lable at ScienceDirect

Journal of Power Sources 276 (2015) 73e79

Contents lists avai

Journal of Power Sources

journal homepage: www.elsevier .com/locate/ jpowsour

Anomalous capacity increase at high-rates in lithium-ion batteryanodes based on silicon-coated vertically aligned carbon nanofibers

Steven A. Klankowski a, Gaind P. Pandey a, Brett A. Cruden b, Jianwei Liu c, Judy Wu c,Ronald A. Rojeski d, Jun Li a, *

a Department of Chemistry, Kansas State University, Manhattan, KS 66506, USAb NASA Ames Center for Nanotechnology, Moffett Field, CA 94035, USAc Department of Physics and Astronomy, The University of Kansas, Lawrence, KS 66045, USAd Catalyst Power Technologies, 200 Carlyn Avenue, Suite C, Campbell, CA 95008, USA

h i g h l i g h t s

* Corresponding author.E-mail address: junli@ksu.edu (J. Li).

http://dx.doi.org/10.1016/j.jpowsour.2014.11.0940378-7753/© 2014 Elsevier B.V. All rights reserved.

g r a p h i c a l a b s t r a c t

� Si-coated VACNFs Li-ion batteryanode is characterized at high cur-rent rates.

� The high-power capability and longcycling efficiency of nano-hybridelectrode is discussed.

� A new phenomenon of increasingcapacity at increasing current at highrates is presented.

a r t i c l e i n f o

Article history:Received 28 July 2014Received in revised form28 October 2014Accepted 21 November 2014Available online 22 November 2014

Keywords:Li-ion battery anodesHigh chargeedischarge ratesElectrochemical impedance spectroscopySilicon-coated vertically aligned carbonnanofibers

a b s t r a c t

This study reports of a multi-scale hierarchical lithium-ion battery (LIB) anode that shows a surprisingincrease in storage capacity at higher current rates from ~3C to ~8C. The anode, composed of forest-likevertically aligned carbon nanofibers coaxially coated with Si shells, is shown to obtain a storage capacityof 3000e3500 mAh (gSi)�1 and greater than 99% coulombic efficiency at a 1C (or C/1) rate, leading toremarkable stability over 500 chargeedischarge cycles. In contrast to other studies, this hierarchical LIBanode shows superior high-rate capability where the capacity decreased by less than 7% from ~C/8 to ~3Crates and, more importantly, increased by a few percent from ~3C to ~8C rates, displaying a new phe-nomenon that becomes more evident after going through long cycles. Electron microscopy, Raman, andelectrochemical impedance spectroscopy reveal that the electrode structure remains stable during longcycling and that this enhanced property is likely associated with the combination of the unique nano-columnar microstructure of the Si coating and the vertical coreeshell architecture. It reveals an excitingpotential to develop high-performance lithium-ion batteries.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

Lithium-ion batteries (LIBs) have become the most importantelectrical energy storage (EES) systems for today's portable elec-tronics due to their high specific energy densities. Furtherincreasing energy density and, more importantly, power density, by

Fig. 1. A SEM image (a) showing a landscape view of the uniform coaxial coating of Sishells on the random CNF array. A TEM image (b) and a schematic (c) that illustrate thenanocolumnar microstructure of Si shell and the conical graphitic stacking structure ofthe VACNF core. Sputtering deposition of Si created the nanocolumnar microstructureextending out from the CNF shaft (highlighted by white dotted lines) at an upwardoblique angle (highlighted by orange dashes) while the Si at the CNF tip showing asolid amorphous feature. A nickel catalyst particle (outlined by the blue dotted line) isshown at the CNF tip. (For interpretation of the references to colour in this figurelegend, the reader is referred to the web version of this article.)

