3.1 Material Synthesis and Characterization
The synthetic procedure for hierarchically porous N-doped CNTs encapsulated with ZnSe/CoSe nanodots (NDs) (ZnSe@CoSe@CN) is illustrated in Fig. 1a. First, the electrospun polyacrylonitrile (PAN) nanofibers containing zinc acetate (PAN@Zn(CH3COO)2) were immersed in 2-methylimidazole methanol solution to form ZIF-8 crystals on the fiber surfaces (PAN@ZIF-8). The scanning electron microscopy (SEM) image shows a rough fiber surface of PAN@ZIF-8, and the fiber diameter increased from ~ 320 to ~ 390 nm, indicating the formed ZIF-8 layer has a thickness of ~ 70 nm (Fig. S1a-b). Second, a ZIF-67 layer was grown on the PAN@ZIF-8 surfaces via a facile solution reaction method (PAN@ZIF-8@ZIF-67), which had no obvious morphology change but a larger diameter of ~ 480 nm, suggesting a ~ 90 nm thickness for the ZIF-67 layer (Fig. S1c). After solvent etching of PAN, the core-shell structured ZIF-8@ZIF-67 NTs were constructed (Fig. S2a-c) and were further calcined withSe powder under H2/Ar. During the selenization process, H2Se gas was first generated from combining H2 with Se vapor and then reacted with the Zn/Co ions released from ZIF-8@ZIF-67 [35], developing abundant uniformly distributed ultrathin ZnSe and CoSe crystals, and the organic ligands in ZIFs were simultaneously carbonized to N-doped carbon, resulting in the formation of ZnSe and CoSe NDs anchored N-doped porous CNTs (ZnSe@CoSe) (Fig. S2b-d). Finally, to further enhance the structural stability and reversible capacity, a polydopamine (PDA) carbonized N-doped carbon layer was coated on the ZnSe@CoSe surface, and obtaining the ZnSe/CoSe doped hierarchical porous CNTs (ZnSe@CoSe@CN).
As indicated by the SEM and transmission electron microscope (TEM) images (Fig. 1b-c), the ZnSe@CoSe@CN nanocomposite has an obvious tubular structure anchored with lots of ultrathin black dots with a diameter of 5–15 nm, which are the ZnSe and CoSe NDs. The powder X-ray diffraction (XRD) pattern of ZnSe@CoSe@CN clearly shows the diffraction peaks of ZnSe (JCPDS No. 88-2345) and CoSe (JCPDS No. 89-2004), confirming the successful synthesis of ZnSe/CoSe crystals in the composite CNTs (Fig. 1d). The high-resolution TEM (HRTEM) image shows the nanodots are closely connected and form relatively large nanoparticles (Fig. 1e and Fig. S3). The crystal lattice spacing of 0.56, 0.32, 0.28, 0.20, and 0.17 nm correspond to the (100), (111), (200), (220), and (311) planes of ZnSe, respectively, while the lattice spacing of 0.202 and 0.27 nm corresponds to the (102) and (101) planes of CoSe [36]. In particular, besides the interfaces between the same component, abound of CoSe/ZnSe heterogeneous interfaces are constructed, and the obvious defects in the interface can generate a large number of active sites, which can enhance the electronic/ionic transmission and reaction kinetics, thus favoring the heterostructured electrodes with much enhanced electrochemical performances [37]. Moreover, graphitic carbon that can promote the conductivity and structural stability is observed in the composite CNTs, including those coated on the ZnSe/CoSe nanocrystals, randomly distributed in the amorphous carbon, and anchored on the composite surface derived from PDA. The selected area electron diffraction (SADE) patterns of ZnSe@CoSe@CN displayed the diffraction rings of crystalline ZnSe and CoSe (Fig. 1f), and the element mapping images show the uniformly dispersed C, N, O, Zn, Co, and Se elements in ZnSe@CoSe@CN (Fig. 1g), further indicating the successful selenization of ZIF-8 and ZIF-67.
