Unveiling the Sodium/Potassium Storage Mechanisms of Nanoporous Indium-Bismuth Anode Using Operando X-ray Diffraction

Material preparation

Microstructural characterizations

Electrochemical measurements

The XRD pattern of the Mg₉₂In₄Bi₄ foils (Fig. 1a) is indexed to Mg₃Bi₂ (JCPDS # 04–0464) and Mg (JCPDS # 35–0821) phases where In solubilizes due to its high solid solubility in Mg as depicted by the phase diagram of the Mg-In system (Fig. S1). ²⁵ The high content (92 at%) of Mg was chosen for not only exceeding the "parting limit" to avoid the compositional and geometric restrictions on dealloying, ^26,27 but also increasing porosity of the dealloyed samples for facilitating the ion transportation and mitigating the volumetric change. Additionally, the precursor composition was chosen to be close to the eutectic point of the Mg-Bi system according to the Mg-Bi phase diagram, in order to refine the size of the eutectic microstructure (Mg(In)/Mg₃Bi₂). After dealloying, the predominant peaks of Mg₃Bi₂ and Mg(In) phases of the Mg₉₂In₄Bi₄ foils disappeared, and the XRD pattern of the np-InBi alloy could be assigned to InBi (JCPDS # 32–0113) and Bi (JCPDS # 44–1246) phases (Fig. 1b). Here, the peaks belonging to Bi are much weaker in comparison with those of InBi. The Rietveld refinement for the XRD pattern of np-InBi was conducted to calculate the phase constitution of np-InBi (Fig. 1b and Fig. S2), showing that the mass fractions of the InBi and Bi phases are 94.7 and 5.3 wt% respectively and thus confirming the small amount of Bi in the np-InBi alloy. The SEM images (Figs. 1c–1d) elucidate that the np-InBi samples show a three-dimensional ligament-channel structure where the mean size of ligaments is 279.0 ± 70.8 nm (Fig. S3). Additionally, the EDX results (Fig. S4) reflect that the actual atomic ratio of the np-InBi alloy is close to its nominal composition. Only a minor amount of Mg is detected, and both the In and Bi elements are homogeneously distributed in the np-InBi alloy. TEM was utilized to further characterize the microstructure of the np-InBi alloy (Figs. 1e–1h). The high-resolution TEM (HRTEM) images (Figs. 1e–1g) reveal that the lattice fringes can be well distinguished, matching with the crystal planes of Bi (003), InBi (111) and (200), further confirming the presence of Bi and InBi in the np-InBi alloy. As depicted in Fig. 1h, the nanoporous feature can be clearly observed with the ligament size of 303.1 ± 109.4 nm (Fig. S3), which is consistent with the SEM results. Herein, the formation of such structure can be attributed to the dealloying procedure. During the dealloying process in the tartaric acid solution, the active element (Mg) in the Mg(In) and Mg₃Bi₂ phases of the Mg-In-Bi precursor was selectively etched away, while the inert elements (In and Bi) were retained and reorganized via surface diffusion to form the nano-sized ligament-channel structure. ²²

Zoom In Zoom Out Reset image size

To evaluate the electrochemical performance of the np-InBi electrode for SIBs, CVs and galvanostatic (dis)charge measurements were performed in the ester-based electrolyte. Figure 2a shows the CVs profiles of the np-InBi electrode during the initial cycles at the scan rate of 0.1 mV s⁻¹. In the first cathodic scan, the peaks at 0.22/0.36/0.52/0.85 V (vs Na⁺/Na) can be attributed to the SEI formation and sodiation of np-InBi, followed by the peaks at 0.09/0.52/0.62/0.73/0.79 V (vs Na⁺/Na) in the anodic scan which are associated with the desodiation of np-InBi. In subsequent cycles, a slight shift of cathodic peaks can be observed accompanied with the well overlapping of the anodic peaks. Figure 2b shows the discharge-charge curves of the np-InBi electrode during the initial cycles at 0.2 A g⁻¹, which delivers the initial discharge capacity of 474.6 mAh g⁻¹ with the initial Coulombic efficiency (ICE) of 89.8% (Fig. S5). Notably, such high ICE is important for the potential industrial application accompanied with the strict electrolyte amount control. ²⁸ However, the cycling stability of the np-InBi electrode is unsatisfactory, and the specific capacity rapidly decays within 15 cycles. The poor cycling stability stems from the pulverization/aggregation and unstable SEI film formation induced by dramatic volumetric variation and ester-based electrolyte. ²⁹