S.A. Klankowski et al. / Journal of Power Sources 276 (2015) 73e7974

improving the structure, composition, and assembly of electrodematerials is critical to broader application, including the ever-increasing fledgling market of electric vehicles. Since inception inthe early 1990's, the vast majority of commercial LIBs utilize a pairof intercalation compounds (graphite and metal oxides such asLiCoO2 or LiFePO4) as the electrodes [1].While being able to providesufficient energy storage capacity and long cycling lifetime, theseelectrodes are not suitable for new applications demanding greaterenergy density at a higher power rate. This has triggered anextensive study on an assortment of novel materials and variousnanoscale electrode architectures [2e5], and of which, Silicon (Si)has been identified as one of the best materials for LIB anodes. Sipresents an extremely high theoretical Lithium (Li) storage capacity(4200 mAh g�1), a very low lithiation potential (0.2e0.4 V vs. Li/Liþ), and is naturally abundant and environmentally benign [6e8].In the amorphic state, the maximum obtainable capacity is~3800 mAh g�1; ten times that of graphite (372 mAh g�1) [8].However, unlike the staged intercalation of graphite anodes, Liforms alloys directly with Si during charging, causing up to 320%expansion in volume that induces severe anisotropic stress [9]. Thisstress leads to fracturing and loss of electrical connection thatconsequently reduces the capacity of Si thin-film electrode afteronly tens of cycles [10e12]. Various nanostructured Si materialsincluding nanoparticles, hollow spheres, egg yolk, etc. have beenemployed to reduce the effect of the internal stress andmake use oftheir large specific surface area and short Liþ diffusion length insolids [7e9]. Among them, long Si nanowires (NWs) were foundable to accommodate the large volumetric changes in both axialand radial direction, which effectively minimizes the fracture,providing greater than 3000 mAh g�1 capacity at C/20 rate anddramatically enhanced cycling lifetime [9]. However, the char-geedischarge rate on Si NW anodes was limited, mostly by the lowelectrical conductivity of amorphorized Si [8,9]. Hybrid coreeshellNWs utilizing a highly conductive and stable NW or nanotube coreto support a Si shell has recently been explored toward improvingthe power rate while maintaining the high capacity and good cyclestability [8,13e20]. However, as the power rate was increased, theobtained capacity still sharply dropped [8,13,16,18], for example by50% or more from ~C/5 to 5C with Si-carbon nanotube (CNT) coreshell structures [19,20].

In our previous study [21], we reported the fabrication andinitial characterizations of coreeshell NW arrays composed of aforest-like arrangement of vertically aligned carbon nanofibers(VACNFs) that were coaxially coated with intrinsic Si by magnetronsputtering. That study was concentrated on the ultra high capacityobtained at normal power rates (C/10 to 2C). This report is focusedon the understanding of the performance at higher rates (up to~10C) of this coreeshell NW anode, detailing how the pine-needle-textured Si shell improves the capacity retention at very high po-wer rates, which is desirable for advanced LIB hybrids.

2. Experimental

Experimental conditions are briefly described below. More de-tails can be found in the supplementary information. Pure copperfoils were deposited with a 300 nm thick chromium layer followedby a 30 nm nickel catalyst layer. These Ni/Cr/Cu sheets were thencut into 18 mm diameter disks for vertically aligned carbon nano-fibers (VACNFs) growth using DC-biased plasma enhanced chemi-cal vapor deposition (PECVD) following the procedure in literature[21e24]. VACNFs have an average length of 3.0 mm, an averagediameter of 150 nm, and ~1.1�109 CNFs cm�2 areal density. Pure Siwas deposited onto the VACNF arrays to form the coreeshell NWarrays by high vacuum magnetron sputtering to a nominal Sithickness of 465 nm (equivalent to the film thickness deposited on

a flat surface). A reusable electrochemical cell (El-Cell, Hamburg,Germany) was used for all half-cell tests with the Si-coated VACNFdisks as the working electrode, a Li disk as the counter electrode,and a Li wire as the reference electrode. A specially designed Kel-Fring was used to separate the working electrode from the poly-ethylene fiberglass spacer to avoid crushing the coreeshell NWs,leaving an exposed working electrode area of 17.5 mm in diameter.The electrolyte consisted of 1.0 M lithium hexafluorophosphate(LiPF6) in a mixture of 1:1:1 volume ratio of ethylene carbonate,ethyl methyl carbonate, and dimethyl carbonate, with 2% vinylenecarbonate added. The chargeedischarge measurements of Si-VACNF hybrids were set from 1.5 to 0.050 V vs (Li/Liþ) to mini-mize Liþ intercalation into VACNF cores. Our previous study hasdemonstrated that the contribution to the total capacity by Liþ

intercalation into VACNF core is negligible (<10%) [21].

