Feilong
Yan
a,
Xuan
Tang
a,
Yuehua
Wei
a,
Libao
Chen
b,
Guozhong
Cao
*cd,
Ming
Zhang
*a and
Taihong
Wang
a
aKey Laboratory for Micro-Nano Optoelectronic Devices of Ministry of Education, School of Physics and Microelectronics, State Key Laboratory for Chemo/Biosensing and Chemometrics, Hunan University, Changsha, 410082, P. R. China. E-mail: zhangming@hnu.edu.cn
bState Key Laboratory for Power Metallurgy, Central South University, Changsha, 10083, P. R. China
cDepartment of Materials Science & Engineering, University of Washington, Seattle, Washington 98195, USA. E-mail: gzcao@uw.edu
dBeijing Institute of Nanoenergy and Nanosystems, Chinese Academy of Sciences, Beijing 100083, China
First published on 6th May 2015
Tin oxides with high theoretical capacities as anodes for lithium-ion batteries always suffer from the electrical disconnect issue of active materials owing to huge volume changes. In this work, flexible mats composed of ultra-small SnOx nanoparticles, graphene, and carbon fibers are synthesized by reducing graphene oxide with stannous ions at room temperature following treatments. SnOx nanoparticles, including SnO and SnO2, with diameters of approximately 4 nm are embedded in the matrix composed of carbon fiber and graphene to form composite fibers that are woven into flexible mats without any binders. As binder-free anodes for lithium-ion batteries, SnOx–graphene–carbon fiber mats can deliver a high reversible capacity of 545 mA h g−1 after 1000 cycles at a current density of 200 mA g−1, which is much better than those of SnOx–carbon fiber mats. The improved performance of SnOx–graphene–carbon fiber mats can be attributed to the ultra-small size of SnOx nanoparticles and the double protection of both graphene and carbon fibers.
However, tin oxides suffer from significant capacity fading due to enormous volume changes during the lithiation/delithiation process.5,10,17 Several approaches have been developed to address this issue. For example, tin oxides were nanosized to minimize the strain during volume changes.4,17–19 This strategy, however, results in a relatively high irreversible capacity in the initial cycle, arising from formation of enlarged solid electrolyte interface (SEI) films because of the high specific area of nanomaterials.20 Another approach is to integrate tin oxides with carbonaceous materials to accommodate their huge volume changes, including amorphous carbon, mesoporous carbon, graphene, carbon nanotubes (CNTs) or carbon nanofiber mats.4,5 For example, SnO2–graphene composites have been synthesized using tin salts and graphene oxide (GO) as raw materials by many methods,11 such as NaBH4 reduction,21 hydrothermal growth,22 and in situ deposition.5 Additional reductants are commonly used in the above routes to reduce GO. Although GO has been demonstrated to be reduced by stannous ions at a relatively high temperature without any reductants,23,24 the reduction of GO by stannous ions at room temperature with little energy consumption has not yet been reported. In addition, GO remains highly resistive even after reduction, which is not optimal for energy storage applications. GO can also control morphology and size of metal oxides.25
To realize the high power density and good cyclic stability of tin oxides based on composites as anodes for LIBs, they should possess two important features: high electrical conductance and fast diffusion channels for Li+.5,26,27 Tin oxides/graphene composites usually have relatively high conductance due to the presence of graphene.26,28 However, diffusion of Li+ may be blocked by the binder (e.g., PVDF, polyvinylidene difluoride) or stacking of graphene sheets in tin oxides/graphene composites.10,29 This issue could be partially solved by introducing some carbon nanoparticles between graphene sheets to enlarge the interspace.30 Li+ diffusion blocked by binders cannot be easily improved without binder-free electrodes,29,31 such as growth nanomaterials on nickel foams. However, the micrometer-sized nickel foam with relatively low specific surface area cannot load nanomaterials at sufficiently high content of active materials; thus, the great interest in exploring tin oxides for fabrication of binder-free anodes for LIBs.29,31
Although modifying LIBs with either carbon fibers or graphene is a useful strategy to maintain structural stability and improve electrochemical properties, the double-protection strategy of tin oxides by both carbon fibers and graphene have not been reported. In this study, flexible hierarchical mats of SnOx–graphene–carbon nanofibers (SGCFs) were synthesized by reducing GO with stannous ions at room temperature, and followed by electrospinning and thermal treatment. The ultra-small tin oxides with diameters of approximately 4 nm were anchored on carbon fibers by chemical bonds. Even though the content of graphene and SnOx in the nanofibers was very low, the flexible mats exhibited a high specific capacity at a large current density, and excellent cyclic stability as binder-free anodes for LIBs. Such superior electrochemical results were ascribed to the carbon and graphene double-protection strategy and the ultra-small size of tin oxides.
