A strategy to overcome the limits of carbon-based materials as lithium-ion battery anodes

CARBON

x x x ( 2 0 1 4 ) x x x –x x x

Available at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/carbon

A strategy of overcoming the limit of carbon-based materials for anode of lithium ion battery Fei Yao a,1, Bing Li a,b,1, Kangpyo So a, Jian Chang a,b, Thuc Hue Ly a,b, An Quoc Vu Hyeona Mun a,b, Costel Sorin Cojocaru c, Hongyan Yue a,d, Sishen Xie e, Young Hee Lee a,b,*

a,b

,

a

Center for Integrated Nanostructure Physics, Institute for Basic Science, Sungkyunkwan University, Suwon 440-746, Republic of Korea Department of Energy Science, Department of Physics, Sungkyunkwan University, Suwon 440-746, Republic of Korea c LPICM CNRS, E´cole polytechnique, Palaiseau 91120, France d School of Materials Science and Engineering, Harbin University of Science and Technology, Harbin 150040, People’s Republic of China e Institute of Physics, Chinese Academy of Sciences, National Centre for Nanoscience and Technology, Beijing 100080, People’s Republic of China b

A R T I C L E I N F O

A B S T R A C T

Article history:

The free-standing Si-coated carbon nanofiber (Si/CNF) mat was fabricated for the anode of

Received 16 May 2014

lithium ion battery through combining electrospun CNF mat with electrodeposited Si layer.

Accepted 8 August 2014

Spaghetti or granule-like Si was obtained by varying the deposition conditions. This Si/CNF

Available online xxxx

mat was directly used as an active material and a current collector as well, which involves neither binders nor additional metal substrate. The best performance was achieved in spaghetti-like Si due to its highly porous nature which can accommodate volume expansion and large surface area which benefit the efficient charge transfer both at Si/CNF interface and at the electrode/electrolyte interface. The optimized Si/CNF mat after annealing at 1000 C delivered a capacity of 870 mA h g1 at 1st discharge and 730 mA h g1 at 50th discharge with a capacity retention of 84%, improving the capacity of pure CNF (280 mA h g1 at the 50th discharge) by almost three times. In addition, corrosion of the current collector no longer exists in our approach. Our X-ray photoemission spectroscopy and electrochemical analysis revealed that the formation of Si–C bond through high temperature annealing can enhance the adhesion between silicon and carbon at the interface which benefits the cyclic performance of anode ultimately.  2014 Elsevier Ltd. All rights reserved.

1.

Introduction

The fast development of modern technology requires matured energy storage devices to meet the demands of ever growing portable electronic and electric vehicle industries.

Rechargeable lithium-ion battery (LIB) has become the most promising candidate due to its high energy density, long cycle life and possibility of compact design compared to any existing battery systems. Graphite has been widely used as a standard anode material for LIB because of its high chemical

* Corresponding author at: Center for Integrated Nanostructure Physics, Institute for Basic Science, Sungkyunkwan University, Suwon 440-746, Republic of Korea. Fax: +82 31 299 6505. E-mail address: [email protected] (Y.H. Lee). 1 These authors contributed equally to the work. http://dx.doi.org/10.1016/j.carbon.2014.08.017 0008-6223/ 2014 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Yao F et al. A strategy of overcoming the limit of carbon-based materials for anode of lithium ion battery. Carbon (2014), http://dx.doi.org/10.1016/j.carbon.2014.08.017

B&M B B&M B&M M M M B&M M No B & M 80% 80.3% 68.2% 90% 89% 90.4% 46.8% 1 — 84%

c

a

b

C, rate is not mentioned in the reference. Capacity is calculated based on silicon mass exclusively. Si NPs were dispersed in polymer solution firstly and then electrospun at the same time with fiber.

1600@55 766@20 750@100 1200@105 2752@100b 773@20 726@40 1300@100 892@50 730@50 500 50 0.5Ca 50 323 100 50 240 35 37 75% 37% 25% 16% 49% 16% 26% 50% 41% 43% CVD CVD CVD Sputtering Sputtering Cospinningc Cospinning Cospinning Cospinning Electro-deposition Si@CNF [8] Si@CNF [9] Si@Hollow CNF [10] Si@CNFs [11] Si@VACNFs [12] Si/CNF [13] Si/CNF [14] Si–CNF core–shell [15] Si/CNF [16] Si/CNF (ours)

2 4 — 0.4 0.2 — 1.6 — — 1.8

Capacity retention Capacity @cycles (mA h g1) Current density (mA g1) Si mass ratio Mass (mg cm2) Si deposition method

