ZnO nanorod arrays on carbon cloth were synthesized as the cores of the precursors by an improved wet chemical reaction method. First, the clean Saibo CC substrates were hung vertically in a 50 mL Teflon-sealed autoclave containing 40 mL of a mixed solution of 0.23 g of Zn(NO₃)₂•6H₂O, 0.12 g of hexamethylenetetramine (C₆H₁₂N₄) and 1.2 mL of ammonia. This Teflon-sealed autoclave was heated at 90°C for 12 h. After the autoclave was naturally cooled to room temperature, the CC substrates coated with white products were repeatedly washed with deionized water and ethanol, and then dried in a vacuum oven.

The prefabricated ZnO nanorod arrays were used as the templates for the growth of SnO₂ shells. In a typical procedure, 10 ml solution of 2.5 mM SnCl₄•5H₂O and 50 mM NaOH were added into 80 mL Teflon-sealed stainless-steel autoclave containing 50 mL of a mixed solvent of ethanol/water (3:2 of v/v). After about 30 min of magnetic stirring, a piece of CC with the ZnO nanorods was immersed in the mixed solution, and maintained at 150°C for 3 h. After that, the specimens were removed from the autoclave and cleaned with distilled water and ethanol.

The obtained CC/ZnO/SnO₂ core-shell nanorods were placed in a porcelain boat, and annealed in N₂ at 600°C for 30 min. The specimens immersed in 0.2 M HCl solution for 2 h to remove the ZnO cores. After that, the final products were cleaned with distillation and ethanol, and dried in air. The above reagents were selected without further purification. They were purchased from Shanghai chemical reagent co. Ltd.

The crystallographic structure of the obtained products was examined by XRD (Panalytical X' Pert, Holland) with Cu-Ka radiation (λ = 1.5418 Å). The structure and morphology of the obtained products were characterized by SEM (JSM-7000F, Japan) and HRTEM (TitanX 60–300, USA). The X-Ray photoelectron spectroscopy (XPS) measurements were performed by using Perkin-Elmer model PHI 5600 XPS System with monochromated aluminum anode as the X-ray source. The mass loading of active materials was calculated by measuring the weight difference between the prepared electrodes and CC substrates using an FA1004 electronic balance. The specific surface area of the specimens was calculated by Brunauer-Emmett-Teller (BET) method based on the measurement of nitrogen adsorption-desorption isotherm on a Micrometics Tristar 3,000 system.

Electrochemical measurements were carried out in coin-type cells (CR2032), which were assembled in an argon-filled glovebox by using a piece of binder-free porous CC/SnO₂ nanotube arrays as the anode material and a lithium-metal circular foil as the counter and reference electrodes. The electrolyte was 1 M solution of LiPF₆ dissolved in ethylene carbonate and Dimethyl carbonate (DMC) (1:1 of v/v). The galvanostatical charge and discharge tests of the cells were executed on the multi-channel battery tester (Neware-CT3008). Cyclic voltammetry (CV) were recorded on a PARSTAT 4000 electrochemical workstation at a scan rate of 0.2 mV s⁻¹ in the range of 3.0–0.005 V vs. Li/Li⁺. Electrochemical impedance spectroscopy (EIS) was carried out in the frequency range of 100 kHz to 0.01 Hz with an ac perturbation voltage of 5 mV.

The growth process of 1D binder-free porous CC/SnO₂ nanotube arrays is schematically illustrated in Figure 1A. First, the ZnO nanorod cores were grown on CC substrates by wet chemical reaction. Secondly, a SnO₂ shell was coated on the surface of the ZnO nanorods by a simple hydrothermal process. Finally, the ZnO nanorod cores were removed from the annealed CC/ZnO/SnO₂ core-shell precursors by HCl solution etching.

The obtained CC/ZnO and CC/SnO₂ products were identified by XRD measurements. As shown in Figure 1B, the broad peaks of both products at about 26 and 43° were caused by the contribution of the CC substrates. CC/ZnO products crystallized into standard hexagonal ZnO (JCPDS No. 36–1451) (Tang et al., 2018), and their lattice parameters were a = b = 3.250 Å and C = 5.207 Å. The peaks of the final CC/SnO₂ products were determined to be the pure tetragonal rutile SnO₂ (JCPDS no. 41–1445) (Dong et al., 2017), and no additional diffraction peaks from ZnO were detectable. The calculated lattice parameters of SnO₂ were a = b = 4.74 Å and C = 3.19 Å, which agrees well with the experimental values (Wang et al., 2012).