Zoom In Zoom Out Reset image size

The ion/electron transfer properties were further probed with EIS and GITT. Figure 2c shows the Nyquist plot of the np-InBi electrode at OCV and the charged state after the 1st cycle, which are further analyzed by fitting equivalent circuits (Fig. S6). The Warburg factor (Fig. 2d) was obtained from the relationship between Z' and ω^−1/2 in the low frequency range and utilized to roughly calculate the diffusion coefficient of Na⁺ (D_Na-eis ) in terms of the following equation, ^30,31

$$\begin{eqnarray}&&Z^{\prime} ={R}_{ct}+{R}_{e}+\sigma {\omega }^{-\displaystyle \frac{1}{2}}\end{eqnarray} \tag{ 1 }$$

$$\begin{eqnarray}&&{D}_{Na-eis}=\displaystyle \frac{{R}^{2}{T}^{2}}{2{A}^{2}{n}^{4}{F}^{4}{C}_{Na}^{2}{\sigma }^{2}}\end{eqnarray} \tag{ 2 }$$

where R, T, A, n, F, C_Na , ω, and σ represent the gas constant, absolute temperature, electrode surface area, number of electrons per molecule, Faraday constant, Na⁺ concentration in the electrolyte (herein, 1 M) utilized to simplify the gradient concentration of Na⁺ which is suitable under the relative high frequency area, ³⁰ angular frequency, and Warburg factor. From the calculated results (Table SI), the charge transfer resistance (R_ct ) becomes smaller and D_Na-eis increases after the 1st cycle, indicating the reduction of impedance and the improved ion diffusion/electron transfer induced by the electrochemical effects. As shown in Figs. 2e–2f, GITT was performed for the np-InBi electrode by intermittently repeated short pulses (current density, 0.05 A g⁻¹) for 1800 s, followed by an open circuit for 7200 s to obtain the equilibrium potential during cycling. The D_Na-gitt values at different potentials were estimated according to the Fick's second law of diffusion and following simplified equation, ^32–34

$$\begin{eqnarray}&&{D}_{Na-gitt}=\displaystyle \frac{4}{\pi \tau }{\left(\displaystyle \frac{{n}_{m}{V}_{m}}{A}\right)}^{2}{\left(\displaystyle \frac{{\rm{\Delta }}{E}_{s}}{{\rm{\Delta }}{E}_{\tau }}\right)}^{2}\end{eqnarray} \tag{ 3 }$$

where τ is the pulse duration (1800 s), n_m is the number of moles, V_m is the molar volume of the alloy obtained from the Materials Project Database (fixed without considering the volume variation), ³⁵ A is the electrode-electrolyte interface area, ΔE_τ and ΔE_s can be obtained from Fig. 2e. The D_Na-gitt values during the discharge process are considerably smaller than those during the charge process, elucidating the facile diffusion capability after the initial electrochemical activation. Meanwhile, the average voltage drop during 0.22−0.54 V (vs Na⁺/Na) is 22 mV, smaller than that (28 mV) during 0.58−0.75 V (vs Na⁺/Na), suggesting that the electrochemical kinetics of NaBi + In + Na₁₅In₂₇ → Na₃Bi + Na₁₅In₂₇ is facile compared with that of InBi + Bi → NaBi + In + Na₁₅In₂₇ during the sodiated process for np-InBi (Fig. 2e). ³⁶ Notably, Lim et al. revealed the possible sluggish kinetics accompanied with the crack formation of NaBi → Na₃Bi compared with that of Bi → NaBi for Bi anode. ³⁷ Hence, such improvement may be related with the formation of Na-In alloy, which inversely accelerates the further sodiation of Bi component.

