decay in LIBs is mainly caused by the Li2 and Li4 stages, namely, the deep lithiation of P and Sn (Figure 4g). This is reasonable because the deep charge or discharge process usually results in huge volume changes, especially for alloy anodes, [31,32] as well as sluggish dynamics (Figure 7g) as discussed in the later section "Dynamics of Sn 4 P 3 anode in AIBs").
For SIBs, the Sn 4 P 3 anode exhibits a general cycling stability with a capacity retention of 78.2% (485 mAh g -1 ) but poor Coulombic efficiency of 97.4% after 50 cycles (Figure 4b). The relevant charge-discharge curves at the 1st and 20th cycles almost coincide, whereas that of the 50th cycle changes greatly, with the appearance of a more obvious charging platform at a low potential (Figure 4b, inset figure). Three valleys can be detected during the desodiation processes (Na3→Na2→Na1) at the 1st and 20th cycles, while the Na2 and Na1 stages merge (Na1+Na2) in the 50th cycle, demonstrating the combination of shallow and deep sodiation of the P stage ( Figure 4e). The statistical data in Figure 4h show that the specific capacity of the Na3 stage is almost constant throughout the entire cycling process, implying that the fading of capacity during the last 30 cycles is mainly attributed to the sodiation of P (Na1+Na2) rather than Sn (Na3). This is due to the sluggish dynamics of P compared with Sn during the insertion-extraction reaction. With respect to PIBs, the "dive" phenomenon (a sudden drop in capacity) emerges at approximately the 15th cycle with a very low Coulomb efficiency of 81.7% (Figure 4c), indicating its pessimal electrochemical performance among AIBs. The shape of the corresponding charge-discharge curves changed sharply, with almost no capacity at the 50th cycle (Figure 4c, inset figure), resulting from the potassiation of both P and Sn (Figure 4i).

SEI Evolution of Sn 4 P 3 Anodes in AIBs
Aside from the evolution of the bulk structure, the transformation of SEI also has a valuable impact on batteries' performance. HRTEM images reveal that the typical thickness of SEI  Figure S10: Supporting Information). The SEM image ( Figure S11, Supporting Information), EDS mapping ( Figure S12, Supporting Information), and weakest Sn 3d XPS signal ( Figure S13, Supporting Information) confirm that PIBs have the largest SEI thickness, because Sn 4 P 3 particles coated with the thick SEI reduce the intensity of XPS signal. The decreasing Sn 3d XPS signature in LIBs indicates the thickening of the SEI during cycling, while the increasing Sn 3d XPS intensity in PIBs demonstrates that a cracked SEI exposes the fresh surface of the Sn 4 P 3 electrode. The C 1s XPS spectra after cycling suggest that the composition of SEI in AIBs mainly incorporates carbon species, including CH 2 -CF 2 (≈290.  Table S2: Supporting Information). These carbon species are primarily decomposed by electrolytes, such as salts (LiPF 6 , NaPF 6 , or KPF 6 ), carbonate solvents (ethylene carbonate (EC) and dimethyl carbonate (DMC)), and additives (fluoroethylene carbonate (FEC)). [33,34] In addition, the polycarbonates (poly (CO 3 ), ≈291 eV) generated in LIBs and SIBs after 50 cycles by the Lewis reaction between PF 5 (Lewis acid, APF 6 →AF + PF 5 ) and EC (monomer) could enhance the flexibility of the SEI, which is significant for an electrode with huge volume expansion. [33] However, no signal indexed to poly (CO 3 ) could be detected in PIBs during the initial 50 cycles, contributing to PIBs having the most unstable SEI character.

