Nanoarchitectonics of the cathode to improve the reversibility of Li–O2 batteries

In pursuit of improving the reversibility of LOBs, a bimetallic ZIF (ZnxCoy) was designed and grown on CNTs via hydrothermal synthesis using Zn and Co acetates, together with 2-methylimidazolate, as described in Figure 1a. The hydrothermal process is beneficial for facilitating the nucleation and growth of bimetallic ZIF as well as reducing the synthesis time. The chemical composition of ZnxCoy particles was controlled by adjusting the ratio of Zn/Co (x/y = 1/4, 1/1, and 4/1) in the starting materials. After carbonization at 900 °C and chemical etching with 1 M H2SO4, bimetallic ZnxCoy–C/CNT composites were successfully obtained to be used as cathode materials for LOBs.

[2190-4286-13-61-1]

Figure 1: (a) Design and synthesis of ZnxCoy–CNT. FESEM images of (b, c, d) as-prepared and (e, f, g) etched composites. (b, e) Zn4Co1–CNT, (c, f) Zn1Co1–CNT, and (d, g) Zn1Co4–CNT.

Figure 1b–d shows the morphologies of bimetallic Zn4Co1, Zn1Co1, and Zn1Co4 particles grown on CNT frameworks, respectively. Field-emission scanning electron microscopy (FESEM) observations confirmed that abundant rhombic dodecahedral ZnxCoy particles with different Zn/Co ratios were successfully integrated into the CNTs. Corresponding energy-dispersive X-ray spectroscopy (EDS) elemental mapping results (Supporting Information File 1, Figure S1) confirm that Zn and Co were uniformly distributed inside the as-grown ZnxCoy particles. We also found that the size of the ZnxCoy particles was decreased by increasing the ratio of Zn/Co during the synthesis due to the distinctive formation mechanisms of the parental ZIF-8 and ZIF-67 particles. Under the same synthesis conditions, the particle size of ZIF-8 is always smaller than that of ZIF-67. This is because the formation of ZIF-67 is proceeded by a fast one-step growth mechanism while ZIF-8 is formed by a slower two-step growth mechanism (i.e., nucleation and growth) [39]. The different formation mechanisms are mainly responsible for determining the particle sizes of ZIF-8 and ZIF-67.

After carbonization and chemical etching processes, we obtained a series of ZnxCoy–C/CNT composites, as shown in Figure 1e–1g, in which the highly porous ZnxCoy–C particles are beneficial for facilitating the electrochemical reactions between Li+ and O2. Moreover, CNT networks allow for sufficient electronic conduction as well as diffusion pathways for O2 and electrolyte in the composites. The atomic ratios of Zn and in the Zn1Co4–C/CNT, Zn1Co1–C/CNT, and Zn4Co1–C/CNT composites were measured by EDS elemental mapping (Supporting Information File 1, Figure S2). Moreover, it should be noted that all the ZnxCoy–C particles exhibited a high concentration of N, mainly induced by thermal decomposition of 2-methylimidazole during the carbonization process. After the carbonization and chemical etching processes, the sizes of the ZnxCoy particles were slightly decreased due to the thermal evaporation of organic linkers and metal ions, maintaining free spaces in the particles.

According to the X-ray diffraction (XRD) patterns of the as-grown ZnxCoy particles on the CNT framework, all reflections are well matched with those of simulated patterns of ZIF-8 and ZIF-67 (Supporting Information File 1, Figure S3). After carbonization and chemical etching, the XRD patterns of ZnxCoy–C/CNT composites exhibit a strong signal for the (002) reflection of graphitic carbon at around 26° with a trace of metallic Co (Figure 2a). Multiple characteristic peaks are detected around 44.1°, 51.4°, and 75.7°, which correspond to (111), (200), and (220) reflections of metallic Co, respectively. Moreover, we found that the (002) peak became obviously sharper in the composites with a decreasing Zn/Co ratio during synthesis. This is because Co facilitates the graphitization of ZnxCoy particles during the carbonization process. From the results, we confirm the critical role of Co for tailoring the microstructure of ZnxCoy–C particles in the composites.

[2190-4286-13-61-2]

Figure 2: (a) XRD patterns and (b) Raman spectra of ZnxCoy–C/CNT composites.

Figure 2b compares Raman spectra of ZnxCoy–C/CNT composites, showing typical Raman bands at ≈1346 cm−1 (D band), ≈1576 cm−1 (G band), and ≈2680 cm−1 (2D band). All the composites show similar Raman scattering without a noticeable difference in full width at half maximum (FWHM) values. Assuming the same content of CNTs in the composites, the differences in the intensity ratio of the D to the G band (Id/Ig) are mainly attributable to the crystallinity of the ZnxCoy–C particles. The Zn1Co4–C/CNT composite has the best crystallinity with a ratio of 1.02, which is lower than that of the Zn1Co1–C/CNT (1.13) and Zn4Co1–C/CNT (1.23) composites.