XPS measurements were executed to analyze the specific elemental composition and chemical bonding state of the final CC/SnO₂ product surface. Figure 2A illustrates that the XPS survey spectrum of the CC/SnO₂ products only contained O, C, and Sn elements, and no other impurities were found. The XPS spectrum of the C1s core-level (Figure 2B) was decomposed into three peaks of 284.5, 285.2, and 289.1 eV, respectively. The highest peak at 284.5 eV was derived from non-oxygenated carbon. The other two weak peaks were mainly contributed by a degree of oxidation produced during the chemical bath and annealing process. The peak of the O 1s spectrum (Figure 2C) consisted of three components. The main peak lied at 530.19 eV was derived from Sn–O–Sn (lattice O²⁻) of stannic oxide. Whereas, the peaks of 531.3 and 532.3 eV represented C=O and C–O, respectively. This is consistent with the results obtained from the C1s spectrum. The binding energies of Sn 3d_5/2 and Sn 3d_3/2 (Figure 2D) were ~486.2 and ~494.7 eV, respectively. These characteristic bands confirm the complete oxidation of Sn on the surface to Sn⁴⁺. This is consistent with the above XRD test results.

The morphologies and the microstructure of the as-prepared precursors were further characterized by SEM and TEM. Figures 3a,b show the SEM images of CC/ZnO nanorod arrays at different magnification. It can be seen that the 1D rod-like ZnO grew vertically and homogeneously on the surfaces of CC substrates, with an average diameter of ~150 nm. The uniform brightness and smooth surface confirm that the microstructure of ZnO nanorods was dense with no noticeable pores (Figure 3c). The HRTEM images shown in Figures 3d,e reveal that the ZnO nanorods have a lattice spacing of 0.247 nm, corresponding to the (101) planes of wurtzite-type ZnO (Ramasamy and Lee, 2011). Figure 3f shows the SEM images of the intermediate products obtained by coating SnO₂ shell on the surface of CC/ZnO nanorod arrays. Although the obtained CC/ZnO/SnO₂ core-shell nanorod arrays were still arranged in an orderly manner, they had a larger mean diameter and a rougher surface than the bare CC/ZnO nanorod arrays.

Figures 4a,b show the morphology of the final products at different magnification. It can be seen that the ZnO nanorod cores of the intermediate products were dissolved and removed after etching in HCl solution. The binder-free CC/SnO₂ nanotube arrays had an open end and no obvious structural defects. Figures 4c–e shows a typical TEM image of the SnO₂ nanotubes disassembled from a CC substrate by ultrasonic vibration. The diameter of SnO₂ nanotubes was about 100–200 nm and the thickness of their shells was about 10–20 nm. The weight difference between the final products and the pure CC substrates indicated that the mass loading of the binder-free CC/SnO₂ nanotube arrays was ~8.3 mg cm⁻². Further HRTEM analysis indicates that the SnO₂ nanotube arrays inherited the rough surface of the intermediate CC/ZnO/SnO₂ core-shell nanorod arrays. Their rough surface provides a rich adsorption interface for absorbing lithium-ions. The walls of the SnO₂ nanotube arrays consisted of nanoparticles with diameters of 2–5 nm. Figure 4f displays the corresponding lattice fringes with an interlayer spacing of 0.335 nm, which was in agreement with the (110) planes of SnO₂ (Ramasamy and Lee, 2010). The EDX mapping shows that Sn and O elements in the SnO₂ nanotube arrays were evenly distributed (Figure 4g). Similar to the XPS results, no other elements except Sn and O were detected in the EDS of SnO₂ nanotube arrays (Figure S1).

The specific surface area and porosity of the 1D binder-free CC/SnO₂ nanotube arrays were further determined by the nitrogen sorption analysis. The nitrogen adsorption-desorption isotherm and pore size distribution of the 1D binder-free CC/SnO₂ nanotube arrays are depicted in Figure 5A. Type IV isotherm validated the mesoporous structure of the obtained nanotube arrays. N₂ Isotherm Tabular Report indicates that their specific surface area was 168.4 m² g⁻¹, which was higher than that of SnO₂ nanoparticles (about 30 m² g⁻¹). Pore size was mainly distributed in the range of 1–12 nm (Figure 5B). Such mesoporous CC/SnO₂ nanotube arrays facilitated the insertion-extraction of lithium-ions.