Alkali Metal Deposition Characteristics on Cu and Sn 4 P 3 Anodes
A combination of the A-Cu battery test and the XPS spectra revealed that the various alkali metal ion deposition characteristics are another significant factor contributing to the diverse electrochemical behaviors of Sn 4 P 3 anodes in AIBs. SEM images (Figure 6a-c) and the inset digital photographs reveal However, only the SEI with black color coating on carbon fibers coming from the separator could be observed in the Li-Cu battery (Figure 6a and Figure S17: Supporting Information). Further electrochemical tests demonstrate that the relatively low overpotential, high Coulombic efficiency, and minor impedance of the Li-Cu battery is ≈14 mV, 95.7%, and 152 Ω, respectively, and that for the Na-Cu battery is ≈20 mV, 80.5%, and 174 Ω, respectively (Figure 6a-c and Figures S17 and S18: Supporting Information). However, both the K-K battery ( Figure S16, Supporting Information) and the K-Cu battery depict a marked overpotential, very low Coulombic efficiency, and pronounced impedance character. Moreover, the Li 1s, Na 1s, and K 2p XPS spectra of the Sn 4 P 3 anode at the 1st, 20th, and 50th cycles represent the typical SEI species, including LiF (≈55.7 eV), NaF (≈1072.7 eV), and KF (2p1/2, ≈295 eV), mainly formed by the decomposition of FEC and APF 6 (Figure 6g-i). [35][36][37] In addition, Sn 4 P 3 anodes in SIBs present a distinct peak shift, indicating the formation of sodium metal dendrites (≈1071 eV). This formation contributes to the relatively low sodium plating overpotential (20 mV) in Na-Cu battery and low discharge voltage platform of Sn 4 P 3 anodes in SIBs (≈0.2 V vs Na/Na + ).

Dynamics of Sn 4 P 3 Anodes in AIBs
Finally, the effect of kinetic factors on the performance of Sn 4 P 3 anode in AIBs was investigated using cyclic voltammetry (CV), galvanostatic intermittent titration technique (GITT), electrochemical impedance spectroscopy (EIS), and C-rate techniques. CV curves with variable scanning speeds from 0.2-2 mV s -1 between 0.005 and 2 V (vs A/A + ) reveal increasing polarization with increasing speed. This means that the oxidation peaks and reduction peaks shift in the positive and negative directions, respectively, in AIBs (Figure 7a-c). The minimum polarization voltage, maximum peak current density (10 A g -1 at 2 mV s -1 ), and symmetric oxidation-reduction peaks of Sn 4 P 3 in LIBs demonstrate their optimum dynamics in lithiation-delithiation systems. In addition, the charge storage mechanism can be determined from the relationship between the scanning speed (v) and the peak current (i) driven by the CV curves. This relationship is expressed as i = av b , where a and b are the slope and y-intercept, respectively, in the lg(v) versus lg(i) curves (Figure 7d). The situations where b = 0.5 and b = 1 indicate that the system is controlled by bulk phase diffusion and surface capacitance, respectively. The contributions of these two mechanisms can be further separated by the relationship i = k 1 v + k 2 v 1/2 (k 1 and k 2 are the ratio for contributions of surface capacitance and bulk phase diffusion respectively) when b is between 0.5 and 1. [38] The supreme b value (0.97) and capacitance contribution (the shaded area in Figure 7e of Sn 4 P 3 in LIBs for all scan rates (Figure 7f) emphasize its mainly capacitancecontrolled mechanism, resulting in its optimal rate performance (Figure 7i and Figure S20: Supporting Information). The ion diffusion coefficient (D) in the solid electrode, which is the rate-determining step (RDS), can be further examined using GITT (see details in Figure S19, Supporting Information). The D values in AIBs all exhibit a downward trend with an increase in the state of charge (SOC) (Figure 7i), indicating the constrained kinetic processes when alkali metal ions are deeply embedded in the Sn 4 P 3 anode, consistent with the above results in the capacity decay analysis section. For various SOC values, both the minimum potassium ion diffusion coefficient, due to the largest ionic radius of K + among A + (0.076 nm for Li + , 0.102 nm for Na + , and 0.138 nm for K + ) [8] , and the marked charge impedance (R ct , Figure 7h and Figure S21: Supporting Information) reveal the sluggish dynamics of Sn 4 P 3 anodes in PIBs.

