In Situ TEM Investigation of Congruent Phase Transition and Structural

Mar 2, 2012 - Fu Sun , Henning Markötter , Ingo Manke , Andre Hilger , Nikolay ...... Xuejun Bai , Yueyang Yu , Harold H. Kung , Biao Wang , Jianming...
1 downloads 0 Views 2MB Size
Download PDF
Letter pubs.acs.org/NanoLett

In Situ TEM Investigation of Congruent Phase Transition and Structural Evolution of Nanostructured Silicon/Carbon Anode for Lithium Ion Batteries Chong-Min Wang,*,† Xiaolin Li,‡ Zhiguo Wang,‡ Wu Xu,§ Jun Liu,‡ Fei Gao,*,‡ Libor Kovarik,† Ji-Guang Zhang,§ Jane Howe,∥ David J. Burton,⊥ Zhongyi Liu,¶ Xingcheng Xiao,# Suntharampillai Thevuthasan,† and Donald R. Baer† †

Environmental Molecular Sciences Laboratory, ‡Fundamental and Computational Science Directorate, and §Energy and Environmental Directorate, Pacific Northwest National Laboratory, Richland, Washington 99352, United States ∥ Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6064, United States ⊥ Applied Sciences, Inc., Cedarville, Ohio 45014-0579, United States ¶ Electrochemical Energy Research Laboratory and #Chemical Sciences and Materials Systems Laboratory, General Motors Global R&D Center, Warren, Michigan 48090, United States S Supporting Information *

ABSTRACT: It is well-known that upon lithiation, both crystalline and amorphous Si transform to an armorphous LixSi phase, which subsequently crystallizes to a (Li, Si) crystalline compound, either Li15Si4 or Li22Si5. Presently, the detailed atomistic mechanism of this phase transformation and the degradation process in nanostructured Si are not fully understood. Here, we report the phase transformation characteristic and microstructural evolution of a specially designed amorphous silicon (a-Si) coated carbon nanofiber (CNF) composite during the charge/discharge process using in situ transmission electron microscopy and density function theory molecular dynamic calculation. We found the crystallization of Li15Si4 from amorphous LixSi is a spontaneous, congruent phase transition process without phase separation or large-scale atomic motion, which is drastically different from what is expected from a classic nucleation and growth process. The a-Si layer is strongly bonded to the CNF and no spallation or cracking is observed during the early stages of cyclic charge/discharge. Reversible volume expansion/contraction upon charge/discharge is fully accommodated along the radial direction. However, with progressive cycling, damage in the form of surface roughness was gradually accumulated on the coating layer, which is believed to be the mechanism for the eventual capacity fade of the composite anode during long-term charge/discharge cycling. KEYWORDS: Si-coated carbon nanofiber anode, Li-ion battery, in situ TEM, DFT-MD, congruent phase transition

S

does the structure evolve and can the volume changes during the cyclic charge/discharge of the battery be accommodated without degrading the overall structure? It is known that upon lithiation both crystalline and amorphous silicon will be transformed to an amorphous LixSi phase and subsequently a crystalline (Li, Si) compound, either Li15Si4 or Li22Si5.14,24−26 It has been observed that the crystallization from the amorphous phase occurs “quickly” based on recent in situ X-ray diffraction (XRD) and transmission electron microscopy (TEM) studies.14,24−26 However, the detailed atomistic mechanism of such phase transitions has not been fully understood. If the phase transformation happens according to the classic nucleation

ilicon, as a candidate anode material, shows several unique characteristics. It has a theoretical gravimetric capacity of ∼4200 mAh g−1 and a volumetric capacity of ∼8500 mAh cm−3. Upon lithiation, crystalline Si will expand more than 300% with dramatic anisotropic elongation along the [110] direction,1−4 leading to the pulverization of the materials. One approach to mitigate these volume change effects is to design and synthesize nanostructures to limit the extent of volume change. A range of nanostructured materials have been investigated, including Si nanowires, nanotubes, nanoparticles of both crystalline and amorphous structures,5−14 and silicon and carbon composite.8,15−23 Recent studies suggest that amorphous Si (a-Si) coated on carbon nanofibers (CNF) could be a promising anode candidate material.17 Two fundamental questions need to be addressed regarding this type of nanocomposite. (1) What are the phase transformations and structural changes during charge and discharge? (2) How © 2012 American Chemical Society