The electrochemical performance of the np-InBi anode was also investigated for PIBs in the ester-based electrolyte. Figure 3a shows the CVs of the np-InBi electrode in the initial five cycles at 0.1 mV s⁻¹. In the 1st cathodic scan, the peaks at 0.09/0.53 V (vs K⁺/K) are assigned to the SEI formation and potassiation process, while the peaks at 0.73/1.00/1.08/1.16 V (vs K⁺/K) are associated with the depotassiation process of the np-InBi electrode. Figure 3b shows the discharge-charge profiles of the np-InBi electrode at 0.2 A g⁻¹, delivering an initial discharge capacity of 458.8 mAh g⁻¹ with an ICE of 65.3% (Fig. S5). Despite the high specific capacity, the np-InBi electrode shows inferior cycling stability, and the discharge capacity decreases to less than 200 mAh g⁻¹ within 15 cycles. Similar to the scenario in SIBs, the poor cycling stability is related with the inferior structural integrity of the electrode and ester-based electrolyte. ³⁸ Hence, the cycling stability of InBi was characterized in the ether-based electrolytes for SIBs and PIBs. As depicted in Fig. S7, the np-InBi anode displays the well improved cycling stability with the capacity retentions of 335.3 and 248.6 mAh g⁻¹ after 100 cycles for SIBs and PIBs respectively, much better than those (less than 200 mAh g⁻¹ merely over 15 cycles) in the ester-based electrolytes. Such results elucidate that the ether-based electrolyte can play the vital role to elevate the electrochemical performance, which may be attributed to the formation of thin and dense SEI film generated by the ether-based electrolyte. In addition, the cycling stability of p-In was characterized in the ester- and ether-based electrolytes for SIBs and PIBs at 0.2 A g⁻¹ (Figs. S8–9), all exhibiting inferior performance compared with that of np-InBi.

Zoom In Zoom Out Reset image size

EIS (Figs. 3c–3d) and GITT (Figs. 3e–3f) analyses were performed to elucidate the ion/charge transportation according to the above-mentioned equations like those of SIBs. From the EIS results (Table SII), R_ct decreases and D_K-eis increases after the 1st cycle, elucidating that the electrochemically-driven enhancement of ion/electron transfer capacity. Meanwhile, the D_K-gitt values of potassiation are substantially smaller than those of the depotassiation process (Fig. 3f). The average voltage drop during 0.17–0.39 V (vs K⁺/K) is 133 mV, much smaller than that (190 mV) during 0.41–0.93 V (vs K⁺/K), suggesting that the electrochemical kinetics of KBi₂ + Bi₃In₅ → K₃Bi + K₈In₁₁ is facile compared with that of InBi + Bi → KBi₂ + Bi₃In₅ during the potassiated process for np-InBi, ³⁶ which implies that the formation of Bi₃In₅ can facilitate the further potassiation of np-InBi (Fig. 3e).

