In Situ Formation of Hierarchical Bismuth Nanodots/Graphene Nanoarchitectures for Ultrahigh‐Rate and Durable Potassium‐Ion S

FULL PAPER

www.small-journal.com In Situ Formation of Hierarchical Bismuth Nanodots/ Graphene Nanoarchitectures for Ultrahigh-Rate and Durable Potassium-Ion Storage

Yuanxin Zhao, Xiaochuan Ren, Zhenjiang Xing, Daming Zhu,* Weifeng Tian, Cairu Guan, Yong Yang, Wenming Qin, Juan Wang, Lili Zhang, Yaobo Huang, Wen Wen,* Xiaolong Li, and Renzhong Tai*

urgent strategies worldwide.[1] Recharge- Metallic bismuth (Bi) has been widely explored as remarkable anode mate- able lithium-ion batteries have achieved rial in alkali-ion batteries due to its high gravimetric/volumetric capacity. great success during the last 40 years, while However, the huge volume expansion up to ≈406% from Bi to full potassia- they gradually display certain limitations tion phase K Bi, inducing the slow kinetics and poor cycling stability, hinders in further large-scale applications, such as 3 high cost, uneven geological distribution its implementation in potassium-ion batteries (PIBs). Here, facile strategy is and short supplies of lithium resources developed to synthesize hierarchical bismuth nanodots/graphene (BiND/G) (0.0017 wt%) around the world. As alternative composites with ultrahigh-rate and durable potassium ion storage derived energy storage sources, sodium ion batteries from an in situ spontaneous reduction of sodium bismuthate/graphene (SIBs) and potassium ion batteries (PIBs) composites. The in situ formed ultrafine BiND ≈( 3 nm) confined in graphene have recently attracted tremendous interest owing to their natural abundance, low cost layers can not only effectively accommodate the volume change during the and environmental friendliness. Despite alloying/dealloying process but can also provide high-speed channels for having a similar abundance with sodium ionic transport to the highly active BiND. The BiND/G electrode provides (sodium and potassium represent 2.36 and a superior rate capability of 200 mA h g−1 at 10 A g−1 and an impressive 2.09 wt% in the Earth’s crust, respectively), reversible capacity of 213 mA h g−1 at 5 A g−1 after 500 cycles with almost no potassium presents some specific advan- + capacity decay. An operando synchrotron radiation-based X-ray diffraction tages. K /K exhibit a lower standard redox potential of −2.93 V (vs E°) compared with reveals distinctively sharp multiphase transitions, suggesting its underlying that of Na+/Na (−2.71 V vs E°), implying a operation mechanisms and superiority in potassium ion storage application. higher working voltage and energy density of PIBs.[2] Moreover, potassium ions have much better conductivity and relatively lower 1. Introduction desolvation energy in organic solvents.[3] These merits of potassium make it a promising low-cost candidate for high-energy and With the tremendous demand for applications from mobile elec- power density energy storage applications. tronic devices, vehicles and stationary power stations, developing The progress of PIBs mainly follows the development of efficient energy storage technologies has become one of the most electrode materials, especially considering the large-size of

Y. X. Zhao, Dr. X. C. Ren, Z. J. Xing, Dr. D. M. Zhu, Dr. W. M. Qin, Y. X. Zhao, C. R. Guan, Prof. Y. Yang Dr. J. Wang, Dr. L. L. Zhang, Prof. Y. B. Huang, Dr. W. Wen, School of Physical Science and Technology Prof. X. L. Li, Prof. R. Z. Tai ShanghaiTech University Shanghai Institute of Applied Physics Shanghai 201210, China Chinese Academy of Sciences Y. X. Zhao, Z. J. Xing, C. R. Guan Shanghai 201800, China University of Chinese Academy of Sciences E-mail: [email protected]; [email protected]; Beijing 100049, China [email protected] W. F. Tian Y. X. Zhao, Z. J. Xing, Dr. D. M. Zhu, Dr. W. M. Qin, Dr. J. Wang, College of Science Dr. L. L. Zhang, Prof. Y. B. Huang, Dr. W. Wen, Prof. X. L. Li, Prof. R. Z. Tai Henan University of Technology Shanghai Synchrotron Radiation Facility Zhengzhou 450001, China Shanghai Advanced Research Institute Chinese Academy of Sciences Shanghai 201204, China The ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10.1002/smll.201905789.