Besides, several contrast samples were prepared to confirm the unique structural and component characters of ZnSe@CoSe@CN. The ZnSe@CN and CoSe@CN composites were prepared in the same process as ZnSe@CoSe@CN, but derived from ZIF-8 and ZIF-67 NTs, respectively. The Zn@Co@CN composite was constructed via direct carbonization of ZIF-8@ZIF-67 without the selenization process, while ZnSe@CoSe was prepared without PDA. These contrast samples show obvious tubular structures (Fig. S4), and the XRD patterns indicate the successful synthesis of CoSe and/or ZnSe crystals in the composites (Fig. S5a). In addition, the Raman spectra show the characteristic D- and G-bonds of carbon, which is related to the graphitic and disordered/defective carbon, respectively. ZnSe@CoSe@CN shows the lowest ID/IG value of 1.08, while ZnSe@CoSe has the highest ID/IG value of 1.18, indicating the PDA-derived N-doped carbon layer favors increasing the graphitic degree, which is conducive to the electrochemical performance (Fig. S5b).
The chemical composition and valence states of the samples were studied by X-ray photoelectron spectroscopy (XPS) (Fig. 2a-f, Fig. S6 and Table S1). The total spectrum of ZnSe@CoSe@CN indicates a close result with the elemental mapping analysis (Fig. 2a). The peaks at 284.7, 286.2, and 288.1 eV in the high-resolution C 1s spectrum represent C-C/C = C, C = N, and C-O bonds, respectively (Fig. 2b). The N 1s spectrum can be deconvolved into three peaks of 398.4, 399.8, and 401.5 eV, corresponding to pyridine-N, pyrrole-N, and graphite-N, respectively (Fig. 2c), and the schematic bonding structure of the three types of N dopants is shown in Fig. 2h. These N dopants can introduce surface defects and bond directly with metal atoms and graphitic carbon, thus producing multiple channels for Li+/Na+ diffusion and strengthening the coupling effect between carbon and metal atoms, which are beneficial to improving the performance of the battery. In the Zn 2p spectrum (Fig. 2d), the peaks located at 1022.2 and 1045.1 eV correspond to Zn 2p3/2 and Zn 2p1/2 of ZnSe [38]. There are two spin-orbit peaks for 2p3/2 (780.88 eV) and 2p1/2 (797.18 eV) in the Co 2p spectrum (Fig. 2e), in which the Co-Se-Co bond at 780.1 eV, the Co-N bond at 782.5 eV, the shake-up satellites of Co 2p at 786.1 and 803.7 eV that resulting from the antibonding orbital between Co and Se atoms can be revealed. Specifically, the peaks at 781.1 and 797.2 eV correspond to Co-O, which may be due to partial surface oxidation [39]. Five peaks can be identified in the Se 3d and Co 3p spectrum (Fig. 2f), including the Se 3d5/2 at 54.7 eV, the Se 3d3/2 at 55.8 eV can be assigned to the Se-Se and Co-Se, the Se-C/O bond at 58.5 eV, the Co 3p1/2 bond at 59.1 eV, and the SeO2 at 61.3 eV [40]. It should be noted that compared to ZnSe@CoSe@CN, ZnSe@CN has lower Zn 2p1/2 and Zn 2p3/2 peaks of 1045.03 and 1022.03 eV (Fig. S6e), and CoSe@CN also shows lower Co 2p1/2 and Co 2p3/2 peaks of 796.58 and 778.68 eV (Fig. S6f). The Zn 2p and Co 2p peaks of ZnSe@CoSe@CN migrated to higher binding energy, resulting from the lattice distortion and defects at the two-phase boundary. These binding energy migrations could accelerate the electron movement from ZnSe to CoSe for charge compensation, resulting in increased electron density of CoSe that give rise to Li+/Na+ absorption and reaction kinetics improvement [23, 41].