S.A. Klankowski et al. / Journal of Power Sources 276 (2015) 73e79 75

3. Results and discussion

3.1. Structural characterization

Our coreeshell NWarray closely resembles that of a forest, withan arrangement of VACNFs that are coaxially coated with intrinsicSi by magnetron sputtering [21]. Each carbon nanofiber (CNF) wasgrown bottom-up and is firmly attached to the copper foil substrate[25,26], providing a highly conductive electron pathway(~2.5 � 105 S/m) between the active Si shell and the current col-lector. The individual CNFs of 50e150 nm in diameter and tunablelength from 3.0 to 5.0 mm, remained fully separated from each otherin a random array. The open space (300e400 nm) between theCNFs allows rather uniform Si deposition and accommodation of Sishell expansion during lithiation, as shown in Figs. 1 and S1.

At the first glance, the individual CNFs after sputter-coating witha Si layer of 500 nm nominal thickness show a cotton swabappearance, with a large Si mass at the tip that tapers graduallydown the CNFs (Fig.1a). A rough and billowy texture extends acrossthe shell surface, representing a distinctive feature of thick sput-tered films [27,28]. More interestingly, transmission electron mi-croscopy (TEM) revealed that the Si-coated CNFs have a uniquepine-leave structure (Fig. 1b), composed of many Si ‘needles’ of~10e20 nm in diameter and 80e100 nm in length protruding fromthe CNF surface at an oblique upward angle. These fine structuresare similar to the columnar microstructure formed in thick sput-tered films on flat surfaces [27e29]. At the conditions where thesubstrate temperature T is much lower than the melting point Tm ofthe coating materials (normally with T/Tm < 0.3) as satisfied in thisstudy (T ¼ 50e150 �C and Tm ¼ 1414 �C for Si), the deposited ma-terials tend to form arrays of vertical nanocolumns separated byvoided or low-density boundaries following Thornton's StructuralZone Model [23]. Such microstructure is a fundamental conse-quence of atomic shadowing acting in concert with the low adatommobility at low T/Tm deposition conditions and the formation ofsuch nanocolumnar structure is known to be enhanced by thesurface irregularities [26], such as the broken graphitic edge at theVACNF sidewall.

It is also well known that VACNFs, grown by plasma enhancedchemical vapor deposition (PECVD), present a unique microstruc-ture similar to a stack of graphitic cones and thus there are manygraphitic edges at the sidewall as schematically illustrated in Fig. 1c[22]. These graphitic edges are active sites facilitating fast electrontransfer at the Si/CNF interface, making it distinct from the smoothbut electrochemically inert graphite-basal-plane-like sidewalls ofCNTs. Here the graphitic edges at the CNF surface may also serve asatomic surface irregularities to facilitate the formation of nano-columnar structures. The sputtered atoms strike the substrate fromthe surface normal, making the incoming deposit flux at a glancingangle relative to the VACNF surface, which significantly augmentedthe atomic shadowing effect. Interestingly, the nanocolumns didnot present at the very top of the CNF tip which directly faced thesputtered atom flux. It is noteworthy that previous studies foundthat Si coating on vertically aligned CNTs by chemical vapordeposition (CVD) formed very different structures, i.e. non-continuous solid clusters anchored on the CNTs similar to a pearlchain structure [18]. A similar nanocolumnar microstructure wasobserved by another group in the Si shells sputter-coated onPECVD-grown VACNFs [20]. This microstructure seems to be aresult from the combination of ion-sputtering and the uniquesidewall structure of VACNFs.

While the nanocolumnar structure with a large contents ofvoids is undesired for many thin-film applications, it provides aunique means here to integrate macro- (Cu foils), micro- (VACNFlength), and nano- (VACNF diameter and Si nanocolumns)

structures for the most efficient Liþ diffusion and electron transportin LIB processes. In fact, the Liþ diffusion length across the solid Si isreduced to less than 10 nm with this hierarchical structure, whosesignificance was not emphasized in previous literature [20]. Ramanspectra of the Si-coated VACNFs (Fig. S2) before half-cell lithiationtests confirmed that it formed the nanocrystalline structure in theSi shell, as indicated by the sharp peak at 480 cm�1 shift from520 cm�1 peak in Si(100) wafer [21,30,31]. After chargeedischargecycles, the nanocrystalline Si is completely converted into amor-phous Si.

3.2. Electrochemical characterization

Fig. 2 shows the insertion/extraction capacity and associatedcolumbic efficiency of the Si-coated VACNFs over 500 char-geedischarge cycles. To evaluate the high-rate properties, thetesting procedure consisted of five power testing (PT) series atvaried C-rates while the electrode was progressively stressed withprolonged 1C chargeedischarge cycles. The expanded details of thefirst 250 cycles and the capacity in relation to the combinedmass ofSi and VACNFs are shown in Fig. S3. Each PT series consisted ofmeasurements at over five different current rates varying betweenC/8 to 8C, with up to five cycles at each specific rate. A few cyclicvoltammograms (CVs) were taken at 0.1e1.0 mV s�1 scan rates aftercompleting PT3 at Cycle #256.