Received: December 27, 2011 Revised: February 13, 2012 Published: March 2, 2012 1624

dx.doi.org/10.1021/nl204559u | Nano Lett. 2012, 12, 1624−1632

Nano Letters

Letter

and growth model, it will lead to composition fluctuation and phase separation in the amorphous that could have serious implications for the stability of the material during charge/ discharge.25,26 In situ TEM observation provides the unique opportunity to reveal the fine details of such complex phase transformations. In this paper, we study hollow CNFs coated with amorphous Si on both the exterior and interior surfaces as an anode. The electrochemical and structure properties of this composite material were investigated with an emphasis on in situ TEM investigation of the structural evolution and phase transformation mechanisms. In addition, density function theory molecular dynamic (DFT-MD) calculations were conducted to understand the crystallization of the amorphous to crystalline phase in lithiated Si. We found the amorphous LixSi crystallizes to Li15Si4 through a spontaneous, congruent process without long distance displacement and diffusion of the atoms. This mechanism is drastically different from what is expected from the classic nucleation and growth process, which necessitates involvement of atomic diffusion, composition fluctuation, and phase separation. Additionally, the unique thin a-Si layer adheres strongly to the carbon nanofiber and shows no spallation or cracking during the early stages of cyclic charging/ discharging. Nevertheless, with progressive cycling, damage gradually accumulated on the a-Si layer, which is believed to be the mechanism for the eventual capacity fading. Amorphous silicon was deposited on hollow structured CNF (Pyrograf-III, PR-25-XT-PS and HHT, Pyrograf Products, Inc., Cedarville, OH) using a chemical vapor deposition (CVD) method with silane as the silicon source at Applied Science Inc. In our work, the silicon to carbon ratio was targeted to be 1:3 by weight, which is expected to provide a specific capacity of ∼1000 mAh/g electrode. The electrochemical properties of the a-Si-CNF were measured in a coin-cell configuration using EC/ DEC/10% FEC as the electrolyte and Li metal as the counter electrode. Microstructural evolutions in the a-Si-CNF composite during the charge/discharge cycle was studied in situ using a nanobattery configuration with a single a-Si-CNF as anode, LiCoO2 as cathode, and an ionic liquid-based electrolyte (illustrated in Figure 1), as previously reported.27−29 A few a-SiCNFs were attached to a gold rod with conductive silver epoxy and a film of LiCoO2 compact powder was attached to a gold rod, respectively serving as the anode and cathode in the nanobattery. One drop of the ionic liquid-based electrolyte (ILE, 10 wt % lithium bis-(trifluoromethylsulfonyl)imide (LiTFSI) dissolved in 1-butyl-1-methylpyrrolidinium bis(trifluoromethylsulfonyl) imide (P14TFSI) was placed on the surface of LiCoO2 film as the electrolyte. A constant potential of −4.0 V was applied to the a-Si-CNF against LiCoO2 upon charging (lithiation of Si) and 0 V upon discharging (delithiation of Si). All in situ electrochemical tests were conducted in a Titan 80-300 scanning transmission electron microscope (STEM) operated at 300 kV with a Nanofactory TEM scanning tunneling microscopy (STM) holder. The general microstructural features of the a-Si-CNF composite are illustrated in Figure 2. The carbon fibers had a hollow structure as shown by the TEM image in Figure 2a. These fibers feature an outer diameter of 100 to 200 nm and a hollow core typically about one-third to one-half of the fiber diameter. Detailed TEM bright-field and dark-field images revealed the fine structural features indicating the crystalline quality of the carbon fiber. These images are presented in

Figure 1. Experimental setup for the in situ TEM testing of the nanobattery. (a) Schematic drawing illustrates the electrical device with a single Si-CNF as the anode, ionic liquid based electrolyte, and a piece of LiCoO2 as the cathode. (b) TEM image shows that the working section of the nanobattery with the Si-CNF immersed into the liquid electrolyte and is against the LiCoO2 cathode. Applying a bias will enable the charging/discharging of this single nanowire battery.

Figure 2b,c. The images clearly demonstrate that the inner wall of the carbon fiber retains graphitic structure, while the outer wall of the carbon fiber shows typical turbostratic carbon structure.30 High-resolution TEM images reveal that the coated silicon layer on CNF was amorphous and had a typical thickness of ∼13 nm as representatively shown in Figure 2d. The coated a-Si layer was further characterized by scanning transmission electron microscopy (STEM) in the high-angle annular dark field (HAADF) imaging mode for which the image contrast is proportional to mass thickness and the square of the atomic number of the element. Therefore, the a-Si shows high brightness in comparison to CNF as illustrated in Figure 2e. This was further confirmed by the Si distribution profile obtained using energy dispersive X-ray spectroscopy (EDS) line scanning analysis as shown in Figure 2f. The STEM-HAADF image and EDS analysis revealed that the outside surface of all CNFs were indeed uniformly coated with an a-Si layer, while the thickness of the a-Si coating layer on the inside wall shows variations from fiber to fiber as it may depend on the end structure of the fiber (compare Figure 2 panels e and f with Figure S1 in the Supporting Information). During the in situ TEM observation, the structural evolution of the a-Si layer on the inside wall of the CNF could not be clearly imaged. Therefore, in this paper we concentrate on the microstructural evolution of the a-Si layer at the outside wall of the CNF. The capacity of this particular a-Si-CNF structure as measured in a half cell configuration was ∼1000 mAh/g at a current density of 1 A/g as illustrated in Figure 3. The structural evolution of the a-Si layer during the charging is illustrated in Figure 4, demonstrating the following characteristics. (1) The a-Si layer is lithiated to form amorphous LixSi (hereafter designated as a-LixSi). (2) The lithiation progresses in a sequential manner, featuring the propagation of the reaction front along the Li+ diffusion direction. This is observed for other materials when tested in a similar configuration as illustrated in Figure 4a−d27,28 and in the corresponding movies, S1 and S2 in Supporting 1625

dx.doi.org/10.1021/nl204559u | Nano Lett. 2012, 12, 1624−1632

Nano Letters

Letter

Figure 3. Specific capacity as a function of cycle number tested in a half cell configuration using Li metal as the counter-electrode and EC/ DEC/10% FEC as the electrolyte.