DOI: 10.1002/smll.201905789

Figure 1. Schematic diagram of the novel anode materials designation. a) 2D-SB laminated with graphene, b) in situ formed BiND/G, and c) KxBiy/ graphene nanocomposites. potassium ions, which draw great efforts to explore stable and These two completely different transition processes indicate high energy density electrode materials. Fortunately, signifi- that the underlying mechanism is strongly associated with the cant advances have been achieved in cathode materials, such unique structure and size of the anode material. Despite this [4] [5] [6] as Prussian blue, Prussian green, amorphous FePO4, great progress, research on the potassium storage properties [7] [8] FeSO4F, and Fe/Mn-based oxides, which exhibit competitive and mechanisms of nanosized Bi is generally in its early stage. energy densities with their counterparts in LIBs and SIBs. How- A facile synthesis of Bi nanoarchitecture and an in-depth explo- ever, as for anode materials, the achieving capacity and cycling ration of their related potassiation/depotassiation mechanisms performance as previously reported are far lower than those in are thus of particular scientific and technological significance. LIBs and SIBs. The former promising candidate for the anode Herein, we developed a novel bismuth nanodots/graphene of PIBs was predicted to be carbonaceous materials in regard (BiND/G) hierarchical composite in situ formed from 2D to their inexpensiveness and high electronic conductivity. sodium bismuthate (2D-SB) for ultrahigh-rate and durable Potassium ions could intercalate into graphite, similar to the potassium ion storage. Sodium bismuthate (SB) is a commer- intercalation behavior of lithium in LIBs,[9] but demonstrated cially available, inexpensive, nontoxic and potent inorganic low energy density because of the limited theoretical capacity oxidant. Due to its strong oxidation characteristics, SB can be of carbonaceous materials. Another series of PIB anodes can naturally reduced to metallic Bi under mild conditions (laser[17] alloy/dealloy with K (such as Sb,[10] Sn,[11] Bi,[12,13] and P[14]) or electron-beam irradiation[18]). Based on these characteristics and display much higher energy density. Among these mate- of SB, we have innovatively succeeded in obtaining 2D-SB by rials, bismuth (Bi) is an ideal choice due to its nontoxicity, large exfoliation and made it laminate with graphene, which is used lattice spacing and high theoretical capacity (385 mA h g−1). as a precursor to synthesize BiND/G (Figure 1a). The 2D-SB Recently, Zhang’s group achieved Bi microparticles as an anode can be easily in situ converted to BiND with isolated ultrafine for potassium storage that could deliver a stable capacity up to granules (≈3 nm) due to the 2D size effect. Spontaneous reduc- 392 mA h g−1 at 0.4 A g−1 after 100 cycles, corresponding to tion of 2D-SB/G to obtain BiND/G greatly simplified the fabri- an almost 97% capacity retention.[13] However, because K+ has cation process, providing a straightforward way to achieve the a much larger ionic radius (1.38 Å) than Na+ (0.97 Å) and Li+ ultrafine Bi nanodots/graphene composite (Figure 1b). The (0.76 Å), the potassiation and depotassiation processes invoke unique structure of BiND/G (shown in Figure 1b,c) efficiently huge volumetric changes and slow kinetics, resulting in fast accommodates the volume change because of the isolated capacity decay and poor rate capability. To address these issues, BiND and graphene and enables fast ionic and electronic trans- the design of Bi anode nanoarchitecture has been demon- port to the high-active BiND during the alloying/dealloying pro- strated as a key strategy that can suppress mechanical fracture. cess (Figure 1c). A robust solid electrolyte interface (SEI) film For instance, Lu et al.[15] developed unique ultrathin carbon can be formed, and better conductivity can be achieved benefit- film/carbon nanorods/Bi nanoparticle materials for a long cycle ting from the super hierarchical nanoarchitecture. As a result, life and exhibited a decent capacity of 121 mA h g−1 at a high BiND/G realizes high-rate capability (200 mA h g−1 at 10 A g−1) current density of 1 A g−1 with 75% capacity retention after and unprecedented cycling stability (213 mA h g−1 at 5 A g−1 700 cycles. Nevertheless, this anode suffered from a complex after 500 cycles with almost no capacity decay). Operando syn- fabrication process and limited electrochemical performance chrotron radiation-based X-ray diffraction (SR-XRD) reveals due to the heavy aggregation during cycling, restricting its fur- that the electrochemical mechanism of the BiND/G undergoes ther application. Exploring a simple and efficient method to sharp and stable phase transitions, which directly contributes to synthesize ultrafine Bi granules can be critical for pushing the its excellent electrochemical performance. Bi anode of potassium ion battery technology to the next level. What is more, the size of crystallite not only influences the electrochemical kinetics process but also may lead to different 2. Results and Discussion electrochemical mechanisms. Zhang’s group revealed that Bi microparticles suffer reversible, stepwise Bi–KBi2–K3Bi2–K3Bi Exfoliated 2D-SB has a well-maintained hexagonal structure transitions after the initial continuous surface potassiation.[13] similar to sodium bismuthate (Figure S1a, Supporting Informa- Guo’s work found that Bi nanoparticles undergo a solid-solution), and their XRD patterns are shown in Figure S1b,c (Sup- tion reaction, followed by two typical two-phase reactions, cor- porting Information). Figure S1d,e (Supporting Information) [16] responding to Bi–Bi(K) and Bi(K)–K5Bi4–K3Bi, respectively. shows that diffraction peaks for (0 0 2) and (1 0 0) of sodium