Improving the Capacity Retention of Sn 4 P 3 Anodes in SIBs through Carbon Hybridization
According to the electrochemical test mentioned above, Sn 4 P 3 anode exhibits the lowest average voltage, best cycling performance, modest specific capacity and coulombic efficiency in SIBs among the explored AIBs, indicating its considerable application potential for sodium ion storage. Sn 4 P 3 /C ( Figure S22, Supporting Information) was further synthesized using the flexible ball milling method to verify the role of carbon hybridization in improving the performance of the Sn 4 P 3 anode in SIBs previously. Pure Sn 4 P 3 shows the "diving" phenomenon after 50 cycles, whereas Sn 4 P 3 /C delivers a much better capacity retention of ≈80.9% after 180 cycles and a more stable Coulomb efficiency, in spite of having a relatively low initial specific capacity of 457 mAh g -1 resulting from the introduction of carbon materials with low capacity (Figure 8a,b and Figure S23: Supporting Information). Based on the charge Figure 6. Alkali metal deposition characteristics on Cu and Sn 4 P 3 anodes. a-c) Morphology of Li (a), Na (b), and K (c) metal plating on Cu foil after 10 cycles at a charged state. Inset: digital photographs of the separator close to the Cu foil. d-f ) The discharge-charge curves at the 2 nd cycle (d), Coulombic efficiencies (e), and corresponding EIS spectra (f ) of Li-Cu, Na-Cu, and K-Cu battery cycled with the discharge capacity limit of 1 mAh cm -2 at 0.5 mA cm -2 . g-i) Li 1s (g), Na 1s (h), and K 2p (i) XPS spectra of Sn 4 P 3 anodes at 1 st , 20 th , and 50 th cycle in LIBs, SIBs, and PIBs.  (Figure 8c) and corresponding V versus dQ/dV plots (Figure 8d) at the 2nd and 100th cycles, both the P and Sn sodiation processes (shown by the dark yellow and blue rectangles, respectively) contribute to the fast capacity-fading of the pure Sn 4 P 3 anode in SIBs, while these two processes present much more stable characteristics for the Sn 4 P 3 /C anode. Notably, the activation phenomenon marked by yellow dots in Figure 8a enhances the capacity retention of the Sn 4 P 3 /C anode, which is primarily caused by the extension of the platform at low potential (Figure 8e,f, blue rectangle) in the 96th and 106th cycles. Nevertheless, the subsequent capacity-fading at the 131st cycle indicates the fluctuating capacity characteristics due to the facile production of dendrites at low voltages for Sn 4 P 3 in SIBs, which has been demonstrated in "Alkali metal deposition characteristics on Cu and Sn 4 P 3 anodes" section. In addition, EIS tests ( Figure 8g) depict a tiny mid-frequency impedance, even after 100 cycles for the Sn 4 P 3 /C anode, whereas this increases sharply for the pure Sn 4 P 3 anode, revealing that the carbon coating strategy can boost the electrical conductivity. Morphological evolution experiments further show severe particle pulverization in the pure Sn 4 P 3 electrode (Figure 8i), owing to its large volume expansion during the extraction-insertion process of Na + , while its structural integrity can be greatly improved through carbon hybridization (Figure 8h).

Conclusions
Using a combination of ex situ XRD, TEM, and DFT calculations, we demonstrate that the Sn 4 P 3 anode initially undergoes segregation of Sn and P, followed by the intercalation of A + in P and then in Sn during the initial discharge processes. The various specific capacities of Sn 4 P 3 anodes in AIBs are mainly attributed to the disparate number of phase transitions and ultimate insertion species (Li 22 Sn 5 and Li 3 P for LIBs; Na 15 Sn 4 and Na 3 P for SIBs; and KSn and K 3 P for PIBs) (Scheme 1a). Furthermore, differential electrochemical curves, ex situ XPS, and SEM revealed that the deep intercalation of A + in P and Sn, especially in P, resulting in the capacity-fading of Sn 4 P 3 anode in AIBs. Serious dendrite growth further induces capacity decay in SIBs, while in PIBs it is the unstable solid electrolyte interphase and sluggish dynamics that lead to capacity decay (Scheme 1b,c). Finally, the electrochemical performance of Sn 4 P 3 anode is greatly enhanced by the carbon hybridization strategy, which improves the structural stability and dynamics of Sn 4 P 3 anode. The understanding of the failure mechanisms, including the structural deterioration, unstable SEI, dendrite growth, and sluggish kinetics, together with the modification strategy and systematic analysis methods used in this study provide theoretical guidance for the development of other alloybased anode materials.