rate is attributed to the good electrical conductivity of the underline carbon in the case of a-Si-CNF. (3) Accompanying the conversion of a-Si to a-LixSi is a significant volume expansion, which is accommodated by the swelling along the radial direction (Figure 4a−d and movies S1 and S2 in the Supporting Information). (4) Continued charging leads to increases in Li concentration in the a-LixSi. When x reaches a critical value, the a-LixSi crystallized to Li15Si4, as clearly identified by the electron diffraction analysis shown in Figure 4e,f (also see the Movie S2 in the Supporting Information). It should be pointed out that Ghassemi et al.14 studied the lithiation behavior of amorphous Si nanorode and reported the formation of crystalline Li22Si5 from a-LixSi. On the basis of electron diffraction analysis, we did not notice the formation of crystalline Li22Si5 in the present case. The crystallization of aLixSi to c-Li15Si4 observed here with the a-Si on a CNF support as a starting material has also been observed for crystalline Si as the starting material based on both in situ TEM and X-ray diffraction analysis.24−26 Therefore, in terms of lithiationinduced formation of a-LixSi and subsequent crystallization of a-LixSi to c-Li15Si4, there are no significant differences between amorphous and crystalline silicon starting materials. One important difference is that the volume expansion for crystalline Si to a-LixSi is strongly anisotropic while the volume expansion for amorphous Si to a-LixSi is isotropic. The formation of c-Li15Si4 from the a-LixSi is very unique, which is characterized by the rapid transformation of the amorphous a-LixSi to crystalline c-Li15Si4 when the x reaches a critical value. It has been observed that following the crystallization of c-Li15Si4 from the a-LixSi, neither phase segregation happens nor residual phase was formed, indicating that the critical value of x for this phase transformation is x = 3.75. Kinetically, we have noticed that the crystallization occurs through the fast propagation of the interface between the crystallized region and the amorphous region (crystallization front sweeping) along the Li+ diffusion direction. The measured crystallization front propagation speed is ∼128 nm/s, which is comparable with the measured lithiation front propagation speed of ∼130 nm/s. The observation of the crystallization happening through a propagation of the crystallization front along the Li+ diffusion direction indicates that the Li enrichment for the crystallization is achieved through the electrochemical driving Li+ diffusion, which happens sequentially along the Li+ diffusion direction. In the case of a classic nucleation and growth process (either heterogeneous or

Figure 2. Microstructural features of the Si-CNF composite. (a) Bright-field TEM shows the general microstructure of the Si-CNF. The Si coating layer is not visible at this magnification. The inset in (a) is the selected area electron diffraction pattern, revealing amorphous structure of the Si-CNF. (b) Bright-field and (c) dark-field TEM images reveal the fine structural features of the CNF. Note the strong contrast of the inner section of the CNF, indicating graphitic structure of the inner wall. The inset in (b) is the EDS spectrum showing the existence of Si. (d) HRTEM image showing the amorphous structure of the Si layer at the outside wall of the CNF. Note the turbostatic structural nature of the outside layer of CNF. (e) STEM-HAADF image showing the spatial distribution of the Si layer. (f) EDS line scanning profile showing the distribution of Si. Both STEM-HAADF imaging and EDS line scanning indicate that for this a-Si coated CNF, the Si coated on the inside wall is much thinner than on the outside wall.

Information. The lithiation front propagates at a speed of ∼130 nm/s, which is comparable with the lithiation rate of 117 nm/s for a carbon-coated single crystalline Si, and is ∼60 times larger than that of intrinsic uncoated crystalline Si.24 The fast charging 1626

dx.doi.org/10.1021/nl204559u | Nano Lett. 2012, 12, 1624−1632

Nano Letters

Letter

Figure 4. Microstructural evolution of the Si-CNF during the charging. The left column is the TEM images, the middle column is a schematic drawing to guide the eye for reading the corresponding TEM image, and the right column is the selected area electron diffraction pattern. (a) TEM image showing the Si-CNF before the charge. (b−f) The time-lapse TEM images showing the structural evolution of Si-CNF. The lithiation is featured by the formation of amorphous a-LixSi phase, which is evidenced by the thickening of a surface layer. The propagation of the lithiation front is marked by the red arrows in (b) and (c). With continued lithiation, a-LixSi crystallizes to a crystalline phase as evidenced by the diffraction contrast shown in (e) and (f). Electron diffraction analysis clearly indicates that the crystallized phase is Li15Si4. To illustrate the difference between the amorphous a-LixSi and possible electron beam induced solidification of the liquid electrolyte, (g) shows the flow of the liquid electrolyte on another fiber surface, leading to the formation of amorphous layer. However, this amorphous layer is featured by a bead shape and shows variation on the thickness and will not be subject to phase transition. The red marks indicate the variation of the thickness of the electrolyte layer.