Figure 2. a) AFM images of 2D-SB; b) corresponding AFM height profiles along lines I, II, III, and IV (a); c) TEM image of 2D-SB; d) HRTEM image of 2D-SB, inset: high-resolution lattice fringes of 2D-SB, corresponding to the (1 0 0) plane; e) TEM images of BiND; f) HRTEM images of BiND, green circles emphasize the apparently isolated BiNDs, inset: high-resolution lattice fringes of BiND, corresponding to the (0 1 2) plane; g) X-ray absorption spectra of 2D-SB, BiND, and Bi powder reference; and h) Raman spectra of 2D-SB and BiND. bismuthate shift 0.36° and 0.3° toward the lower 2θ angle direc- Figure 2c,d) and display unfixed sheet structures. The inset of tion after exfoliation, indicating lattice expansion along both the HRTEM image (Figure 2d) displays an interplanar spacing the c- and a-axes and demonstrating the noteworthy size reduc- of ≈0.48 nm, corresponding to the (1 0 0) crystal plane of the tion. The evident Tyndall effect also suggests the successful hexagonal sodium bismuthate. Due to the strong oxidation preparation of samples with nanosized thickness in solution characteristics of 2D-SB, it can naturally in situ formed BiND (Figure S2, Supporting Information). To directly estimate the under mild condition.[17,18] Even in the cell, 2D-SB can be easily thickness of 2D-SB, atomic force microscopy (AFM) analysis in situ reduced to BiND by spontaneous electron reduction was performed, as displayed in Figure 2a,b. The average topo- (detailed mechanisms in the Supporting Information). To fur- graphic height is ≈1.2 nm, corresponding to 3–5 atomic layers. ther understand the internal structure of reduced BiND, TEM The morphology and lattice fringes of 2D-SB are investigated images are taken and displayed in Figure 2e,f. Figure 2e clearly using transmission electron microscopy (TEM) (as shown in shows that the original homogenous sheet structure converts

Small 2019, 1905789 1905789 (3 of 9) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.small-journal.com to isolated ultrafine nanodots with a size of≈ 3 nm, as empha- of 0.5 A g−1) are shown in Figure 3b, and display a stable dis- sized by the green circles in Figure 2f. The HRTEM image of charge capacity of 240 mA h g−1 even after 50 cycles with an the BiND presents apparent parallel fringes of 0.33 nm corre- excellent Coulombic efficiency of≈ 100%. The capacity loss sponding to the metallic bismuth (0 1 2) crystal plane (inset in during the initial several cycles may be due to the activation Figure 2f). The Bi LIII X-ray absorption spectra of 2D-SB, BiND process, namely, the formation of SEI and high-speed channels and Bi powder are displayed in Figure 2g, where the absorp- for ionic and electronic transport. Figure 3c displays the voltage tion edge of BiND is the same as that of the metallic bismuth, profiles of the BiND/G at different current densities ranging demonstrating that the 2D-SB was reduced to BiND with zero from 0.1 to 10 A g−1, where they are of similar shapes, implying valence. The Raman spectra of the reduced samples (Figure 2h) little polarization effect. The discharge plateaus at 0.92, 0.48, display two distinct scattering peaks at 73 and 95 cm−1, corre- and 0.35 V can still be clearly observed at a high current den- −1 sponding to the well-known modes Eg and A1g of Bi, respec- sity of 10 A g , suggesting the ultrafast kinetic process. The tively.[19] The disappearance of the Raman peak located at BiND/G structure guarantees a short ionic transportation path 650 cm−1, which is assigned to 2D-SB, further confirms that and abundant potassium transfer channels, which benefit ionic 2D-SB has completely converted to BiND.[20] The relatively conductivities. The ultrafast kinetics can be further observed low melting point of Bi usually leads to serious agglomeration in Figure 3d, where the rate performance of BiND/G is dis- during the formation process, even under the low-temperature played and there are outstanding reversible capacities of 320, hydrothermal reduction reaction or ball-milling, so that the 260, 240, 230, 225, and 215 mA h g−1 at current densities of 0.1, size of the currently reported Bi nanoarchitecture material is 0.2, 0.5, 1, 2, and 5 A g−1, respectively. Even at an ultrahigh cur- more than 50 nm.