A high insertion capacity of ~4500 mAh (gSi)�1 (normalized tothe Si mass) was observed in the first cycle at C/8 rate (due to SolidElectrolyte Interphase (SEI) formation, side reactions with residualoxygen, and transformations of crystalline Si to amorphous Si) [21]and then the electrode became fairly stable with an extraction ca-pacity of ~3140 mAh (gSi)�1 in the second cycle at the same C-rate.The current rate was then incrementally increased to 5C at the 12thcycle, showing an extraction capacity still as high as 2973 mAh(gSi)�1 and a coulombic efficiency of ~99%. Overall, the Si-coatedVACNF electrode showed remarkable stability. Even after stress-ing at 5Ce8C during the five power tests, the 1C capacity onlyslowly dropped from 3078 mAh (gSi)�1 (cycle #8) to 1266 mAh(gSi)�1 (cycle #505), with average columbic efficiency as high as99.8%. Even the end capacity is 3.4 times of the maximum bygraphite anodes.

The power testing results are summarized in Fig. 3. The firstseries of 12 cycles consisted of 2 cycles each at C/8, C/4, C/2, C/1, 2Cand 5C where the initial current density for C-rates was calculatedfrom the average capacity of 3535 mAh (gSi)�1 from our previousstudies [21]. As the C-rate was increased, it did show a gradualdecrease in capacity, but the total loss was only 5.3% from C/4 to 5C,rarely seen in Si LIB anodes [8]. The small difference is shown in thegalvanostatic profiles in Fig. 3a. The end portions of the extractionprofiles are enlarged in Fig. S4a for better view of the small changeof extraction capacity from 3140 mAh (gSi)�1 at C/4 rate to2973 mAh (gSi)�1 at 5C rate. The profiles show plateaus at0.4e0.05 V for insertion and 0.2e0.6 V for extraction, consistentwith the characteristics of other nanostructured Si anodes [8].

To our surprise, the C-rate capability was significantly improvedafter prolonged cycling at 1C rate. As shown in Fig. 3c, after 100 full1C cycles following PT1, the electrode was still able to provide anextraction capacity of 2885 mAh (gSi)�1 at C/4 rate in the first cycleof PT2, or 91.8% of that in the 3rd cycle at the same C-rate. Moreimportantly, it slightly increased from 2644 mAh (gSi)�1 at ~2.67Cto 2717 mAh (gSi)�1 at ~6.5C in this power test (PT2). To validatethis, the 3rd power test (PT3) was performed after another 100cycles at 1C (i.e. at Cycle #200), but using 5 cycles at each C-rates toensure the reliability. A similar trend as in PT2 was observed: thecapacity decreased first and then increased (by ~2.8%) while thepower rate was increased from C/10 to 8C. A minimum was found

Fig. 2. Insertion and extraction capacities and Coulombic efficiency over ~500 chargeedischarge cycles (see caption of Fig. S3 for detailed conditions). PT: Power Test series at variedC-rates; CV: cyclic voltammetry measurements.

Fig. 3. Galvanostatic chargeedischarge profiles of Si-coated VACNFs at selected C-ratesduring (a) the 1st PT series (cycles 1e16) and (b) the 4th PT series (after 300 cycles). (c)Specific extraction capacities versus the C-rate during PT1 (cycles 1e16) ( ), PT2 (cycles101e125) ( ), PT3 (cycles 201e255) ( ), PT4 (cycles 301e355) ( ) and PT5 (cycles401e455) ( ) are summarized. The anomalous capacity increase at high C-rates(~3Ce8C) is highlighted in the dashed box. For ease of discussion in different context,either the notation xC or C/(1/x) is used in referring to the same C-rate.