homogeneous nucleation),31 fluctuation of local chemical composition through diffusion will lead to a random spatial distribution of the new phases along the CNF and the formation of a second phase as in the case of lithiation-induced phase transformation of SnO2.29 Therefore, the formation of cLi15Si4 from the a-LixSi is a congruent process, which is

characterized by a pure phase transformation of amorphous to crystalline structure of the same chemical composition. This phase transformation involves neither a long-range diffusion of the chemical species, nor any local chemical composition fluctuations; rather it is a local rearrangement of atomic configuration.32,33 The Li enrichment for the formation of the 1627

dx.doi.org/10.1021/nl204559u | Nano Lett. 2012, 12, 1624−1632

Nano Letters

Letter

Figure 5. (a) Calculated formation energy of both crystalline and amorphous LixSi as a function of x. (b) The driving force for the crystallization of amorphous LixSi to crystalline phase as a function of x. (c) Atomic structure model of the amorphous structure Li15Si4. (d) Atomic structure model of crystalline Li15Si4.

the amorphous phase may crystallize in proper conditions. It is evident that crystalline Li15Si4 and Li21Si5 have the lowest formation energy. The driving force for crystallization is the difference of Gibbs free energy between the amorphous and crystalline phase. Since ΔG = Gcrystalline − Gamorphous = ΔE + PΔS − TΔS where ΔE, the change of internal energy, is of the order of 0.3−1.0 eV/LixSi; at 0 K with pressure equal to 0 GPa, we can make the approximation of ΔG ≈ ΔE. Figure 5b shows the ΔG as a function of x. Figure 5b indicates that the largest driving force, ΔG, corresponds to the formation of crystalline Li15Si4 from the amorphous phase, theoretically validating the reason why the crystalline Li115Si4, rather than other (Si, Li) compounds form congruently from the amorphous LixSi (x = 3.75). One of the major questions related to the performance of the Si-based anode materials is how the microstructure evolves with the progression of the cyclic charging and discharging. Microstructural evolution of the Si layer during the cyclic charging/discharging is shown in Figure 6. During discharging (delithiation), the c-Li15Si4 first transforms to a-LixSi and eventually to a-Si. Therefore, the overall phase transformation can be written as a-Si → a-LixSi → c-Li15Si4 for charging (lithiation), and a-LixSi/c-Li15Si4 → a-LixSi → a-Si for discharging (delithiation). The phase transformation of a-Si ↔ a-LixSi is featured by a volume change of ∼310% as determined in this study, which is approximately equal to the volume change of crystalline Si when it is lithiated to aLixSi.6,23,39−42 No significant volume change has been observed accompanying the crystallization of a-LixSi to crystalline c-

crystalline Li15Si4 from the amorphous LixSi is controlled by the electrochemically driven directional Li diffusion rather than Li random walk fluctuation. This crystallization process can be termed as “electrochemical driving congruent crystallization,” which is the reverse process of electrochemical driving solidstate amorphization,34 and is drastically different from the classic nucleation and growth process for which local chemical composition fluctuation and phase separation occurs. It should be pointed out that although formation of c-Li15Si4 from a-LixSi has been reported in the literature,2,26 the congruent nature associated with this phase transformation has never been previously realized. It is known that for crystalline Si and Ge, lithiation-induced a-LixM (M = Si and Ge) has a large solubility range for lithium, and at higher x the a-LixM shows similar chemical properties to the crystalline Li15M4.2,24,26,35−38 The aforementioned experimental observation is supported by the DFT-MD calculations of the formation energy of LixSi in both amorphous and crystalline forms as illustrated in Figure 5 (the detailed calculation method is described in Supporting Information). The formation energy is calculated as Ef(x) = ELixSi − xELi − ESi where x is the number of Li atoms per Si atom and Ef is the formation energy. ELixSi is the total energy of the LinSim (such as Li15Si4) structure divided by the number of Si atoms, and ELi and ESi are the energies of Li atoms in a bodycentered cubic Li and Si atom in a diamond structure. Figure 5a shows the formation energy of amorphous and crystalline LixSi as a function of x. The formation energies of amorphous alloys are higher than that of the crystalline ones, which indicates that 1628

dx.doi.org/10.1021/nl204559u | Nano Lett. 2012, 12, 1624−1632

Nano Letters

Letter

Figure 6. TEM images showing the structural evolution of the Si-CNF during the cyclic charging and discharging. On (a−d), the left column is charged and the right column is discharged. Note that with cyclic charging and discharging, the surface of the coating layer is gradually crumpling (compare with Figure 4a−d). The two particles shown in this CNF are pre-existing particles. With a very limited number of in situ cycling in the TEM column, the change of the coating layer is not significant. However, as a general trend, we noticed that even with a limited number of cyclic charging/discharging, the surface indeed begins to become rough with the progress of the cycling. To illustrate this point, we magnified the images (a) 1st charged and (d) 4th charged and shown as (e) and (f), respectively. Note the slightly increased surface roughness in (f) as compared with (e).