Graphene oxide (GO) was prepared according to our techniques described in the literature.29 GO was dispersed in DMF at a concentration of about 0.4 mg mL−1, and the suspension was processed with ultrasonic waves for 30 min. To prepare the precursor solution for electrospinning, 0.43 g PAN were dissolved in GO–DMF solution and vigorously stirred for 3 h at 30 °C, forming the brown solution; 1.5 mmol SnCl2·2H2O were dissolved in DMF and vigorously stirred for 1 h at room temperature, forming the transparent solution in Fig. 1a. Next, the vigorously stirred solution was added in the brown solution with a dropper. To prepare comparative carbon fibers, a precursor without GO was also prepared. A precursor for nanofibers without graphene was prepared in a similar way, also without GO. All of the precursor solution was transferred into a 5 mL syringe with a stainless steel needle (with 0.71 mm inner diameter). A syringe pump controlled the flow rate of the precursor solution to about 0.2 mL h−1. A piece of aluminum foil used as the collector was vertically positioned about 15 cm away from the needle. The needle was connected to a high-voltage DC power supply to obtain 11–14 kV. Under these conditions, PAN–SnCl2·2H2O and PAN–GO–SnCl2·2H2O nanofibers were generated and formed mats. After preoxidation at 230 °C in air for 6 h, the resulting brown films (PAN–GO–SnCl2·2H2O) were treated at 600 °C and 700 °C in argon for 2 h in order to carbonize the PAN and decompose SnCl2·2H2O, and the product is denoted as SGCF-600 and SGCF-700, respectively. The SnOx–C nanofibers (SCFs) were treated in a similar way at 700 °C in order to obtain the corresponding samples, and the product is denoted as SCF-700.
Sn2+ + 2H2O ⇔ Sn(OH)2 + 2H+ | (1) |
Sn(OH)2 + GO → SnO2 + G + 2H+ | (2) |
The detailed morphological and structural features of the SGCFs were also characterized by TEM. Fig. 3a shows a typical TEM image of SGCF-700. No aggregated SnOx nanoparticles are found on the smooth surface of the fibers, indicating that the particles were embedded in SGCFs. This phenomenon can be confirmed by more TEM images (Fig. S3†) at different magnifications. The fracture of SGCFs could be ascribed to ultrasonication during sample preparation for the TEM test. The high-magnification TEM image is shown in Fig. 3b. The lighter color in the fibers represents the carbon matrix and the darker color represents SnOx. This image reveals that the ultra-small nanoparticles with diameters of approximate 4 nm (black dots) were well dispersed in SGCFs, suggesting good compatibility between carbon and nanoparticles. The size of SnOx is very close to the perfect diameter of SnO2 nanoparticles as anodes for LIBs reported in a previous paper.36 To further explore composite structure, high-resolution TEM (HRTEM) studies were carried out, as presented in Fig. 3c and d. An ultra-small SnOx was embedded in the SGCFs. Two planes with d-spacing of 0.275 nm in Fig. 3c and 0.26 nm in Fig. 3d were observed on the surface of SGCFs. Those values are highly consistent with the (011) planes of SnO and the (002) planes of SnO2,5,37 indicating the presence of ultra-small SnOx nanoparticles in SGCFs and transformation of SnO2 to SnO during the treatment at 700 °C. The ultra-small size of SnOx nanoparticles could be attributed to the inhibition of carbon matrix and graphene on nanoparticle growth.35 The presence and distribution of Sn elements were further verified by TEM element mapping (Fig. S4a–d†). Fig. S4c† confirmed the homogenous distribution of Sn on CNFs. Other elements, such as C, N, and O, also were uniformly distributed on CNFs. Taking into account the TEM results, it can be concluded that ultrafine SnOx nanoparticles are homogenously distributed on CNFs.