The electrochemical properties of the binder-free porous CC/SnO₂ nanotube arrays were tested using lithium foil as the counter electrode. Figure 6A illustrates the first three CV curves of these nanotube arrays at a scanning rate of 0.2 mVs⁻¹. In the first cathodic step, the reduction peak in the range of 0–0.4 V corresponded to the alloying reaction of Sn and Li, while the peak at ~0.95 V was mainly ascribed to the formation of the solid-state electrolyte interface (SEI) film and the reduction reaction of SnO₂ (Zhou et al., 2012). The anodic peak at ~0.4 V was contributed to the dealloying process of Li_xSn alloy. The electrochemical reaction mechanism between Li and SnO₂ electrode is as follows (1)SnO2+4Li++4e-⇒Sn+2Li2O (2)xLi++xe-+Sn⇔LixSn In the following cycles, no significant changes were found except the disappearance of the mapping peak of SEI film. The second and third CV curves overlapped well and decreased slightly compared to the first CV curve, indicating that the electrochemical reaction had good reversibility (Zhou et al., 2012; Qi et al., 2018).

Figure 6B depicts the charge-discharge profiles of the binder-free porous CC/SnO₂ nanotube arrays at the current density of 200 mA g⁻¹. An evident plateau was observed at ~0.95 V in the first discharge curve. This discharge plateau is due to the formation of the SEI film (Zhou et al., 2012). The initial discharge and charge capacities of the binder-free porous CC/SnO₂ nanotube arrays were ~1149.6 and ~831.8 mA h g⁻¹, respectively, corresponding to a lower irreversible capacity loss (~12.4 %) caused by the incomplete conversion reactions and the formation of a SEI layer. The charge-discharge profiles of the second, third, fourth and fifth curves almost overlapped with each other, indicating a stable electrochemical performance. This is consistent with the CV testing results. Figure 6C shows the cycling performance of the electrodes made from the binder-free porous CC/SnO₂ nanotube arrays and the commercial SnO₂ nanoparticles, respectively. After 500 cycles, the binder-free porous CC/SnO₂ nanotube arrays maintained a capacity of 595.7 mA h g⁻¹, while the capacity of the commercial SnO₂ nanoparticles decayed to almost zero after 23 cycles. Here, it should be noted that the initial capacity of the nanotube arrays is slightly higher than their theoretical capacity due to the contribution of CC substrate. The SEM image of the 500-cycled electrodes demonstrated that the binder-free porous CC/SnO₂ nanotube arrays (Figure S2) were tightly attached to the CC surface and remained intact after 500 cycles. The above-mentioned electrodes were cycled at various current densities (100–1,800 mA g⁻¹) to further verify their rate capability. Compared with nanoparticles, the binder-free porous CC/SnO₂ nanotube arrays exhibited an excellent rate performance (Figure 6D). Even at the high current densities of 1,800 mA g⁻¹, the stable capacity of 380 mA h g⁻¹ could be maintained. In contrast, the SnO₂ nanoparticle electrodes with a mass loading of 3.06 mg cm⁻² failed completely before the 200 mA g⁻¹ current cycle. According to the previous literature (Gogotsi and Simon, 2011; Xia et al., 2016; Qi et al., 2018), the large reversible capacity is contributed mainly by the high mass loading and large specific surface area, while the high rate capability and remarkable cycling stability are mainly due to the binder-free porous 1D hollow architecture and their thin walls.

The electrochemical impedance spectroscopy (EIS) measurements were further performed to explore trends in electrochemical activity of the binder-free porous CC/SnO₂ nanotube arrays. The Nyquist plots of the binder-free porous CC/SnO₂ nanotube arrays and the commercial SnO₂ nanoparticles are displayed in Figure 7. In the Nyquist plots of the first and 50th cycled binder-free porous CC/SnO₂ nanotube array electrodes, two semicircles were found in the high to middle frequency range, and a straight sloping line in the low frequency range. According to the Randles equivalent circuit inserted into Figure 7A, the semicircle in the high frequency region corresponds to the resistance (R_SEI) and capacitance (C_SEI) of the SEI layer and the semicircle with a large radius in the middle frequency region represents the double-layer capacitance (C_DI) and the charge transfer process (R_CT), while the sloping line in the low frequency region reflects the Warburg impedance (Zw) related to the solid-state diffusion of lithium-ions (Qi et al., 2018). In contrast, the resistance (R_SEI) and capacitance (C_SEI) of the SEI layer contributed less to the Nyquist plots of commercial SnO₂ nanoparticles owing to their small specific surface area and low electron transfer (Figure 7B). The corresponding Randles equivalent circuit was also inserted in Figure 7B. The binder-free porous CC/SnO₂ nanotube arrays had a lower charge transfer resistance than the commercial SnO₂ nanoparticles, with only a slight increase after 50 cycles. Their charge transfer resistance was 6.65 and 8.84 Ω, respectively, for the first and 50th cycled electrodes. This indicates that the binder-free porous CC/SnO₂ nanotube array electrodes exhibited an excellent conductivity due to the faster rate of charge transfer (lithium-ion migration) on the 1D hollow electrodes. This leads to better cycle stability and rate capability.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.