while it changes to tensile upon delithation. Three factors may be suggested to account for the observation that lithiation/ delithiation-induced volume changes are fully accommodated along the radial direction: (1) Good adhesion between the a-Si coating layer and the CNF (as discussed in the subsequent paragraph). (2) The elastic modulus of CNF is about 10 times of LixSi.43 Therefore, the CNF imposes a strong mechanical

Li15Si4. It has been observed that the volume change during the charge/discharge is primarily accommodated in the radial direction as clearly revealed by the TEM images shown in Figures 4 and 6 and the movies S1 and S2 in the Supporting Information. During the lithiation/delithiation, both axial and radial stress, as well as hoops stress, is produced in the LixSi.17 During the lithiation, the axial stress in the LixSi is compressive, 1629

dx.doi.org/10.1021/nl204559u | Nano Lett. 2012, 12, 1624−1632

Nano Letters

Letter

charge/discharge. These results support the view that the gradual capacity fading of the battery using this material is partially related to the gradual accumulation of damage of the coating layer following each cyclic charge/discharge. In addition, the a-Si coating adheres well to carbon fiber and the adhesion between the a-Si layer and the carbon fiber is demonstrated in the following in situ work as described below. In the present in situ experimental setup using a single nanowire, the surface of the nanowire was wetted by a thin layer of the liquid electrolyte.27−29 During imaging, this layer of electrolyte provided the fast channel for the diffusion of Li+ along the surface. When we carried out the in situ experiment, we generally kept the electron beam spread to avoid the gelation of the wetting layer of electrolyte. However, with a slightly increased electron dose, this wetting layer of electrolyte becomes gelled, forming an amorphous shell and covering the a-LixSi as illustrated in Figure 7 with the gelled electrolyte layer thickness of ∼25 nm (also see the movie S3 in the Supporting Information).Because of the amorphous structure of both the aLixSi and the gelled electrolyte, the boundary between these two layers is marginally visible. Following the crystallization of a-LixSi to c-Li15Si4 as shown in Figure 7c, the gelled electrolyte remains as amorphous and is distinguishable from the c-Li15Si4. During discharging, the c-Li15Si4 ↔ a-LixSi ↔ a-Si is accompanied by a shrinkage along the radial direction. During this transition we were surprised to notice that a crack forms between the gelled electrolyte layer and the a-LixSi and the crack propagated against the Li+ diffusion direction as indicated by the red arrows in Figure 7d−f and the movie S4 in the Supporting Information. The crack propagates with a speed of 37 nm/s, which is directly proportional to the Li+ diffusivity during the discharging. Remarkably, there is no separation between the a-LixSi and the carbon fiber, demonstrating that the adhesion between a-LixSi and the carbon fiber is stronger than that between a-LixSi and the gelled electrolyte. The recharging behavior of this discharged Si-CNF is shown in Figure 7g−j and the movie S5 in the Supporting Information. Two features can be seen during the recharging. (1) The a-Si ↔ a-LixSi showed uniform radial swelling and no reaction front propagation is noticed as demonstrated by the movie S5 in Supporting Information. (2) The a-LixSi eventually crystallized to Li15Si4 and filled up the gap formed during the discharging. This observation firmly demonstrates that the defect generated during the delithiation is fully recovered on the free surface. Composite structures based on amorphous Si coated on both outside and inside of CNF walls have emerged as a promising microstructural designing concept for tailoring structures that take advantage of the high capacity of Si and reliable cycle life for Li ion battery. The in situ TEM study of one such composite structure indicates that the a-Si layer coated on the outside wall of the CNF by the CVD method is very well adhered to the CNF surface. During the cyclic charge/ discharge, the a-Si undergoes the following phase transformation process: a-Si → a-LixSi → c-Li15Si4 for charging, and a-LixSi/c-Li15Si4 → a-LixSi → a-Si for discharging. Crystallization of a-LixSi to Li15Si4 is in nature a congruent process, which does not involve large-scale atomic motion and phase separation. During the cyclic charge/discharge, the composite structured a-Si-CNF shows no large-scale morphological changes, such as distortion or bending, which is in marked contrast with the case of a monolithic single crystal that is featured by distortion and significant anisotropic shape accommodation upon lithiation. Instead, the thin coating layer