Experimental Section
Synthesis of Sn 4 P 3 and Sn 4 P 3 /C: Stoichiometric amounts of Sn powder (Aladdin, 99.999%) and red P (Aladdin, 99.999%) were loaded into a stainless steel jar (50 mL) with a ball (size: 1 mm)-to-powder mass ratio of 100:1. The loading process was conducted in a glove box (MIKROUNA) under an Ar atmosphere with the quantity of H 2 O and O 2 below 0.1 ppm. Thereafter, the jar was clamped to avoid contacting with air. It was then removed from the glove box and installed in a highenergy mechanical milling (HEMM) machine (AM400, Ants Scientific Instruments) at 500 rpm for 3 h. The Sn 4 P 3 /C composites were also prepared using HEMM under the same conditions, with the prepared Sn 4 P 3 samples and Super P as raw materials in a mass ratio of 7:3. All results presented herein were repeated at least three times.
Electrode Preparation: The active materials were mixed with Super P (Super C65, Timcal) and sodium alginate (SA, Aladdin) binder in the mass ratio of 7:2:1 in deionized water. The obtained slurry was then Figure 8. Improving the capacity retention of Sn 4 P 3 anodes through carbon hybridization in SIBs. a,b) Reversible specific capacity (a) and Coulomb efficiency (b) of Sn 4 P 3 /C and Sn 4 P 3 at 0.5 C (1 C = 600 mA g -1 for Sn 4 P 3 , 1 C = 500 mA g -1 for Sn 4 P 3 /C) during 180 cycles. c,d) Charge curves (c) and corresponding V versus dQ/dV plots (d) of Sn 4 P 3 /C and Sn 4 P 3 at the 2 nd and 100 th cycle. The sodiation processes at a relatively high potential, attributed to P, and at a low potential, resulting from Sn, are labeled with a dark yellow and blue rectangle, respectively. e,f) Charge curves (e) and corresponding V versus dQ/dV plots (f) of Sn 4 P 3 /C at the 96 th , 106 th , and 131 st cycle. These special cycles are marked in (a) for capacity activation analysis. g) Nyquist patterns of Sn 4 P 3 /C and Sn 4 P 3 at the 2 nd and 100 th cycle. h,i) SEM images of Sn 4 P 3 /C (h) and Sn 4 P 3 (i) at the 2 nd and 100 th cycle.
cast on the copper current collector at drying temperature of 80 °C. The electrodes were cut into ϕ14 disks with active material loading of ≈2 mg (thickness of about 35 µm) and further dried at 120 °C prior to battery assembly.
Battery Assembly: Electrochemical tests were conducted in 2032 half coin cells with active materials as the working electrode and glass fiber (Whatman GF/D) as the separator in AIBs. Li, Na, and K metal disks acted as counter electrodes in the LIBs, SIBs, and PIBs respectively, and the thickness of these disks are ≈0.5 mm. LiPF 6 (1 m) in an EC-DMC-FEC mixture (4.5/4.5/1 v/v/v), NaPF 6 (1 m) in an EC-DMC-FEC mixture (4.5/4.5/1 v/v/v), and KPF 6 (0.8 m) in an EC-DMC-FEC mixture (4.5/4.5/1 v/v/v) acted as electrolytes (the amount of electrolyte in each battery is 80 µL) in the LIBs, SIBs, and PIBs, respectively.
Electrochemical Measurement: Galvanostatic charge-discharge cycles at various C-rates (1 C = 1100 mA g -1 for LIBs, 1 C = 600 mA g -1 for SIBs, 1 C = 400 mA g -1 for PIBs) between 0.005-2 V at 25 °C and the GITT (pulse of 0.1 C for 30 min, with a 3 h interruption between each pulse) were conducted using a CT3001A system (Wuhan LAND). The specific capacity of AIBs are calculated based on the mass of Sn 4 P 3 for pure Sn 4 P 3 system or Sn 4 P 3 /C for Sn 4 P 3 /C composites. CV (voltage range: 0.005-2 V, scan rate: 0.2, 0.5, 1 and 2 mV s -1 ) and EIS (frequency range: 1 000 000-0.01 Hz, bias voltage: 0.005 V) were performed using a CHI800D system (CH Instruments). Alkali symmetric batteries with the same alkali metal as the working and counter electrodes were measured at the current density of 0.5 mA cm -2 with the capacity limit of 1 mAh cm -2 . Alkali metal-Cu batteries were measured with the current density of 0.5 mA cm -2 and the capacity limit of 1 mAh cm -2 during discharge, and then charged with the current density of 0.5 mA cm -2 and the voltage limit of 2 V.