The chemical composition at the surface of SGCF-700 was also investigated by X-ray photoelectron spectroscopy (XPS). As shown in Fig. 4a, the Sn 3d5/2 peak can be deconvoluted to two peaks. The peaks at 486.7 and 487.4 eV are ascribed to Sn2+ and Sn4+, corresponding to atom ratios of 61.3% and 38.7%, respectively.1,5 This result is consistent with the TEM conclusion that there are two tin-based compounds in the SGCF-700. The O 1s spectrum is shown in Fig. 4b, which could be deconvoluted to four peaks. The peaks at 531.3, and 533.2 eV are assigned to oxygen in O2 and H2O, respectively, which should be attributed to the adsorbed H2O and O2 on the carbon matrix–graphite.1 Another peak at 533.6 eV originated from C–O bonding indicates that the oxygen also bonds with carbon. Other peaks at 531.3 and 533.2 eV can be indexed to oxygen and water.1 The remaining peak may be evidence of Sn–O bonding (532.4 eV), which was directly explained by the anchoring of ultra-small tin oxides with 4 nm diameters on carbon fibers by chemical bonds.1 The full XPS spectrum of SGCF-700 in Fig. 4c demonstrated the presence of nitrogen and carbon in SGCFs, but not tin and oxygen. The fine XPS spectrum of N 1s shows two peaks at about 398.4 and 400.8 eV that could be indexed to pyridine-type and conjugated nitrogen. Both of the above nitrogen types have positive effects on the storage of lithium ions, especially pyridine-type nitrogen.29 In addition, nitrogen doping could improve carbon conductivity, and result in enhanced properties of lithium ion storage.1,4 The EDS spectrum in Fig. 4d shows the presence of carbon, oxygen, tin, and nitrogen in SGCFs, which is consistent with XPS results in Fig. 4c. The copper can be assigned to the sample holder for TEM characterization. Thus, SnOx nanoparticles (including SnO2 and SnO) were chemically embedded in nitrogen-doped carbon fibers.
Thermogravimetric analysis (TGA) was carried out to gain insight into the chemical composition of SGCF-700 in air; results are shown in Fig. 5. From the TGA curve, SGCF-700 has 68.22% weight loss from 300 to 633 °C that can be assigned to combustion of graphene and carbon in air forming CO2.4 This carbon weight loss agrees well with XPS analysis results. The final product of TGA is SnO2 with weight of 33.8%. Considering the transformation of SnO to SnO2 with weight increase of 11.9% and Sn2+ with an atomic ratio of 61.3%, the weight ratio can be estimated at 18.5 wt% for SnO, 13.1 wt% for SnO2, 67.3 wt% for carbon fibers, and 0.1 wt% for graphene. Calculation details are shown in the ESI.†
Fig. 5 Thermogravimetric analysis (TGA) of SGCF-700 at heating rate of 3.5 °C min−1 in air with flow rate of 20 mL min−1. |
Nitrogen adsorption studies were carried out in order to investigate the surface area and pore volume of as-prepared porous SGCF-700 mats (Fig. S5†). BET results show that porous SGCF-700 nanofibers have a surface area of about 14.9 m2 g−1. In addition, porous SGCF-700 nanofibers also have a pore volume of about 0.208 cm3 g−1. Fig. S5b† demonstrates that most of the pores are smaller than 50 nm in diameter, and that SGCF-700 mats are mainly composed of mesopores (1–50 nm). The BET results demonstrate the formation of porous structure in the SGCF-700 mat.