stability, good conductivity, and low Li reaction potential versus Li/Li+ which is a favorable factor for high output cell voltage. Furthermore, the well-defined layered structure of graphite exhibits only 10% volume change during Li intercalation and therefore benefits the anode stability and the cell life time [1]. Nevertheless, a rather small theoretical maximum Li storage capacity of 372 mA h g1 for LiC6 stage is far from the demand of modern society [2]. Moreover, typical graphite anode fabrication process usually involves metal current collector and binder. The uses of these two extra components not only cause a corrosion-related issue in a long run (environmentally assisted cracks will eventually form on Cu substrate) but also affect the effective mass of the electrode more seriously [3]. Considering the more realistic picture of the battery performance, current collector and binder always need to be taken into account which is adverse to the specific and volumetric capacity of the cell eventually [4]. However, the anode capacity reported so far is usually considered as for pure active materials exclusively without concerning the use of metal substrate, as indicated in the last column in Table 1. In this case, the real capacity of the anode has been overestimated by a factor of 3 approximately compared to that of a whole electrode including metal support [4]. Therefore, a high capacity and metal substrate-free anode is highly demanding. In order to overcome these issues, tremendous efforts have been made on different carbon structures to replace the graphite anode. Compared to carbon nanotubes/graphene thin films which involve complicated fabrication procedures and often post-treatment processes, the free-standing carbon nanofiber (CNF) mat can be easily fabricated into a large area by a robust electrospinning method. Conductivity of the CNF mat is 10 S cm1 and BET surface area is 33 m2 g1 [5]. This method is easy and cost effective, and the generated CNF mat as a free-standing film can be directly used as an electrode without involving metal current collectors or binders. However, the typical Li storage capacity of CNFs is still relatively low. Recently, silicon, a high lithium storage capacity material (specific capacity of 3572 mA h g1 at room temperature, corresponding to Li15Si4) has been proposed [6]. Yet, large volume expansion up to 400% during charge/discharge causes a severe structural pulverization, making this material impractical. For example, the Si thin film deposited on metal substrate by chemical vapor deposition leaves crack formation during cycling and therefore a contact loss between active Si material and current collector takes place, leading to a poor cyclability [7]. Owing to these difficulties, several Si/carbon fiber (CF) composite structures have been proposed, as summarized in Table 1. Silicon has been successfully deposited onto CFs or CNFs through chemical vapor deposition or sputtering method [8–12]. Although Li storage capacity was improved due to the contribution of the deposited Si layers, inhomogeneous deposition of Si atoms on fibers along the depth of the film diminishes the effect of Si layers. Si nanoparticles have also been deposited on CNFs by dispersing them in organic solution and then co-spinning onto metal substrate [13–16]. This method causes undefined nature of adhesion between Si nanoparticles and CNFs, which is closely

Structural features

x x x ( 2 0 1 4 ) x x x –x x x

Structure

CARBON

Table 1 – Anode performance comparison of silicon/CNF composites fabricated through different methods. B and M in the last column indicate the use of binder and metal substrate, respectively. It is of note that capacity is calculated based on active material only without considering the mass of metal electrode in which the capacity is overstated in the case of previous publications.

2

Please cite this article in press as: Yao F et al. A strategy of overcoming the limit of carbon-based materials for anode of lithium ion battery. Carbon (2014), http://dx.doi.org/10.1016/j.carbon.2014.08.017

CARBON

xxx (2014) xxx–xxx

related to the inefficient charge transport across the interface and the low capacity retention. Electrochemical deposition of silicon onto CNF substrate is rather promising, since the liquid reaction is easy to handle with low cost and also the shape of silicon can be controlled by varying the deposition conditions. In this study, the free-standing CNF mat was fabricated by using electrospinning of polymer solution followed by stabilization and carbonization. Si was deposited on the surface of CNFs by electrodeposition. By varying the deposition conditions, a spaghetti-like Si layer with high surface area and porosity was formed uniformly over nanofibers independent of the depth of the film. High temperature annealing of 1000 C was performed to improve material purity and construct stable Si and CNF interface by forming Si–C bond. This free-standing Si-coated CNF mat was directly used as an anode material for LIB without using any additional metal current collector which is distinct from previous works (Table 1). The capacity of Si/CNF mat anode was clearly improved by almost three times compared to that of graphite material. The detailed electrochemical analysis was provided in conjunction with structural properties.

3

the solution conductivity. Electrodeposition of Si was carried out though cyclic voltametry technique. Si deposition of 200, 500, 1000 and 1500 cycles was achieved by scanning the voltage from 2.7 to 0 V vs. Ag/Ag+. The as-deposited Si/CNF mat was firstly rinsed by PC solution to remove the toxic residue and then annealed at 1000 C for 3 h at a vacuum level of 1 · 106 Torr.

2.3.

Material characterization

2.

Experimental section

The morphology and structure of the as-synthesized and electrochemically tested Si/CNF mat were observed by field emission scanning electron microscopy (JEOL, FESEM 7600F) and transmission electron microscopy (JEOL, JEM3010, 300 kV). The CNF or Si/CNF mats were directly used for SEM observations. For TEM observations, the samples were firstly sonicated in ethanol and then transferred onto holey carbon membranes supported by copper grids. Raman spectroscopy (Renishaw, RM-1000 Invia) with an excitation energy of 2.41 eV (514 nm, Ar+ ion laser) was used to characterize the optical properties of the samples. The Si/CNF mat was also characterized by atomic force microscopy (AFM, SPA-400, Seiko, Japan) and X-ray photoelectron spectroscopy (XPS, ESCA 2000, VG Microtech, England).

2.1.

Synthesis of CNF mat

2.4.