To clarify the sodiation/desodiation mechanisms, the phase transition of the np-InBi electrode was probed by operando XRD technique during the initial discharge-charge-discharge processes between 0.01–2.0 V (vs Na⁺/Na) at 0.025 A g⁻¹ (Fig. 4). At the beginning of the discharge, the peaks at 25.7°/31.4°/35.8°/37.8°/41.8°/44.6° and 27.2°/37.9°/39.6°/44.6°/45.9°/46.0°/48.7° can be attributed to InBi and Bi, respectively, while the peak at 45.9° is assigned to the stainless-steel mesh. Meanwhile, the peaks at 38.6°/41.3°/44.0° belong to BeO (JCPDS # 43–1000) stemming from the oxidation of Be surface facing to the air. ³⁹ In Stage 1 (1.78−0.39 V (vs Na⁺/Na)), the peaks belonging to InBi and Bi diminish and eventually disappear, accompanied with the appearance and gradual increase of peaks at 18.2°/25.5°/31.6°/36.5°/37.4°/41.2°/45.8°, 19.2°/20.1°/20.9°/21.6°/22.4°/23.3°/32.5°/32.9°/36.7°/37.2°/38.5° and 33.0°/36.3°/39.2° denoted to the NaBi (JCPDS # 04–0701), Na₁₅In₂₇ (JCPDS # 65–1163) and In (JCPDS # 05–0642), respectively. As the discharge proceeding in Stage 2 (0.39–0.01 V (vs Na⁺/Na)), the peaks of NaBi and In gradually fade away and finally vanish, accompanied by the emergence and strengthening of peaks at 18.4°/18.8°/20.9°/26.3°/32.8°/33.7°/37.0°/37.8°/39.1°/47.6° attributed to Na₃Bi (JCPDS # 04–0351). Meanwhile, the intensities of peaks assigned to Na₁₅In₂₇ further increase, related to the enhancement of crystalline degree induced by the electrochemical effects. Hence, this process is associated with the complete transformation from InBi + Bi to Na₃Bi + Na₁₅In₂₇ via the immediate product of NaBi + Na₁₅In₂₇ + In. When being charged back to 2.0 V (vs Na⁺/Na) (Stage 3 and 4), inverse dealloying reactions occur, correlated with the dealloying of Na₃Bi + Na₁₅In₂₇ to generate InBi + Bi via NaBi + Na₁₅In₂₇ + In. Regarding the 2nd discharge, the similar alloying process of InBi + Bi → NaBi + Na₁₅In₂₇ + In → Na₃Bi + Na₁₅In₂₇ can be detected, implying the high reversibility of such reaction. The contour plot (Fig. 4a) schematically demonstrates the corresponding phase evolution process of the np-InBi electrode in SIBs.

Zoom In Zoom Out Reset image size

Notably, In is generally regarded as the "relatively inactive material," which delivers the limited storage capability. For example, Veith et al. reported that the pure In film exhibits the low initial reversible capacities (75 mAh g⁻¹) and needs tens of cycles to be activated to reach the highest capacity of 125 mAh g⁻¹, merely a quarter of its theoretical value (467 mAh g⁻¹). ¹⁴ From the above-mentioned results, the Na₁₅In₂₇ phase can form in the 1st cycle without any activation and correspond to the specific capacity of 129 mAh g⁻¹, comparable to that of pure In after being activated. Such results suggest that the alloying strategy via introducing the traditional Na-active material (Bi) can significantly accelerate the activation of In. Lim et al. have reported the delayed structural transformation of Bi during the sodiation, interpreting that the NaBi can tolerate a considerable amount of excess Na⁺ ions in a metastable structure before transforming into Na₃Bi structure. ³⁷ Hence, NaBi can serve as the stable Na-containing channels, facilitating the sodiation of In with the high sodiophilicity of NaBi. ⁴⁰