[15,16,21] Here, in situ formed BiND are pre- rent density of 10 A g−1, a reversible capacity of 200 mA h g−1 sented as disconnected ultrafine granules with a size of only is still delivered. When the current density is switched back to ≈3 nm. The unique structure may yield excellent electrochem- 0.5 A g−1 after cycling at various current densities for 35 cycles, ical performance in PIB and forming a hierarchical structure the specific capacity recovers back to 240 mA h −g 1 with almost with graphene could further improve material stability and no capacity decay, indicating excellent structural stability of the ionic transport ability and suppress agglomeration. The 2D-SB electrode under a wide range of current (shown in Figure S7, synthesized via exfoliation can be effectively assembled layer Supporting Information). Figure 3e presents the galvanostatic by layer with graphene due to their similar 2D structure. In electrochemical cycling performance of BiND/G at 5 A g−1, this work, we obtain a novel 2D-SB/G nanocomposite film where it can deliver a reversible capacity of 213 mA h g−1 with (Figure S3, Supporting Information) by filtrating a 2D-SB col- high capacity retention of 99% over 500 cycles (The capacity con- loidal solution and graphene (the Raman spectrum is displayed tribution of graphene is shown in Figure S8, Supporting Infor- in Figure S4, Supporting Information) together, and a hierar- mation). When the current density increased to 10 A g−1, the chical BiND/G nanostructure is in situ formed after battery specific capacity of BiND/G could still stabilize at 180 mA h g−1 assembly. Figure S5 (Supporting Information) shows the mor- over 500 cycles, corresponding to a capacity decay of ≈0.024% phology and elemental distribution of the 2D-SB/G nanocom- per cycle (Figure 3f). The Coulombic efficiencies are surpris- posite, where Na, Bi and O elements are uniformly distributed. ingly ≈100%, even at this high current density. After the initial Moreover, Figure S6 (Supporting Information) displays a uni- activation process, a high reversible capacity and outstanding form distribution of 2D-SB in the cross section and suggests Coulombic efficiency could be maintained, indicating rapid a well-integrated structure with graphene. The advantages of stabilization of the electrode–electrolyte interface. Such state- this unique hierarchical structure are profound: 1) ultrafine of-the-art stable capacity and rate performance are attributed to nanoscale Bi granules could offer a large specific surface area, the highly active BiND and hierarchical structure of BiND/G. high activity and superior contact with the electrolyte, 2) iso- To further evaluate the electrochemical behavior of BiND/G, lated BiND and graphene composites could provide a large the selected initial cyclic voltammetry (CV) curves at a scan space to buffer volumetric expansion and inhibit agglomera- rate of 0.2 mV s−1 are shown in Figure 4. The initial cathodic tion, and 3) ordered hierarchical structures could enhance fast process presents a gradual curve from 0.9 to 0.5 V and can be ionic transport channels and ensure short transport paths. assigned to SEI layer formation on the BiND/G surface (shown The electrochemical potassium storage performance of the in Figure S9, Supporting Information). The broad reduction BiND/G composite is evaluated in Figure 3. Figure 3a displays peak at 0.26 V can be attributed to the alloy reaction between the galvanostatic charge/discharge curves of BiND/G for the Bi and K. During the following cathodic processes, three reduc- initial three cycles at a current density of 100 mA g−1 with a tion peaks at 0.36, 0.48, and 0.86 V can be observed and are voltage window of 0.1–2 V. The first discharge–charge capaci- attributed to a stepwise discharge platform located at 0.3, 0.48, ties are 429/337 mA h g−1, indicating an initial coulombic effi- and 0.88 V, respectively. The anodic process has a large oxida- ciency of 78%, which can be ascribed to SEI film formation on tion peak at 0.59 V and a small shoulder peak at 0.46 V, which the electrode surface, electrolyte decomposition and irrevers- is attributed to the charge platform located at 0.50 V. Upon ible potassium ion insertion in defective sites. The following further depotassiation, the anodic peaks located at 0.72 and electrochemical cycling profiles almost overlap with a small 1.16 V appeared, and these peaks correspond to charge plateaus capacity decay and voltage hysteresis. The galvanostatic charge/ located at 0.7and 1.12 V, respectively, as shown in Figure 3a. The discharge profiles of BiND/G depicted three plateaus in both cathodic and anodic current peaks kept their shapes during the the charge and discharge processes, corresponding to the step- subsequent scans, implying identically reversible redox reac- wise potassiation and depotassiation reactions. Galvanostatic tions. Furthermore, the response current of the fifth CV curve electrochemical cycling curves of BiND/G (at a current density shows a slight increase when compared with the third cycle,