Electrochemical measurement

Poly(amic acid) (PAA) was synthesized by pyromellitic dianhydride (PMDA, Sigma Aldrich) and oxydianiline (ODA). PMDA of 4.4 g was added into ODA (4.0 g) pre-dissolved DMF solution (21 g). The mixture was stirred for 30 min with a magnetic bar. 413 lL triethyl amine (TEA) was then added to control the molecular weight. The as-prepared solution was then electrospun into PAA nanofibers onto a cylindrical collector wrapped with aluminum foil. The separation distance between the needle and collector, DC bias voltage, and solution flow rate were 15 cm, 20 kV, and 0.2 mL h1, respectively. The PAA nanofiber mat with aluminum foil was then put into stabilization oven and converted into polyimide (PI) mat after seven different oxidation steps at a rate of 1 C min1 [17]. The PI mat was then peeled off from the aluminum foil and transferred into high temperature furnace. CNF mat was formed by annealing the PI mat according to three steps annealing procedures (firstly from room temperature to 600 C in 1 h, then 600 C to 1000 C in 1.3 h, and finally maintaining in 1000 C for 1 h) under argon gas environment. The detailed procedures have been published elsewhere [5].

Bare CNF and Si/CNF mats were carefully weighted after 1000 C annealing by using the A&D BM-22 microbalance located inside the dry room. Electrochemical measurements were carried out with a CR 2032 coin cell using VMP3 instrument (BioLogic Science Instruments). The cell was assembled in a dry room using CR 2032 cell case with bare CNF mat or silicon/CNF paper as a working electrode, lithium metal foil as a counter/reference electrode, and 1 M of LiPF6 in a 1:1 (v/v) mixture of ethylene carbonate (EC) and diethyl carbonate (DEC) as electrolyte. No binder or metal substrate was used. A glassy carbon microfiber was used as a separator. The cells were charged and discharged galvanostatically between 2.0 and 0.01 V vs. Li/Li+. Here, we defined 1 C to be 372 mA h g1. The cyclic voltametry measurement of the half cell battery was performed at a rate of 0.1 mV s1. The AC impedance spectra were obtained by applying a sine wave with an amplitude of 10 mV over a frequency range of 100 kHz to 10 mHz to the assembled cells before cycling.

2.2.

A schematic of a silicon-coated CNF (Si/CNF) mat fabrication processes was shown in Fig. 1. The poly(amic acid) (PAA) solution with 1 wt% of triethyl amine as a catalyst to polymerization was electrospun into nanofibers through a Taylor cone at the tip of the needle onto an aluminum foil wrapped cylindrical collector, as shown in Fig. 1a. The separation distance between needle and collector was 15 cm. DC bias voltage of 20 kV and solution flow rate of 0.2 mL h1 were applied. The as-synthesized PAA nanofiber mat was then peeled off from the aluminum foil substrate and then eventually converted into CNF mat (Fig. 1b) after stabilization and carbonization

Electrodeposition of silicon onto CNF mat

Silicon electrodeposition was conducted using a home-made three-electrode system in an Ar-filled glove box (humidity and oxygen contents are below 1.0 ppm). The working electrode is CNF mat (diameter of 15 mm), counter electrode is platinum wire and reference electrode is Ag/Ag+ (tetra-nbutylammonium perchlorate (TBAP)/acetonitrile). Propylene carbonate (PC) was chosen as a non-aqueous solvent with 0.5 M SiCl4 as Si source. 0.1 M tetra-n-butylammonium chloride (TBACl) was added into the organic electrolyte to improve

3.

Results and discussion

Please cite this article in press as: Yao F et al. A strategy of overcoming the limit of carbon-based materials for anode of lithium ion battery. Carbon (2014), http://dx.doi.org/10.1016/j.carbon.2014.08.017

4

CARBON

x x x ( 2 0 1 4 ) x x x –x x x

Fig. 1 – Schematic of a silicon-coated CNF mat fabrication processes: (a) schematic of electrospinning apparatus, (b) the fabricated nanofiber network, (c) apparatus of electrodeposition of Si, and (d) the deposited Si/CNF mat. The inset shows cross section of coaxial type Si/CNF. (A color version of this figure can be viewed online.)

processes. The detailed process was described elsewhere [5]. Silicon was deposited electrochemically on the CNF mat (working electrode) in a home-made three-electrode cell (Fig. 1c). 0.5 M SiCl4/propylene carbonate (PC) was used as an electrolyte with 0.1 M TBACl as a supporting electrolyte to increase the solution conductivity [18]. Platinum wire was used as a counter electrode and Ag/Ag+ (TBAP/acetonitrile) electrode as a reference electrode. A voltage sweeping range of 2.7 to 0 V vs. Ag/Ag+ was applied so that Si can be deposited. This Si/CNF mat, as shown in Fig. 1d, was finally formed after a high vacuum (1 · 106 Torr) annealing at 1000 C for 3 h. The CNF-Si core–shell structure is indicated in the inset of Fig. 1d. Fig. 2a shows typical cyclic voltammograms (CVs) at a scan rate of 20 mV s1 for CNF in PC electrolyte with/without adding SiCl4. It is clear to see a reduction peak centered at around 2.0 V only in the case of electrolyte containing SiCl4. This suggests that Si ion was reduced into Si and deposited onto CNF mat during CV test following the electrochemical reaction in Eq. (1): SiCl4 þ 4e ! Si þ 4e