constraint on the axial deformation of the LixSi. (3) The a-Si coating layer is very thin (13 nm) as compared to the length of the CNF (several micrometers), which corresponds to a plane strain configuration in the term of continuum mechanics. It should be realized that during the lithiation of Si, the CNF is also lithiated and contributes to the observed structural changes. Lithiation of graphite leads to a basal plane space changing from 0.34 to 0.36 nm,44 which corresponds to a 5.9% radial expansion, much less compared with a volume change of ∼300% for the case of lithiation of Si. Therefore, the structural evolution we observed is dominated by the lithiation of Si. Microscopically, we found that the adhesion between the coated a-Si layer and the CNF was very strong as evidenced by the TEM images shown in Figure 6 with cyclic charging/ discharging and the movies S3, S4, and S5 in the Supporting Information. With each cyclic charge/discharge, the a-Si layer went through the phase transformation as described previously and showed uniform expand/shrink along the radial direction. No cracking or peeling off of the a-Si layer from the carbon fiber was noticed during the initial cyclic charge/discharge of a single a-Si-CNF. However, with the cyclic charge/discharge, the surface of the coating layer gradually became rough, indicating an accumulation of defects within and on the surface of the coating layer. It is known that for bulk crystalline Si and Ge, the charge/discharge cycling will lead to the formation of vacancy clusters, nanopores, and cracks.2,17,40,45 Formation of vacancy clusters and nanopores is the consequence of clustering of vacancies formed by the delithiation as likewise occurs during the deallowing process.46 For the present composite structure, the coating layer only has a thickness of ∼13 nm, which allows the effective escape of many vacancies to the free surface, minimizing the formation of macroscopic voids or cracks. This accounts for the fact that we do not observe any cavity or nanopores in the coating layer following the cyclic charge and discharge. Instead, we only notice gradual accumulation of surface unevenness. Three reasons may be suggested to account for the robustness of the coating layer on the carbon fiber. (1) The coating layer only has a thickness of ∼13 nm, which is thin enough to accommodate the volume changes. There exists a critical value for the thickness of the coating layer above which the volume change induced stress cannot be fully relaxed, possibly resulting in the formation of cracks. The critical thickness of the coating layer has been estimated to be ∼50 nm by Hu et al.17 For a coating layer thickness of 30 nm, Hu et al. noticed the formation of the nanopores, which is also related to the stress distribution within the coating layer.17,47 (2) The amorphous structure of the Si, which corresponds to the isotropic deformation of the coating layer, can be somewhat directed and constrained by the presence of the CNF core. (3) The rigid confinement of the C fiber. Hu et al. concluded that the volume change accompanying the a-Si ↔ a-LixSi transformation was characterized by inelastic flow along the radial direction.17 The total number of cycles that can be accommodated during in situ testing configuration within the TEM column is limited and longer-term tests were conducted with a button cell. It can be difficult to directly correlate what we observed in situ on the structural evolution for a few cycles with the gradual capacity fading of up to 100 cycles of the battery. For the same material used in the present work, We conducted postmodem analysis of the material microstructure both in half- and full-cell configuration and noted the coating layer appeared to be segmented on the carbon fiber surface following the 100 cyclic 1630

dx.doi.org/10.1021/nl204559u | Nano Lett. 2012, 12, 1624−1632

Nano Letters

Letter

possibly the diameter of the CNFs as well as the bonding between the coating layer and the CNFs can become design parameters to optimize the lifetime of the composite anode based on Si and carbon.

■

ASSOCIATED CONTENT

S Supporting Information *

Additional information, figures, and movies (movie 1, si_002.avi; movie 2, si_003.avi; movie 3, si_004.avi; movie 4, si_005.avi; movie 5, si_006.avi.) This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: (C.-M.W.) [email protected]; (F.G.) Fei. [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by the Laboratory Directed Research and Development (LDRD) program of the Pacific Northwest National Laboratory (PNNL). The work was conducted in the William R. Wiley Environmental Molecular Sciences Laboratory (EMSL), a national scientific user facility sponsored by DOE’s Office of Biological and Environmental Research and located at PNNL. PNNL is operated by Battelle for the DOE under Contract DE-AC05-76RLO1830. The work at Oak Ridge National Laboratory, managed by U.T. Battelle, LLC, for the U.S. Department of Energy under contract DE-AC0500OR22725 was sponsored by the Vehicle Technologies program for the Office of Energy Efficiency and Renewable Energy.

■

Figure 7. TEM images showing the wetting and electron beaminduced solidification of the electrolyte layer (EL) on the Si-CNF surface during the charging and discharging. The insets in a, b, c, f, g, and j show the selected area diffraction pattern from Si-CNF. The drawing on the top of the images (a−c) is a schematic to guide the eye for reading the images. (a) Microstructure of the Si-CNF before charging. (b) Charging induced the formation of amorphous a-LixSi. (c) Continued enrichment of Li in a-LixSi leads to the instantaneous transformation of a-LixSi to crystalline Li15Si4, while the EL remains amorphous. (d−f) The microstructural evolution during the discharging of the nanowire shown in (c). Note the time scale is restarted from 0 s for the image (d). The discharging is featured by transformation of Li15Si4 to a-LixSi and continued shrinkage of the LixSi with the decreasing x. Note the formation of crack (indicated by the arrows) between the EL and the a-LixSi, indicating that the bonding between the a-LixSi and C wall is stronger than that between the EL and the a-LixSi. (g−j) shows the microstructure evolution during the recharging of the nanowire in (f). Note the filling of the gap between the EL and the a-LixSi, and eventual formation of crystalline Li15Si4. Note the continued thickening of the EL. Image (j) is captured further along the Li diffusion direction and this accounts for the thinner gelled electrolyte thickness in (j) than in (i).