Ex Situ XRD Measurement: 2032 coin batteries were first charged or discharged to the special potential marked in Figure 2a-c, and then disassembled in the glove box to extract the working electrode. After that, the electrode was rinsed with DMC to remove the residual electrolyte, and dried at room temperature. The dried electrode was then fixed on the XRD specimen stage and coated with Kapton tape to protect the sample from air exposure. XRD patterns were recorded using a Bruker D8 Advance diffractometer with a Cu K α radiation source operating at 40 KV and 40 mA. The test range was 10-80°, the step size was 0.02° and the dwell time was 0.1 s.
Ex Situ SEM Measurement: The dried electrode after disassembling in a glove box was first fixed on an SEM specimen stage, and sealed in a valve bag to protect the sample from air exposure. After that, the sample Scheme 1. Schematic of ion storage mechanism and failure theory of Sn 4 P 3 anode in AIBs. a) Illustration of various final discharge products of Sn 4 P 3 anodes in AIBs. The height of the column for each compound represents its theoretical specific capacity. b) Illustration of the dynamics and structural deterioration of Sn 4 P 3 anodes in AIBs. PIBs show sluggish dynamics compared with LIBs and SIBs, resulting from the larger radius of K + . The Sn 4 P 3 anode exhibits obvious phase separation behavior after cycling, i.e., Sn nanocrystalline surrounded by the amorphous P matrix. c) Illustration of the unstable SEI and dendrite growth of Sn 4 P 3 anodes in AIBs. The unstable SEI in PIBs and the serious dendrite growth in SIBs further result in capacityfading of Sn 4 P 3 anodes.
was transferred into the SEM sample chamber quickly with a very short exposure time to air (<30 s). SEM and corresponding EDS mapping images were taken using the Hitachi SU8010 SEM operating at 10 KV, 10 mA and 15 KV, 15 mA, respectively.
Ex Situ TEM Measurement: The sample powder was first obtained by scraping the dried electrode in a glove box, and then dispersed in ethyl alcohol. After that, the dispersion was dropped in to a standard Cu grid, and dried at room temperature. TEM images were taken using spherical aberration-corrected FEI Titan Themis Cubed G2 300 TEM operating at 300 KV.
Ex Situ XPS Measurement: The dried electrode after disassembling in the glove box was first fixed on to the XPS specimen stage, and then put in a Vacuum transfer chamber to avoid exposure to moisture. XPS patterns were recorded using a Thermo Fisher Escalab 250Xi XPS with Al K α radiation source. The binding energy was calibrated based on carbon contamination using the C1s peak at 284.8 eV.
Computational Method: DFT calculations were performed using the projector-augmented wave (PAW) potentials [39] and the Perdew-Burke-Ernzerhof (PBE) exchange-correlation functional [40] , as implemented in the Vienna Ab-initio Simulation Package. [41] The cut-off of the planewave kinetic energies was set to 500 eV. The amorphous Sn 4 P 3 structure was simulated using a periodic cubic supercell containing 24 Sn and 18 P atoms. The liquid-quench method was used to create amorphous structures via ab-initio molecular dynamics simulations. The heating temperature was 1000 K. The equations of motion were integrated with the Verlet algorithm using a time step of 1 fs, and the temperature was controlled by the velocity of rescaling and a canonical ensemble (NVT) using a Nosé-Hoover thermostat. Similar calculation schemes have been employed successfully in previous studies. [42,43] The formation energy of A x Sn 4 P 3 is defined as: (1) where E(A x Sn 4 P 3 ), E(Sn 4 P 3 ), and E(A) are the total energies of A x Sn 4 P 3 , Sn 4 P 3 , and bulk (bcc) A, respectively. The specific capacity C (mAh g -1 ) is calculated using the following equation: where n is the valence number of A, F is the Faraday constant at 26.8 Ah mol -1 , and M is the molecular mass of Sn 4 P 3 . The average voltage V is calculated as follows: where E(A x1 Sn 4 P 3 ), E(A x2 Sn 4 P 3 ), and E(A) are the total energies of the A x1 Sn 4 P 3 , A x2 Sn 4 P 3 , and Im3m alkali metals, respectively.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.