ð1Þ

Si loading amount on CNF mat was controlled by varying number of CV deposition cycles, as shown in Fig. 2b. Mass (thickness) of Si/CNF mat keeps increasing from 1 to 4 mg (from 25 to 130 lm), as the number of Si deposition cycles increased to 1000 cycles. A series of structure characterizations of Si/CNF mat with 200 cycles CV deposition are shown in Fig. 3. Micro-Raman spectra of bare CNF mat and pristine Si/CNF mat (without annealing, indicated as Si-200-p) in Fig. 3a clearly show

G-band near 1592 cm1, which is related to the optical E2g phonon at the Brillouin zone center indicating sp2 hybridization of carbon network and D-band near 1352 cm1, which corresponds to transverse optical phonon near the K point and indicates sp3 hybridization of carbon network [19]. The intensity ratio of D band to G band (ID/IG) remained unchanged (0.83) between bare CNF and Si-200-p after electrodeposition of Si, indicating that carbon material is remarkably stable compared to traditional metal substrate in severe electrochemical environment [20]. Thus, the issues for induced corrosion of common metal current collector can be avoided in the case of CNF. Nevertheless, the fairly high value of ID/IG implies relatively low conductivity of CNFs, which may lead to poor electrochemical performance [5]. This value could be improved by optimizing carbonization conditions. In the case of Si/CNF mat after 1000 C annealing (indicated as Si-200-a), the value of ID/IG slightly decreased to 0.79, suggesting an improved graphitization in the CNF network. It is also worth noting that D and G bands are broadened slightly due to the formation of Si–C bonds, which will be discussed in the next paragraph. It is of note that no Si-related peak can be found in Si-200-p. This could be ascribed to the highly disordered deposited Si layer which is caused by electrostatic clustering with alkyl terminators and also to the presence of deposited electrolyte residues on the surface (See Fig. S1 in the Supplementary data) [21]. On the other hand, three additional Si-related peaks were shown in the spectrum of Si-200-a. It is known that the first order transverse-optical (TO) phonon mode of crystalline Si (c-Si) will display a sharp peak at 520 cm1 which usually becomes broadened and is

Please cite this article in press as: Yao F et al. A strategy of overcoming the limit of carbon-based materials for anode of lithium ion battery. Carbon (2014), http://dx.doi.org/10.1016/j.carbon.2014.08.017

CARBON

xxx (2014) xxx–xxx

Fig. 2 – (a) Cyclic voltammograms of silicon electrodeposition in PC solution with/without SiCl4 at a scan rate of 20 mV s1. (b) Mass and thickness of Si/CNF mat with respect to different silicon deposition cycles. The error bar is added in the figure. (A color version of this figure can be viewed online.)

downshifted when the long-range order in Si is lost [22]. In our case, the peak located at around 500 cm1 was assigned to microcrystalline or nanocrystalline (lc/nc) Si and a broad band at the low energy side originated from the presence of amorphous Si (a-Si). The peak near 300 cm1 resembles transverse acoustic (TA) phonon mode of c-Si and could be softened in a-Si [22–24]. In addition, c-Si usually exhibits a small peak at 950 cm1 which is related to the chemisorption of atomic/molecular oxygen species [25]. Here, a softening (920 cm1) was also observed in Si-200-a sample. A red shift of 30 cm1 is possibly caused by the existence of a-Si [24]. All of these factors demonstrate that the as-deposited Si is completely disordered and evolves into more distinct a-Si and lc/nc-Si with additional oxygen species after high temperature annealing. Fig. 3b plots the XPS spectra of Si/CNF mat with 200 cycles deposition before and after1000 C annealing. It is obvious to see that the intensities of Si 2s and Si 2p peaks increased clearly while the C 1s peak relatively decreased. In addition, Cl 2p peak which appeared in Si-200-p disappeared after annealing. After Si deposition, certain amount of electrolyte could be decomposed and remained on the Si surface. After annealing, the residual film which mainly contained C, O, and Cl was removed, as seen in Fig. S1. This is also in good corroboration with Fig. 3a.