Fig. 4. TEM images of the Si-coated VACNF taken out from the cell in the (a) lithiated(charged) state after the 138th cycle and (b) delithiated (discharged) state after 750th

S.A. Klankowski et al. / Journal of Power Sources 276 (2015) 73e7976

around 3C. The C-rate dependence is more evident in extraction

profiles than insertion (Fig. 3b and S4b). After 200 cycles, theextraction capacity still remained ~2548 mAh (gSi)�1 at C/4 rate or81.1% of the initial extraction capacity. The CVs after 256 cycles(Fig. S5) are similar to the literature on Si-coated NWs [11,18]. Thespecific capacity versus the current density normalized to both ofthe surface area and Si mass (Fig. S6) showed the similar trends asthe C-rate dependence in Fig. 3c, excluding the possibility that thecapacity increase at high C-rates was artifacts by resetting C-ratecurrents after each 100 cycles.

cycle. Carbon/Formvar coated grids were used for TEM.

S.A. Klankowski et al. / Journal of Power Sources 276 (2015) 73e79 77

The 4th and 5th power tests were carried after 300 and 400cycles, respectively, which showed similar phenomenon, where thecapacity increased by 5.6% from 3C to 8C (PT4) and 6.8% from 4C to8C (PT5), respectively. Overall, in each PT series, the capacitychanges were very small (<10%) while the C-rate was increased bygreater than 60 fold. The overall capacity only slightly dropped aftereach addition 100 cycles. Preconditioning at low C-rates at thebeginning was found critical to stabilize the electrode to achievethe high performance. Applying over 5C rate at the initial cyclestended to irreversibly damage the electrode and lower theperformance.

Fig. 3b also shows that, after long cycles, the extraction plateauwas lowered by ~90 mV at 8C rate from those at other rates, indi-cating that it requires less energy to extract the inserted Li. This isunlikely due to Li plating since the low potential limit was setat þ50 mV versus Li/Liþ reference, higher than the usual Li platingpotential. TEM examination after long cycles did not show anyevidence of Li plating, but did show that the nanocolumnar struc-ture of the Si shell was preserved (Fig. 4).

Fig. 5. Nyquist plots of Si-coated VACNF LIB anode after (a) 115th cycle and (b) 200thcycle at different static potential from 0.55 V to 0.05 V versus Li/Liþ. Solid lines arefitting curves using the equivalent circuit model shown in (c).

To our knowledge, the capacity increase at high currents has notbeen reported in literature. The Li storage capacity of most LIBelectrodes drops significantly as the rate is increased. We attributethis to the unique VACNF-Si coreeshell architecture, which consistsof the nanocolumnar Si shell that remains intact after long cycling(Fig. 4). The void space between the nanocolumnar Si acts as gal-leries for fast Liþ diffusion in electrolyte and significantly shortenedthe Liþ path length in solid Si. In addition, an expected thin SEI veilforms over the individual Si needles, providing a flexible sheaththat prevents the Si nanoneedles from coalescing or breaking offduring lithiation/delithiation-induced expansion/contraction. ThisSEI layer could be further enhanced by the addition of 2% vinylenecarbonate (VC) to the electrolyte, since VC is known to be poly-merized by electrochemical reduction at the potential below 0.8 V(vs. Li/Liþ), which forms a thin and reliable coating to enhance theoverall stability and cycling performance of Si electrodes [32,33].The TEM image (Fig. 4a) of a SieCNF taken out from the cell in thelithiated form (at 0.05 V) after a long cycling test indeed shows thatthe Si nanocolumns in the shell was elongated but remainedanchored on CNF core. A very thin (20e30 nm) low-density ma-terial, likely so-called SEI, can be seen filling the void space betweenSi nanocolumns and hold them in place. In contrast, the TEM image(Fig. 4b) of the SieCNF sample taken out from the cell in the deli-thiated form (at 1.5 V) after long cycling shows shorter nano-columnar structure with a clean surface, very similar to that of thepristine sample before used for LIB tests (see Fig. 1b). These resultsare consistent with the SEM observation of the reversible expan-sion/contraction in our previous study [21]. More TEM/SEM imagesare provided in Fig. S7 in Supplementary Information.