Figure 3. Electrochemical performance of BiND/G as an anode for PIBs. a) Galvanostatic charge–discharge curves at 0.1 A g−1, b) the cycling performance at 0.5 A g−1, c) charge–discharge profiles for the battery cycled at various current densities (0.1–10 A g−1); d) rate performance at various current densities; e) the cycling performance at a large current density of 5 A g−1 after five cycles of activation by 1 A −g 1; and f) the cycling performance at a large current density of 10 A g−1 after five cycles of activation by 1 A g−1. The specific capacity was calculated based on the total mass of the original 2D-SB/G composite. which corresponds to the capacity activation process during (high frequency region) and the inclined line (low frequency the first few cycles and coincides with the galvanostatic charge– region) can be related to charge-transfer resistance and the dif- discharge observation in Figure 3. Benefitting from the novel fusion of K ions into BiND/G, respectively. Compared to the 1st structure of BiND/G, K ions make better contact with the active cycle, the resistance of BiND/G does not change significantly material. These merits are further elucidated in Figure S10 after the 10th cycle, and barely increases after the 100th cycle, (Supporting Information), where the redox peaks show similar indicating that the robust SEI layer was formed after the first −1 shapes at various scan rates from 0.2 to 3 mV s , suggesting cycle. The BiND/G still delivers a very low Rct value after 100th superior conductivity and ultrahigh ionic transport speed. To cycle (the value is shown in Table S1, Supporting Information), gain further insights into the electrochemical kinetics of the which could be attributed to the stable SEI film and preeminent BiND/G electrode, electrochemical impedance spectroscopy conductivity of the nanoscale electrode material due to the short (EIS) measurements were conducted. The Nyquist plots are ionic transport path and high-speed channels. Furthermore, the displayed in Figure 4b, where the depressed semicircle section ionic diffusion resistance decreases as the number of cycles

Figure 4. a) Selected CV curves at a sweeping rate of 0.2 mV s−1 for BiND/G. b) EIS studies of BiND/G at different cycles. The inset shows the EIS plots at the high frequency region and equivalent circuit of the EIS fitting, Rs represent Ohmic resistance of electrode, RSEI is the SEI resistance, CPE1 correspond to the constant phase element of SEI, Rct is the charge transfer resistance, CPE2 correspond to the constant phase element of charge transfer, Zw represents the Warburg impedance related to the diffusion of potassium ions. c) Comparison of the capacity retention from previously reported PIB anodes with that of our work. increases and stabilizes after the 10th cycle, further suggesting variation nor peak shift in the XRD pattern. In the meantime, fast potassium diffusion kinetics into the electrode, which no diffraction peaks related to the Bi-K alloy structure arise, agrees well with the rate performance results mentioned above. which is contrary to the surface potassiation route of Bi micro- The phenomenon can be ascribed to the hierarchical nanoscale particles accompanied by the coexistence of Bi, KBi2 and K3Bi2 structure, which facilitates ultrafast K+ ion diffusion and ena- phases.[13] This is mainly due to the formation of a thick SEI bles superior contact between the active material and electrolyte. layer directly on the wrapped graphene layers, instead of the Consequently, the outstanding electrochemical performance individual BiND, which greatly favors the structural integrity of BiND/G is superior to pure Bi (shown in Figure S11, Sup- of the whole BiND/G electrode.[15] During this long period porting Information) and guaranteed at a high current density, of discharge process, excess potassium ions are confined in as shown in Figure S12 (Supporting Information). The BiND/G the robust SEI layer and interface layer and have no interac- anode presented an unprecedented electrochemical perfor- tion with the BiND. When the cell discharge is ≈0.27 V, a new −1 mance of 99% capacity retention after 500 cycles at 5 A g . XRD pattern, which can be referred to as K3Bi, emerged with Even at 10 A g−1, it still delivers 88% capacity retention after the concurrent disappearance of Bi. This distinctly sharp phase 500 cycles, as shown in Figure 4c, which is superior to most transition can be attributed to the ultrathin granules of BiND reported anode materials in PIBs and indicates the state-of-the- and the superior hierarchical nanoarchitectures, demonstrating art cycling stability.[12–14,22–31] the high activity of BiND and the fast ion and electron trans- To reveal the potassiation/depotassiation mechanisms of port capability of BiND/G. The direct transformation from BiND/G, operando SR-XRD was performed during the elec- Bi to K3Bi differs from the previously reported Bi-based electrochemical cycling process between 0.1–2.0 V at 0.1 A g−1 trode,[13,15,16] further showing the potassiation process is mainly (Figure 5a,b). The structure of bismuth remains stable over a controlled by kinetics instead of thermodynamics.[13] Reducing long period before discharge to 0.27 V, with neither intensity the size of Bi to the nanodot scale and laminating well with