5

Fig. 3c is the SEM image of the bare CNF mat with an average diameter of 150 nm, where the surface of CNFs was smooth. On the other hand, the Si-200-a sample displayed a rough spaghetti-like surface, as shown in Fig. 3d. The cross sectional view of Si–CNF core–shell structure was shown in the inset of Fig. 3d. It is of note that the core–shell structure was formed uniformly independent of the depth over hundred micrometers, which is in good contrast with other methods such as sputtering and CVD, in which Si is not uniformly deposited along the depth of the sample. AFM morphology of the same sample was provided in Fig. 3e with an amplified phase image in Fig. 3f, again demonstrating rough Si surface on the surface of CNFs. This unique spaghetti-like Si structure provided large surface area compared to the flat Si thin film. The volume expansion is therefore expected to be accommodated to a certain degree by the high porosity of Si under the condition of 200 cycles deposition which is certainly better than the thick Si layers. Fig. 3g is the TEM image of the Si-200-a sample. The layer thickness of the deposited Si was 20 nm in this case. The existence of Si layer on the surface of CNF was again confirmed by EELS line profile along the dashed line in the TEM figure, as shown in the bottom panel in Fig. 3g. The spaghetti shape of Si was not visible here probably due to the structure damage during TEM sample preparation process with sonication. Since the thin film of electrodeposited silicon is highly active and therefore can be oxidized immediately upon exposure to air during transfer from the glove box to TEM holder, or X-ray diffraction (XRD) measurements, and also the portion of lc/nc Si is rather small, the crystalline nature of the electrodeposited Si film is unlikely to be directly observed, as shown in Fig. S2 [26]. To obtain information for interface between CNF and Si, C 1s and Si 2p peaks in XPS were deconvoluted, as shown in Fig. 4. C 1s peaks before and after annealing were clearly distinct with each other. Clear Si–C peak near 282.9 eV was visible in addition to sp2 and sp3 peaks after annealing, while only intense sp2 and sp3 peaks were shown before annealing [27,28]. It is of note that the ratio of sp3 to sp2 peak was reduced after annealing (from 41% reduced to 31%), revealing similar trend to the change of D/G ratio in Raman spectra, as shown in Fig. 3a. In addition to this, considerable amount of Si–C bonds (14%) was generated after annealing. In the case of Si 2p, the main peak near 102.2 eV was SiOx peak with additional Si–Si peak near 99.3 eV before annealing [29]. After annealing, SiOx content was slightly increased due to some remaining oxygen content in the chamber. More importantly, Si–C peak near 100.8 eV appeared after annealing. Both XPS data clearly indicates that the Si–C bonds were formed at the Si–CNF interface after annealing. This peak is small due to narrow interface region, which is hardly observable by Raman spectroscopy (see Fig. 3a). This can be also confirmed by the XRD pattern in Fig. S2. The presence of such Si–C bonds at the interface may not contribute to Li storage but plays an important role in strengthening adhesion of Si layer to CNFs and consequently facilitating efficient charge transfer at the interface during lithiation/delithiation process [30,31]. This will improve cyclibility of the device eventually. The electrochemical performance of Si/CNF mat was investigated in LiPF6/EC + DEC solution. Fig. 5a shows the CV

Please cite this article in press as: Yao F et al. A strategy of overcoming the limit of carbon-based materials for anode of lithium ion battery. Carbon (2014), http://dx.doi.org/10.1016/j.carbon.2014.08.017

6

CARBON

x x x ( 2 0 1 4 ) x x x –x x x

Fig. 3 – (a) Micro-Raman spectra of bare CNF mat and Si/CNF mat with 200 cycles of Si deposition before/after annealing, indicated as Si-200-p and as Si-200-a in the figure. (b) XPS spectra of the electrode surface with active materials consisting of Si-200-p and Si-200-a, respectively. (c) SEM images of as-synthesized bare CNFs and (d) Si-200-a. The cross-sectional images are shown in the insets. (e) AFM image of Si-200-a. The high resolution image of dashed square in (e) is shown in (f). (g) TEM image of Si-200-a. The EELS line profile along the dashed line is shown in the bottom panel. (A color version of this figure can be viewed online.)

profile comparison between bare CNF mat and Si/CNF mat with 200 cycles Si deposition. For better comparison, the bare CNF mat was annealed (indicated as CNF-a) at the same conditions as the composite mat. The curves were recorded after 1st CV scan between 0.01 and 2 V at a scan rate of 0.1 mV s1. In the case of bare CNF-a (solid), lithiation/delithiation occurred below 0.3 V in the cathodic/anodic scan which resembled the characteristic of hard carbon [32,33]. On the other hand, the cathodic peak below 0.1 V and the anodic peak at 0.15 V belong to the characteristic of Li intercalation/deintercalation into graphitic layers [34,35]. (For electrochemical performance comparison between CNF-a and nonannealed CNF, see Fig. S3.) This manifests that our CNF mat contains a certain degree of graphitization and disordered phase which is consistent with Fig. 3a and Fig. 4. In the case of Si/CNF mat, no appreciable peaks related to LixSi alloy formation were observed in Si-200-p (dashed line). On the contrary, two pairs of redox reaction peaks were observed in Si-200-a (symbol). The sharp cathodic peak at 0.01 V can be attributed to a combination effect of CNF mat and c-Si/ a-Si. The cathodic peak (Li alloy) at 0.2 V and anodic peaks

(Li dealloy) at 0.37 and 0.52 V are due to the formation of amorphous LixSi phase and delithiation back to a-Si, respectively [36,37]. The increase of the peak intensities of a-Si with the scanning cycle numbers can be ascribed to the conversion of the lc/nc-Si into amorphous phase during the repeated CV scans. The electrochemical analysis in Fig. 5a provides us further understanding of the Si crystallinity which is again in good agreement with Fig. 3a. The galvanostatic charge/discharge (CD) profiles of 1, 10, 30, 50th cycles for CNF-a mat (line) and Si-200-a (symbol) are shown in Fig. 5b. (CD profiles of Si/CNF mat with 200 cycles deposition before/after 1000 C annealing are plotted in Fig. S4a.) The CD profiles of CNF-a mat show a gradual change in a broad voltage window during charge/discharge, revealing a V-shape feature. This is in good contrast with a U-shaped graphite CD curve due to the existence of disordered carbon phase in our CNF mat [32,38]. In the 1st charge of CNF-a, a plateau near 0.7 V vs. Li/Li+ can be attributed to the formation of solid-electrolyte interface (SEI) via electrolyte decomposition [32]. In the discharge process, the slope of the curve started approximately at 0.3 V vs. Li/Li+ and