REFERENCES

(1) Lee, S. W.; McDowell, M. T.; Choi, J. W.; Cui, Y. Nano Lett 2011, 11, 3034. (2) Liu, X. H.; Zheng, H.; Zhong, L.; Huang, S.; Karki, K.; Zhang, L. Q.; Liu, Y.; Kushima, A.; Liang, W. T.; Wang, J. W.; Cho, J.-H.; Epstein, E.; Dayeh, S. A.; Picraux, S. T.; Zhu, T.; Li, J.; Sullivan, J. P.; Cumings, J.; Wang, C.; Mao, S. X.; Ye, Z. Z.; Zhang, S.; Huang, J. Y. Nano Lett 2011, 11, 3312. (3) Boukamp, B. A.; Lesh, G. C.; Huggins, R. A. J. Electrochem. Soc. 1981, 128, 725. (4) Larcher, D.; Beattie, S.; Morcrette, M.; Edstroem, K.; Jumas, J. C.; Tarascon, J. M. J. Mater. Chem. 2007, 17, 3759. (5) Chan, C. K.; Peng, H.; Liu, G.; McIlwrath, K.; Zhang, X. F.; Huggins, R. A.; Cui, Y. Nat. Nanotechnol. 2007, 3, 31. (6) Park, M. H.; Kim, M. G.; Joo, J.; Kim, K.; Kim, J.; Ahn, S.; Cui, Y.; Cho, J. Nano Lett 2009, 9, 3844. (7) Song, T.; Xia, J. L.; Lee, J. H.; Lee, D. H.; Kwon, M. S.; Choi, J. M.; Wu, J.; Doo, S. K.; Chang, H.; Park, W. I.; Zang, D. S.; Kim, H.; Huang, Y. G.; Hwang, K. C.; Rogers, J. A.; Paik, U. Nano Lett. 2010, 10, 1710. (8) Cui, L. F. H.; Hu, L. B.; J., L.; Choi, J. W.; Cui, Y. ACS Nano 2010, 4, 3671. (9) Cho, J. J. Mater. Chem. 2010, 20, 4009. (10) Kim, H.; Han, B.; Choo, J.; Cho, J. Angew. Chem., Int. Ed. 2008, 47, 10151. (11) Zhang, J. G.; Liu, J.; Wang, D. H.; Choi, D.; Fifield, L. S.; Wang, C. M.; Xia, G.; Nie, Z. M.; Yang, Z. G.; Pederson, L. R.; Graff, G. J. Power Sources 2010, 195, 1691. (12) Magasinski, A.; Dixon, P.; Hertzberg, B.; Kvit, A.; Ayala, J.; Yushin, G. Nat. Mater. 2010, 9, 353. (13) Zhou, S.; Liu, X. H.; Wang, D. W. Nano Lett. 2010, 10, 860.

shows instantaneous swelling/shrinking only along the radial direction during the charging and discharging. Associated with the overall thin nature of the coating layer (∼ 13 nm) at the outside surface of the CNF, no nanopores and crack develop within the coating layer, which attributes to the observed slow capacity fading of this material when tested in the half cell. It can be generally realized that the thickness of coatings and 1631

