Dimensional Gradient Structure of CoSe2@CNTs–MXene Anode Assisted by Ether for High-Capacity, Stable Sodium Storage

Highlights Dimensional gradient structure of sheet–tube–dots was constructed with CoSe2@CNTs–MXene for fast ion and electron transportation. Density functional theory study discloses the electrochemical difference of CoSe2@CNTs–MXene in ether/ester electrolyte system. Phase transformation of CoSe2@CNTs–MXene was analyzed by in situ XRD. The full cell based on CoSe2@CNTs–MXene anode was also assembled. Electronic supplementary material The online version of this article (10.1007/s40820-020-00562-7) contains supplementary material, which is available to authorized users.


Introduction
Recently, emerging sodium-ion batteries (SIBs) have been deemed to be the new-generation battery for largescale energy storage applications benefitting for abundant resources, low cost and low standard hydrogen potential (− 2.71 V) [1][2][3][4]. In the last ten years, researchers have developed a great variety of anode electrode materials and different kinds of cathode materials for SIBs [5,6]. Meanwhile, advanced SIBs systems, such as Na-S, Na-metal and Na-O 2 batteries, have also achieved prominent energy density in the effort of researchers [7][8][9]. However, owing to its large ionic radius (1.02 Å) compared to Li ion (0.76 Å) and dense molar mass, poor reaction kinetics, bad rate performance and weak cycling performance limit the large-scale application of SIBs [10]. Therefore, constructing high specific capacity anode materials with good reaction kinetics is an effective method to build extraordinary energy density sodium-ion full cells. Meanwhile, appropriate electrolyte systems are another way to ensure long-term and high cycling performance for full batteries. So, it is necessary to develop compatible electrode materials and electrolyte system to promote the application of SIBs.
Transition metal chalcogenides (TMCs), a kind of narrow/zero bandgap materials with low cost, high electric conductivity and unique electrochemical properties, have been widely investigated as the potential active materials for alkali-ion batteries [11][12][13][14]. In most cases, pulverization caused by huge volume change and dissolution of intermediate in electrochemical processes leads to poor cycling performance. A generic strategy is encapsulating TMCs in functionalized carbon transferred from metal-organic frameworks (MOFs) [15]. MOFs can provide high surface area and controllable microstructure. After carbonization and selenylation process, the hollow functionalized carbon shell can tolerate tremendous phase change stress [16]. Meanwhile, heterogeneous element-doped carbon shell can also restrain the dissolution of sodium selenide in electrolyte [17]. Xu et al. synthesized ZnSe/N-doped hollow carbon architectures for SIBs with a revisable capacity of 250.8 mAh g −1 at 1 A g −1 exhibiting excellent stability [18]. CoSe 2 @N-CF/ CNTs can give a capacity of 428 mAh g −1 at 1 A g −1 after cycling for 500 times [19]. Synergistic effect in hierarchical gradient structure can greatly enhance electron/ion diffusion, as well as carbon shell encapsulation strategy can guarantee the stabilization of active materials and impede the strain during electrochemical processes. Therefore, constructing hierarchical structure of TMCs transferring from MOFs is a potential method to acquire high-performance sodium-ion anode materials.
Electrolyte, an ion conductor, is another critical component of the rechargeable battery system [20]. Common carbonate ester-based solvents such as ethylene carbonate (EC) and propylene carbonate (PC) are widely applied in SIBs. These electrolyte systems can obtain high ionic conductivity, wide electrochemical window and stable solid electrolyte interphase (SEI) film on the surface of electrode materials [21]. However, low initial coulombic efficiency (ICE), poor cycling stability and the dissolution of reaction intermediate are severe challenges that need to be addressed. Compared with LIBs, the working voltage is usually lower than 4 V, which makes it possible for the adoption of ester electrolyte system [22]. On the other hand, the ester-based electrolyte can significantly improve ICE and rate performance of SIBs [23,24]. Transition metal sulfides (TMDC) like ZnS showed a superior rate capability and outstanding long-term cyclability assisted by ether electrolyte [25]. Cu 2 MoS 4 -RGO exhibited excellent cycling stability (215 mAh g −1 after 2000 cycles) and good full cell performance (75.5% after 500 cycles) [26].
Herein, we report a dimensional gradient structure building with CoSe 2 @CNTs-MXene anode materials for SIBs by exploiting a NaPF 6 in a new DEGDME electrolyte system. Cobalt (Co)-MOFs are deposited on the MXene by an easy coprecipitation method. Carbon nanotubes (CNTs) grow on the surface of MXene in the catalysis of Co particles, followed by selenylation process. MXene, acting as the flexible matrix, not just promotes the fast ion and electronic transmission by constructing a "sheet-tube-dots" hierarchical structure, but also impedes the dissolution of Na 2 Se in electrochemical processes. The CoSe 2 @CNTs-MXene in ether electrolyte maintains an outstanding cycling performance of 400 mAh g −1 after 200 cycles at 2 A g −1 with a high ICE of 81.7% and excellent rate stability of 347.5 mAh g −1 at 5 A g −1 , which is much better than electrochemical behaviors in ester system (only 27 mAh g −1 for 200 cycles). The great electrochemical contrast of CoSe 2 @CNTs-MXene in disparate electrolyte systems is evidenced by DFT calculations. Meanwhile, phase transformation of CoSe 2 @ CNTs-MXene in the first cycle was successfully analyzed by in situ XRD and dynamic electrochemical impedance spectroscopy (EIS) analysis. Importantly, CoSe 2 @CNTs-MXene 1 3 and Na 3 V 2 (PO 4 ) 3 /C full cell is assembled, delivering outstanding cycle performance with a capacity of 280 mAh g −1 after 50 cycles at 100 mA g −1 .
2 Experimental Section 2.1 Materials 2-Methylimidazole, Co (NO) 3 6H 2 O, methanol, LiF and HCl were purchased from Shanghai Aladdin Bio-Chem Technology Co., Ltd. Ti 3 AlC 2 powders were purchased from Jilin 11 Technology Co. Ltd., China. All chemical reagents were used without further purification. The ultrapure water was used throughout the experiment process.