Please cite this article in press as: Yao F et al. A strategy of overcoming the limit of carbon-based materials for anode of lithium ion battery. Carbon (2014), http://dx.doi.org/10.1016/j.carbon.2014.08.017

CARBON

xxx (2014) xxx–xxx

7

Fig. 4 – High-resolution XPS spectra of Si/CNF with 200 cycles of Si deposition before and after annealing. Figure (a) and (c) are C 1s and Si 2p fitted peaks before annealing. (b) and (d) are C 1s and Si 2p fitted peaks after 1000 C annealing. Peak positions and relative ratios are shown in the figure. (A color version of this figure can be viewed online.)

delivered a large specific capacity below 0.1 V vs. Li/Li+. The capacity from the potential region above 0.1 V may be ascribed to the faradic capacitance on the surface of CNFs and the capacity from the region lower than 0.1 V can be related to the lithium intercalation into CNFs [20,33,38]. This is in good agreements with Fig. 5a. The CNF-a mat delivered a discharge capacity of 458 mA h g1 at the end of 1st cycle and decreased to 280 mA h g1 after 50 cycles. Compared to CNF-a mat, the Si-200-a sample showed a large capacity of 870 mA h g1 in the 1st discharge and a capacity of 730 mA h g1 after 50th discharge which is almost three times higher than that of CNF-a. (The rate performance analysis and Coulombic efficiency of Si-200-a sample can be seen in Fig. S4b.) However, the large capacity loss between the 1st charge (1650 mA h g1) and discharge related to the formation of the SEI layer was observed, even though no distinguished SEI-related plateau was present. Compared to the Si thin film deposited on the two dimensional Cu substrate reported earlier, the improved cyclic life was attributed to the highly porous three dimensional CNF substrate and also the unique Si spaghetti structure, in which both can accommodate Si volume expansion. However, the capacity of the whole film is relatively low compared to some

of the previous works (see Table 1). Nevertheless, the direct comparison is not applicable due to the previous results which took into account only the active material. In our case, the result is rather reasonable and more practical in terms of real battery application by avoiding the use of the metal current collector. Thus, metal corrosion-induced long term stability issues and overestimation of specific/volumetric capacity can be circumvented [3,4]. Unfortunately, a relatively higher resistance compared to metal current collector resulted from the existing fairly large amount of sp3 bonds in CNF is inevitable. (Nyquist plots of Si/CNF mat with different Si loading masses were shown in Fig. S5.) Conductivity of the CNF could be improved by incorporating carbon nanotubes within CNF [39]. For better understanding of the structure-related Li storage capacity, different Si amount was deposited on the CNF mat by varying the number of CV electrodeposition cycles. The capacity retention and rate performance of Si/CNF mats after 1000 C annealing upon the different Si deposition cycle numbers are summarized in Fig. 5c. Here, the specific capacity was expressed in terms of total mass of Si and carbon. The capacity of the pristine CNF and Si/CNF without heat treatment was plotted as a reference. It is clear to see that the

Please cite this article in press as: Yao F et al. A strategy of overcoming the limit of carbon-based materials for anode of lithium ion battery. Carbon (2014), http://dx.doi.org/10.1016/j.carbon.2014.08.017

8

CARBON

x x x ( 2 0 1 4 ) x x x –x x x

bonds are formed easily at the interface between Si and CNF after high temperature annealing (See Fig. S2). This is not only helpful for efficient charge transport at the interface (See Fig. S5) but also structure stabilization against the increased deformation energy. Nevertheless, the deformation energy is proportional to the Si thickness, while the interfacial cohesive energy of Si–C bonds remains constant. When the deformation energy exceeds interfacial cohesive energy at critical layer thickness of Si, the structure of outshell Si breaks down. Thus, as the Si thickness increases, more crack initiation and propagation are triggered. In other words, in the case of thicker Si layers, crack generation caused by volume variation is more significant [40]. Therefore, thicker Si layer peels off easily and eventually the capacity degradation occurs more severely than that of thinner Si layer. Furthermore, Si shape transformation was observed from spaghetti-like to granulelike when the CV deposition cycles keep increasing from 200 to 1500, as shown in Fig. S6a. The thin Si layer fabricated with 200 cycles CV deposition revealed a highly porous structure (See Fig. 3d and f), and therefore the volume expansion can be minimized compared to the thick granular shape Si layers. The structural analysis of Si-200-a/Si-1500-a after charge/discharge test can be found in Figs. S6b–d.

4.