CV tests were performed to investigate the mechanism of electrochemical reactions, as shown in Fig. 6. The cathodic peak of SGCF-700 at about 0.57 V in the first cycle (Fig. 6a) is ascribed to formation of an SEI film and the lithiation of SnO2 (eqn (3)), which disappears in the following cycles. Another peak in the range of 0–0.3 V can be indexed to intercalation of lithium ions into carbon (eqn (5)) and formation of Li–Sn alloys (eqn (4)).38–40
During the anodic scan, three peaks were found. The peak at about 0.1 V could be attributed to the delithiation of graphite carbon. Another anodic peak at 0.65 V can be ascribed to delithiation of Li–Sn alloys (eqn (4)). The anodic peak at about 1.35 V may be deconvoluted to two aspects. The first is the reversible transfer of Sn0 to SnO2, which has been observed in some SnO2–carbon–graphene composites for LIB.23,41,42 Another possible reaction representing this peak is the reversible lithium ions extracting from defective sites or micropores of carbon matrix.43 Further investigation needs to be carried out to clarify details of the mechanism for this peak. Although the CV curves of SCF-700 are similar to those of SGCF-700, there are some differences between them. First, changes of anodic peaks for SGCF-700 at 0.65 and 1.35 V are smaller than those of SCF-700 in the initial four cycles, showing an improved cyclic stability of SGCF-700 by introducing graphene, although both show good cyclic stability from the second to the fourth cycle. Second, the peak current ratios of peak 3 to peak 2 of SGCF-700 are higher than those of SCF-700, indicating a possible improved reversibility of the transfer of Sn0 to SnO2 and lithium ions extracting from the defective sites or carbon matrix micropores.44
SnOx + 2xLi+ + 2xe− → Sn + xLi2O | (3) |
Sn + xLi+ + xe− ⇔ LixSn (0 ≤ x ≤ 4.4) | (4) |
6C + yLi+ + ye− ⇔ LiyC6 (0 ≤ y ≤ 1) | (5) |
The charge–discharge curves of SGCF-700 mats are shown in Fig. 7a. It is worth noting that the poorly defined plateau at about 0.8 V associated with the reduction of SnO2 to Sn becomes indiscernible in the following cycles, which suggests irreversibility of this reaction within the narrow voltage window.45 The potential profiles of SGCF-700 electrodes at a current density of 200 mA g−1 presented excellent discharge and charge capacities of 1272.6 and 828.7 mA h g−1 in the first cycle corresponding to a Coulombic efficiency (CE) of 65.1%. While increasing the number of cycles, the discharge capacity gradually decayed to 891.6, 639.8, 601.6, and 598 mA h g−1 in the 2nd, 100th, 200th, and 400th cycles, respectively. The large capacity loss (34.9%) in the first cycle is generally attributed to irreversible formation of the SEI film during the first discharge process.
Fig. 7b displays the cyclic properties of SGCF-700 mats in the potential range of 0.005–3.0 V at a current density of 200 mA g−1. The CE of the second cycle exceeds 92.3%. The high CE indicates that embedding SnOx nanoparticles in the carbon matrix can largely prevent detrimental reactions between SnOx and the electrolyte, which is in good agreement with the CV results. High capacity should benefit from the ultra-small SnOx nanoparticles and porous carbon network structure, which enable full utilization of SnOx nanoparticles. More importantly, after several cycles the reversible capacity is stabilized at 100%.
The slow shrinkage of charge–discharge curves after the first cycle implies the high cycling stability of the as-prepared composite. The reasons for this enhancement are caused by the synergistic effect of electrospinning and optimal heat treatment. For comparison, the SGCF-600 composite anode delivered an initial discharge capacity of 1277 mA h g−1 with a CE of 70.9%. The properties of SGCF-700 are just a little better than those of SGCF-600. But after 135 cycles, SGCF-600 anodes only delivered a capacity of 496 mA h g−1, which is much lower than SGCF-700 anodes. The inferior properties of SGCF-600 could be attributed to the low degree of carbonization and poor nanofiber conductance. This result agrees with a recent study on PAN-based carbon for LIBs.29 The porous carbon nanofiber mats prepared by electrospinning act as a buffering matrix that can prevent the agglomeration of SnO2 particles, volume expansion, and mechanical stress during Li alloying and dealloying.31 The porous carbon nanofiber is used as an electrical conductor to compensate for the drawbacks of semiconductors such as SnO2 and Li2O. And it helps promote decomposition of SnOx as a semiconductor into Sn and Li2O. Moreover, the porous carbon nanofibers allow the charge transfer and mass transfer to facilitate lithiation/delithiation due to high electronic conductivity.5,29