Synthesis of Ti 3 C 2 T x MXene Nanosheets
Firstly, 1 g LiF was dissolved in 10 mL 9 M HCl at room temperature. Then, 1 g Ti 3 AlC 2 powders were added into LiF/HCl solution for 10 min to avoid overheating. The next, mixed solution was transferred into a Teflon autoclave and kept at 60 ℃ for 24 h. After cooling to room temperature, the etching product was washed with 3 M HCl and ultrapure water successively, until the pH of solution reached 7. Finally, the black jelly was sonicated in water under the protection of argon for 2 h and centrifuged at 3500 rpm. The concentration of final obtained Ti 3 C 2 T x MXene nanosheets colloidal solution was about 8 mg mL −1 .

Synthesis of CoSe 2 @CNTs-MXene
In a popular method, 0.6 mmol Co (NO) 3 ·6H 2 O and 10 mL MXene colloidal solution were mixed in 40 mL methanol. Then, 6 mmol 2-methylimidazole in 50 mL methanol with 12 μL triethylamine was poured into the above solution followed by stirring continuously for 1 h. After washing with methanol, the ZIF-67@MXene was dried in vacuum at 80 ℃ for 12 h. Next, ZIF-67@MXene was annealed under Ar/H 2 atmosphere at 700 ℃ for 2 h with a heating rate for 2 ℃ min −1 , the Co@CNTs-MXene. At last, Co@ CNTs-MXene was mixed with Se powder at a weight ratio of 1:3 and heated under Ar atmosphere at 500 ℃ for 3 h. The preparation of CoSe 2 @CNTs was synthesized by a similar method without MXene.

Materials Characterization
X-ray diffraction (XRD) was tested by Rigaku SmartLab SE with Cu Kα radiation (λ = 1.5406 Å). Thermogravimetric analysis (TGA) was measured by STA 449 F5 Jupiter. Raman spectrum was obtained by LabRAM HR Evolution using 532-nm laser. The surface composition of samples was analyzed by X-ray photoelectron spectroscopy (XPS, ESCALAB250Xi). Morphology was acquired by a field emission scanning electron microscope (FESEM, ZEISS SIGMA) and a field emission transmission electron microscope (FETEM, JEOL JEM-2100F); both of them were equipped with an EDX spectrometer (Oxford Instruments). In situ XRD was represented by a special electrochemical reaction unit with a beryllium (Be) window.