Fig. 5 – (a) The 2nd and 10th cyclic voltammograms of CNF-a (solid line), Si-200-p (dashed line) and Si-200-a (symbol) mats between 0.01 and 2 V at a scan rate of 0.1 mV s1. (b) Voltage profiles of CNF-a/Si-200-a between 0.01 and 2 V at a charging rate of 0.1 C. The cycle numbers are indicated in the figure. (c) Charge (filled symbols)/discharge (open symbols) capacity in terms of different numbers of silicon deposition cycles. Sample indications and different charge rate (numbers) are shown in the figure. (A color version of this figure can be viewed online.) capacity retention became poorer as the Si loading amount increased. For example, the capacity of 1500 Si electrodeposition cycles was 986 mA h g1 at the 1st discharge and decreased to 610 mA h g1 at the 50th discharge. Compared to the capacity retention of 84% in Si-200-a, the value was clearly reduced to 62% in the case of Si-1500-a. This phenomenon can be explained as follows: Due to the existing sp3 bonds in CNF, as discussed in Figs. 3a and 4, Si–C

Conclusions

We have proposed a method to improve Li storage capacity of carbon materials by introducing Si layers. Silicon-coated carbon nanofiber mat was fabricated by combining electrospun carbon nanofiber mat with electrodeposited Si layer. The structure involves neither a metal substrate nor binders. This could be useful in effectively improving the specific/volumetric capacitance for the whole electrode and designing anode structures without metal corrosion issues in a cost-effective way. Thermal annealing of the combined mat at 1000 C was necessary to remove undesired residues formed during electrodeposition process and to induce strong Si–C bonds at the interface between Si layer and CNFs, which improve adhesion of Si to CNF and consequently facilitate efficient charge transport. This results in clear improvement of the capacity of carbon materials by almost three times for most cases in practical applications. Therefore, the free-standing CNF mat itself is a reasonable platform as an active material (Si layer) supporter. Optimization to improve composite structures for higher capacity, lower resistance, and longer cycle life is further required for industry applications.

Acknowledgements This work was supported by the Institute for Basic Science (IBS, EM1304) and in part by BK-Plus through Ministry of Education, Korea.

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.carbon. 2014.08.017.

Please cite this article in press as: Yao F et al. A strategy of overcoming the limit of carbon-based materials for anode of lithium ion battery. Carbon (2014), http://dx.doi.org/10.1016/j.carbon.2014.08.017

CARBON

xxx (2014) xxx–xxx

R E F E R E N C E S

[1] Tarascon JM, Armand M. Issues and challenges facing rechargeable lithium batteries. Nature 2001;414(6861):359–67. [2] Winter M, Besenhard JO, Spahr ME, Nova´k P. Insertion electrodematerials for rechargeable lithium batteries. Adv Mater 1998;10(10):725–63. [3] Wang Y, Liao H, Wang J, Qian X, Zhu Y, Cheng S. Effects of current collectors on electrochemical performance of FeS2 for Li-ion battery. Int J Elechtrochem Sci 2013;8(3):4002–9. [4] Gogotsi Y, Simon P. True performance metrics in electrochemical energy storage. Science 2011;334(6058):917–8. [5] Xuyen NT, Ra EJ, Geng HZ, Kim KK, An KH, Lee YH. Enhancement of conductivity by diameter control of polyimide-based electrospun carbon nanofibers. J Phys Chem B 2007;111(39):11350–3. [6] Hatchard TD, Dahn JR. In situ XRD and electrochemical study of the reaction of lithium with amorphous silicon. J Electrochem Soc 2004;151(6):A838–42. [7] Maranchi JP, Hepp AF, Evans AG, Nuhfer NT, Kumta PN. Interfacial properties of the a-Si/Cu: active–inactive thin-film anode system for lithium-ion batteries. J Electrochem Soc 2006;153(6):A1246–53. [8] Cui LF, Yang Y, Hsu CM, Cui Y. Carbon–silicon core–shell nanowires as high capacity electrode for lithium ion batteries. Nano Lett 2009;9(9):3370–4. [9] Simona GK, Maruyama B, Durstock MF, Burton DJ, Goswami T. Silicon-coated carbon nanofiber hierarchical nanostructures for improved lithium-ion battery anodes. J Power Sources 2011;196(23):10254–7. [10] Howe JY, Burton DJ, Qi Y, Meyer III HM, Nazri M, Nazri GA, et al. Improving microstructure of silicon/carbon nanofiber composites as a Li battery anode. J Power Sources 2013;221:455–61. [11] Chen PC, Xu J, Chen HT, Zhou CW. Hybrid silicon–carbon nanostructured composites as superior anodes for lithium ion batteries. Nano Res 2011;4(3):290–6. [12] Klankowski SA, Rojeski RA, Cruden BA, Liu JW, Wu J, Li J. A high-performance lithium-ion battery anode based on the core–shell heterostructure of silicon-coated vertically aligned carbon nanofibers. J Mater Chem A 2013;1(4):1055–64. [13] Ji LW, Jung KH, Medford AJ, Zhang XW. Electrospun polyacrylonitrile fibers with dispersed Si nanoparticles and their electrochemical behaviors after carbonization. J Mater Chem 2009;19(28):4992–7. [14] Ji LW, Zhang XW. Evaluation of Si/carbon composite nanofiber-based insertion anodes for new-generation rechargeable lithium-ion batteries. Energy Environ Sci 2010;3(1):124–9. [15] Hwang TH, Lee YM, Kong BS, Seo JS, Choi JW. Electrospun core–shell fibers for robust silicon nanoparticle-based lithium ion battery anodes. Nano Lett 2012;12(2):802–7. [16] Fan X, Zou L, Zheng YP, Kang FY, Shen WC. Electrospinning preparation of nanosilicon/disordered carbon composite as anode materials in Li-ion battery. Electrochem Solid-State Lett 2009;12(10):A199–201. [17] Macossay J, Leal JH, Kuang AX, Jones RE. Electrospun fibers from poly(methyl methacrylate)/vapor grown carbon nanofibers. Polym Adv Technol 2006;17(5):391–4. [18] Nicholson JP. Electrodeposition of silicon from nonaqueous solvents. J Electrochem Soc 2005;152(12):C795–802. [19] Casiraghi C, Pisana S, Novoselov KS, Geim AK, Ferrari AC. Raman fingerprint of charged impurities in graphene. Appl Phys Lett 2007;91(23):233108. [20] Yao F, Gu¨nes F, Ta HQ, Lee SM, Chae SJ, Sheem KY, et al. Diffusion mechanism of lithium ion through basal plane of layered graphene. J Am Chem Soc 2012;134(20):8646–54.