Figure 5. Phase transition of the BiND/G electrode. a) Galvanostatic charge–discharge curves between 0.1–2.0 V at 0.1 A g−1 during operando observation, b) corresponding operando XRD pattern, c) the changes in phase fractions in the potassiation/depotassiation reactions, d) crystalline structure transition in the first potassiation process, and e) crystalline structure transition of the alloy/dealloy reaction in the following potassiation/depotassiation process. graphene strongly influences the interface structure and phase the 1st depotassiation process, its XRD peaks become slightly transitions, which dominates the exceptionally high-rate of the broader compared with that of the starting one, indicating potassium ion storage property. that the metallic Bi nanodots do not agglomerate during the Unlike the first discharge process, the stepwise K3Bi–K3Bi2– electrochemical cycling process due to the unique integrated KBi2–Bi transition governs the reversible electrochemical cycling nanostructure. This is very different from previous studies of afterward under the shield of the robust SEI layer (Figure 5e). Bi-based anodes in lithium batteries, where fast Bi agglomera- [32,33] K3Bi is first converted into 3K Bi2 at 0.5 V (2K3Bi → K3Bi2 + tion is usually observed and results in poor cycling stability. + – 3K + 3e ) and K3Bi2 is transformed into KBi2 at 1.1 V (K3Bi2 → Actually, all phases that appeared during potassiation/depotassi- + – KBi2 + 2K + 2e ), which is fully converted into metallic Bi at ation (Bi, KBi2, K3Bi2, K3Bi) are nanoscale, which always ensures + – + 1.6 V (KBi2 → 2Bi + K + e ). The reverse phase transition is a short K ionic transportation pathway. Hence, an excellent also observed in the following discharging process, demon- electrochemical rate capacity and stability can be achieved. strating a highly reversible alloy-dealloy reaction. Benefitting Metallic Bi holds the R-3m space group (Figure 5d), which from the characteristic of the in situ formed BiND/G anode, the is a typical layered compound if viewing along the c-axis, with ultrafine nanoscale granules ensure ultrafast kinetics and lead d spacing of 3.93 Å along the (0 0 3) plane. During the 1st to a sharp and sudden phase transition, as shown in Figure 5c. discharge process, metallic Bi is transformed into K3Bi, which The advantage of the nanoscale electrode is further confirmed holds a hexagonal structure similar to that of graphite with a in Figure 5b. When metallic Bi is formed again at the end of space group of P63/mmc (Figure 5d). This is a well-known

Small 2019, 1905789 1905789 (7 of 9) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.small-journal.com layered compound that allows fast ionic intercalation/de-inter- The 2D-SB was obtained by liquid phase exfoliation. The uniform calation. During the 2nd discharge reaction, the diffraction mixture of sodium bismuthate (150 mg, Alfa Aesar) and ethanol patterns revealed by Figure 5b indicate that the potassium con- (1500 mL, Sinopharm) was vigorously sonicated for 72 h to form a homogeneous colloidal solution. The solid product was then collected tent gradually increases as the discharging process moves for- by suction filtration and washed a few times with deionized water and ward, which is consistent with the nature of the potassiation ethanol. After drying in a vacuum oven overnight (at 50 °C), 2D-SB was process. For example, the K:Bi ratio in Bi–KBi2–K3Bi2–K3Bi is finally obtained. The 2D-SB/G was obtained following a similar route. 0, 0.5, 1,5, and 3. Actually, cubic KBi2 holds an Fd-3m space The 2D-SB and graphene were dispersed thoroughly in ethanol under group (Figure 5e), which can also be viewed along the c axis vigorous stirring. These two suspensions (2D-SB and graphene with a direction with layered packing of K and Bi. The K Bi holds a mass ratio of 9:1, shown in Figure S13, Supporting Information) were 3 2 alternately added into the filter paper (under suction filtration) in a monoclinic structure with a space group of C2/c (Figure 5e), dropwise manner. Then, the solid product was collected by suction with layered atomic packing along the b axis direction as well. filtration and washed a few times with deionized water and ethanol. After These layered structural features of Bi and alloyed forms pro- drying in a vacuum oven (at 50 °C) overnight, the 2D-SB/G was finally vide suitable ion diffusion pathways, intrinsically ensuring fast obtained. The BiND/G composite could be easily obtained through ionic transport kinetics. The high conductivity of the alloyed spontaneous electron reduction of 2D-SB/G in the assembled battery. structure and graphene layers further facilitates fast electronic Specifically, the 2D-SB/G was processed as the normal electrode (details in the Electrochemical Measurements section) assembled in the battery transport. and could be spontaneously reduced to BiND/G during the standing process. Materials Characterization: The morphologies and element distribution 3. Conclusion of BiND/G were characterized by scanning electron microscopy (Gemini SEM 300). The microstructure and high-resolution morphologies of In summary, 2D-SB has been reported as a PIB anode for the the 2D-SB and BiND were studied by TEM (JEM-2100). XRD patterns first time, and BiND/G obtained by in situ formation exhibits of SB and 2D-SB were measured at BL14B1 using a Mythen1K linear detector.