Transmission electron microscopy (TEM) observations with corresponding fast Fourier transform (FFT) patterns confirm the different crystallinity of ZnxCoy–C particles, depending on the concentration of Co during the synthesis. After the carbonization, the crystallinity of ZnxCoy–C particles can be enhanced by decreasing the Zn/Co ratio during synthesis [35]. Unlike Zn4Co1–C particles (Figure 3a), which have a typical amorphous carbon structure, both Zn1Co1–C (Figure 3b) and Zn1Co4–C (Figure 3c) particles contain some short-range graphitic carbon structures with a lattice (d)-spacing of 0.34 nm. Even after chemical etching, we still found residual metallic Co in the ZnxCoy–C/CNT composites with a d-spacing of 0.176 nm, corresponding to (200) crystal planes of the face-centered cubic structure (Figure 3d–f). The amount of residual metallic Co is dependent on the chemical composition of ZnxCoy particles. It is expected that it would play an important role as an electrocatalyst in the composites for lowering the overpotential for the given electrochemical reactions.

[2190-4286-13-61-3]

Figure 3: TEM images of (a, d, g) Zn4Co1–C/CNT, (b, e, h) Zn1Co1–C/CNT, (c, f, i) Zn1Co4–C/CNT. The insets in the bottom row are the corresponding selected area diffraction (SAED) patterns.

For further inspection of the structural characteristics, the surface chemistry of ZnxCoy–C/CNT composites was further investigated by using X-ray photoelectron spectroscopy (XPS), as shown in Figure 4. The XPS spectra were carefully deconvoluted, based on the excitation of C 1s at the binding energy of 284.5 eV. According to the Co 2p spectra collected from the ZnxCoy–C/CNT composites (Figure 4a), strong signals were observed at binding energies of 778.9 and 794 eV, corresponding to the Co 2p3/2 and Co 2p1/2 orbitals of metallic Co, respectively, regardless of the Zn/Co ratio. In contrast, the signal of Zn was only detected in the Zn 2p spectrum of the Zn4Co1–C/CNT composite (Figure 4b), indicating that Zn was easily evaporated during the carbonization process. From the C 1s and N 1s spectra (Supporting Information File 1, Figure S4), we found that N species could be spontaneously doped by the thermal decomposition of 2-methylimidazolate during carbonization. In practice, Zn1Co4/CNT composites show the highest sp2-carbon content induced by the highest crystallinity among the ZnxCoy–C particles, in which the N content was measured to be 2.3 atom %. It is expected that such structural features would promote high electrical conductivity of the ZnxCoy–C particles. Therefore, it can be inferred that the crystallinity and N content in the ZnxCoy–C particles can be increased by decreasing the Zn/Co ratio, since the metallic Co facilitates graphitization and N doping at a given temperature.

[2190-4286-13-61-4]

Figure 4: XPS spectra of ZnxCoy–C/CNT composites: (a) Co 2p and (b) Zn 2p.

Figure 5a shows the electrical conductivities of ZnxCoy–C/CNT composites, indicating that the Zn1Co4–C/CNT composite with the smallest Zn/Co ratio showed a superior electrical conductivity of 48.5 S·cm−1. This is attributed to the highest crystallinity of the Zn1Co4–C/CNT composite, which contains the highest fraction of metallic Co among the given structures. The pore structures of the ZnxCoy–C/CNT composites were characterized by N2 isotherms, as presented in Figure 5b. The specific surface areas and total pore volumes of ZnxCoy–C/CNT composites were calculated by using the Brunauer–Emmett–Teller (BET) model. Note that the pore structure of the ZnxCoy–C/CNT composites is highly dependent on the Zn/Co ratio during synthesis. The composite with higher Zn content has a larger BET surface area and micropore volume but a smaller mesopore volume, as compared in Figure 5c and Supporting Information File 1, Table S1. According to the adsorption–desorption hysteresis curves, the Zn1Co4–C/CNT composite has the lowest BET surface area (305 m2·g−1) but the highest mesopore volume (1.23 cm3·g−1). In contrast, the Zn4Co1–C/CNT composite shows the highest BET surface area (489 m2·g−1) with the lowest mesopore volume (0.88 cm3·g−1). This reveals that ZnxCoy–C particles possess typical characteristics of both ZIF-8- and ZIF-67-derived carbon materials. During the carbonization process, Zn forms a microporous domain with a large surface area while Co forms a mesoporous domain with a small surface area in the carbon matrix. Therefore, the surface area can be reduced with an increase in mesopore volume by increasing the Co concentration. Thus, we note that the BET surface area and mesopore volume of ZnxCoy–C/CNT composite can be easily tailored by controlling the Zn/Co ratio.

[2190-4286-13-61-5]

Figure 5: (a) Electrical conductivity, (b) N2 sorption isotherms, (c) Dollimore–Heal desorption pore size distributions, and (d) TGA curves of ZnxCoy–C/CNT.

On the other hand, the thermal stability of ZnxCoy–C/CNT composites was also examined by thermogravimetric analysis (TGA), as presented in Figure 5d. From the TGA curves, the weight losses were measured to be 6.1%, 9.2%, and 17.4% for Zn1Co4–C/CNT, Zn1Co1–C/CNT, and Zn4Co1–C/CNT composites, respectively. These weight losses were mainly due to thermal evaporation of adsorbed moisture and Zn, together with thermal decomposition of amorphous carbon in the ZnxCoy–C particles. In this respect, the Zn1Co4–C/CNT composite was the most thermally stable because of its relatively higher fraction of robust graphitic carbon structure.