Electrochemical Measurements
For half-cell, active materials (AC), Super P and PVDF were stirred with a weight ratio of 8:1:1 by a high-speed homogenizer. The slurry was coated on copper foil with a scraper, and the mass load of AC is about 1.1 mg cm −2 . Sodium foil and glass fiber (Whatman) were used as a counter electrode and a separator. Two different kinds of system ether (1 M NaPF 6 in DEGDME) and ester (1 M NaPF 6 in PC) were adopted as electrolyte, respectively. And the ratio of electrolyte/electrode in the half-cell is about 10 uL mg −1 . For full cell, Na 3 V 2 (PO 4 ) 3 /C was adopted as cathode materials. The capacity matching was achieved by controlling the coating thickness and promoting the anode capacity having a 5-10% surplus compared with cathode capacity. The mass loading of cathode electrode slice is about 6 mg cm −2 . Before assembling full cell, CoSe 2 @CNTs-MXene was activated for three cycles in advance. Cyclic voltammetry (CV) curves were obtained by CHI 660D workstation. A Neware BTS-4008 system was employed for charge/discharge and rate performance. Electrochemical impedance spectroscopy (EIS) was evaluated by using a Zahner IM6 system (0.01-10 5 Hz).

Computational Details
Quantum ESPRESSO v6.4.1 software packages were adopted for all density functional theory (DFT) calculation [27]. Perdew-Burke-Ernzerhof (PBE) in generalized gradient approximation (GGA) considering van der Waals force (VDW) was employed to analyze the exchange functional [28]. The kinetic energy cutoffs for the wavefunction were 60 Ry. 5 × 5 × 1 C-N surface and 3 × 3 × 1 Ti 3 C 2 O 2 surface built with a 20 Å vacuum region. A 3 × 3 × 1 k-point mesh of Brillouin zone was put to use. The binding energy (E bind ) of Na 2 Se on different matrix could be calculated by E bind = E Na2Se-matrix -E Na2Se -E matrix . For all structure and adsorption models, the energy convergence accuracy was within 1 × 10 -7 eV and 0.001 eV Å for force. Rietveld method was used to refine the XRD data of Na 3 V 2 (PO 4 ) 3 /C [29].