To further understand the high-power capability, electro-chemical impedance spectroscopy (EIS) was employed to charac-terize the Si/VACNFs at various states of charge (SOC) and after longcycles. Fig. 5 shows the Nyquist plots measured at potentials from0.55 V (~30% SOC) to 0.05 V (~100% SOC) after 115 and 200 cycles at1C. The solid lines are the fitting curves using a Voigt-type equiv-alent circuit (Fig. 5c) based on the Frumkin and MelikeGaykazyanmodel [34], which has an extra serial element QDIFF when comparedto a study on Si NW LIB anodes [35]. At high frequencies (>10 kHz),the impedance is dominated by the series resistance RSERIES fromexternal circuits, Si/substrate interface, and the ionic conduction inelectrolytes. In the medium frequency region (10 kHze~10 Hz) it isdominated by the parallel RC elements including the surfaceresistance RSEI for Liþ migration through the SEI film at the outersurface of the Si shell and a constant phase element (CPE) QDL ac-counting for the fast pseudocapacitive Li reaction with the active Sishell [36,37]. CPE instead of a capacitor is necessary in order to fitthe vertically depressed semicircle caused by the porous structure.A serial Warburg diffusion element (W) is used to account for theslow Liþ diffusion interior Si, represented by the 45� slope (from~10 Hz to ~1.0 Hz) at end of the semicircle. A second CPE (QDIFF) isused to fit the nearly vertical line at the low frequency end(<1.0 Hz), accounting for the very slow Liþ diffusion in thicker Simaterial (diffusion coefficient DLiþ ¼ ~3 � 10�14 cm2/s in bulk solidSi vs. 5.1 � 10�12 cm2/s in nano-Si [38] and ~1 � 10�5 cm2/s involatile organic electrolyte [39]).

The above model fits all EIS curves very well (c2 ¼ ~1 � 10�3).The fitting parameters are summarized in Tables S1eS3 ofSupplementary Information. Clearly, RSERIES is very small and onlyslightly increases with SOC. The value of W is also relatively stableand increases slightlywith SOC after long cycles. QDL values are verysimilar in two sets of experiments. Themost prominent changes areshown in RSEI, which gradually decreases upon lithiation from 7.25to 4.05 U after 115 cycles and from 14.09 to 4.32 U after 200 cycles.This trend is opposite to the observation in amorphous Si thin films[40] and single-phase Si NWs [35], in which RSEI increased as SOC

S.A. Klankowski et al. / Journal of Power Sources 276 (2015) 73e7978

was increased. It is likely that, during Li insertion process as theelectrode potential is lowered, the expansion of the LieSi alloy shellin this multi-scale architecture stretches the SEI layer that has beenformed at higher potential and causes it to thin, thus allowing forbetter Liþ transport [8]. The admittance factor YO of QDIFF increaseswith SOC, which reflects a decrease in impedance as Si is lithiated.This can be related to the expanded porous structure of the Sinanocolumns during alloying as well as the doping effects.

The stability of the SEI layer at Si electrodes upon cycling is oneof the important issue to realizing the practical electrochemicalperformance of Si based electrodes [41,42]. Recently, it has beenreported that formation/deposition of SEI products is a dynamicprocess varying with the electrode potential; in competition withthe charge transfer process. Franger and co-workers [42] studiedthe SEI behavior on nano-Si electrode by EIS and observed an in-duction loop at ~1 Hz which appears only when the potential of theelectrode is below 0.35 V vs Li/Liþ (or <0.20 V vs Li/Liþ, if electrolytecontains additive like fluoroethylene carbonate). The inductionphenomenon indicated that both processes, SEI formation andlithiation of the Si electrode, take place simultaneously at low po-tentials. They concluded that it was mostly the instability of the SEIon the surface of the Si-particles that caused the appearance of theindicative loop on the corresponding EIS spectra. On carefully ex-amination of our EIS data, as presented in Fig. 5, a similar inductiveloop phenomenon below 0.2 V vs. Li/Liþ is observed after 200 cy-cles. The expanded EIS spectra in the high-frequency region at lowpotentials are shown in Fig. 6. In case of the EIS data recorded after115 cycles, inductive loop appears only at �0.1 V vs. Li/Liþ (Fig. 6a).This indicates that the SEI layer is stable during initial tens ofchargeedischarge cycles. Asmentioned earlier, the SEI layer formed

Fig. 6. Expanded Nyquist plots of the electrochemical impedance spectra at high frequency0.10 and 0.05 V (vs. Li/Liþ), respectively. All measurements were taken at steady potentials

in VC containing electrolyte possessed improved properties andbetter resistance to the stress caused by the large volume variationsduring lithiation/delithiation processes [32,33]. However, after 200cycles, the onset potential of the inductive loop moved to 0.2 V (vs.Li/Liþ), as shown in Fig. 6bed. This inductive loop became moreprominent at low potentials (0.1 and 0.05 V). This phenomenonindicates that the expansion of the Si nanocolumns due to lithiation(at low potential) generated some uncovered LixSiy surface and,hence induced the formation/deposition of more SEI components.The appearance of inductive loop after 200 cycles also indicatespartial electrode degradation, most likely because lithiation (hence,expansion) process is faster compared to the SEI formation/deposition.