Figure 6a–c shows the galvanostatic discharge profiles of the LOBs assembled with ZnxCoy–C/CNT cathodes at a current density of 50 mA·g−1. All of the cathodes showed a well-defined voltage plateau at ≈2.7 V vs Li/Li+ for Li2O2 formation. The initial discharge capacities of Zn4Co1–C/CNT, Zn1Co1–C/CNT, and Zn1Co4–C/CNT cathodes were measured to be 16,000 mAh·g−1, ≈15,100 mAh·g−1, and ≈17,900 mAh·g−1, respectively. These results support the premise that the specific surface area and total pore volume of ZnxCoy–C/CNT directly affect the formation of Li2O2 in the cathodes. In particular, the highest initial discharge capacity of the Zn1Co4–C/CNT cathode is mainly attributable to the large total pore volume of (1.23 cm3·g−1) rather than to the specific surface area (305 m2·g−1). This is because Li2O2 is formed on the cathode surface at the initial state of discharge and then deposited in the pores during the discharge process.

[2190-4286-13-61-6]

Figure 6: Galvanostatic discharge profiles (a, b, c) of the LOBs with (a) Zn4Co1–C/CNT, (b) Zn1Co1–C/CNT, and (c) Zn1Co4–C/CNT electrodes. The discharge measurements were performed in pure O2 at a current density of 50 mA·g−1. Plots of discharge curves (d, e, f) of (d) Zn4Co1–C/CNT, (e) Zn1Co1–C/CNT, and (f) Zn1Co4–C/CNT electrodes. The cells were discharged in pure O2 with a limited capacity of 500 mAh·g−1 at various current densities in the range of 50–2,000 mA·g−1. The terminal discharge and charge potentials during cycling (g, h, i) measured on (g) Zn4Co1–C/CNT, (h) Zn1Co1–C/CNT, and (i) Zn1Co4–C/CNT electrodes. The measurements were performed in pure O2 at a current density of 200 mA·g–1 with a cutoff capacity of 500 mAh·g−1.

The rate capability of ZnxCoy–C/CNT cathodes was also investigated, as shown in Figure 6d–f. The galvanostatic discharge profiles of ZnxCoy–C/CNT cathodes were recorded with a limited capacity of 500 mAh·g−1 at various current densities (iapp) ranging from 50 to 2,000 mA·g−1. Interestingly, ZnxCoy–C/CNT cathodes exhibited different trends in their rate capability compared with the discharge capacity. With an increasing current density, the specific surface area of ZnxCoy–C/CNT cathodes becomes more crucial for boosting electrochemical reactions and lowering the overpotential for Li2O2 formation. In practice, the Zn4Co1–C/CNT cathode with the highest specific surface area maintained a stable electrochemical performance even at a high current density of 2,000 mA·g−1, while the Zn1Co1–C/CNT and Zn1Co4–C/CNT cathodes showed significant fading in their performance. According to plots of the overpotential (η) values for various cathodes against iapp (Supporting Information File 1, Figure S5), we note that the specific surface area of ZnxCoy–C/CNT is the predominant factor determining the rate capability of LOBs, because at high current densities, the oxygen cannot diffuse deeply into the pores and Li2O2 tends to form film-like particles on the surfaces of cathodes [40].

Even though the Zn4Co1–C/CNT cathode has the lowest electrical conductivity, it still exhibited the lowest η value. This implies that the overpotential is not simply dependent on the electrical conductivity of the cathode, but is also related to the nature of the discharge product, Li2O2, as well as to electrocatalytic effects. In the Zn4Co1–C/CNT cathode with a large specific surface area, the formation of amorphous Li2O2 film could be favored rather than that of toroid-like Li2O2. Compared with toroid-like Li2O2, amorphous Li2O2 film has better ionic conductivity; therefore, it is able to effectively reduce the charge transfer resistance. In addition, the electrocatalytic activities of metallic Co and N could facilitate the decomposition of Li2O2 during the next charge process, so that the overpotential of Zn4Co1–C/CNT cathode could be reduced.

The cycling performance of ZnxCoy–C/CNT cathodes was examined at a current density of 200 mA·g−1, as shown in Figure 6g–i. The Zn4Co1–C/CNT cathode exhibited superior cycling stability (≈137 cycles) compared to that of the Zn1Co1–C/CNT cathode (≈70 cycles) and Zn1Co4–C/CNT cathode (≈120 cycles). Even after 100 cycles, the Zn4Co1–C particles still maintained their morphologies and microstructures without significant structural deterioration (Supporting Information File 1, Figure S6). This could be attributed to the relatively low overpotential of the Zn4Co1–C/CNT cathode during cycling, which facilitates the reversible formation and decomposition of Li2O2. Thus, it should be emphasized that the pore structure control of cathode materials is the main responsible for improving the long-term cycling performance of LOBs (Supporting Information File 1, Table S2).

留言 (0)

沒有登入
gif