[35,36] The powder sample was loaded into a quartz capillary outstanding electrochemical performance. The facile synthesis with a diameter of 0.5 mm, which was continuously rotated during X-ray strategy immensely simplified the fabrication process while data acquisition. The wavelength of the X-rays was calibrated using the obtaining ultrafine BiND with isolated granules ≈( 3 nm) by LaB6 standard from NIST (660b) and was 0.6887 Å. Raman scattering spontaneous reduction of 2D-SB. Moreover, BiND/G, as an spectra of Bi powder, 2D-SB, BiND, and graphene were recorded using a anode of PIB, exhibits a superior rate capability of 200 mA h g−1 Renishaw in vivo microscope equipped with a 532 nm wavelength laser. at 10 A g−1 and an impressive reversible capacity of 213 mA h g−1 To avoid laser-induced transformation, Raman spectra were recorded at −1 a relatively low power (0.5%). AFM analysis of 2D-SB was carried out at 5 A g after 500 cycles with almost no capacity decay. The using a Bruker Multimode 8 SPM. remarkable electrochemical performance of BiND/G is ascribed Electrochemical Measurements: Electrochemical performance of to its novel structure, which not only efficiently accommodates BiND/G versus K batteries was tested via CR2025 coin cells, which were the volume change but also enables high-speed channels for assembled in an Ar-filled glove box. The working electrode slurry was ionic and electronic transportation. This novel laminated struc- prepared by mixing 2D-SB/G, carbon black, and sodium carboxymethyl ture is the key to enhancing conductivity and buffer volume cellulose binder (CMC) in a mass ratio of 80:10:10 in deionized water. The slurry was first coated on Cu foil and then dried at 60 C overnight changes and achieving stable SEI during the electrochemical ° under vacuum to eliminate residual solvent. The electrode was punched cycling process. Operando SR-XRD reveals distinctively sharp −2 into disks with a loading of ≈1 mg cm . The electrolyte used 1 m KPF6 two- phase transitions (Bi–K3Bi) after SEI formation in the in 1,2-dimethoxyethane. The LANHE CT2001A battery testing system first potassiation process and reversible multiphase transitions was used to obtain charge/discharge profiles with a voltage window of (namely, Bi–KBi2–K3Bi2–K3Bi) in the following cycles. The 0.1–2 V. The specific capacity was calculated based on the total mass of excellent electrochemical performance and unique multiphase 2D-SB/G. The CV and EIS measurements were obtained using CHI 760E transitions suggest that the Bi-based nanoarchitecture as a PIB (Chenhua Instrument Company, Shanghai, China) and PARSTAT 2273 (Princeton Applied Research) electrochemical workstations. anode has potential in large-scale electric energy storage and Operando Synchrotron Radiation-Based X-Ray Diffraction Experiment: will attract increasing academic and industrial research interest In operando XRD experiments were performed at beam line BL14B1 in the near future. of Shanghai Synchrotron Radiation Facility with an X-ray wavelength of 0.6887 Å wavelength. The operando XRD study was carried out using a flat-plate transmission geometry. The XRD signal was acquired by a Mythen1K linear detector with a typical time of ≈6 min for one scan. 4. Experimental Section A specially made operando cell was prepared for operando XRD Materials Synthesis: Graphene was obtained by electrochemical experiments with an airtight sandwich-type feature in which the Mylar exfoliation, where natural graphite flakes (Alfa Aesar) were employed window was located on both the cathodic and anodic side. There was a as electrodes and sources of graphene for electrochemical exfoliation. hole with a radius of ≈2 mm in the middle of the current collectors made Electrochemical exfoliation of graphite was performed in a two-electrode of Cu foil, guaranteeing that X-ray could penetrate through the active system using platinum as the counter electrode and a graphite flake materials while cycling. The electrochemical current density used for the −1 as the working electrode. A grounded Pt wire was placed parallel to operando experiment was 0.1 A g . the graphite flake with a distance of 5 cm. The ionic solution was 4.8 g sulfuric acid diluted in 100 mL deionized water. The electrochemical exfoliation process was carried out by applying DC bias on a graphite electrode (from −10 to 10 V). To remove unwanted large graphite Supporting Information particles produced during exfoliation, the suspension was subjected to centrifugation at 2500 rpm. After drying, they were dispersed in DMF Supporting Information is available from the Wiley Online Library or (Sinopharm) solution by sonication.[34] from the author.