Synthesis and Characterization of CoSe 2 @CNTs-MXene
CoSe 2 @CNTs-MXene was prepared by a universal strategy followed by carbonization and selenylation processes, which is displayed in Fig. 1a. Typically, single-layer Ti 3 C 2 T x was synthesized through a fluoride-based salt etchants method reported by Ghidiu [30,31]. To avoid the oxidization of Ti 3 C 2 T x during usage, appropriate sodium L-ascorbate was added in the colloidal solution [32]. XRD patterns (Fig. S1) indicate successful synthesis of Ti 3 C 2 T x MXene. Microstructure of single-layer Ti 3 C 2 T x characterized by TEM is shown in Fig. S2, and the individual Ti 3 C 2 T x MXene looks ultrathin and transparent, with a size range from 100 to 700 nm [33]. Selected area electron diffraction (SAED) patterns reveal excellent crystallinity and unique hexagonal structure of single-layer Ti 3 C 2 T x MXene [34]. Then, ZIF-67 nanocubes were prepared through a conventional co-precipitation method at ambient temperature [35]. SEM images of ZIF-67 ( well impede the reunion of MXene benefitting for construction the conductive network. XRD analysis in Fig. S3 also clearly shows the similarity between ZIF-67 and ZIF-67/ MXene. After annealing treatment at 800 ℃ under Ar/H 2 atmosphere, a novel "tube-on-nanosheet" structure of Co@ CNTs-MXene was obtained (Fig. 1e). CNTs with hundreds of nanometers grew neatly on the surface of MXene. Co 2+ ions were reduced to elemental cobalt (ICPDS No.15-0806) under the effect of H 2 (Fig. S4). Pure cobalt nanoparticles constantly catalyze the growth of carbon nanotubes during the carbonization process of ZIF-67 [37]. The final product CoSe 2 @CNTs-MXene was acquired after selenylation at 500 ℃ for 3 h, as displayed in Fig. 1f, g. CoSe 2 particles were encapsulated on the top of CNTs. Two crystal interplanar spacings, 0.253 and 0.319 nm, could be well fitted with the (221) of CoSe 2 and (103) of carbon, respectively. The elemental mapping of CoSe 2 @CNTs-MXene is also displayed in Fig. S5. For comparison, CoSe 2 @CNTs was also prepared without the addition of MXene (Fig. S6). The construction of a "sheet-tube-dots" hierarchical structure can significantly enhance the transport of electrons and sodium ions. The XRD patterns of CoSe 2 @CNTs and CoSe 2 @ CNTs-MXene are displayed in Fig. 2a. Both of them clearly show a single-phase composition of cubic CoSe 2 (ICPDS no. 65-3327, space group: Pa -3 ). No additional reflection appearing in the spectrogram means the high purity of products. At the same time, characteristic peaks of MXene at 5-10° could not be found in the diagram. This phenomenon indicates MXene has been evenly dispersed in CoSe 2 @ CNTs-MXene. Raman spectroscopy (Fig. 2b) of CoSe 2 @ CNTs-MXene shows two peaks at 186 and 673 cm −1 , respectively, owning to the A g and A 1g modes for CoSe 2 . Some tiny peaks near 500 cm −1 can be attributable to the slight oxidation of CoSe 2 surface. D and G peaks for sp 3 and sp 2 carbon are situated at 1345 and 1586 cm −1 , respectively. The I D /I G scale is about 1.07, indicating a good degree of graphitization, which is conducive to the transfer of electrons and sodium ions.
TGA of CoSe 2 @CNTs-MXene was tested in hot air from 25 to 800 ℃ with a heating rate of 10 ℃ min −1 , as shown in Fig. 2c. The mass increases between 400 and 500 ℃ owing to the formation of SeO 2 and oxidization of MXene, respectively. The primary mass loss after 600 ℃ contributes to the transformation from CoSe 2 to Co 3 O 4 and the sublimation of SeO 2 [19]. Based on the above TGA tests, the calculated mass percentage of CoSe 2 in CoSe 2 @CNTs-MXene is about 63%. XPS was measured to invesitage chemical states of CoSe 2 @CNTs-MXene, total spectrogram of CoSe 2 @ CNTs-MXene (Fig. 2d) is clearly observed the coexistence of Co, Se, Ti, C and N, relatively. Figure 2e-i exhibits the fine spectrogram of Co 2p, Se 3d, Ti 2p, C 1 s and N 1 s, respectively. The signal of Co 2p (Fig. 2e) shows two broad peaks locating at 781.1 and 797.2 eV corresponding to Co 2p 3/2 and Co 2p 1/2 . Peaks at 778.4 and 793.7 eV can be corresponded to the Co-O bond due to the surface oxidization of CoSe 2 [38]. Owing to the orbital between Co atoms and Se atoms, two satellite peaks also can be found in the spectrogram [35]. The Se 3d spectra (Fig. 2f) can be split into two main peaks at 54.5 and 55.4 eV for Se 3d 5/2 and Se 3d 3/2 , respectively. Peaks between 57 and 62 eV belong to the existence of CoSe 2 and SeO 2 [39]. In the high-resolution Ti 2p spectra (Fig. 2g), peaks at 458.55 and 464.3 eV correspond to Ti 2p 3/2 and Ti 2p 1/2 [40]. For C 1 s in Fig. 2h, three peaks locating at 284.8, 286.6 and 287.55 eV can be assigned to sp 2 C, C-O and C = O, respectively. Peaks from left to right for N 1 s are attributed to graphitic, pyrrolic and pyridinic nitrogen, in turn [18].