Fig. 7c compares the cyclic stability and specific capacities of SGCF-700 and SCF-700 mats. It is clear that SGCF-700 mats exhibit much better cycling stability than SCF-700 electrodes. SCF-700 mats delivered an initial discharge capacity of 1637 mA h g−1 with a CE of 43.0%. Although the initial discharge of SGCF-700, 1273 mA h g−1, is lower than that of SCF-700. SGCF-700, a discharge capacity of 640 mA h g−1 in the 100th cycle, 1.25 times that of SCF-700 in the same cycle. Even after 480 cycles, SGCF-700 mats still delivered a discharge capacity of 600 mA h g−1, which is almost 50% higher than that of traditional graphitic anode (372 mA h g−1). In addition, the CE for SGCF-700 was also higher than that of SCF-700 in the fourth cycle in spite of CE being almost 100% in subsequent cycles. Besides, the high CE in SGCF-700 anodes after four cycles suggests high reversibility of SGCF-700 mats for the storage of lithium ions. Considering the difference between SGCF-700 and SCF-700, the improved properties of SGCF-700 could be attributed to the presence of graphene, which can affect the growth of nanoparticles resulting in ultra-small nanoparticles and suppress the agglomeration of tin alloys.10,28 Another reason for the high properties of SGCF-700 may be the ultra-small diameter (4 nm) of SnOx nanoparticles, which is close to the optimizing size (about 3–4 nm) of SnO2 for Li+ storage.36
Fig. 7d shows the rate capacities of SGCF-700 and SCF-700. SGCF-700 maintains reversible capacities of 850, 737, 613, 504, and 368 mA h g−1 at current densities of 100, 200, 500, 1000, and 2000 mA g−1, respectively. These values are higher than those of SCF-700 at corresponding current densities. This improvement could be attributed to the enhanced conductance arising from graphene. The synergistic effects arising from the ultra-small SnOx particles within the porous carbon nanofibers and graphene are further discussed for comparison. It is proven that the graphene can be used to improve the mechanical properties of porous carbon nanofiber mats made by electrospinning.
To demonstrate the superior electrochemical performance of SnOx–carbon-nanofiber (CNF) freestanding electrodes developed in this study, Table 1 compares the capacities and cycles of the electrodes prepared using similar SnOx-based materials and approaches taken from recent reports. Indeed, the SnOx–graphene–carbon fiber composites obtained in this study present relatively high capacities and superior cyclic stability.
Material | Capacity (mA h g−1) | Current density | Ref. |
---|---|---|---|
SnOx–CNF | 674, 100th cycle | 500 mA g−1 | 5 |
SnOx–CNF | 510, 40th cycle | 30 mA g−1 | 14 |
SnOx–CNF | 450, 30th cycle | 25 mA g−1 | 46 |
SnOx–CNF | 649, 100th cycle | 30 mA g−1 | 47 |
SnOx–CNF | 608, 200th cycle | 500 mA g−1 | 48 |
SnOx–CNF | 520, 100th cycle | 200 mA g−1 | 49 |
SnOx–CNF | 512, 200th cycle | 200 mA g−1 | 50 |
SnOx–G–CNF | 600, 480th cycle | 200 mA g−1 | This study |
SnOx–G–CNF | 545, 1000th cycle | 200 mA g−1 |
In order to investigate the lithium-driven structural and morphological changes during the charge and discharge processes, we performed ex situ SEM analyses on the SGCF-700 electrodes before cycling and after the 1000th cycle and SCF-700 electrodes after the 100th cycle at a current density of 200 mA g−1; the SEM images are shown in Fig. 8. It can be seen in Fig. 8 that the structure and morphology of SGCF-700 electrodes maintain integrity (Fig. 8a and S5a†). However, in the case of SCF-700 electrodes, much of the aggregate was on the porous carbon nanofiber surfaces (Fig. 8b), and protuberant particles with diameters of 1–4 μm were dispersed in SGCFs (Fig. S5b†). The stark contrast between the two samples demonstrates that graphene not only can obviously improve specific capacity, but also can maintain cycling stability for LIBs.
Fig. 8 (a) SEM image of SGCF-700 after 1000 cycles. (b) SEM image of SCF-700 after 100 cycles. Both SGCF-700 and SCF-700 were tested at current density of 200 mA g−1. |
Footnote |
† Electronic supplementary information (ESI) available: SEM images of diverse samples, FT-IR spectra for rGO and GO and TEM images of the SGCF-700. See DOI: 10.1039/c5ta02107c |
This journal is © The Royal Society of Chemistry 2015 |