9

[21] Kim HJ, Seo MH, Park MH, Cho JP. A critical size of silicon nano-anodes for lithium rechargeable batteries. Angew Chem Int Ed 2010;49(12):2146–9. [22] Holzapfel M, Buqa H, Hardwick LJ, Hahn M, Wu¨rsig A, Scheifele W, et al. Nano silicon for lithium-ion batteries. Electrochim Acta 2006;52(3):973–8. [23] Kanellis G, Morhange JF, Balkanski M. Effect of dimensions on the vibrational frequencies of thin slabs of silicon. Phys Rev B 1980;21(4):1543–8. [24] Smith JE, Brodsky MH, Crowder BL, Nathan MI, Pinczuk A. Raman spectra of amorphous Si and related tetrahedrally bonded semiconductors. Phys Rev Lett 1971;26(11): 642–6. [25] Nakamoto K. Infrared and Raman spectra of inorganic and coordination compounds: Part A: theory and applications in inorganic chemistry. 6th ed. Hoboken NJ: John Wiley & Sons Inc; 2009. [26] Chen X, Gerasopoulos K, Guo J, Brown A, Wang C, Ghodssi R, et al. A patterned 3D silicon anode fabricated by electrodeposition on a virus-structured current collector. Adv Funct Mater 2011;21(2):380–7. [27] Contarini S, Howlett SP, Rizzo C, Angelis BAD. XPS study on the dispersion of carbon additives in silicon carbide powders. Appl Surf Sci 1991;51(3–4):177–83. [28] Rybachuk M, Bell JM. Electronic states of trans-polyacetylene, poly(p-phenylene vinylene) and sp-hybridised carbon species in amorphous hydrogenated carbon probed by resonant Raman scattering. Carbon 2009;47(10):2481–90. [29] Fourches N, Turban G, Grolleau B. Study of DLC/silicon interfaces by XPS and in-situ ellipsometry. Appl Surf Sci 1993;68(1):149–60. [30] Wilson AM, Dahn JR. Lithium insertion in carbons containing nanodispersed silicon. J Electrochem Soc 1995;142(2):326–32. [31] Wang W, Kumta PN. Reversible high capacity nanocomposite anodes of Si/C/SWNTs for rechargeable Li-ion batteries. J Power Sources 2007;172(2):650–8. [32] Bruce PG, Scrosati B, Tarascon JM. Nanomaterials for rechargeable lithium batteries. Angew Chem Int Ed 2008;47(16):2930–46. [33] Li CC, Yin XM, Chen LB, Li QH, Wang TH. Porous carbon nanofibers derived from conducting polymer: synthesis and application in lithium-ion batteries with high-rate capability. J Phys Chem C 2009;113(30):13438–42. [34] Striebel KA, Sierra A, Shim J, Wang CW, Sastry AM. The effect of compression on natural graphite anode performance and matrix conductivity. J Power Sources 2004;134(2):241–51. [35] Levi MD, Wang C, Gnanaraj JS, Aurbach D. Electrochemical behavior of graphite anode at elevated temperatures in organic carbonate solutions. J Power Sources 2003;119– 121:538–42. [36] Chen XL, Li XL, Ding F, Xu W, Xiao J, Cao YL, et al. Conductive rigid skeleton supported silicon as high-performance Li-ion battery anodes. Nano Lett 2012;12(8):4124–30. [37] Yao Y, McDowell MT, Ryu I, Wu H, Liu N, Hu LB, et al. Interconnected silicon hollow nanospheres for lithium-ion battery anodes with long cycle life. Nano Lett 2011;11(7):2949–54. [38] Yoo EJ, Kim J, Hosono E, Zhou HS, Kudo T, Honma I. Large reversible Li storage of graphene nanosheet families for use in rechargeable lithium ion batteries. Nano Lett 2008;8(8):2277–82. [39] Ra EJ, An KH, Kim KK, Jeong SY, Lee YH. Anisotropic electrical conductivity of MWCNT/PAN nanofiber paper. Chem Phys Lett 2005;413(1–3):188–93. [40] Wu H, Cui Y. Designing nanostructured Si anodes for high energy lithium ion batteries. Nano Today 2012;7(5): 414–29.

Please cite this article in press as: Yao F et al. A strategy of overcoming the limit of carbon-based materials for anode of lithium ion battery. Carbon (2014), http://dx.doi.org/10.1016/j.carbon.2014.08.017