dx.doi.org/10.1021/nl204559u | Nano Lett. 2012, 12, 1624−1632

Nano Letters

Letter

(14) Ghassemi, H.; Au, M.; Chen, N.; Heiden, P. A.; Yassar, R. S. ACS Nano 2011, 5, 7805. (15) Cui, L. F.; Yang, Y.; Hsu, C. M.; Cui, Y. Nano Lett 2009, 9, 3370. (16) Wang, W.; Kumta, P. N. ACS Nano 2010, 4, 2233. (17) Hu, L.; Wu, H.; Gao, Y.; Cao, A.; Li, H.; McDough, J.; Xie, X.; Zhou, M.; Cui, Y. Adv. Energy Mater. 2011, 1, 523. (18) Ng, S. H.; Wang, J.; Wexler, D.; Chew, S. Y.; Liu, H. K. J. Phys. Chem. C 2007, 111, 11131. (19) Wang, W.; Kumta, P. N. Acs Nano 2010, 4, 2233. (20) Yang, J.; Wang, B. F.; Wang, K.; Liu, Y.; Xie, J. Y.; Wen, Z. S. Electrochem. Solid State 2003, 6, A154. (21) Xue, J. S.; Myrtle, K.; Dahn, J. R. J. Electrochem. Soc. 1995, 142, 2927. (22) Jang, S. M.; Miyawaki, J.; Tsuji, M.; Mochida, I.; Yoon, S. H. Carbon 2009, 47, 3383. (23) Cui, L. F.; Ruffo, R.; Chan, C. K.; Peng, H. L.; Cui, Y. Nano Lett 2009, 9, 491. (24) Liu, X. H.; Zhang, L. Q.; Zhong, L.; Liu, Y.; Zheng, H.; Wang, J. W.; Cho, J. H.; Dayeh, S. A.; Picraux, S. T.; Sullivan, J. P.; Mao, S. X.; Ye, Z. Z.; Huang, J. Y. Nano Lett 2011, 11, 2251. (25) Li, J.; Dahn, J. R. J. Electrochem. Soc. 2007, 154, A156. (26) Obrovac, M. N.; Christensen, L. Electrochem. Solid-State Lett. 2004, 7, A93. (27) Huang, J. Y.; Zhong, L.; Wang, C. M.; Sullivan, J. P.; Xu, W.; Zhang, L. Q.; Mao, S. X.; Hudak, N.; Liu, X. H.; Subramanian; Fan, H. Y.; Qi, L.; Kushima, A.; Li, J. Science 2010, 330, 1515. (28) Wang, C. M.; Xu, W.; J., L.; Choi, D. W.; Arey, B.; Saraf, L. V.; Zhang, J. G.; Yang, Z. G.; Thevuthasan, S.; Baer, D. R.; Salmon, N. J. Mater. Res. 2010, 25, 1541. (29) Wang, C.-M.; Xu, W.; Liu, J.; Zhang, J.-G.; Saraf, L. V.; Arey, B. W.; Choi, D.; Yang, Z.-G.; Xiao, J.; Thevuthasan, S.; Baer, D. R. Nano Lett 2011, 11, 1874. (30) Tibbetts, G. G.; Lake, M. L.; Strong, K. L.; Rice, B. P. Compos. Sci. Technol. 2007, 67, 10. (31) Porter, D. A.; Easterling, K. E. Phase Transformations in Metals and Alloys, 2nd ed.; CRC Press: Cheltenham, UK, 1992. (32) Zubarev, E. N. Physics-Usp. 2011, 54, 473. (33) Gonzalez-Hernandez, J.; Chao, B. S.; Strand, D.; Ovshinsky, S. R.; Pawlik, D.; Gasionrowski, P. Appl. Phy. Comm. 1992, 11, 557. (34) Limthongkul, P.; Jang, Y.; Nancy, J. D.; Chiang, Y., M Acta Mater. 2003, 51, 1103. (35) Gladyshevskii, E. I.; Kripyakevich, P. I. Sov. Phys. Crystallogr. 1960, 5, 549. (36) Johnson, Q.; Smith, G. S.; Wood, D. Acta Crystallogr. 1965, 18, 131. (37) Hatchard, T. D.; Dahn, J. R. J. Electrochem. Soc. 2004, 151, A838. (38) Obrovac, M. N.; Krause, L. J. J. Electrochem. Soc. 2007, 154, A103. (39) Arico, A. S.; Bruce, P.; Scrosati, B.; Tarascon, J. M.; Van Schalkwijk, W. Nat. Mater. 2005, 4, 366. (40) Chan, C. K.; Peng, H. L.; Liu, G.; McIlwrath, K.; Zhang, X. F.; Huggins, R. A.; Cui, Y. Nat. Nanotechnol. 2008, 3, 31. (41) Beaulieu, L. Y.; Eberman, K. W.; Turner, R. L.; Krause, L. J.; Dahn, J. R. Electrochem Solid State 2001, 4, A137. (42) Zhang, Q.; Zhang, W.; Wan, W.; Cui, Y.; Wang, E. Nano Lett. 2010, 10, 3243. (43) Sdemczyk, B. G.; Wang, Y. M.; Cumings, J.; Hetman, M.; Han, W.; Zettl, A.; Ritchie, R. O. Mater. Sci. Eng., A 2002, 334, 173. (44) Liu, Y.; Zheng, H.; Liu, X. H.; Huang, S.; Zhu, T.; Wang, J. W.; Kushima, A.; Hudak, N. S.; Huang, X.; Zhang, S. L.; Mao, S. X.; Qian, X. F.; Li, J.; Huang, J. Y. ACS Nano 2011, 5, 7245. (45) Liu, X. H.; Huang, S.; Picraux, S. T.; Li, J.; Zhu, T.; Huang, J. Y. Nano Lett 2011, 11, 3991. (46) Erlebacher, J.; Aziz, M. J.; Karma, A.; Dimmitrov, N.; Sieradzki, K. Nature 2001, 410, 450. (47) McDowell, M. T.; Lee, S. W.; Ryu, I.; Wu, H.; Nix, W. D.; Choi, J. W.; Cui, Y. Nano Lett 2011, 11, 4018.

1632

dx.doi.org/10.1021/nl204559u | Nano Lett. 2012, 12, 1624−1632