However, the small shell thickness and strong mechanicalattachment to highly conductive VACNF core makes the outersurface of LixSiy highly active so that the SEI was reversibly oxidizedin the discharge process, thus maintaining a high columbic effi-ciency. Even though it requires extensive further study to reveal thedetailed mechanism, particularly with in-situ techniques such asRaman spectroscopy and synchrotron X-ray scattering or absorp-tion, the consistent results from our ex-situ TEM/SEM and EIS atdiscrete SOCs illustrate a possible explanation to the seeminglyanomalous phenomenon of capacity increase at very high rates.Bringing this to the attention of the EES community is expected tostimulate research toward developing stable high-power LIBmaterials.

region: (a) after the 115th cycle at 0.10 V; (b), (c) and (d) after the 200th cycle at 0.15,in the lithium insertion region.

S.A. Klankowski et al. / Journal of Power Sources 276 (2015) 73e79 79

4. Conclusions

The hybrid electrode of pine-leave nanocolumnar structuredSilicon-shell on VACNFs has been demonstrated as a remarkablehigh-power anode for Lithium-ion batteries. The Silicon shellspresent the full Lithium storage capacity of amorphous Silicon,~3000 to 3500 mAh (gSi)�1 at normal rates (up to ~3C). The elec-trode showed very little degradation in over 500 cycles even aftergoing through five heavily stressed power tests, retaining~1200 mAh (gSi)�1 at 1C rate at the end. In addition, increasing theC-rate from C/8 to 8C during the power tests only caused very smallchange (<7%) in the capacity. More importantly, the measured ca-pacity even slightly increased from 3C to 8C rates. Such anomaloushigh-rate capability is a new phenomenon that has not been re-ported in literature and needs high attention. Further studies onsuch new phenomenon are required to understand the responsiblemechanism, which is imperative to developing high-power per-formance Lithium ion batteries.

Acknowledgments

The work at Kansas State University was supported by NSF grantCMMI-1100830, NASA grant NNX13AD42A, NSF EPSCoR AwardEPS-0903806, and matching funds provided by the State of Kansasfor the latter two.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.jpowsour.2014.11.094.

References

[1] M. Winter, J.O. Besenhard, M.E. Spahr, P. Novak, Adv. Mater. 10 (1998)725e763.

[2] J.W. Long, B. Dunn, D.R. Rolison, H.S. White, Chem. Rev. 104 (2004)4463e4492.

[3] J.M. Tarascon, M. Armand, Nature 414 (2001) 359e367.[4] J.F.M. Oudenhoven, L. Baggetto, P.H.L. Notten, Adv. Energy Mater. 1 (2011)

10e33.[5] W. van Schalkwijk, B. Scrosati, Advances in Lithium-ion Batteries, Springer,

2002.[6] S. Ohara, J. Suzuki, K. Sekine, T. Takamura, J. Power Sources 136 (2004)

303e306.[7] A.S. Arico, P. Bruce, B. Scrosati, J.M. Tarascon, W. Van Schalkwijk, Nat. Mater. 4

(2005) 366e377.

[8] M.R. Zamfir, H.T. Nguyen, E. Moyen, Y.H. Lee, D. Pribat, J. Mater. Chem. A 1(2013) 9566e9586.

[9] C.K. Chan, H. Peng, G. Liu, K. McIlwrath, X.F. Zhang, R.A. Huggins, Y. Cui, Nat.Nanotechnol. 3 (2008) 31e35.

[10] B.A. Boukamp, G.C. Lesh, R.A. Huggins, J. Electrochem. Soc. 128 (1981)725e729.