Acknowledgements [12] K. Lei, C. Wang, L. Liu, Y. Luo, C. Mu, F. Li, J. Chen, Angew. Chem. 2018, 130, 4777. Y.Z., X.R., and Z.X. contributed equally to this work. This work [13] J. Huang, X. Lin, H. Tan, B. Zhang, Adv. Energy Mater. 2018, 8, is supported by the National Basic Research Program of China 1703496. (973 Program) (Nos. 2017YFA0403400, 2017YFA0402800, and [14] W. Zhang, J. Mao, S. Li, Z. Chen, Z. Guo, J. Am. Chem. Soc. 2017, 2016YFA0401000), National Science Foundation of China (NSFC 139, 3316. grants: U1732121, U1932201, 21203235, and 21303147), Shanghai [15] S. Su, Q. Liu, J. Wang, L. Fan, R. Ma, S. Chen, X. Han, B. Lu, ACS Sailing Program (No. 19YF1452700), CAS Pioneer “Hundred Talents Appl. Mater. Interfaces 2019, 11, 22474. Program” (type C), and Henan Haizhisen Energy Technology Co., LTD. [16] Q. Zhang, J. Mao, W. Pang, T. Zheng, V. Sencadas, Y. Chen, Y. Liu, The authors also thank beamlines BL14B1, BL15U, BL08U, and BL19U of SSRF (Shanghai Synchrotron Radiation Facility) for providing the Z. Guo, Adv. Energy Mater. 2018, 8, 1703288. beam time. [17] M. Assis, E. Cordoncillo, R. Torres-Mendieta, H. Beltrán-Mir, G. Mínguez-Vega, A. F. Gouveia, E. Leite, J. Andrés, E. Longo, Phys. Chem. Chem. Phys. 2018, 20, 13693. [18] S. Sepulveda-Guzman, N. Elizondo-Villarreal, D. Ferrer, Conflict of Interest A. Torres-Castro, X. Gao, J. Zhou, M. Jose-Yacaman, Nanotechnology 2007, 18, 335604. The authors declare no conflict of interest. [19] S. Onari, M. Miura, K. Matsuishi, Appl. Surf. Sci. 2002, 197, 615. [20] A. Diasa, R. L. Moreira, J. Raman Spectrosc. 2010, 41, 698. [21] T. R. Penki, G. Valurouthu, S. Shivakumara, V. A. Sethuraman, Keywords N. Munichandraiah, New J. Chem. 2018, 42, 5996. [22] X. Wu, W. Zhao, H. Wang, X. Qi, Z. Xing, Q. Zhuang, Z. Ju, J. Power alloy anode, bismuth nanodots, operando X-ray diffraction, Sources 2018, 378, 460. potassium-ion battery, 2D sodium bismuthate [23] X. Cheng, D. Li, Y. Wu, R. Xu, Y. Yu, J. Mater. Chem. A 2019, 7, Received: October 10, 2019 4913. Revised: November 6, 2019 [24] J. Zheng, Y. Yang, X. Fan, G. Ji, X. Ji, H. Wang, S. Hou, Published online: M. R. Zacharia, C. Wang, Energy Environ. Sci. 2019, 12, 615. [25] Y. Liu, Z. Tai, J. Zhang, W. Pang, Q. Zhang, H. Feng, K. Konstantinov, Z. Guo, H. Liu, Nat. Commun. 2018, 9, 3645. [26] Z. Yi, N. Lin, W. Zhang, W. Wang, Y. Zhu, Y. Qian, Nanoscale 2018, [1] V. Palomares, P. Serras, I. Villaluenga, K. B. Hueso, 10, 13236. J. Carretero-González, T. Rojo, Energy Environ. Sci. 2012, 5, 5884. [27] H. Wang, X. Wu, X. Qi, W. Zhao, Z. Ju, Mater. Res. Bull. 2018, 103, [2] J. C. Pramudita, D. Sehrawat, D. Goonetilleke, N. Sharma, Adv. 32. Energy Mater. 2017, 414, 1602911. [28] K. Huang, Z. Xing, L. Wang, X. Wu, W. Zhao, X. Qi, H. Wang, Z. Ju, [3] S. Komaba, T. Hasegawa, M. Dahbi, K. Kubota, Electrochem. J. Mater. Chem. A 2018, 6, 434. Commun. 2015, 60, 172. [29] Y. An, Y. Tian, L. Ci, S. Xiong, J. Feng, Y. Qian, ACS Nano 2018, 12, [4] A. Eftekhari, J. Power Sources 2004, 126, 221. 12932. [5] P. Padigi, J. Thiebes, M. Swan, G. Goncher, D. Evans, R. Solanki, [30] I. Sultana, M. M. Rahman, T. Ramireddy, Y. Chen, A. M. Glushenkov, Electrochim. Acta 2015, 166, 32. J. Mater. Chem. A 2017, 5, 23506. [6] V. Mathew, S. Kim, J. Kang, J. Gim, J. Song, J. P. Baboo, W. Park, [31] V. Lakshmi, Y. Chen, A. A. Mikhaylov, A. G. Medvedev, I. Sultana, D. Ahn, J. Han, L. Gu, Y. Wang, Y. S. Hu, Y. K. Sun, J. Kim, NPG Asia M. M. Rahman, O. Lev, P. V. Prikhodchenko, A. M. Glushenkov, Mater. 2014, 6, e138. Chem. Commun. 2017, 53, 8272. [7] N. Recham, G. Rousse, M. T. Sougrati, J. N. Chotard, C. Frayret, [32] Y. Zhong, B. Li, S. Li, S. Xu, Z. Pan, Q. Huang, L. Xing, C. Wang, S. Mariyappan, B. C. Melot, J. C. Jumas, J. M. Tarascon, Chem. W. Li, Nano-Micro Lett. 2018, 10, 56. Mater. 2012, 24, 4363. [33] J. Chen, X. Fan, X. Ji, T. Gao, S. Hou, X. Zhou, L. Wang, F. Wang, [8] X. Wang, X. Xu, C. Niu, J. Meng, M. Huang, X. Liu, Z. Liu, L. Mai, C. Yang, L. Chen, C. Wang, Energy Environ. Sci. 2018, 11, 1218. Nano Lett. 2017, 17, 544. [34] K. Parvez, Z. Wu, R. Li, X. Liu, R. Graf, X. Feng, K. Müllen, J. Am. [9] W. Luo, J. Wan, B. Ozdemir, W. Bao, Y. Chen, J. Dai, H. Lin, Y. Xu, Chem. Soc. 2014, 136, 6083. F. Gu, V. Barone, L. Hu, Nano Lett. 2015, 15, 7671. [35] T. Yang, W. Wen, G. Yin, X. Li, M. Gao, Y. Gu, L Li, Y Lu, H. Lin, [10] W. D. McCulloch, X. Ren, M. Yu, Z. Huang, Y. Wu, ACS Appl. Mater. X. Zhang, B. Zhao, T. Liu, Y. Yang, Z. Li, X. Zhou, X. Gao, Nucl. Sci. Interfaces 2015, 7, 26158. Technol. 2015, 26, 020101. [11] I. Sultana, T. Ramireddy, M. M. Rahman, Y. Chen, A. M. Glushenkov, [36] M. Gao, Y. Gu, Z. Gong, L. Li, X. Gao, W. Wen, J. Appl. Crystallogr. Chem. Commun. 2016, 52, 9279. 2016, 49, 1182.