Electrochemical Performance of CoSe 2 @CNTs-MXene
The unique structure of CoSe 2 @CNTs-MXene makes it a promising application in energy storage fields. Electrochemical performances of CoSe 2 @CNTs-MXene were investigated by assembling half-cell countering with sodium foil. Cyclic voltammetry (CV) curves of CoSe 2 @ CNTs-MXene (Fig. 3a) in ether electrolyte were measured in 0.1-3 V at a scan rate of 0.1 mV s −1 . In the first discharge, a broad cathodic peak at 1.065 V corresponds to the insertion of sodium ion into CoSe 2 , leading to the formation of Na x CoSe 2 . With further discharge, Na x CoSe 2 was eventually broken into Na 2 Se and Co till to 0.1 V. Then, in the anodic process, peaks locating at 1.7-2.0 V stand for Na 2 Se, which regenerated to amorphous CoSe 2 clusters [38,41,42]. For comparison, CV curves of CoSe 2 @ CNTs are also displayed in Fig. S7. In order to reveal the electrochemical reaction progress of the first cycle, in situ XRD was also tested, which is displayed in Fig. 4 (2) Na x CoSe 2 + (4 − x) Na + ↔ Co + Na 2 Se. be discovered in the spectra. In order to illustrate this mechanism clearly, ex situ Raman was measured. After charging to 3 V, the characteristic peak of CoSe 2 is found in Fig. S8. All these pieces of evidence can prove the CoSe 2 -Na 2 Se-CoSe 2 conversion process. Figure 3b exhibits the CV graphs of CoSe 2 @ CNTs-MXene in the ester environment, which presents a distinct electrochemical process. The cathodic peak in 0.8 V in the first cycle shifts left about 0.26 V than ether electrolyte, and a loose anodic peak exists around 2 V. However, reaction peaks could not fit commendably and disappeared gradually after the second cycle. Galvanostatic charge/discharge profiles of CoSe 2 @CNTs-MXene in ether  CoSe 2 @CNTs and pure MXene, the revisable capacity is only 215.34 mAh g −1 after 200 cycles (Figs. S9 and S10), respectively. Rate performances of CoSe 2 @CNTs-MXene in the ether (Fig. 3f) can obtain invertible capacities for 450.5, 423, 391.5, 366 and 347.5 mAh g −1 at a current of 1, 2, 3, 4 and 5 A g −1 , respectively, which is superior to some other MXene-based anodes ( Table S1). All of these performance tests indicate CoSe 2 @CNTs-MXene can exhibit great electrochemical properties in ether electrolyte. In order to explain the huge electrochemical performance discrepancy of CoSe 2 @CNTs-MXene in different electrolyte systems, XPS investigation for C 1 s, O 1 s and F 1 s of anode materials (after the first cycle) is shown in Fig. S11. The SEI layer comprises sodium inorganic/organic complex, organic matter and fluoride. For ether, an obvious peak locating at 686.7 eV can be contributed to the C-F bond. Such high fluorine content guarantees prominent mechanical strength of SEI layer, which can keep the cycling stability during electrochemical processes [7]. After cycling for 100 times, separators were disassembled for XPS analysis. Spectra in Fig. S12 can also verify the dissolution of selenide. A weak peak ranging from 58 to 60 eV can be found in the ester spectra, which belongs to selenide signals. In contrast, no signal was detected in the ether separator.

DFT Calculations
Density functional theory (DFT) was used to describe the polyselenide shuttling constraint mechanism in different electrolytes. The space group of cubic CoSe 2 (Fig. 5a) is Pa are shown in Fig. S13. Energy gaps of PC and DEGDME are 6.0363 and 5.2606 eV, respectively; both of them can satisfy the requirement of SIBs. Relaxed adsorption geometries of Na 2 Se/PC, Na 2 Se/DEGDME, Na 2 Se/C-N and Na 2 Se/ Ti 3 C 2 O 2 are demonstrated in Fig. 5e-h. The binding energy between PC, DEGDME and Na 2 Se is 0.69 and 0.6 eV. It follows that PC as the electrolyte solvent has an interaction with sodium selenide. DFT calculation further confirmed the binding strength of Na 2 Se on C-N nanosheet is only 0.62 eV. So, when Na x CoSe 2 transferred into Na 2 Se and pure Co in the first discharge process, abundant Na 2 Se dissolved in the electrolyte, leading to the low initial coulombic efficiency and serious capacity loss. Charge density difference of Na 2 Se on Ti 3 C 2 O 2 (Fig. S14) shows an electron-loss region around selenium atoms and an electron-rich region around MXene surface. Exposed Ti atoms and O atoms on MXene could capture electrons from selenium atoms under the effect of the Lewis acidity property [43]. Ti-Se bond and Se-O bond guarantee Na 2 Se can be anchored tightly (2.04 eV) on the surface of MXene [42]. The enormous binding energy prevents the dissolution of Na 2 Se in the process of electrochemical processes, which is in good agreement with experimental results.