The porous structure characterized by the N2 sorption isotherm shows an IV-type curve with an evident hysteresis loop (Fig. 2g and Fig. S7), indicating the as-synthesized samples have a mesoporous structure with a similar pore size distribution, and the pore size centered at ~ 0.5 nm suggests the samples also have a microporous structure. Among the samples, Zn@Co@CN has the highest specific Brunauer–Emmett–Teller (BET) surface area of 275.49 m2·g− 1 and the largest total pore volume of 5.693 cm3·g− 1. Although the selenidation process may decrease the BET surface area and pore volume, ZnSe@CoSe@CN has a high specific BET surface area of 191.267 m2·g− 1, which is higher than the samples of ZnSe@CN, CoSe@CN, and ZnSe@CoSe, confirming the core-shell ZIF-8@ZIF-67 precursor and the CN coating layer are conductive to increase the specific surface area. Besides, the ZnSe@CoSe@CN NT also has a large total pore volume of 4.454 cm3·g− 1, which is beneficial to the contact between active materials and electrolytes, shortening the Li+/Na+ diffusion pathways and relieving the inevitable volume change of ZnSe/CoSe during the charge/discharge process (Table S2) [42–44].
The electrochemical performance of the as-prepared hierarchical composite NTs was investigated as the anode material for LIBs. As illustrated in the typical cyclic voltammetry (CV) curves of the ZnSe@CoSe@CN anode (Fig. S8a), the cathodic peak at 1.27 V can be related to the Li+ insertion and formation of LixCoSe (CoSe → LixCoSe). The peak at 0.69 V corresponds to the formation of solid electrolyte interface (SEI) films and the conversion reaction of LixCoSe to Li2Se and metallic Co (CoSe → Co + Li2Se ), and the peak at 0.56 V corresponds to the solid electrolyte interface (SEI) layer formation and the reduction of ZnSe (ZnSe → Li2Se + Zn → Li2Se + LixZn → Li2Se + LiZn). Correspondingly, the oxidation peak at 1.36 V in the first charge process is ascribed to the multi-step dealloying reaction of LixZn alloy phase and the oxidation of LiZn to ZnSe, and the broad oxidation peak at 2.09 V can be assigned to the reversible selenization of Co to CoSe (Co + Li2Se → LixCoSe → CoSe) [40]. During the second cycle, the reduction peaks shift to 1.41 V and a broad peak of 0.91 V, and the oxidation peaks remain unchanged. The peak position change may result from the irreversible side reactions and activation processes caused by the initial lithium insertion/desorption in the first cycle. The following scan showed maintained CV curves, suggesting that the conversion process of ZnSe@CoSe@CN is highly reversible and stable. ZnSe@CN and CoSe@CN displayed the corresponding CV curves, demonstrating the above-mentioned mechanism analysis is reasonable (Fig. S8b-e). Moreover, the galvanostatic discharge/charge (GDC) curve of the first discharge process revealed two obvious platforms at 1.25 and 0.56 V, in which the former was attributed to the lithiation of CoSe while the latter was the lithiation of ZnSe and the formation of SEI film. For the first charge process, the platform at 1.36 and 2.0-2.5 V can be attributed to the formation of ZnSe and CoSe, respectively, consistent with the electrochemical reaction at the corresponding peak position in the CV curve. ZnSe@CoSe@CN delivered high initial discharge/charge capacities of 2739 and 1583 mAh g− 1, respectively, with an initial coulombic efficiency (CE) of 57.79%, the low CE may be caused by the SEI films formation and irreversible decomposition of the electrolyte [45]. In particular, the discharge capacity can be conserved at 1500 mAh g− 1 after 100 cycles (Fig. 3a), much higher than the other contrast samples (Fig. S9). The high lithium storage performance confirms the unique component and structural advantages of ZnSe@CoSe@CN.
The rate performance of the as-prepared electrodes is further investigated (Fig. 3b). Compared to the contrast electrodes, ZnSe@CoSe@CN shows the highest capacity of 1216.2, 1032.8, 960.4, 891.8, 807.6, and 708.1 mAh g− 1 at 0.1, 0.2, 0.5, 1, 2, and 3 A g− 1, respectively, and a high capacity of 650.8 mAh g− 1 can be delivered even at an ultrahigh current density of 5 A g− 1. More importantly, the capacity can be maintained at 1230 mAh g− 1 when the current density is back to 0.1 A g− 1, confirming a high reversible rate capability. The striking rate performance of ZnSe@CoSe@CN results from the rapid Li+ diffusion through the ZnSe and CoSe interlayers, and the conductive 1D porous structure promoted Li+ diffusion kinetics.