Simultaneously, operando XRD was utilized to investigate the (de)potassiated mechanism of the np-InBi electrode in PIBs (Fig. 5). During the initial discharge in Stage 1 (1.61–0.40 V (vs K⁺/K)), the peaks belonging to InBi and Bi gradually diminish and finally disappear, accompanied with the appearance and proliferation of peaks at 16.1°/31.2°/32.6°/37.8°/49.8° and 28.3°/29.5°/31.4°/33.0°/42.7°/43.9°/47.4°/48.8° relevant to KBi₂ (JCPDS # 03–0698) and Bi₃In₅ (JCPDS # 23–0850), respectively. With the continuous discharge in Stage 2 (0.40–0.01 V (vs K⁺/K)), the peaks at 28.9°/29.6°/33.2° and 30.6°/30.9°/31.9°/32.4°/35.7°/40.8°/41.4°/41.8°/43.6° indexed to K₃Bi (JCPDS # 04–0642) and K₈In₁₁ (JCPDS # 37–0002) respectively, start to appear and strengthen at the expense of the sustaining declination of KBi₂ and Bi₃In₅. Such procedure is correlated with complete transformation from InBi + Bi to K₃Bi + K₈In₁₁ through the immediate products of KBi₂ + Bi₃In₅. Upon the desodiation process (Stage 3, 0.01–0.7 V (vs K⁺/K)), the reoccurrence and continuous growth of the peaks of KBi₂ and Bi₃In₅ can be probed, accompanying with the intensities of peaks indexed to K₃Bi and K₈In₁₁ steadily decreasing and finally vanishing, indicative of the depotassiated conversion from K₃Bi + K₈In₁₁ to KBi₂ + Bi₃In₅. When being fully charged (Stage 4, 0.7–2.0 V (vs K⁺/K)), the peaks belonging to KBi₂ and Bi₃In₅ begin to decrease and eventually disappear with the occurrence and enhancement of peaks attributed to InBi and Bi. In the subsequent 2nd discharge (Stage 5–6) processes, similar alloying procedure of InBi + Bi → KBi₂ + Bi₃In₅ → K₃Bi + K₈In₁₁ can be found, but the peak intensities of the discharged products are markedly smaller than those during the first discharge process. Hence, the (de)potassiation mechanism of the np-InBi electrode is correlated with the (de)alloying processes of InBi + Bi ↔ KBi₂ + Bi₃In₅ ↔ K₃Bi + K₈In₁₁. Particularly, compared with the scenario in SIBs, the K₈In₁₁ forms in PIBs and is relevant to the higher specific capacity of 169.8 mAh g⁻¹ even under the more sluggish kinetics of K⁺ in contrast to that of Na⁺. ⁴¹ The improved performance may be related to the generation of intermetallic phase (Bi₃In₅) except for KBi₂ during the middle stage of (dis)charge, differing from the phenomenon (NaBi + In) in SIBs. The formation of such intermetallic phase can realize the superior potassiation depth of In by achieving more uniform mixing of Bi and In compared with pure In in SIBs. In addition, only the crystalline (de)sodiated/(de)potassiated products of np-InBi can be detected due to the limited energy density in the lab-scale XRD equipment, where the amorphous phases need to be evidenced by high-resolution synchrotron in the following work.

Zoom In Zoom Out Reset image size

In summary, we fabricated the np-InBi alloy by chemical dealloying of the Mg-In-Bi precursor. The np-InBi alloy exhibits a three-dimensional ligament-channel structure with the average ligament size of 279.0 ± 70.8 nm (based upon the SEM results). As an anode for SIBs, the np-InBi anode shows the initial discharge capacity of 474.6 mAh g⁻¹ with the ICE of 89.8%. Even in PIBs, the np-InBi anode could display the initial discharge capacity of 458.8 mAh g⁻¹. By utilizing the operando XRD, the (de)sodiated/(de)potassiated mechanisms (crystalline products) of np-InBi were proposed, involving the (de)alloying processes of InBi + Bi ↔ NaBi + In + Na₁₅In₂₇ ↔ Na₃Bi + Na₁₅In₂₇ ((de)sodiation) and InBi + Bi ↔ KBi₂ + Bi₃In₅ ↔ K₃Bi + K₈In₁₁ ((de)potassiation). The present results could provide valuable information on understanding of electrochemical processes and design of advanced alloy anode materials for SIBs and PIBs.

The authors gratefully acknowledge financial support by the National Natural Science Foundation of China (51871133), Taishan Scholar Foundation of Shandong Province, the Key Research and Development Program of Shandong Province (2021ZLGX01), and the program of Jinan Science and Technology Bureau (2019GXRC001).