Kinetics Analysis of CoSe 2 @CNTs-MXene
Capacitive/diffusion behavior analysis, galvanostatic intermittent titration technique (GITT) and electrochemical impedance spectroscopy (EIS) are adopted to investigate the excellent electrochemical kinetics of CoSe 2 @ CNTs-MXene in detail. The cathode and anode peaks of CoSe 2 @CNTs-MXene (Fig. 6a) at different scan rates (from 0.5 to 2.5 mV s −1 ) show no significant deviation and can be repeatable at a high scan rate. This evidence can also explain the reason for outstanding rate performances of CoSe 2 @CNTs-MXene. The plot log (i) against log (v) and the fitting line are displayed in Fig. 5b. Calculation formulas of the capacitive contribution are Eqs. S1 and S2 described detail in supporting information. Capacitive   (Fig. 6c). Higher contribution of capacitance at high scan rate is conducive to favorable cycle and rate properties [44,45]. Diffusion stability of sodium ion in CoSe 2 @CNTs-MXene under different electrolytes could get further insight by GITT. Figure 6d shows the discharge curve from 3.0 to 0.1 V under an identical current of 100 mA g −1 for 5 min followed by a relaxation step for 30 min until the cutoff voltage. The ether battery shows evident cycling stability than ester than the good reaction kinetics of CoSe 2 @CNTs-MXene in ether electrolyte.
Electrochemical impedance spectroscopy analysis of CoSe 2 @CNTs-MXene in different electrolytes was also explored. Nyquist plots of CoSe 2 @CNTs-MXene in ether and ester are displayed in Fig. 6e-f. EIS of CoSe 2 @CNTs was also tested (Fig. S15). EIS curves can be divided into three parts, inductive reactance and semicircle related to the resistance of electrode in high-frequency region and the linear part for ion diffusion in low frequency [46]. Both curves can be well fitted with the classic equivalent circuit model as shown in Fig. S16. Charge-transfer resistance between interfaces for CoSe 2 @CNTs-MXene in ether is 9.77 Ω, which is much lower than the ester system  Table S2. This phenomenon also proves the favorable invasion at interface. Low resistance for the interface of electrode and electrolyte contributes to high sodium-ion diffusion rate during electrochemical processes. Dynamic EIS analysis of CoSe 2 @CNTs-MXene in different kinds of electrolyte at first discharge/charge cycle (Fig. S17) can also prove the excellent reaction kinetics of ether. The addition of MXene can also improve interface situations owing to the contact of the solid-liquid interface.

Conclusions
Dimensional gradient structure of CoSe 2 @CNTs-MXene transferred from ZIF-67/MXene has been successfully designed. A particular "sheet-tube-dots" hierarchical structure can greatly promote the fast ion/electronic transmission and keep the stability of CoSe 2 nanoparticles. Meanwhile, electrochemical performances of CoSe 2 @CNTs-MXene in two-electrode systems, ether/ester electrolyte systems were systematically explored. CoSe 2 @CNTs-MXene in ether exhibits outstanding cycling properties which can obtain 400 mAh g −1 after 200 cycles at 2 A g −1 with a high ICE of 81.7%, and excellent rate stability of 347.5 mAh g −1 at 5 A g −1 is much better than electrochemical behaviors in ester system (only 27 mAh g −1 for 200 cycles). Transformation mechanisms of CoSe 2 were also explored by in situ XRD and ex situ Raman. Density functional theory study discloses that the CoSe 2 @CNTs-MXene in ether electrolyte system contributes to stable sodium storage performance owing to the strong adsorption force from hierarchical structure and weak interaction between electrolyte/electrode. For full cell, CoSe 2 @CNTs-MXene//Na 3 V 2 (PO 4 ) 3 /C can also afford a reversible capacity of 280 mAh g −1 over 50 cycles at a current of 100 mA g −1 . Briefly, unique dimensional gradient structure and suitable electrolyte design promote potential application of CoSe 2 @CNTs-MXene in sodium storage fields.