Long-term cyclic stability is another required feature for ideal anodes in LIBs. During the first several cycles at 0.1 A g− 1, all of the samples displayed decreased capacity and enlarged CE (Fig. 3c). After that, the anodes show steady capacity. ZnSe@CoSe@CN exhibited a high capacity of 1595.8 mAh g− 1 after 100 cycles, considerably higher than that of ZnSe@CN (919.4 mAh g− 1), CoSe@CN (1178.1 mAh g− 1), Zn@Co@CN (1189.6 mAh g− 1), and ZnSe@CoSe (884.8 mAh g− 1). Besides, the ZnSe@CoSe@CN anode still exhibited the highest capacity of 1087.4 mAh g− 1 after 200 cycles at 1 A g− 1 and 813.7 mAh g− 1 after 400 cycles at 2 A g− 1, respectively (Fig. S10). Particularly, a remarkable capacity of 600 mAh g− 1 was maintained even after 1000 cycles at 5 A g− 1, while ZnSe@CoSe cannot maintain 300 cycles at this high current density because the structure collapsed during the cyclic process, further indicating the importance of the PDA-derived carbon layer on structural stability (Fig. 3d) These results confirm the superior long-term cyclic stability of ZnSe@CoSe@CN. It should be noted that the ZnSe@CoSe@CN anode prepared in this work is also among the best anodes based on the comparison with other reported metal selenide-based anodes for LIBs (Fig. 3e and Table S3) [36, 45–51].
According to the previous reports, the corresponding electrode storage mechanism can be revealed according to the relationship between peak current (i) and scan rate (v): i = avb, in which the calculated b value of 0.5 and 1.0 indicate the electrochemical reaction is dominantly through diffusion behavior and pseudo-capacitive behavior, respectively [46]. In this work, the CV curves were tested at different scan rates from 0.2 to 1.0 mV s− 1 (Fig. 4a), the fitted b-values of recurring peak 1 to peak 4 for ZnSe@CoSe@CN are 0.98, 0.759, 0.847, and 0.786, respectively (Fig. 4b), confirming a pseudo-capacitive dominant behavior during the redox process. The capacity contribution ratio can be calculated by the following equation: i = k1v + k2v1/2, where k1v and k2v1/2 correspond to the capacity and diffusion contributions. The calculated capacity contribution is increased from 70.52% at 0.2 mV s− 1 to 83.84% at 1 mV s− 1 (Fig. 4c-d), further indicating a pseudo-capacitance dominant process in the whole lithium storage capacity, which benefits the lithium storage performance [48].
Furthermore, the diffusion coefficient of Li+ (DLi+) in the ZnSe@CoSe@CN anode upon the lithiation and delithiation process was measured by the galvanostatic intermittent titration technique (GITT), which suggests fast Li+ diffusion kinetics (Fig. 4e-f). The DLi+ value range of ZnSe@CoSe@CN in the discharging and charging process is 3.12×10− 14 to 6.29×10− 13 and 5.31×10− 14 to 6.97×10− 13, respectively, indicating a good diffusion ability. This should attribute to the ZnSe/CoSe phase interface generated lattice defects leading to the fast Li+ migration and storage, and the hierarchical porous tubular structure assists sufficient contact of electrolyte and electron transport, resulting in the high pseudo-capacitive contribution and the minimal charge transfer resistance of ZnSe@CoSe@CN. As indicated by the Nyquist plots measured via the electrochemical impedance spectroscopy (EIS), ZnSe@CoSe@CN has a Re value of 4.43 Ω and Rct value of 115.70 Ω, which are the lowest resistance among the as-prepared samples (Fig. 4g), suggesting a much fast ion diffusion and reaction kinetics, hence superior rate performance [50]. Therefore, the construction of heterostructured bimetallic selenides within the hierarchical porous CNTs is conducive to improving the pseudo-capacitive contribution, fasting the electron transfer rate and reaction kinetics, and finally to the enhancement of the electrochemical performance.