[11] H. Kim, J. Cho, Nano Lett. 8 (2008) 3688e3691.[12] K. Peng, J. Jie, W. Zhang, S.-T. Lee, Appl. Phys. Lett. 93 (2008) 033105-1-3.[13] L.-F. Cui, Y. Yang, C.-M. Hsu, Y. Cui, Nano Lett. 9 (2009) 3370e3374.[14] L.-F. Cui, R. Ruffo, C.K. Chan, H. Peng, Y. Cui, Nano Lett. 9 (2008) 491e495.[15] P.-C. Chen, G. Shen, Y. Shi, H. Chen, C. Zhou, ACS Nano 4 (2010) 4403e4411.[16] J. Qu, H.Q. Li, J.J. Henry, S.K. Martha, N.J. Dudney, H.B. Xu, M.F. Chi, M.J. Lance,

S.M. Mahurin, T.M. Besmann, S. Dai, J. Power Sources 198 (2012) 312e317.[17] J.R. Szczech, S. Jin, Energy Environ. Sci. 4 (2011) 56e72.[18] W. Wang, R. Epur, P.N. Kumta, Electrochem. Comm. 13 (2011) 429e432.[19] A. Gohier, B. Laïk, K.H. Kim, J.L. Maurice, J.P. Pereira-Ramos, C.S. Cojocaru,

P.T. Van, Adv. Mater. 24 (2012) 2592e2597.[20] Y. Fan, Q. Zhang, Q.Z. Xiao, X.H. Wang, K. Huang, Carbon 59 (2013) 264e269.[21] S.A. Klankowski, R.A. Rojeski, B.A. Cruden, J. Liu, J. Wu, J. Li, J. Mater. Chem. A 1

(2013) 1055e1064.[22] A.V. Melechko, V.I. Merkulov, T.E. McKnight, M.A. Guillorn, K.L. Klein,

D.H. Lowndes, M.L. Simpson, J. Appl. Phys. 97 (2005) 41301-1-39.[23] B.A. Cruden, A.M. Cassell, Q. Ye, M. Meyyappan, J. Appl. Phys. 94 (2003)

4070e4078.[24] J. Liu, J. Essner, J. Li, Chem. Mater. 22 (2010) 5022e5030.[25] M. Meyyappan, L. Delzeit, A. Cassell, D. Hash, Plasma Sources Sci. Technol. 12

(2003) 205e216.[26] Q. Ngo, A.M. Cassell, A.J. Austin, J. Li, S. Krishnan, M. Meyyappan, C.Y. Yang,

IEEE Electr. Device Lett. 27 (2006) 221e224.[27] J.A. Thornton, J. Vac. Sci. Technol. A 4 (1986) 3059e3065.[28] R. Messier, A. Giri, R. Roy, J. Vac. Sci. Technol. A 2 (1984) 500e503.[29] I. Petrov, P.B. Barna, L. Hultman, J.E. Greene, J. Vac. Sci. Technol. A 21 (2003)

S117eS128.[30] I. Abdulhalim, R. Beserman, R. Weil, Phys. Rev. B 39 (1989) 1081e1091.[31] B. Li, D. Yu, S.-L. Zhang, Phys. Rev. B 59 (1999) 1645e1648.[32] L. Martin, H. Martinez, M. Ulldemolins, B. Pecquenard, F. Le Cras, Solid State

Ionics 215 (2012) 36e44.[33] L. Chen, K. Wang, X. Xie, J. Xie, J. Power Sources 174 (2007) 538e543.[34] J.R. Macdonald, Interpretation of AC impedance measurements in solids, in:

G. Mahan, W. Roth (Eds.), Superionic Conductors, Springer, US, 1976, pp.81e97.

[35] R. Ruffo, S.S. Hong, C.K. Chan, R.A. Huggins, Y. Cui, J. Phys. Chem. C. 113 (2009)11390e11398.

[36] T. Zhang, H.P. Zhang, L.C. Yang, B. Wang, Y.P. Wu, T. Takamura, Electrochim.Acta 53 (2008) 5660e5664.

[37] B. Markovsky, M.D. Levi, D. Aurbach, Electrochim. Acta 43 (1998) 2287e2304.[38] N. Ding, J. Xu, Y.X. Yao, G. Wegner, X. Fang, C.H. Chen, I. Lieberwirth, Solid

State Ionics 180 (2009) 222e225.[39] S.I. Lee, U.H. Jung, Y.S. Kim, M.H. Kim, D.J. Ahn, H.S. Chun, Korean J. Chem. Eng.

19 (2002) 638e644.[40] Y.M. Lee, J.Y. Lee, H.T. Shim, J.K. Lee, J.K. Park, J. Electrochem. Soc. 154 (2007)

A515eA519.[41] J.-H. Cho, S.T. Picraux, Nano Lett. 14 (2014) 3088e3095.[42] E. Radvanyi, K. Van Havenbergh, W. Porcher, S. Jouanneau, J.-S. Bridel, S. Put,

S. Franger, Electrochim. Acta 137 (2014) 751e757.