Meanwhile, the reaction mechanism of ZnSe@CoSe@CN was analyzed via ex-situ XRD tests. As shown in the ex-situ XRD patterns with corresponding discharge and charge voltages (Fig. 4h-i), during the discharge process from stage A to stage B, the ZnSe phase can still be detected, while the strength of the CoSe phase decreases significantly, and the signals of Co and Li2Se can be observed, indicating that CoSe has experienced the insertion of Li+ at a higher potential than the ZnSe phase. When further discharging to 0.01 V (stage D), the peaks of ZnSe and CoSe phases disappear, while the signals of Co, Zn, and Li2Se become obvious, indicating that both ZnSe and CoSe have gone through a complete lithification process except the alloying reaction between Li and Zn/Co. Besides, the broad peak centered at 24° is the amorphous and the graphitic plane (002) of carbon [35]. During the delithiation process, when the electrode is charged to 0.75 V (stage E), the ZnSe signal can be clearly observed and CoSe signals appear in weak intensity, while the Co and Li2Se phases still exist, indicating ZnSe experienced the Li+ intercalation process, while CoSe only observed part of lithium intercalation. In the fully charged state (stage G), the Li2Se phase disappears, and the ZnSe and CoSe phases can be detected, indicating that both ZnSe and CoSe phases have gone through the complete delithiation reformulation process and returned to the initial state, further confirming the high reversibility of the ZnSe@CoSe@CN anode.
Moreover, the SEM and TEM images of the ZnSe@CoSe@CN anode after 200 cycles at 1 A g− 1 show the obvious 1D porous morphology, the HRTEM image and XRD pattern indicate the existence of ZnSe and CoSe crystals, and the EDX spectrum and element mapping images confirm the existence and even distribution of the elements (Fig. S11), indicating the high structural integrity of ZnSe@CoSe@CN.
The sodium storage performance also has been investigated to further confirm the structural and composition advantages of the hierarchical porous ZnSe@CoSe@CN NTs. The CV curves tested at 0.1 mV s− 1 within 0.01-3 V revealed the electrochemical reaction details (Fig. S12). Taking ZnSe@CoSe@CN as an example, in the first cathodic scan, the peaks at around 0.51, 1.19, and 1.48 V are related to the SEI film formation and Na+ insertion into CoSe to form NaxCoSe, and the reduction of NaxCoSe to Co and Na2Se processes (CoSe → NaxCoSe → Co + Na2Se) [52]. The reduction peak at 0.38 V is corresponded to the conversion reaction between ZnSe and Na to form Zn and Na2Se (ZnSe → Zn + Na2Se), and the peak at 0.05 V is attributed to the formation of alloying NaZn13 [53]. For the first anodic scan, the distinct peak at 1.03 V is due to the removal of Na+ and conversion reaction of Zn to ZnSe, and the peaks of 1.51 and 1.77 V are ascribed to the multistep electrochemical reaction between Co and Na2Se to form CoSe [38, 40]. The CV curves overlap well from the second cycle, suggesting high reaction reversibility. The plateaus in the GDC profile correspond to the peaks in the CV curves, and there is no significant difference in the subsequent cycles, further confirming the excellent repeatability of ZnSe@CoSe@CN (Fig. 5a). The first discharge and charge capacities of ZnSe@CoSe@CN are 660 and 559 mAh g− 1, respectively, with a CE value of 84.6%. In sharp contrast, the control samples deliver much lower repeatability and charge/discharge capacities (Fig. S13).
The prepared tubular ZnSe@CoSe@CN anode exhibits excellent rate capability, a surprisingly reversible capacity of 523.2, 501.3, 481.2, 457.9, 423.1, 396.4, and 331 mAh g− 1 can be retained when the current density is 0.1, 0.2, 0.5, 1.0, 2.0, 3.0, and 5.0 A g− 1 (Fig. 5b), respectively, and the capacity can be recovered to 515 mAh g− 1 when the current density is returned to 0.1 A g− 1, higher than the contrast samples, suggesting the unique structure and component advantages of ZnSe@CoSe@CN for Na+ insertion and desertion. Moreover, the ZnSe@CoSe@CN anode exhibits superior long-term cycling stability, a reversible capacity of 622.3 mAh g− 1 after 100 cycles at 0.1 A g− 1 can be maintained (Fig. 5c), which is much higher than ZnSe@CN (329.7 mAh g− 1), CoSe@CN (312.8 mAh g− 1), Zn@Co@CN (237.2 mAh g− 1), and ZnSe@CoSe (212.8 mAh g− 1). Meanwhile, a considerable capacity of 429.6 and 397.1 mAh g− 1 can be maintained after 200 cycles at 1 A g− 1 and 400 cycles at 2 A g− 1, respectively (Fig. S13), even achieving a high capacity of 397.3 mAh g− 1 after 1000 cycles at 5 A g− 1 (Fig. 5d). ZnSe@CoSe@CN is among the best selenide-based anode materials for SIBs (Fig. 5e and Table S4) [37, 39, 53–59].
To interpret the high-rate capability, the electrochemical kinetics of the ZnSe@CoSe@CN anode was investigated by a series of measurements at 0.2 ~ 1.0 mV s− 1 (Fig. 6). The CV curves present similar shapes at different scan rates of 0.2-1.0 mV s− 1, and have quantified b-values of 0.87, 0.95, 0.98, 0.97, and 0.87 for anodic and cathodic peaks, indicating a capacitive dominant characteristic of ZnSe@CoSe@CN (Fig. 6a-d). As shown in Fig. 6c, a dominant capacitive contribution of 93.62% can be quantified in the CV curve at 1.0 mV s− 1. As the scan rate decreases, the capacitive contribution further decreases to a minimum value of 86.54% at 0.2 mV s− 1 (Fig. 6d).
The GITT curve indicates a high calculated diffusion coefficient of Na+ (DNa+) values of 2.19×10− 14 ~ 6.62×10− 13 and 5.81×10− 14 ~ 6.77×10− 13 during the discharging/charging process for ZnSe@CoSe@CN (Fig. 6e-f), indicating fast reaction kinetics. As indicated by the EIS spectra, the ZnSe@CoSe@CN anode has the lowest Re value of 5.79 Ω and Rct value of 37.91 Ω among the as-prepared samples (Fig. 6g), which is in agreement with the GITT results. These results suggest that in addition to Li+, the unique 1D hierarchical porous tubular architecture anchored with heterostructured bimetallic selenides also has a high Na+ diffusion ability, benefiting the high capacitance contribution of ZnSe@CoSe@CN, which efficiently enhance the cycling and rate capability for SIB. In addition, the possible reaction mechanism of ZnSe@CoSe@CN for SIBs anode was also analyzed by ex-situ XRD tests. As shown in the ex-situ XRD patterns with corresponding discharge and charge voltages (Fig. 6i-h), during the discharge process, the CoSe and ZnSe diffraction peaks gradually disappeared and new peaks appear that correspond to Na2Se, Co, and NaZn13. During the desertion process, the CoSe and ZnSe peaks reappeared, indicating the high reversibility of the ZnSe@CoSe@CN anode.
The remarkable lithium/sodium storage performance could be attributed to the electrospinning-assisted assembly effect on the 1D hierarchical porous tubular architecture, in which the hollow porous structure can accommodate the volume expansion generated during the conversion reaction, which is beneficial to preserving the structural stability of the anode. Besides, the porous structure with a thin wall and high specific surface area accelerates the Li+/Na+ insertion/desertion ability and favors the contact with electrolytes. Also, downsizing ZnSe and CoSe to nanoscale effectively reduces the Li+/Na+ diffusion distance and the mechanical stress during the charging/discharging process, and the heterogeneous phase interface with lattice defects accelerates the electrons transportation and the diffusion rate of Li+ and Na+, hence the tubular ZnSe@CoSe@CN composite exhibits fast reaction kinetics. Moreover, the PDA-derive N-doped carbon coating layer prevents pulverization and increases the conductivity of the material, further enhancing the structure stability and lithium/sodium storage performance.